Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
A case/parental/sibling control study of Ewing's sarcoma/peripheral primitive neuroectodermal tumor (pPNET)
(USC Thesis Other)
A case/parental/sibling control study of Ewing's sarcoma/peripheral primitive neuroectodermal tumor (pPNET)
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UM I films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UM I a complete manuscript and there are missing pages, these w ill be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. ProQuest Information and Learning 300 North Zeeb Road, Ann Arbor, M l 48106-1346 USA 800-521-0600 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A CASE/PARENTAL/SIBLING CONTROL STUDY OF EWING’S SARCOMA/PERIPHERAL PRIMITIVE NEUROECTODERMAL TUMOR (pPNET) Copyright 2002 by Jeffrey P. Dayton A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment o f the Requirements of the Degree DOCTOR OF PHILOSOPHY (EPIDEMIOLOGY) May 2002 Jeffrey Paul Dayton Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3073769 __ ___ __® UMI UMI Microform 3073769 Copyright 2003 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA 90007 This dissertation, written by .................................................... under the direction of h..iX. Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY Date ...H9 or..AP#...?.Q52 DISSERTATION COMMITTEE Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS List of Tables viii List of Figures be Abstract x Preface xii Chapter 1: Epidemiologic and relevant molecular biologic review o f Ewing’s sarcoma/peripheral primitive neuroectodermal tumor (pPNET) 1 I. Introduction 1 II. Epidemiology o f Ewing’s sarcoma 1 IIA. Ewing’s Sarcoma and Ethnicity 2 II.B. Ewing’s Sarcoma in Siblings 8 II.C. Ewing’s Sarcoma and Stature 9 II.D. Other Risk Factors Associated with Ewing’s Sarcoma 11 II.D.l. Parental Exposures and Risk of Ewing’s Sarcoma 12 II.D.l.i. Agriculture 12 ILD. 1 .ii. Maternal Exposures During Pregnancy 14 ILD.l.iii. Smoking 15 ILD. 1 .iv. Other Parental Exposures 16 II.D.2. Childhood Exposures 16 H.D.3. Birth Disorders/Cancer Syndromes Related to Ewing’s sarcoma 17 II.D.3.L Birth Disorders 17 H.D.3.iL Cancer Syndromes 20 II.D.4. Prior Bone Trauma and Metal Implants 22 ii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. II.D.5. Case Reports of Ewing's Sarcoma Patients 23 n.D.6. Geographic/Seasonal Variation in Incidence 24 II.D.7. HLA Antigens and Ewing's Sarcoma 25 II.D.8. Adenovirus, cytomegalovirus and Ewing's sarcoma 25 IE. Origin and Histogenesis 28 III.A. Neural Markers and Evidence of Neural Differentiation 28 III.B. Histogenesis and Differentiation 29 IV. Molecular Biology of Ewing’s Sarcoma/pPNET 30 IV. A. Translocation 30 IV.A.1. EWSGene 34 IV.A.2. FLU and Other ets Family Genes Involved in Ewing's Sarcoma 36 IV.A.3. Properties of the EWS/FLI1 Fusion Protein 38 IV.A.4. Mechanism of Translocation in Ewing's Sarcoma 41 FV.A.5. Translocations Related to the Ewing’s Sarcoma t(l 1 ;22) 45 FV.A.6. Ethnic Specificity o f the Ewing's Translocation 48 IV.A.7. t( 11 ;22) Translocations in Other Tumors 49 IV.A.8. EWS Gene Involved in Other Translocations 53 IV.B. Transformation of Cell Lines by Ewing’s Sarcoma Fusion Proteins 57 IV.C. Target Genes of the Ewing’s Sarcoma Fusion Protein 58 IV.D. Mutations in Tumor Suppressor Genes and Oncogenes 61 IVJE. Insulin-like Growth Factor-I Pathway 62 IV.E.1. The Role of IGF-I 62 iii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IV.E.l.i. IGF-I and its Receptor 62 IV.E. 1 .ii. IGF-I in Bone Remodeling 64 IV.E.2. Insulin-like Growth Factor-I and Puberty 65 IV.E.3. IGF-I pathway and Ewing's Sarcoma 65 IV.E.4. IGF-I and Tumor Biology 71 IV.F. Nerve Growth Factor and Receptors 72 IV.F.l. Role of Nerve Growth Factor 72 IV.F.2. Expression of Nerve Growth Factor Receptors in Ewing’s Sarcoma 73 V. Clinical Prognostic Factors 75 V.A. Translocation Heterogeneity and Prognosis 75 V.B. Other Cytogenetic Changes and Prognosis 77 VI. Conclusions 77 Chapter 2: Analysis of Case Family Control Study of Ewing’s Sarcoma 79 VII. Introduction 79 VIII. Available Data for Analysis 80 VIII.A. Statistics of Parental Control Studies (both parents available) 82 VIII.B. Statistics for Use with Only One Parent 91 VUI.C. Statistics for Sibling Control Studies 94 VUI.D. Comparison o f Statistics for Family-based Study Designs 98 VIII.E. Statistics for Family Based Controls Studies 103 VTII.F. Considerations in Use of Younger Siblings 106 EX. Implementation of Case/Parent/Sibling Study 107 iv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IX.A. Sample Size Estimates 107 IX.B. Sources of Participants 107 IX.C. Participation/Enrollment Issues and Rates 108 IX.D. Sources of DNA for the Study 108 X. Analysis of Current Study 110 X.A. Test Using All Available Participants 111 X.B. Test Using Living Patients Only 113 X.C. Full Versus Half Siblings 115 X.D. Test Using Parents Only and Siblings Only 116 X.E. Test Using Caucasian Cases and Hispanic/Latino Separately 119 X.F. Discussion of Pertinent Results 122 X.F.L NGFGene 122 X.F.2. IGF-I Gene 123 X.F.3. IRS1 Gene 123 X.F.4. IGF-IR Gene 124 X.F.5. EWS Gene 125 X.F.6. Interaction between the EWS gene and IRS 1 gene 126 Chapter 3: Grant Proposal for a Case/Sibling/Parental Control Study of Ewing’s Sarcoma 127 I. Abstract 127 II. Background and Purpose 129 III. Hypothesis & Specific Aims o f Research 138 III. A. Hypothesis 139 v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. III.B. Specific Aims 139 m.B.1. Primary Aims 139 III.B.2. Secondary Aims 139 IV. Experimental Design and Methods 142 IV.A. Study Design 142 IV. A. 1. Case-Parental/S ibling Contro 1 142 V. Data Collection 144 V.A. Basic Demographic and Clinical Data 144 V.B. Contact Procedures and Specimen Collection 144 V.B.1. Family Controls and Cases (Prevalent & Incident) (Figure 3) 144 V.B.2. Specimen collection 146 VI. Laboratory Procedures 149 VI.A. Biologic Samples 149 VLB. Genetic Analysis 149 VI.B.l. Specific Methods for Genotyping of the Insulin-like Growth Factor-I Gene CA Repeat Polymorphism 149 VI.B.2. Specific Methods for Genotyping of the Insulin-like Growth Factor-I Receptor Gene AGG Repeat Polymorphism 150 VI.B.3. Specific Methods for Genotyping of the Insulin-like growth factor I receptor exon 16 M nll polymorphism 151 VT.B.4. Specific Methods for Genotyping of the Insulin Receptor Substrate 1 Gene CA Repeat Polymorphism 152 VI.B.5. Specific Methods for Genotyping of the Nerve Growth Factor B glll Restriction Enzyme Polymorphism 152 vi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VI.B.6. Specific Methods for Genotyping o f the EWS gene M nll Restriction Enzyme Polymorphism 153 VII. Quality Controls 154 VIII. Preliminary Data 155 IX. Statistical Considerations 156 EX.A. Case Parental/Sibling Control Design 156 IX.A. 1. The Family Based Association Test (FBAT) software 159 IX. A. 1 .i. Limitations of the FBAT program 160 IX.A.2. Sample Size Estimates 160 IX.A.3. Survival Advantage versus Disease Association 162 IX.A.4. Use o f Siblings as Controls 163 IX.A.5. Future Studies 164 X. Advantages and Limitations of the Study 164 XI. Significance of the Study 172 XII. Human Subjects 173 References (Numerical) 179 References (Alphabetical) 198 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES Chapter 1 Table 1: Studies of the ethnic specific incidence of Ewing’s sarcoma 8 Table 2: EWS in non-Ewing's sarcoma tumors and ethnicity. 56 Table 3: IGF-I and tumor pathophysiology in Ewing’s sarcoma 71 Chapter 2 Table 1: Available Families for Analysis: 81 Table 2: 2 x 2 Table for case-parental control study 84 Table 3: Stratification of exposure in case-parental control study 84 Table 4: Transmission disequilibrium test 86 Table 5: Example o f coding alleles versus genotypes for K = 3 alleles 89 Table 6: Sample size estimates for a case-parental control study design 90 Table 7: 1-TDT for missing parental controls 93 Table 8: Tests of association using all available study participants 112 Table 9: Tests o f association using living patients only 114 Table 10: Tests of association using parental controls only 117 Table 11: Tests of association using sibling controls only 119 Table 12: Tests of association for Caucasian cases 120 Table 13: Tests of association for Hispanic cases 121 Chapter 3 Table 1: Genetic markers and method of genotyping 143 Table 2: Sample size estimates 161 viii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES Chapter 2 Figure 1: Coding of half siblings in to the model 115 Chapter 3 Figure I: Worldwide incidence rates o f Ewing’s sarcoma 131 Figure 2: Genotypes used fbrcase family control study 141 Figure 3: Enrollment of Participants 148 Figure 4: Example of marker frequency that is population specific 170 Figure 5: Example o f marker frequency that is not population specific 170 ix Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Jeffrey P. Dayton Jonathan D. Buckley ABSTRACT A CASE/PARENTAL/SIBLING CONTROL STUDY OF EWING’S SARCOMA/PERIPHERAL PRIMITIVE NEUROECTODERMAL TUMOR (pPNET) In the United States, Ewing’s sarcoma is the second most common bone tumor after osteosarcoma, affecting mostly adolescents and young adults. Worldwide there is a well-established difference in incidence o f this disease geographically and ethnically and in some populations there is a virtual absence of this cancer. Lack of a change in incidence as a given population moves to a different geographic location has long suggested involvement of a genetic factor(s). Although much is known about the molecular genetics of this disease, the responsible genetic factor(s) have not been identified. Ewing’s sarcoma family of tumors is characterized by a reciprocal translocation between the EWS gene and members of the ets-transcription factor family o f genes. In ninety eight percent of Ewing’s sarcoma cases there is evidence o f such a translocation. Additional sites of genetic control include the insulin-like growth factor-I (IGF-I) and nerve growth factor (NGF) pathways. IGF-I is part of the growth hormone (GH)/IGF-I axis that is upregulated during puberty, a time when the incidence o f Ewing’s sarcoma peaks. IGF-I and its downstream pathway are involved in the tumor biology of Ewing’s sarcoma. Vitamin D induced expression o f NGF by osteoblasts may account for the predilection of Ewing’s sarcoma to bone tissue and this pathway may be related to disease etiology. x Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. We hypothesize that the genetic factors) responsible for the ethnic specific incidence of Ewing’s sarcoma may be involved in the translocation event, the insulin-like growth factor pathway, and/or the nerve growth factor pathway. In order to test our hypothesis, we enrolled 97 cases with Ewing’s sarcoma identified through Childrens Hospital Los Angeles, the California Cancer Registry, and via a study internet site along with both their parents and siblings as ethnically matched controls. Analysis was carried out using the Family Based Association Test (FBAT) program which utilizes the entire family structure in order to determine if any association exists between the disease and a maker. We found a significant association between Ewing’s sarcoma a single nucleotide polymorphism located upstream of the EWS gene (p=0.003, recessive model) and the IRS gene (134 bp allele, p=0.016, recessive model). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PREFACE Ewing's sarcoma is a relatively rare cancer that afflicts most often children and young adults. This is an especially devastating point in ones life to be diagnosed with cancer. Many o f the individuals that volunteered to participate in this study made significant alterations in their daily life in order to under go chemotherapy and surgery to treat their disease. Many also lost their lives after a valiant fight. It is the hopes of the author that their lives will affect the lives of future victims o f this disease. Along with this study and others, insight in to the disease process may in the future bring about new modalities to treat this disease. This body of work represents an initial attempt to characterize the possible genetic elements involved in the disease process. The first section is a review of the relavant epidemiologic and molecular biology needed to develop and substantiate a hypothesis. In the second section, analysis of the study results is displayed and discussed. The final section is a grant proposal for the work presented in the first two sections. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 1: Epidemiologic and relevant molecular biologic review o f Ewing's sarcoma/peripheral primitive neuroectodermal tumor (pPNET) L Introduction Ewing’s sarcoma is a rare tumor that occurs most frequently during childhood and adolescence. Histopathologically it is characterized as a small round blue cell tumor, with evidence of neural markers and differentiation along with a characteristic genetic change, a translocation, t(l 1 £2). In the past decade, there has been considerable changes in the classification and grouping of Ewing’s sarcoma and related tumors in to the Ewing’s sarcoma family o f tumors (ESFT).l ESFT refers to osseous and extraosseous Ewing’s sarcoma, osseous or extraosseous peripheral primitive neuroectodermal tumor (pPNET), and Askin’s tumor J Several factors render the study of Ewing’s sarcoma a scientific challenge: previous studies classified both Ewing’s sarcoma and osteosarcoma together under the classification of bone tumors; Ewing’s sarcoma is rare and until recently, the diagnosis has been one o f exclusion; lastly, the cell o f origin is still debated today. Despite considerable research on Ewing's sarcoma, few findings shed light on its etiology and only a few consistent characteristics of this disease are reported. Its most remarkable and consistent findings are its ethnic specific incidence, peak in incidence during adolescence, and specific molecular genetic characteristics. P . Epidemiology of Ewing's sarcoma The prior combined classification o f osteosarcoma and Ewing’s sarcoma as 'bone cancers’ complicates the epidemiologic study of Ewing’s sarcoma. The I Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cancers originate from different cells, osteosarcoma from bone forming osteoblasts and Ewing's sarcoma hypothesized from neuroectodermal cells. Furthermore, little emphasis has been placed on differentiating Ewing’ s sarcoma from other small round blue cell tumors occurring in the bone. This often led to misclassification of the disease prior to the mid 1980s. Reviewing the literature on Ewing’s sarcoma indicates a large number o f inconsistent associations and relatively few consistent findings. However, the consistent findings, specifically the characteristic translocation and the ethnic specific incidence pattern, may point to an underlying etiology for this disease. II.A. Ewing’s Sarcoma and Ethnicity One of the first and perhaps most striking findings in the study of Ewing’s sarcoma is the very low incidence o f this disease in several ethnic groups, namely those of African and Asian descent. This association has consistently been noted in numerous studies (Table 1 below). The relative absence of this disease in non- Caucasian populations may be the most convincing evidence pointing to a genetic underlying etiology. Yet, to date, no factor(s) have been identified that fully explain the incidence pattern of Ewing’s sarcoma. One of the first reports on the ethnic specific incidence of Ewing’s sarcoma came in 1970, when Joseph Fraumeni and Andrew Glass reported a deficit of Ewing’s sarcoma among African-Americans in the United States.^ Based on death certificates of all U.S. children who died o f bone cancer between 1960 and 1966, the authors found only 12 non-Caucasian (10 African-American, 1 American Indian, 1 Asian) cases among the 482 cases total. They hypothesized that “racial differences 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in the occurrence of cancer may give insight into its etiology”. Although the possibility o f higher fatality among Caucasians could explain these findings, later studies conclusively determined that the incidence was higher in Caucasians than non-Caucasians.3*5 Alternatively, the diagnosis of Ewing’s sarcoma may be more likely in Caucasians than non-Caucasians due to difficulty of diagnosis and difference in availability o f health care among ethnic groups. Support for a genetic rather than environmental basis for the difference in incidence can be found in the rarity of Ewing’s sarcoma in tumor registries o f Uganda and Jamaica. Fraumeni and Glass conclude that Africans may be genetically resistant to Ewing’s sarcoma. Jensen and Drake studied bone tumors among the 1.3 million specimens submitted to the Armed Forces Institute of Pathology (AFIP) in Washington D.C.^ O f the 239 cases of Ewing’s sarcoma, 2 were in African-Americans (0.8%). Their work supports the distinction between ethnic differences in incidence data versus the previously reported mortality data of Fraumeni and Glass. Jensen and Drake further hypothesized that Ewing’s sarcoma was an etiologically distinct tumor based on the ethnic difference in incidence pattern. Linden and Dunn reported findings similar to those of Drake and Jensen using the California Tumor Registry.^ Only 2 out of 147 reported Ewing’s sarcoma cases occured in African-American patients. In addition, only three cases of Ewing’s sarcoma were reported in other non-Caucasian groups (2 Chinese cases and 1 American Indian case). 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Using the New York State population based cancer registry, and selecting the years 1975 to 1980, Polednak reported finding only 5 African-American compared to 124 Caucasian-American cases o f Ewing’s sarcoma (4%). 5 The age adjusted rates were 1.5 per million for Caucasian-Americans and 0.3 per million for African- Americans, all ages. The author puts forth the hypothesis that the translocation, t(l 1;22) involved in Ewing’s may occur less frequently in African-Americans. Gurney et al. reported on the one-year age specific incidence rates of childhood cancer by race, sex and histologic type using SEER data (1974-1989) for children under the age of 15.6 Incidence rates for Ewing's sarcoma in Caucasians was 3.3 per million while rates for African-Americans was 0.3 per million for children under 15 years of age. Overall the maletfemale ratio was 1.08:1 while the overall Caucasian: African-American ratio was 11.0:1.0. No other childhood tumor investigated had a Caucasian:African-American ratio above 2. Incidence peaked at age thirteen in both males and females (8.3 and 5.3/million respectively). While these reports on the incidence of Ewing's sarcoma were from the United States, data on Ewing’s sarcoma in other geographic areas world wide showed similarly low incidence patterns. Interpretation of epidemiological studies conducted in areas without modem diagnostic expertise and/or absence of population based tumor registries should be done with caution. However, there is consistently a large difference in the ethnic specific incidence of Ewing’s sarcoma in all populations examined and hence, one can conclude with some certainty that the ethnic difference in incidence is real. 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Edington et al. found that no cases of Ewing’s sarcoma were diagnosed at the University of Ibadan, Nigeria cancer registry from 1960-1966.7 A subsequent study from the same area reported on cases diagnosed between the years 1960 to 1976 and found only 3 cases (1.8%) of Ewing’s sarcoma among 170 Nigerians diagnosed with cancers of the bone.8 Problems with reporting and diagnosing in a poorly developed country must be considered in this situation. Li et al. using the Shanghai Cancer Registry, reported that Ewing’s sarcoma was rare in the Chinese population.^ Only 10 cases o f Ewing’s sarcoma were diagnosed in Shanghai and Beijing with the incidence calculated to be 0.5 per million in Shanghai for children ages 0 to 19. Perhaps the most comprehensive international study of Ewing’s sarcoma incidence rates was conducted by Parkin et al. 10 The incidence of childhood bone cancer (osteosarcoma and Ewing’s sarcoma) was calculated for approximately 50 countries where the ascertainment and classification o f these different tumors was possible. Age-standardized rates were calculated by the direct method using the world standard population. Data was collected for the approximate time period of 1970-1979. When the population at risk was unknown, (as was the case in several registries from developing countries) comparisons were based on relative proportions of bone tumors and their distribution by age and anatomical site. The age standardized incidence rates for Caucasians was around 2 per million in children ages 0-14. Except for S3o Paulo, Brazil (2.5 per million ages 0-14), the incidence in Hispanic populations appears to be tower than in Caucasians (between 1.1 in Cuba to 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.5 in Puerto Rico and Los Angeles). In African populations, the incidence is significantly lower. For the combined tumor registries in the United States (Delaware Valley, Los Angeles, New York and the SEER registries), the incidence in African-Americans is 0.35 per million ages 0-14. For the three combined tumor registries from Africa (Uganda, Nigeria, and Zimbabwe) in which population at risk was known, the incidence rate was 0.22 per million ages 0-14, and based on only one case. In Asian populations, the incidence rates per million for China and Japan were 0.4 and 0.51 respectively for ages 0-14. Interestingly, in Western Asia and the Middle East, the incidence rates are not as low. In Bombay, India and Israeli Jews, the incidence rates are 2.1 and 2.0 respectively. Although based on only a small number of cases, the proportion of osteosarcoma to Ewing’s sarcoma in the Pacific Islands suggests that Ewing’s sarcoma is not all that rare (7:6 in Papua New Guinea, 3:1 in Fiji and 2:5 in Maoris). Furthermore, ratios from North Africa suggest that Ewing’s is also not as rare in these areas (1.4:1). The incidence data for a given ethnic population shows some variation from study to study for several reasons. First of all, the age group included in the calculation of incidence varies across studies. Secondly, Ewing's sarcoma is a rare tumor and small differences in the number of cases produces large fluctuations in incidence. Thirdly, the study population and time period for a given study is unique to that population. Finally, different ethnic groups can be classified in a different manner by each investigator, also contributing to variation in rates. Although there is some variation across studies, there is a consistent ethnic pattern of incidence. 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Taken together, a review of these studies confirms two findings. First, the incidence of Ewing’s sarcoma is significantly lower in several non-Caucasian populations compared to Caucasian populations. This low incidence in these populations is retained after migration to a geographic location where the incidence is generally higher. The lack of significant change in incidence upon migration points to the presence of a genetic factor. Lastly, it is of interest to point out the incidence of Ewing’s sarcoma in African-Americans in the United States is 0.35 per million ages 0-14 and the incidence for the three combined tumor registries from Africa was 0.22 per million ages 0-14. This slight increase in incidence in African- Americans could be explained by the admixture of Caucasian genes (estimated to be approximately 2 3 -2 5 %).* 1’12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 1: Studies of the ethnic specific incidence of Ewing's sarcoma Results Comments Reference Deficit of African-American cases o f Ewing’s sarcoma Based on death certificates Fraumeni & Glass (1970) Only 2 out o f239 cases in African-Americans Based on incidence data Jensen & Drake (1970) May also be rare in other ethnic groups (Asian and American- Indian) Based in incidence data from California Linden & Dunn (1970) Incidence rare in African- Americans Based on New York State cancer registry Polendak (1985) 11:1 ratio of Caucasian to African-American cases Based on several cancer registries across U.S. Gurney et al. (1995) Ewing’s sarcoma is rare in Africa From Nigeria Edington et al. (1970), Oyemade & Aboiye (1982) Ewing’s sarcoma is rare in China From Shanghai cancer registry Li et al. (1980) Incidence rates of Ewing’s sarcoma in several ethnic/geographic areas World wide incidence of Ewing’s sarcoma Parkin et a. 1993) fl.B . Ewing’s Sarcoma in Siblings If the ethnic difference in the incidence o f Ewing’s sarcoma points to an underlying genetic predisposition, one would assume that a familial pattern of inheritance might exist. However, this is not the case. Ewing’s sarcoma in siblings or other relatives is extremely rare. In fact, only four cases of siblings with Ewing’s sarcoma have been reported in the literature. Three sibling cases were in sisters and one sibling case was in brothers. 13-16 reports of Ewing’s sarcoma in other family members has been reported. This finding may be explained by the rarity of this tumor. 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. II.C. Ewing’s Sarcoma and Stature The incidence of Ewing’s sarcoma increases during childhood and peaks in adolescence.1 This increase in incidence may be due to pubertal factors such as sex hormones or growth factors which increase during puberty. Ewing’s sarcoma cases may be exposed to higher levels o f these factors which would lead to earlier puberty, increased height, and/or increased weight compared to normal individuals. However, an association between these any of these and Ewing's sarcoma has not been consistently found. Unlike osteosarcoma which occurs in the area of the growth plate in bones and is associated with increased stature, there is less evidence for an association between increased stature and Ewing’s sarcoma, which occurs most often in the diaphysis of the bone. Except for an early report by Fraumeni who found significantly increased stature o f Caucasian Ewing’s cases compared to Caucasian hospital controls for all ages and both sexes combined, follow-up studies have not confirmed this finding. ^ The use of hospital controls and relatively loose matching on age (or another indicator of puberty) questions the validity o f this early study. Other investigators did not find an increased stature in Ewing’s sarcoma cases relative to controls. Pendergrass et al. compared the height and weight at diagnosis of 291 Ewing’s sarcoma patients to age and sex specific norms from the National Center for Health Statistics.^ No difference in height or weight was found between male cases and the male population mean heights and weights or distributions of height and weight. However, female patients were smaller than the 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. female population mean, but this difference only approached marginal significance (mean difference — 1.4 cm, p=0.06). Furthermore, while the male cases showed no difference from the population norm with respect to weight, females weighed significantly less than the population norm (mean difference — 1.5 kg, p=0.03). Because females were smaller at diagnosis than expected, the authors hypothesized that this may be due to a nutritional deprivation from tumor burden and proposed that females would have lower survival rates than expected. However, no significant difference in survival between males and females or between sex-specific age groups for height or weight could be found. This study used study records of height and weight measured at diagnosis which may not be an adequate marker for growth changes due to effects of the disease and variability in the time between onset of symptoms and diagnosis. Buckley et al. completed a case control comparison o f 153 Ewing’s sarcoma pairs diagnosed between 1983 to 1985.19 Controls were selected by random digit dialing and matched on birth date (within 2 years for cases older than two years of age and within 1 year for those younger than two) and race. No difference in indirect measurements of body size, parental height and subjects birth length or weight between cases and controls was found. Although the data suggested that male Ewing’s sarcoma cases began and ended puberty at an earlier age (p=0.12 and 0.07 respectively), these same male cases reported less weight and height gain than controls (p=0.002 and 0.02 respectively). Parents reported height and weight gain 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and onset of secondary sexual characteristics. In conclusion, no consistent pattern was observed across genders that would suggest a biologic basis. Holly et al. found no significant difference in height between male and female cases and controls (p=0.24 and 0.81 respectively), and no significant difference in birth weight or height were noted.20 Winn et al. also found no difference between case’s birth weight or height and that of both sibling and population-based control groups.^ Pui et al. investigated the height at diagnosis of 3,657 children under the age o f 18 with a variety of cancers and made comparisons with published standards.^ The authors found no significant deviations for any of the cancers studied which included 113 Ewing's sarcoma cases (65 males 48 females). The association between stature and Ewing’s sarcoma is difficult to address and the results o f previous studies are inconsistent. Measuring height at diagnosis for any disease is a problem because one can not rule out the effect of the disease process on achieved height, especially in growing children. Using parental estimates of height and height gain is subject to error. However, the incidence of Ewing’s sarcoma increases and peaks during adolescence and other factors involved during this time period may be increasing the likelihood o f disease. ILD. Other Risk Factors Associated with Ewing’s Sarcoma A number of studies have investigated several possible risk factors for Ewing’s sarcoma. The only conclusion from these studies has been that no one factor is identified as a risk factor in a consistent fashion. 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. II.D .l. Parental Exposures and Risk of Ewing’s Sarcoma H.D.1.L Agriculture One of the more interesting associations between an environmental exposure and Ewing’s sarcoma is parental exposure to farming. Agents related to fanning that may increase the risk o f Ewing’s sarcoma include pesticides and viruses. Two studies found increased risks for cases whose fathers worked on forms or had agriculture related occupations and one study found increased risks for cases whose mothers worked in agriculture related o c c u p a t i o n s . ^ ^ 1,23 Holly et al. found that after adjustment for other factors, fathers employed in agriculture related areas 6 months prior to the conception of the case had increased relative risks o f 8.8 (95% Cl 1.8 - 42.7).20 Further supporting this finding is the observation that after adjustment for occupation, paternal exposure to pesticides, herbicides, or fertilizers was related to an increased relative risk of 6.1 (95% Cl 1.7 - 21.9). This study was a mixture of population-based cases identified through SEER in the San Francisco area (N=28) and 15 patients from Western states referred to the study from a pediatric oncologist and 193 controls. Because these 15 patients were not population based, the study may include biases due to characteristics specific to these cases. However, subset analysis o f the two groups revealed that relative risk estimates for paternal occupation in agriculture was higher in the population based case series (RR 8.8, 95% Cl 1.04 - 66.2) compared to the 15 referred patients (RR 4.3,95% Cl 0.28 - 71.1). Furthermore, paternal exposure was assessed through interview o f the mother, not the father, so the accuracy of exposure assessment is 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. questionable. Recall bias may lead to increased reporting of perceived risk factors in cases compared to controls. Although not significant, Winn et al. found that fathers o f cases were 3 times as likely to have farming as the usual occupation (RR 3.1,95% Cl 0.9 - 9.5) and two times more likely to be working on a farm during the pregnancy of the case (RR 2.2, 95% Cl -0.7 - 6.5) than regionally matched c o n t r o l s . 2 1 However, there was no association between living on a farm or ranch, or exposure to pets in the household and risk of Ewing’s sarcoma in this case-control study.21 This study used two different control groups, population based and sibling controls. While a number of factors were found to be risk factors in each subgroup, no one factor increased the risk of Ewing’s sarcoma in both control groups. The study design was a case control comparison o f208 cases aged 5 months to 22 years matched to a sibling control and an age-matched regional population control^ 1 Siblings had to have the same parents as the case and attempts were made to use same sex and older siblings when possible. Hum et al. found that mother’s occupation in farming increased the risk of Ewing’s sarcoma compared to population controls.-3 One hundred and eighty six cases with bone cancer (Ewing’s sarcoma or osteosarcoma) were enrolled on the study and matched on sex and age with 919 controls. Cases were selected from the Ontario Cancer Registry and diagnosed with bone cancer between 1980 - 1988 and were under the age of 25. Mothers answered the questionnaires about parental employment prior to the birth of the case (fathers) and prior to and including the year 13 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. o f the case birth (mothers). After excluding incomplete responses and adoptions, the final data set included 152 cases o f bone cancer and 713 controls for paternal exposures and 115 cases and 600 controls for maternal exposure. For paternal exposure, no association between Ewing's sarcoma and farming was found (OR 0.5, 95% Cl 0.1-3.9); however, significantly increased risk of Ewing's sarcoma in a child was found if the mother was employed in farming, horticulture or animal husbandry (OR 7.8,95%CI 1.9-31.7). Although parental occupation in agriculture was not investigated specifically, Buckley et al. found no relationship between Ewing's sarcoma and household exposures to such substances as insecticides, paints, and petroleum products. *9 Although several studies have shown an increased risk of Ewing’s sarcoma in relation to parental agricultural exposure, there are inconsistencies pertaining to the agent responsible. A large variety o f agricultural activities are reported with no evident agent common to all that would explain the increased risk. Despite reports o f clustering in one agricultural area of Western Australia, reported by Holman et al., others have found no significant difference in incidence between rural and urban areas.24,25 Furthermore, parental exposure to agriculture most likely does not explain the ethnic incidence pattern of Ewing’s sarcoma. II.D. l.iL Maternal Exposures During Pregnancy Maternal exposures during pregnancy do not show any consistent risk patterns for the development o f Ewing's sarcoma. Holly et al. reported increased but nonsignificant relative risks for thyroid hormone replacement (RR 3.4 95% Cl 0.8- 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14.7) and antibiotics during pregnancy (RR 2.7 95% Cl 0.8-9.0).20 Winn et al. reported that mothers were more likely to have used medications for nausea and vomiting during the case’s pregnancy than the sibling control (OR 2.6; 95%CI 1.2 - 5.9) however, there was n o significant difference for the population c o n t r o l s .21 Buckley et al. found no differences in maternal factors related to childbirth between cases and controls except for fewer numbers o f case mothers reporting threatened abortions (p=0.004), low-calorie diets (p=0.02), and use of high dose vitamins (p=0.003) during the pregnancy of the case. 19 Hum et al. found significantly increased risks for mothers employed in teaching fields (OR 3.1, 95%CI 1.1-8.7) and as noted previously, the authors also found increased risk for mothers employed in farming, horticulture and animal husbandry (OR 7.8, 95%CI 1.9-31.7).23 II.D .l.iiL Sm oking One study found an association between parental smoking and risk of Ewing’s sarcoma; however, these results have not been reproduced by other investigations. Winn et al. found an increased risk of Ewing’s sarcoma if the mother and father smoked during pregnancy with the case compared to their sibling c o n t r o l s .21 The risk increased in a dose dependent manner in the mothers; however, it was not significant in the lower outpoints. The relationship with smoking was not observed in unrelated region-matched controls and could be due to bias as parents are more likely to recall cancer related exposures in the case compared to the child without Ewing’s sarcoma. However, this may be a real risk factor if certain individuals are more genetically susceptible to the effects o f cigarettes in utero. 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Two other studies could not confirm this association. Holly et al. found no relationship between parental smoking status and risk of Ewing’s sarcoma and Buckley et al. found no difference in parental smoking between cases and controls. 19,20 1I.D.1. iv. Other Parental Exposures Holly et al. found no association with parental education, use of alcohol, medications taken at any time, or parental weight.20 Winn et al. found no association between cancer among relatives, religion, income, and parent's height and weight.21 Buckley et al. found no association between parental alcohol history and Ewing's s a r c o m a . *9 Hum et al. found that fathers of Ewing’s sarcoma cases employed in social sciences had increased risk compared to controls (OR 6.2,95%CI 1.6-24.5).23 N o explanation for this association is evident. n.D .2. Childhood Exposures Besides parental exposures, childhood exposures may influence the risk of Ewing’s sarcoma. Holly et al. found increased risks for intake of poison or overdose of medication in cases compared to controls (RR 4.4 - 95% Cl 1.4 - 13.5).20 In an attempt to exclude recall bias by including only those that had seen a physician, the relative risk actually increased to 9.3 (95% Cl 1.9-46.5). However, no single substance or class of substances could be identified and instead, a wide range of substances had been ingested by the cases. Winn et al. found no association between prior bone trauma and risk o f Ewing's s a r c o m a .21 16 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. One case o f Ewing’s sarcoma occurring after successful treatment and remission of acute lymphoblastic leukemia (ALL) has been described in the literature.26 At 8 years of age she was diagnosed and treated for ALL. Eleven years later she was diagnosed with Ewing’s sarcoma. Although, the existence of secondary Ewing’s sarcoma is extremely rare, the authors conclude that there was a relationship between the two malignancies because some of the cytogenetic abnormalities found in secondary leukemias (especially partial monosomy 7) were found in the subsequent Ewing’s sarcoma tumor. II.D J. Birth Disorders/Cancer Syndromes Related to Ewing’s sarcoma II.D.3.L Birth Disorders Several studies report an association between birth disorders and Ewing’s sarcoma. If such an association exists, then it could point to the involvement of a genetic factor, perhaps the same or a related genetic factor explaining the incidence pattern of Ewing’s sarcoma. However, again, no consistent birth anomaly or related group of anomalies is found in the literature. In a large case series, McKeen et al. found a large number of anomalies of renal development.^? Nineteen of 154 patients reviewed had evidence of a urogenital defect. The authors note that the mesenchymal origin o f these defects supported the hypothesis of a mesenchymal origin o f Ewing’s sarcoma proposed at that time. However, no control group was used and it is difficult to determine if the findings are only a characteristic of this set of cases. 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In the case control study done by Winn et al., no increase in the number of urinary tract disorders was noted. The authors did find an increased number o f inguinal and umbilical hernias in cases compared to region and age matched controls.^ ^ However, there was no excess o f hernias when compared to the sibling controls and only a small portion of the cases had hernias (approximately 10%). There was a lack of association when using sibling controls which may be due to overmatching on a common genetic predisposition to hernia risk. Winn et al. also found an increased number o f heart disorders among cases relative to their sibling which they suggest was due to increased examination and medical treatment in the cases compared to the sibling.21 Cope et al. found a higher than expected frequency of inguinal hernias in Ewing’s sarcoma cases relative to population estimates.28 Medical records for 306 patients with Ewing’s sarcoma/pPNET treated at the NIH from 1960-1992 were reviewed for presence of congenital anomalies. The frequency o f anomalies was compared to expected values obtained from the Collaborative Perinatal Project from 1966 which contained values for Caucasian children (96% of cases were Caucasian). Sixty-seven percent of the cases had congenital anomalies while 13 cases had a history of inguinal hernia (4 females and 9 males). The relative risk of inguinal hernia was calculated to be 13.3 (95% Cl 3.6-34.1) for females and 6.67 (95%CI 2.67-13.7) for males. The authors argue that the excess of congenital anom alies may point to a ‘disruption in normal embryologic development’ in Ewing’s sarcoma cases. However, only 13 cases out o f306 display an excess o f a specific anomaly 18 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and the authors also noted that the population rates of congenital anomalies varied between studies. Narod et al. reported on rates of congenital anomalies among 20,304 children diagnosed with cancer in Great Britain from 1971 to 1986.29 Ewing’s sarcoma had one of the highest rates (5.8%) and was second only to Wilm’s tumor (8.1%). Comparison was made with the frequency obtained from the British Columbia Health Surveillance Registry. There was a significant increase in the number of Ewing’s sarcoma cases with spina bifida (N=2), cataracts (N=2), optic deformities (N=2), spinal deformities (N=2) and osteodystrophy (N=2). There was no excess of genitourinary malformations among Ewing’s sarcoma cases. The comparison to a population registry in another geographic location coupled with the low number of each anomaly brings the results in to question; however, Ewing’s sarcoma was more commonly associated with anomalies than most other childhood cancers. An additional case report by Fein-Levy et al. found further interest in optic deformities.^® The authors report on a sixteen-year-old female patient with congenital optic atrophy and Ewing’s sarcoma.^® Congenital optic atrophy is a rare autosomal dominant disorder (1 in 50,000) characterized by loss of optic nerve tissue and preliminary linkage to a gene on the long arm of chromosome 13 (OPA1). The significance of the association with optic deformities is uncertain at this time and may be only coincidence. Several case control studies found no association between Ewing’s sarcoma and birth disorders. Nakissa et cd. found no congenital abnormalities in 18 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Caucasian patients with Ewing's sarcoma treated at the University o f Rochester Medical Center.31 Buckley et al. did not find any difference in the number of birth defects between cases and controls; however, they did report finding no diabetes in first degree relatives of Ewing’s sarcoma cases while diabetes was found in 10 first degree relatives o f controls (p=0.001).^ Holly et al. reported no relationship to birth disorders or cancer in family members of Ewing’s sarcoma patients.^® II.D.3AL Cancer Syndromes Any association between Ewing’s sarcoma and other cancers could provide insight in to a common etio logic factor, either environmental, genetic or both. However, no consistent associations are found across studies. Novakovic et al. reported an association between stomach cancer in family members o f patients with neuroectodermal tumors.32 Cancer in first and second degree relatives o f patients diagnosed with Ewing's sarcoma family of tumors was assessed and compared to Connecticut tumor registry sex, age and 5-year calendar interval rates. Cases were taken from patients treated at the Pediatric Branch and the Radiation Oncology Branch o f the National Cancer Institute from 1965-1992. Risk o f all cancers was not found to be significantly different from expected, although several cancers were found to be significantly increased in family members o f the patients. These included stomach cancer (O/E = 2.0,95% Cl 1.4-2.8, obs=34), melanoma (O/E - 1.9,95% Cl 1.2-2.8, obs=23), brain tumor (O/E = 1.9, 95% Cl 1.1-3.0, obs=18), and bone cancer (O/E = 4.2,95% Cl 1.7-8.6, obs=7). The maternal side of the family had higher, but nonsignificant, risks of these cancers than the paternal side. Several 20 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cancers were reported less often in patients than expected including bladder cancer (O/E = 0.2,95% Cl 0.1-0.5, obs=5) and rectal cancer (O/E = 0.0, 95% Cl 0.0-0.1, obs=0). A partial confirmation of diagnosis was undertaken by cross reference to death certificates in those cancers that were reported in excess (80% confirmation rate). Of the 310 Caucasian cases of Ewing's sarcoma family of tumors, 256 were enrolled in to the study. Potential bias in this study includes recall bias and participation bias. The reporting was not completely confirmed and thus, families with one cancer victim may be more inclined to report diseases in other members as cancers even if this is not the case. Families with more cases of cancer are more likely to be concerned about the issue and thus more likely to participate in this type of study. The authors note that increased risk o f the tumors found to be in excess on the maternal side might be related to the translocation process. This parallels the inheritance of the Philadelphia chromosome in which chromosome 22 that is involved more often comes from the m other.33 Hartley et al. report that mothers o f Ewing’ s sarcoma patients were not at an increased risk of cancer, particularly breast cancer, compared to expected rates calculated from an area of England where most resided.34 Sixty-one mothers out of 62 mothers eligible (1 case was adopted) were included in the study. The children were diagnosed with Ewing's sarcoma between 1954 to 1986. The observed/expected ratio of all cancers was 4/4.6 = 0.87 while the O/E ratio for breast cancer was 2/1.4 =1.4 and both were nonsignificant. 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hartley et al. reported no increase in the number of observed cancers among the first degree relatives of Ewing’s sarcoma patients. 3 5 There was also no increased risk of any specific tumor and no association between age of diagnosis, sex or tumor site and cancer in a first degree relative. Results were similar for cancer in second degree relatives with a decreased but nonsignificant number of observed cancers (O/E = 23/31.4). Cases for this study were obtained from the Manchester Children’s Tumor Registry from 1965-1988. All reports of cancer were confirmed and histopathological material was reviewed in first degree relatives. Expected numbers were calculated from the North Western Regional Cancer Registry for age, sex and time period rates. Fifty-six families of Ewing’s sarcoma cases were interviewed. It has to be noted that in all these studies, the use of diagnosis prior to 1985 is likely to introduce some cancers that are not Ewing’s sarcoma. This may dilute the association between Ewing’s sarcoma and any other cancer or group of cancers. However, only one study finds any association and this study found associations with a number of different cancers, questioning the results. n.D .4. Prior Bone Trauma and Metal Implants The occurrence o f Ewing's sarcoma in and around the area o f the bone leads one to consider physical injury or other alteration to the bone. Based on case reports, only limited evidence for any association exists. Winn et al. found no association between prior bone trauma (broken bones or bone disorders) and risk o f Ewing’s sarcoma in their analysis when compared to 22 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. either control group.21 Buckley et al. also reports no association between Ewing’s sarcoma and trauma in the site of tumor. ^ McDougal reported the first case of Ewing’s sarcoma occurring at the site of a metal plate.36 a second case report of Ewing’s sarcoma occurring in the site where a previous metal plate was removed 7.5 years prior also provides limited evidence for an association between the tw o.37 The metal consisted of mostly chromium (27-30%) molybdenum (5-7%), and nickel (2.5%) and no dissimilar metals were used for the plates and screws eliminating the possibility of electrolysis. II.D.5. Case Reports o f Ewing's Sarcoma Patients Case reports can provide a starting point for further investigation of potential risk factors. Few conclusions can be drawn from these studies until later case control or cohort studies are conducted. Perhaps the most interesting case report comes from Holman et al. who reported six cases of Ewing's sarcoma in a rural area of Western Australia between June 1979 and May 1981.24 Several characteristics of the cases point to possible etiologic clues. All six were male, Caucasian, and ranged in age from 12 to 34. All but one patient was interviewed directly. Each had exposure to farm animals; although exposure was to different animals in each case, there may be a common factor such as pesticide or flea exposure. Two patients had personal contact, playing on the same high school sports team. Three patients had exposure to chemicals used in sheep dipping. While all six lived in rural towns, 84% of individuals living in the 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. same region resided outside of rural towns. This may be a case of random clustering and not point to any etiologic clues, but is interesting none the less. Austin et a l reported on a similar clustering of Ewing's sarcoma in the San Francisco Bay area which had 30 cases in an 8-year period and of these 30 cases, 6 (20%) occurred within a 30 block diameter.^® This may have represented an outbreak due to a previously unidentified factor or another case of clustering of this disease. n.D .6. Geographic / Seasonal Variation in Incidence Larsson and Lorentzon found significantly higher incidence rates of Ewing's sarcoma in the southern part of Sweden than the rest of the country (middle and northern).25 The study period covered 1958 to 1968 and included 74 Ewing’s sarcoma cases reported to the Swedish Cancer Registry. Contrary to reports of involvement o f an agricultural related factor, there was no difference in incidence between rural and urban areas. The sample size was relatively small; however all hypotheses were tested at the 1% level of significance. McWhirter et al. report that Australia has a higher incidence of Ewing’s sarcoma than osteosarcoma with a ratio of 1.21.39 The results of their study are taken from the Australian Paediatric Cancer Registry from 1982 to 1991 and included individuals up to age 15. Precise ethnic data were not available for the authors to comment on. The incidence rate of Ewing’s sarcoma was calculated as 0.3 per 100,000. 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ross et al. report on the seasonal variation in incidence of childhood cancer diagnoses.^ There were 20,949 childhood cancer cases diagnosed in the U.S. between 1989 and 1991. The authors foimd no trend in the monthly diagnosis of Ewing’s sarcoma. n.D.7. HLA Antigens and Ewing's Sarcoma Chan et al. reported that Ewing's sarcoma patients were slightly more likely to carry HLA-B8 and Awl9 class I antigens than local controls, although the results were not significant. ^1 Thirty-eight Caucasian cases and 262 Caucasian controls were used in this series. This report is the only one of its kind and no follow-up study is noted in the literature, thus the significance of this finding is unknown and no biological explanation is put forth by the authors. HLA information was collected on two siblings with Ewing’s sarcoma and their family members. * 5 Neither sibling carried the B8 or Awl9 antigens, but the mother and one unaffected sibling of the cases carried B8. Lipinski et al. showed that Ewing’s sarcoma cell lines did not display evidence o f HLA class II expression and only variable expression of HLA class I antigens.^ n.D .8. Adenovirus, cytomegalovirus and Ewing's sarcoma Sanchez-Prieto et al. published a report of the adenovirus early region 1A (ElA) gene inducing the formation o f EWS/FLI1 translocation.43 They studied the formation of the translocation via RT-PCR and Southern blot analysis in both normal human keratinocytes and fibroblasts. The authors cite a similar morphological appearance between malignant cells formed after expression of the E1A genes (289R 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and 243R) and Ewing’s sarcoma cells (both are small round cell populations). The authors found that cell lines immortalized by adenovirus (HEK 293 human kidney cell line) also contained the EWS-FLI1 fusion gene. FISH showed translocation in the HEK293 as well as multiple copies o f the 12S El A gene present on a number of chromosomes. The authors propose that there is a remote chance that adenovirus infection of humans may result in the production of a specific translocation. Although provacative, this hypothesis has not been reproducibled despite attempts by other i n v e s t i g a t o r s . ^ * ^ Among them, Kovar et al. note that only a monoclonal fusion EWS-FLI1 transcript was found in each of the cell lines by Sanchez-Prieto, instead of a polyclonal fu sio n .^ This implies that if EA1 is involved, there is a very low frequency of EA1 induced recombination. Furthermore, Kovar et a l found no evidence of the EWS-FLI1 translocation in short and long El A transfected cell lines. In addition, the authors studied 27 Ewing’s sarcoma cell lines (from 19 original tumors) and found no evidence of El A sequences.^ However, Sanchez-Prieto et al. argued that they found evidence of El A sequences in 3 Ewing’s sarcoma cell lines and 14 of 32 primary tumors.^? They also noted that the results of their findings were variable and dependent upon many factors, which may account for the discrepant results. Meric et al. found no evidence of the EWS-FLI1 gene product in breast and ovarian tumor cell lines.48 El A, through its function as a tumor suppressor gene, is being studied as a therapeutic treatment of these cancers. Worried that the ELA gene may also produce unwanted translocation between EWS and FLI1, the authors studied 13 E1A transfected cell lines via RT-PCR and Western blotting and could not detect the presence of the fusion transcript or protein. 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cope suggests that Ewing’s sarcoma may be a ‘rare sequel’ o f perinatal cytomegalovirus (CMV) infection.^ The author sites that CMV has been associated with many of the congenital anomalies found in some of the Ewing’s sarcoma studies (see above). The author points out several additional arguments for this hypothesis. CMV infects only human hosts and Ewing’s sarcoma is present in only humans. CMV seropositivity occurs later in Caucasians than individuals o f African or Asian descent and this implies that more Caucasian infants could be infected in utero because of a lack of maternal passive immunity. CMV virus infects both neural and bone tissue similar to the presentation of Ewing’s sarcoma. This hypothesis lacks any biologic evidence at this point in time. In addition, the rates o f congenital CMV infection in different ethnic groups are not well established by the author. In feet, a study o f285 cases o f congenital CMV infection, 59% were Caucasian-American and 33% African-American.50 Thus, the risk of CMV infection in Caucasian-Americans appears to be twice the risk o f African-Americans. However, the disease rates o f Ewing’s are higher (1:5 to 1:10). Congenital CMV infection is also much more common in lower income families which should make low socio-economic status a risk factor for the disease. At least for Los Angeles County, this does not seem to be the case.51 Boppana et al. show that seropositivity is not completely protective o f congenital CMV infection because o f infections due to a different strain of the virus.52 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IP . Origin and Histogenesis Ewing’s sarcoma was first described in 1921 by James Ew ing.^ Although he postulated that the tissue of origin was endothelial, this argument has been carried on for nearly seventy years. Today it is believed that Ewing’s sarcoma is derived from the primitive neuroectoderm, specifically neural crest cells and that the immediate precursor cell is the postganglionic parasympathetic neuron. Recent work indicates that the precursor cell may actually have both myogenic and neurogenic potential as several biphenotypic tumors have been described which possess both Ewing’s sarcoma/neural markers and myogenic markers. m .A . Neural Markers and Evidence of Neural Differentiation The cell of origin in Ewing's sarcoma is believed to be neural due to the staining of antibodies to neural markers, which are found on ceils o f neural lineage. Lipinski et al. found evidence o f expression of several surface antigens associated with neuroectodermal lineage on Ewing’s sarcoma cell lines.42 These markers included ganglioside Gd2> an acidic glyco lipid, nerve growth factor receptor, and neural cell adhesion molecule (N-CAM). All cell lines tested expressed nerve growth factor receptor and 80% of the cell lines expressed N-CAM. van Valen et al. describe finding functional Y t receptors for neuropeptide Y in 11 cell lines of Ewing’s sarcom a.^ Upon stimulation o f these receptors with neuropeptide Y, there was inhibition of cyclic AMP (cAMP) formation. All 11 cell lines were derived from typical Ewing’s sarcomas while one cell line that was considered to be atypical was found to not express receptors for neuropeptide Y. 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lee et al. investigated the diagnostic usefulness of MIC2 staining and RT-PCR for the EWS/FLI1 fusion gene (discussed below).55 Of 34 cases, 26 (76%) showed evidence of EWS/FLI1 fusion transcript via RT-PCR 32 of 33 cases (95%) stained positive for the MIC2 cell surface glycoprotein and all 26 cases with the EWS/FLI 1 gene stained positive for MIC2. MIC2 can also be expressed in a number of non- Ewing’s tumors, notably rhabdomyosarcoma and lymphomas, also SRBCTs. Sugimoto et al. found that Ewing’s sarcoma cell lines without immunohistochemical and electron microscopy evidence of a neural phenotype are able to express a neural phenotype upon stimulation with dibutyryl cyclic A M P .56 Navarro et al. found evidence for neural differentiation in Ewing’s cell lines.57 Using scanning electron microscopy and dibutyryl cyclic A M P to induce differentiation, the authors found development of neuritic processes. Furthermore, no significant difference in appearance of neural features was noted between peripheral neuroepithelioma (pPNET) and Ewing’s sarcoma, strengthening the current hypothesis that these two tumors are related. m.B. Histogenesis and Differentiation Theile presents a unified view o f the origin of Ewing’s sarcoma, peripheral neuroepithelioma tumors, and neuroblastoma in the context of neural crest development.58 The author expresses the view that each tumor represents a cell that is derived from a certain stage during the developmental pathway o f neural crest cells. The observation that neuroblastoma can exhibit characteristics o f several neural crest derivatives such as neurons, Schwann cells, adrenal m edullar cells and 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. melanocytes indicates that the cell of origin may be a neural crest stem cell. Peripheral neuroepitheliomas may be derived from committed parasympathetic cells that are incapable o f further differentiation based on the expression of choline acetyl transferase used by parasympathetic neurons to make acetylcholine, while Ewing’s sarcoma may correspond to the same cell type that is capable of further differentiation. IV. Molecular Biology o f Ewing’s Sarcoma/pPNET IV. A. Translocation The most consistent finding in Ewing's sarcoma is the presence o f a balanced translocation between the EWS gene on chromosome 22ql2 and a member o f the ets DNA binding family. This translocation is found in 95-98% of all Ewing's sarcoma family o f tumors and is an obvious candidate for the genetic factor that could explain the incidence pattern on Ewing's sarcoma. In 1983, two groups published the finding of a t(l 1;22) translocation in Ewing’s sarcoma. The first group, Aurias et al., found translocations involving chromosome 22ql2 in all four cases studied. ^ 9 Two cases displayed the characteristic t(ll;22) while the other cases displayed more complex translocations. Turc-Carel et al. found sim ilar results in five cell lines developed from relapsed Ewing’s sarcoma patients.60 In four out of five cell lines a reciprocal t(ll ;22)(q24;ql2) translocation was found. In three o f these cell lines, every cell kept the der(l 1) and der(22) chromosomes while the other cell line lost the der(l 1) in a majority of the cells. 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Since then, several groups have confirmed and/or expanded on these original findings. Whang-Peng et al. discovered the presence o f the t(l I;22)(q24;ql2) translocation in two Caucasian patients with peripheral neuroepithelioma/pPNET.^i The authors noted that this tumor type typically occurs in older children and younger adults and because of the translocation, may be related to Ewing’s sarcoma or undergo a similar pathogenesis. Zucman et al. found no difference in the t(l 1 ;22) breakpoint regions o f Ewing's sarcoma and peripheral neuroepithelioma.^^ The investigators searched for the genes involved in the translocation which had been mapped on chromosome 22 between D22S1 and D22S15. Using fluorescence in situ hybridization, a chromosome 22 cosmid library, and chromosome walking procedure, the authors narrowed down the region involved in the breakpoint (EWSR1 -Ewing's sarcoma breakpoint region 1) on chromosome 22 to a 7 kb region from 17 tumors. Similar methods mapped the chromosome 11 breakpoint (EWSR2- Ewing’s sarcoma breakpoint region 2) to a region 40 kb in length from 17 tumors. Using probes that flanked the breakpoint, the authors determined that only the der(22) chromosome expressed a transcript and that the transcript was only found in tumor tissue, not the normal tissue from the patient. Plougastel et al. determined the gene involved in the EWSR1 translocation to be the EWS (Ewing’s sarcoma) gene.63 Mao et al. characterized the EWSR2 gene involved in the translocation, FLU (Friend murine leukemia integration 1) which is a member of the ets gene family.64 The mouse 31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. homologue of FLI1 is the site o f integration for Friend murine leukemia virus which leads to activation of FLI1 in 75% of erythroleukemias induced by this virus.65 McKeon et al. provided additional evidence for the similarity o f Ewing’s sarcoma and peripheral neuroepithelioma in the pattern of proto-oncogene expression and common neural origin.66 Although peripheral neuroepithelioma and Ewing’s sarcoma differ in histology (peripheral neuroepithelioma displays neural differentiation while Ewing’s sarcoma is undifferentiated) these authors found that both express the same pattern o f the following proto-oncogenes: c-myc (most abundant), N-myc, c-myb, and c-mil/raf-1 (all expressed at similar levels between the two tumor types). No expression o f c-fes or c-sis was detected in these tumors and c-ets-1 expression varied. Tumor specimens displayed similar results as cell lines, but varied more in the mRNA levels for each oncogene than the cell lines. Five of the Ewing’s sarcoma cell lines investigated were found to have choline acetyl transferase activity which is responsible for the synthesis o f acetylcholine, a parasympathetic neurotransmitter. This pattern of expression o f choline acetyl transferase was also found in peripheral neuroepithelioma. This is the basis for postulating a similar cell o f origin since both express an enzyme found in the parasympathetic neuron. Dunn et al. report finding a fusion protein between EWS and another member of the ets-family genes named ERG (Ets Related Gene) in one Ewing’s sarcoma cell line.67 In addition, they report finding the EWS/FLI 1 fusion transcript in an Askin’s tumor (neuroepithelioma o f the chest wall). In the same year, Sorensen et al. found 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. fusion of EWS with ERG.^8 The authors investigated two Ewing's sarcoma cells lines that did not contain a t(l 1;22) rearrangement in order to determine if EWS was involved in translocations to other genes. Using cytogenetic analysis, fluorescence in situ hybridization, RT-PCR and immunoprecipitation experiments, the authors concluded that the N-terminal portion of EWS was fused to the car boxy terminal portion of ERG and that functional protein was expressed from this transcript. ERG is 68% identical to FLI1 at the amino acid level and 98% identical at the DNA binding domain. The authors estimated that the EWS/ERG fusion gene is present in approximately 7% of Ewing’s sarcomas. Ishida et al. report the fusion o f EWS to another member of the ets family, ETV4/E1 AF.69 Two tumors with t(17;22)(ql2;ql2) were examined to determine if the EWS region was translocated to the ETV4 gene. The ETV4 gene is also known as E1AF and was originally described because it was bound to enhancer elements of the adenovirus type 5 El A gene, PEA3 (human homologue of Pea3). ETV4 is located on 17ql2 and codes for the adenovirus El A enhancer-binding protein. ETV4 is thought to be involved in invasion and metastasis in cancer by increasing expression of matrix metalloproteinases. The breakpoints for the tumor samples were in exon 8 or intron 7 o f EWS and the same intron in ETV4. An Alu repeat was found within the breakpoint region in the intron of ETV4. Delattre et aL investigated the translocations t(l 1^22) and t(21;22) in a large series of tumors (N=l 14 - 87 Ewing’s family, 12 non-Ewing’s family, 15 u n specified).^ Using RT-PCR design to identify both translocations, 95% of the 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ewing’ s/pPNET tumors (83/87) had either translocation. Eleven of the Ewing’s family of tumors contained EWS/ERG transcripts comprised of four different types of junctions. The authors also pointed out that 15% of the tumors had fusion transcripts despite containing normal appearing chromosomes. IV.A.1. EWS Gene Plougastel et al. characterized the genomic structure of the EWS (Ewing’s sarcoma) gene on chromosome 22.63 The entire gene covers a region approximately 40 kb in length, orientated centromere to telomere, and contains 17 exons ranging in size from 33 bp (exon 10) to 395 bp (exon 17). Introns range in size from 127 bp (intron 2) to at least 5 kb (intron 4). The EWS gene can be separated in to several domains: (1) the N-terminal domain (exons 1-7), (2) three glycine and arginine rich regions (exon 8 and 9, and exons 14 and 16), (3) RNA binding domain (exons 11 to 13). The location of 19 Ewing's sarcoma tumor breakpoints were found to be in introns 7 or 8 in 18 cases and in intron 10 in one case. Delattre et al. identified a region o f the amino terminal domain of the EWS gene that is 40% homologous to the carboxy terminal domain of eukaryotic RNA polymerase II proteins. 71 The authors also note that the remaining portion of the EWS gene is similar to single stranded nucleic acid binding proteins. They speculate that the EWS gene is involved in RNA synthesis and processing. Stolow et aL identified a gene, Cabeza (expressed in the adult fly head), located on the X chromosome of Drosophilia melanogaster (fruit fly) that is approximately 50% homologous to E W S .72 Cabeza was shown to bind RNA and is 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. hypothesized to be involved in early development and embryogenesis of the central nervous system. The authors show that EWS and Cabeza share a similar RNA recognition motif (RRM) (50% homology) and a zinc finger domain (65% homology). Aman et al. reported that the EWS gene displays homology to another gene, FUS(TLS), translocated with CHOP in myxoid liposarcomas (see below ).73 Several regions in both genes showed similar amino acid composition prompting the authors to suggest that these genes may have originated from a common ancestor gene. Both genes are believed to be housekeeping genes as they were found to be expressed in all tissues investigated (heart, brain, skeletal muscle, lung, liver, blood lymphocytes, kidney, ovary, testes, colon, spleen, thymus - bone was not studied). Bertolotti et al. reported that normal EWS, and not EWS-FLI1, interacts with the transcriptional complex (TF11D) and the RNA polymerase II complex.74 The TFIID complex is involved in preinitiation of transcription while RNA polymerase II complex is involved in transcription elongation. The authors show that EWS interacts with both of these complexes while EWS/FLI1 does not, suggesting they play different roles. Previously, the same group identified another protein with homology to EWS, hTAFn68 (70%), and grouped EWS, FUS/TLS and hTAFn68 into a family all with similar transcriptional machinery interactions.75 Petermann et al. further investigated the interaction between EWS and EW S/FLI1 and the RNA polymerase H.76 The amino terminal portion of E W S is shown to bind to a subunit o f the RNA polymerase II complex, hsRPB7. A similar 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. interaction was shown for EWS/FLI1, but not for the normal FLI1 protein. Replacement of the EWS portion of the chimeric EWS/FLI1 protein with the hsRPB7 was able to demonstrate similar gene activation as the EWS/FLI1 gene itself. The authors hypothesize that since the hsRPB7 portion of the RNA polymerase is in the regulatory portion of the complex, that the interaction of EWS/FLI1 may play a role in target gene promoter selection. Rossow and Janknecht described EWS functioning as a 'transcriptional cofactor’ by binding with another set transcriptional factors, CBP (CREB-binding protein) and p300.?7 The authors further demonstrated that EWS localizes to the nucleus and that the carboxy terminal portion of the gene negatively regulates the transcriptional activating potential of the amino terminal portion. The authors found that EWS was able to upregulate transcription from various promoters o f c-fos and Erb2 (HER2/Neu). This upregulation was dependent upon the presence of CBP/p300. IV.A.2. FLU and Other ets Family Genes Involved in Ewing's Sarcoma The ets (erythroblastosis virus-transforming sequence) family of genes has been characterized in the mouse.78 This family of genes contains a similar DNA binding domain in the 3’ portion of each protein. The viral oncogene v-ets has two homologues in the mouse and humans, ets-1 and ets-2. In humans, ets-1 maps to chromosome 11 (located near FLI-1), whereas ets-2 maps to 21 (located near ERG). Protein and mRNA levels are higher for both ets-1 and ets-2 and are detected in young murine tissues compared to adult tissue. Ets-2 mRNA is up regulated after 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. partial hepatectomy and precedes DNA synthesis, while ets-1 is not up regulated after partial hepatectomy. Ets-2 mRNA is able to accumulate in the presence o f a protein synthesis inhibitor, implying that ets-2 induction does not require de novo protein synthesis, like other proto-oncogenes. This research indicates that ets-2 can be added to a list of proto-oncogenes that allow cells to emerge from the Go phase of the cell cycle. Ets-1 may not be involved in murine liver cell cycle control, but may be involved in other tissues. The entire sequence o f the FLI1 gene is composed of 9 exons, which span approximately 120 kb on chromosome 1 lq24 and encode a 452 amino acid protein with 80% homology to ERG.^9,80 Ra0 et a l show that FLI1 recognizes and binds to ets specific DNA sequences and is involved in transcriptional activation.81 Siddique et al. show that ERG functions in a similar manner. 82 The 3’ end of ERG functions in DNA binding while two other domains (amino terminal and car boxy terminal regions) function in transcriptional activation. Yi et al. report that expression o f normal FLU and ERG inhibit apoptosis in cell culture. 8 3 The respective EWS/ets fusion genes also showed a similar ability to inhibit apoptosis and the use o f antisense RNA was used to increase the susceptibility to apoptosis in cell culture. Troung and Ben-David reviewed the normal function o f FLI1 as a transcriptional activator.84 FLI1 binds to a specific sequence and several target genes were identified that are involved in megakaryopoesis, angiogenesis, and cell cycle control. FLI1 is also known to interact with other proteins such as the retinoic 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. acid receptor, thyroid receptor and glucocorticoid receptor. FLI1 is involved in development of endothelial cells and hematopoesis. Overall, the author suggested that FLU is involved in apoptosis, differentiation, and cellular proliferation. Mager et al. found that FLI1 is expressed in developing neural crest cells that will become mesenchyme in quails.85 Specifically, FLI1 is expressed later in neural crest cells after they have reached their target locations and only in me so ectodermal neural crest cells, not neurogenic or melanogenic neural crest cells. Neural crest cells are the possible cell of origin of Ewing’s sarcoma. rV.A.3. Properties of the EWS/FLI1 Fusion Protein Mao et al. reported that EWS/FLI1 and normal FLI1 recognize and bind to similar DNA sequence specific sites.^ The authors used randomly synthesized oligonucleotides, and alignment of the oligonucleotides that bound normal FLU to determine the optimal binding sequence. It was found that FLI1 binds to the sequence ACCGGAAGa/T/c with the internal sequence GGAA being found in all Ets containing DNA binding proteins. Truncated forms o f FLI1 proved that the ets- domain was responsible for the DNA binding specificity. EWS/FLI1 bound to the same sequence as normal FLI1 and with a similar affinity. FLU is expressed in resting mature T cells while stimulated T cells express high levels o f EWS mRNA. The authors conclude that the cell cycle specificity of both proteins coupled with the same binding properties points to a role in EWS regulating the chimeric EWS/FLI1 protein differently than the normal regulatory portion of FLI1. It may be that EWS causes dysregulation o f the same target genes of FLU. FLI1 and ERG share 98% 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. homology in DNA binding domain, which may account for the similar tumorogenic ability o f EWS/FU1 and EWS/ERG. Ohno et al. found that the EWS/FLIl gene is a transcriptional activator and the EWS domain of the chimeric protein functions as a “modulatory/regulatory” domain.86 The authors used a chloramphenicol acetyl transferase (CAT) reporter assay to show that the EWS/FLIl increased transcription 9.5 fold whereas FLI1 increased transcription 6.5 fold (nonsignificant difference). EWS/FLIl had no effect when the E74 sequence (recognized by FLI1) was removed pointing to sequence specific activity. Deletion analysis o f the EWS domain in the chimeric protein suggested a modulatory/regulatory role o f this domain on the car boxy terminal transcriptional activation domain (CTA) of FLI1. Bailly et al. characterized the DNA binding properties of the EWS-FLI1 fusion protein.^? The authors determined that the DNA binding specificity of the EWS-FLI1 fusion protein was similar to the normal FLI1 and that both of these proteins localize to the nucleus. However, the authors found that the EWS-FLI1 protein transcriptional activation was 10 fold better than the transcriptional activation properties of the FLI1 protein in reporter assays. Furthermore, they hypothesized that overexpression o f c-myc in the Ewing’s family o f tumors compared to other neural tumors such as neuroblastoma, could be related to the transcriptional activation by EWS-FLI1. The EWS-FLI1 fusion protein was shown to increase expression from the promoter region o f c-myc in a CAT reporter assay, although normal FLI1 was also able to increase expression. 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lessnick et al. studied the function of several domains o f EWS when fused to FLI1.88 Two distinct domains (an amino terminal and the distal portion) of the translocated section of the EWS gene were found to have separate functions. The amino terminal domain enabled the fusion protein to transform NIH3T3 cells (amino acids 1-82). The distal portion of the EWS gene (amino acids 83-265) transformed less efficiently, but is a strong transcriptional activator of a reporter gene assay. The authors hypothesize that EWS/FLIl functions as an aberrant transcription factor in Ewing’s sarcoma tumors. Based on a single case report, Knezevich et al. hypothesized that EWS/FLIl may inhibit differentiation of tumor cells.89 a ten-year-old female was diagnosed with a Ewing sarcoma of the left arm which expressed EWS/FLIl. After treatment consisting of surgery, radiation and chemotherapy, the remaining tumor displayed a different phenotype. EWS/FLIl expression was non-detectable and the tumor histopathologically resembled a ‘well-differentiated peripheral neural tumor’. Assessment of clonality using methylation pattern of the FMR1 gene on the X chromosome showed that the two tumors were clonally derived. However, the authors were not able to further study this issue in other genes because o f a limited supply of the residual tumor. The authors hypothesized that a residual tumor somehow lost the fusion gene and resulted in a more differentiated and slower growing tumor. Olsen and Hinrichs demonstrated the importance of post translational modification in the function o f the fusion protein EWS/FLIl .90 Phosphorylation of 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. an IQ domain (located in exon 10 of EWS) of EWS/FLIl is necessary for proper transcriptional activation o f the fusion protein. The phosphorylated protein was better able to bind DNA (2.2 fold) and better able to transcribe a reporter gene (1.6 fold). Instead of localizing to the nucleus, the unphosphorylated protein localized to both the cytoplasm and the nucleus. Approximate^ 85% of EWS/FLIl translocations do not involve the portion of EWS where the IQ domain exists, questioning the relevance of this result. However, the authors note that translocations, which do involve the domain, have a poorer prognosis which may result from the properties o f the phosphorylated IQ domain.91-93 IV.A.4. Mechanism of Translocation in Ewing's Sarcoma One hypothesized reason that Ewing's sarcoma could be more prevalent in Caucasian populations is that this population may form the translocation product while other populations do not form the translocation. Several studies have investigated the mechanism of translocation in Ewing’s sarcoma and to date, the reason why this translocation occurs is not folly understood. Besides simple balanced translocation events, more complex translocation involving other chromosomes and numerous DNA breaks have been described. Desmaze et al. investigated the complex translocations using fluorescence in situ hybridization with probes that flanked each o f the genes believed to be involved in the Ewing's translocation.^ Three cases were analyzed for complex translocations with the following results. In the first case, a t(2;l 1 ;22) translocation occurred, in which the EWS and FLI1 genes were fused and the whole downstream segment of 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. chromosome 22 was translocated on to the short arm of chromosome 2. The second case exhibited an EWS/FLIl fusion that was found on chromosome 11, the result of an insertion of an EWS containing fragment from chromosome 22. The distal portion of chromosome 22 was translocated to chromosome 12. The third case demonstrated an EWS/ERG translocation in which the ERG portion o f chromosome 21 was translocated to chromosome 22 while this chromosome was split in to two large fragments. The authors noted that the EWS and ERG genes were orientated in opposite directions and the EWS/ERG gene rearrangement was the result o f a complex rearrangement. Since the incidence of Ewing's sarcoma is ethnic specific, and the translocation event is a possible genetic factor which could explain this observation, then perhaps the region of DNA involved in the translocation is different, in different ethnic populations. If this is the case, then one likely expectation would be the presence of a narrow region o f DNA that is involved in the translocation. However, the actual breakpoint region on EWS or FLI1 is very heterogeneous and occurs over a relatively large region. The involvement of a relatively large region points to a random translocation event rather than one that is genetically predisposed and sequence specific. However, this observation does not rule out the possibility of ethnic differences in chromatin conformation that predispose to the translocation event. Delattre et al. investigated the translocations t(l 1 ;22) and t(21 £2) in a large series o f tum ors.^ A majority o f the transcripts contained previously described 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. fusions between exon 7 of EWS and exon 6 of FLU (type 1, N=42) and between exon 7 of EWS and exon 5 of FLIl (type 2, N=15). Overall, the several possible EWS/FLIl combinations included joining exons 7 to 10 of EWS (2-3 kb region) to exons 4 to 8 of FLIl (30-40 kb region). Clearly, there is a heterogeneous mixture of possible fusion transcripts. Zucman-Rossi et al. analyzed the sequence surrounding the breakpoint regions on EWS and FLIl for characteristic sequence patterns in order to determine the possible mechanism by which the translocation occurs.95 The authors sequenced the entire EWSR1 and EWSR2 regions, then designed sets of primers to amplify the junction region of the translocation using 15 PCR reactions. After screening 77 Ewing’s family tumors which were known to contain the EWS/FLIl translocation, sequences surrounding the junction points were analyzed for possible elements that would point to a mechanism o f translocation. The chromosome 22 junctions occurred in two areas, EWSR1A (3.0 kb) followed by a 1.2 kb region o f intron 8 that was uninvolved, and then a second region, EWSR1B, which was 1.2 kb in length. No explanation for the lack o f recombination sites in the 1.2 kb region was found (no DNase sensitive sites or other sites). The corresponding junction sites of chromosome 11 spanned a region from exon 2 to exon 8 with a peak density in intron 4 (3 fold increase). There was no correlation between the chromosome 22 and 11 breakpoints (r=0.06). The authors analyzed a 120 bp region surrounding each breakpoint on the der(22) chromosome using a dot matrix program to find direct or inverted repeats o f greater than or equal to 4 bp. No obvious homology between the 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 and 22 chromosomes was found. No known recombination sites were found which included a search for topoisomerase I and II, translin, heptamer/nonamer, chi consensus, Alu, alternating purine/pyrimidine sequences, palindromic, polypurine, and polypyrimidine sequences. A similar approach was taken to amplify the der(l 1) breakpoint region with subsequent sequence analysis. Overall, four tumors showed a more complex process which involved the insertion o f a locally derived inverted sequence (LDIS), while the rest were balanced (+/- < 5 bp). The authors hypothesized a model o f illegitimate recombination that would explain both the complex and balanced translocations observed. In essence, balanced translocations form when double strand breaks occur and the resulting fragments (single stranded overhang) are either expanded by DNA polymerase or reduced by 5’-3’ exonuclease. In the case o f a LDIS, the presence o f inverted repeats makes it possible for a single stranded fragment o f a double strand break to fold back on itself and form a hairpin structure. Subsequent cutting o f the other end o f the hairpin inverts the sequence upon itself resulting in a single stranded overhang with the original sequence inverted. After inspection o f a large series o f tumors, the authors conclude that unlike previous mechanisms of recombination, site-specific signal sequences and homologous stretches between the two chromosomes, illegitimate recombination is the mechanism by which the t( 11 ;22) chromosomal translocation takes place. However, there is no clear reason why some ethnic groups would be more protected against illegitimate recombination than others. Several hypothesis can be proposed: (1) The genomic structure in the region translocated is ethnically different; (2) the 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. EWS/FLIl gene acts differently depending on ethnicity; (3) Caucasians are at a higher risk for the inciting event which causes the translocation; (4) and/or the downstream pathway of the EWS/FLIl gene is inhibited in non-Caucasians. To add to the diversity o f translocation products, Kovar et al. describe the inclusion o f ‘FLIl cryptic exons’ found in 3 tumors of greater than 300 screened t u m o r s . 9 6 i n two cases, the authors found inclusion of portions of intron 5 of FLIl as exons in the fusion product and in one case, they found a portion of intron 6 from FLIl. Obata et al. examined the breakpoint region of 12 Ewing’s sarcoma cases (presumably Japanese) by PCR amplification and subsequent sequencing.97 \ consensus sequence of 5’ AGAAAARGRR-3’ was found in four cases and sequences suggestive o f Aiu repeats and topoisomerase cleavage sites were found in the vicinity of almost all cases studied. Although only a limited number of cancers were studied, the authors conclude that multiple mechanisms may be involved in the recombination event in the population studied. IV.A.5. Translocations Related to the Ewing’s Sarcoma t(ll;2 2 ) Interestingly, the long arm of chromosome 22 is involved in a number of translocation events, supporting the notion that this area may be “highly recombinogenic”.98 if this is the case, these other translocation events may be more common in Caucasians, however this does not appear to be the case. A t(l 1 ;22) translocation has been found to be present in all cells of some individuals, compared to only tumors cells o f Ewing’s sarcoma patients. Budarf et 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. al. reported that this is the “most common non-Robertsonian” translocation in humans and characterized the breakpoint region in relation to Ewing's sarcoma/pPNET tumor associated breakpoints along with several other common tumor breakpoints.98 On chromosome 11, the constitutional breakpoint was found to be proximal to the Ewing’s breakpoint and separated by a number o f markers indicating that they are not close to one another. On chromosome 22, the constitutional breakpoint was found to be proximal to the X . light chain locus (involved in the t(8;22) o f variant Burkitt lymphoma) and the BCR1 region (involved in the t(9;22) of acute lymphocyte leukemia) on chromosome 22 which were all proximal to the breakpoint region involved in Ewing's sarcoma. The BCR1 region and the Ewing's sarcoma breakpoint were also distant from one another. The authors concluded that because o f the involvement of chromosome 22 in a number of translocations and the existence o f the X light chain locus on chromosome 22 which is involved in somatic rearrangement o f immunoglobulin genes, that this chromosome may be “highly recombinogenic”. If this is true, and chromosome 22 is prone to recombination, then there could be two likely possibilities for the Ewing’s translocation. First, the recombination occurs at a much higher frequency in Caucasians than non-Caucasians, making the translocation a race specific event. Secondly, the recombination occurs at the same frequency in all populations, but either the inciting agent or the downstream pathway are more permissive in Caucasians compared to non-Caucasians. 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Zackai and Emanuel investigated the constitutional t(l 1;22) translocation in 32 unrelated f a m i l i e s . 9 9 The authors argued that it is unlikely that a founder effect is taking place. They support this with evidence of de novo translocation in a case where the grandparent’s genotypes were available and normal Furthermore, there are heteromorphic variants of the der(22) in three families who were extensively studied. Finally, the constitutional translocation is found in different ethnic groups and of the 32 families studied, one was Black (3.1%), two Hispanic (6.3%), and one Pakistani (3.1%). However, if there is no bias in the reporting and investigation of cases suspected of carrying a translocation event, this still indicates that a large majority of translocation events occur in Caucasians. Fraccaro et al. investigated the t( 11 ;22) translocation in 43 unrelated families in Europe and New Zealand (1 c a s e ) . in this study the authors do not note the ethnicity of any families and it is assumed that all were Caucasian. In comparison to the heterogeneity of the Ewing’s sarcoma breakpoints, those involved in the t(9;22) or Philadelphia chromosome found in chronic myelocytic leukemia (CML) are similar in size.*01 The breakpoint region on chromosome 22 of 19 patients with CML were found to span a region of 5.8 kb termed the “breakpoint cluster region” (her). In comparison, the chromosome 9 breakpoint region was found to be fairly large, at least 27 kb, similar to the chromosome 11 breakpoint in Ewing’s sarcoma. The Philadelphia chromosome is found in 96% of CML cases and as with the EWS/FLIl gene, is hypothesized to be involved in the pathogenesis o f this disease. 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IV.A.6. Ethnic Specificity of the Ewing's Translocation If the translocation event or its fusion protein product are the cause o f the ethnic specific incidence pattern found in Ewing’s sarcoma then the most likely component of this ethnic specific translocation would be the EWS gene. This gene is found in every case while several members of ets genes can be involved. Due to the postulated genetic factor involved in Ewing’s sarcoma and the presence of an almost ubiquitous translocation, Zucman-Rossi et a l investigated the possible interethnic variation in the region o f the breakpoint in the EWS g e n e . 102 The investigators searched for polymorphisms in EWSR1 (EWS breakpoint region) and EWSR2 (FLIl breakpoint region). The authors used genomic DNA from normal Caucasian and African individuals and probed with a variety of fragments from ESWR1 and ESWR2. The only difference noted in any o f the Southern blots was a different size fragment found in three African individuals when probed with three fragments of ESR1. This polymorphism corresponded to a 2480 bp deletion in the Alu rich intron 6 of EWS. The group reported finding this polymorphism exclusively in individuals o f African descent (49/576 alleles studied, only 2 homozygous for mutation) and none of the 81 Caucasians studied. The deletion is hypothesized to have arisen as a result of a deletion between two Alu repeats that have 80% homology. The timing of this deletion is also believed to have occurred 100,000 years ago. The allele is found only in populations o f African descent. Thus, the authors hypothesize that this mutant allele could account for some of the genetic protection against the occurrence of Ewing's sarcoma in individuals o f African 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. descent. However, this is based on a disproportionate number of African individuals studied. Furthermore, the mutation has a prevalence o f only 8% in the African populations which could be argued to only protect a small number of individuals and hence is unlikely to account for the large discrepancy of Ewing’s across ethnicities. Furthermore, Kovar points out that a majority of Ewing’s translocation rearrangements take place in intron 7 not intron 6, however, he states that the Alu repeats may be involved in chromosomal structure and s t a b i l i z a t i o n . ^ 3 One can conclude that if the intron 6 polymorphism plays a role, it does not by itself account for the large ethnic difference in Ewing’s sarcoma cases. Therefore, there must be other genetic elements that account for the ethnically disproportionate incidence of Ewing’s sarcoma. IV.A.7. t(ll;2 2 ) Translocations in Other Tumors The t(l 1 ;22)(q24;q 12) translocation has been found in several other tumor types besides Ewing’s sarcoma and peripheral primitive neuroectodermal tumor (pPNET). An explanation for this finding could be that these tumors are also related to the Ewing’s family o f tumors. Of interest is the finding of EWS/FLIl in biphenotypic or polyphenotypic tumors because it questions what the true cell of origin is for these tumors. Whang-Peng et al. noted the presence of a t(l 1 ;22) translocation in esthesioneuroblastoma, or olfactory neuroblastoma, a rare tumor derived from neuroectodermal stem cells of the olfactory epithelium. The cell line investigated contained other cytogenetic aberrations; however, the t(l I;22)(q24;ql2) 49 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. translocation was indistinguishable from the translocation involved in the Ewing’s family o f tumors and found in 70% of the cells studied. It is interesting to note that the cell line was derived from a 22-year-old Caucasian male similar in ethnicity and age distribution to Ewing's sarcoma patients. Sorensen et al. also investigated the presence of the EWS/FLIl fusion transcript in esthesio neuro blasto ma, and determined that two cell lines investigated contained transcripts fusing EWS and FLIl similar to Ewing’s family of tumors. *05 The cell lines were derived from a 22- year-old male and a 22-year-old female. Four of six primary esthesio neuroblastoma tumors contained EWS/FLIl by RT-PCR. The two tumors without EWS/FLIl expression were from two males aged 50 and 54. The authors note that esthesioneuroblastoma has a bimodal age distribution (late adolescence/early adulthood and 55 years of age) with the late peak possibly representing another cell of origin. The authors also note that one of the cell lines investigated expressed insulin growth factor -I, IGF-I (see below for more discussion). Other investigators have foiled to find evidence of a translocation event in esthesioneuroblastoma. Nelson et al. did not find expression of the MIC2 antigen, which is characteristic of Ewing sarcoma family tumors, in 18 esthesio neuro blastomas, but they did find evidence of other neural markers. 106 The authors hypothesize that esthesioneuroblastoma may be of neural origin, but do not find evidence to suggest that it is part of the Ewing’s sarcoma family o f tumors. Further investigations done by Argani et al. found no evidence for MIC2 expression (0/17 studied) or presence of EWS/FLIl (0/11 studied) in 20 olfactory 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. neuroblastomas. *07 Mezzelani et al. found no evidence of a translocation between EWS/FLIl, EWS/ERG, or EWS/FEV in five frozen esthesioneuroblastoma tumor specimens via RT-PCR, Southern blot, or FISH. 108 Sorensen et al. found expression of EWS/FLIl in a group of tumors termed biphenotypic sarcomas because they display both myogenic and neural features. 109 The authors noted that Ewing's sarcoma and peripheral neuroectodermal tumors display neural features while rhabdomyosarcomas (RMS), another small round blue cell tumor of childhood which does not have an ethnic incidence pattern like Ewing's sarcoma, displays myogenic features. A group of tumors displays features o f both type along with the expression o f EWS/FLIl. The expression o f EWS/FLIl was determined by RT-PCR and Northern blots in 5/5 biphenotypic sarcoma cell lines while RT-PCR proved that the fusion gene was also present in 3/3 primary tumor samples available. The fusion gene in the biphenotypic sarcomas was indistinguishable from the fusion gene o f the Ewing's family o f tumors. The authors hypothesize that the Ewing’s family of tumors and these biphenotypic tumors are related, and that the presence o f the EWS/FLIl gene causes a neural phenotype. All cell lines and tumor samples were derived from Caucasian cases except for one for which the ethnicity is unknown (Dr. Triche, personal communications). 110 Thomer et a l described the presence o f EWS/FLIl in two polyphenotypic tumors and two embryonal/alveolar RMS.l 11 The cases were as follows: a girl age 5 and a boy age 37 months with polyphenotypic tumors; and a girl age 20 months and a girl age 4 V z with RMS. RT-PCR detected the expression of EWS/FLIl in all 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. four tumor specimens from the cases. The authors note that unrelated tumors may adopt the same tumorogenic pathway by expressing EWS/FLIl. No information on the ethnicity o f these tumors is presented. Burchill et al. found expression o f the EWS/FLIl gene by RT-PCR in 6/6 Ewing's sarcoma patients and also in 2/12 neuroblastoma patients.* *2 Both neuroblastoma patients were male (ages 3 and S 14) and both presented with adrenal masses besides other lesions. Catecholamine expression was positive, MIC-2 staining was negative, and no N-myc amplification was noted in either case. The 3- year-old patient had a normal appearing bone marrow chromosome spread whereas the 5-14 year old had a “complex hyperdiploid karyotype” with changes to the chromosome lq. The authors suggest that both tumors define a subset o f tumors with mixed Ewing’s sarcoma and neuroblastoma characteristics indicating that both are derived from a neural crest cell capable o f developing into both tumor cell types. These results may represent false positive findings or the presence of the translocation in a subset of tumor cells as cyto logical data did not note a translocation. Furthermore, it is not known whether the RT-PCR results corresponded with actual protein expression. Hawkins et al. found the t(l I;22)q24;ql2) translocation in a 17-year-old male patient with acute nonlymphocytic leukemia who had received a bone marrow transplantation (BMT) from a matched donor. * *3 After the BMT, a t(l 1 ;22) translocation was noted and pre-BMT samples from the patient revealed that the translocation was present in approximately 50% of the cells. No prior cytologic 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. information was available and no conclusion could be made as to when the translocation occurred during the disease process. In every case where the ethnicity of the patient with the EWS/FLIl fusion gene has been found in non-Ewing’s sarcoma tumors, it has been Caucasian. Either there is a bias in the reporting and detection of this fusion gene in the limited number of cases or the fusion gene is synonymous with the ethnic specificity of these tumors. IV.A.8. EWS Gene Involved in Other Translocations The EWS gene is involved in a number of other translocation events in other tumors besides the Ewing’s sarcoma family of tumors (Table 2 below). As with EWS/ets translocations, these translocations involve the amino terminal portion of EWS fused with a DNA binding domain of the other fusion product. If the EWS gene itself is responsible for the ethnically disproportionate incidence rates of Ewing’s sarcoma, one could hypothesize that these other tumors would show a similar ethnic incidence pattern. Gerald et al. characterized 109 patients with desmoplastic small round-cell tumor (DSRCT) containing a t(l 1 ;22)(p13;q 12) translocation fusing the EWS gene with the Wilm’s tumor gene (WT1)J ^ Ewing’s sarcoma and DSRCT have several characteristics in common; both occur in children and young adults, both are poorly differentiated small round cell tumors and both contain the N-terminal domain o f EWS fused to a DNA binding protein. Of the 109 cases, 19 were female (17.4%) and 90 were male (82.6%). The median age at diagnosis was 22 (range 6-49) and females were diagnosed on average 4 years earlier than males. The EWS breakpoints occurred in intron 7 with one case occurring in 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. intron 8 and one in intron 9. The WT1 breakpoints all occurred within intron 7, which resulted in the fusion of the N-terminal domain o f EWS with the DNA binding zinc-finger domain o f WT1. No information on the ethnic background of the 109 patients is provided in the text. However, DSCRT is seen in a wide number of ethnically diverse patients (Gerald WL, personal communication).! 15 Zucman et al. described the presence of a t(12;22Xql3;ql2) in four cases with malignant melanoma of soft parts which results in an EWS/ATF-1 fusion product. 11 6 Malignant melanoma of soft parts (also called clear cell sarcoma) occurs primarily in young adults and is more common in females. It occurs in the deep soft tissues usually near tendons and aponeuroses. The translocation leads to fusion of the N-terminal domain o f the EWS gene to the bZIP DNA binding domain ofATF-1. ATF-1 is a cAMP-dependent transcription factor. The fusion protein contains a large portion of the ATF-1 gene, but is lacking a regulatory cAMP dependent site. The reciprocal gene fusion was found to be expressed by RT-PCR and its contribution to pathogenesis is unknown. Epstein et al. previously characterized one cell line, SU-CCS-1, used in the Zucman et al. study. 11 ? The cell line came from the pleural effusion of a 16-year-old Caucasian female. A second cell line used by Zucman et al. was derived from a 36-year-old male from Sweden. 118 The authors note that in this case, the 12q- marker was lost in 30% of the cells and that the t(12;22) may be quite prevalent in malignant melanoma of soft parts (3/5 cases described). Although the ethnicity when specified has been Caucasian, the incidence of malignant melanoma of soft parts appears to be similar 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in Caucasians and African-Americans (0.1/million in Caucasian males, 0.1/million in African-American males, 0.2/million in Caucasian females, 0.1 /million in African- American females) (SEER 1 9 8 3 - 1 9 8 7 ).! 19 Panagopoulos et al. have described a translocation t(12;22)(ql3;ql2) which fuses EWS and CHOP in a subset of myxoid l i p o s a r c o m a .120 Normally, the t(12;16)(ql3;pl 1) translocation which fuses FUS and CHOP characterizes myxoid liposarcoma. However, in at least two patients characterized by the authors, the EWS/CHOP fusion was found. The patients were 40 and 45-year-old males with no ethnic information, although they can be assumed Caucasian since the study was from Sweden. CHOP is a member of the bZIP family o f proteins and contains a leucine zipper that can bind DNA. The fusion gene contained the N-terminal domain of EWS fused to the exon 2 of CHOP. The CHOP gene is transcribed telomere to centromere and the authors hypothesize that this may be reason why it is not frequently found compared to FUS/CHOP. 120 Although the ethnicity of the two individuals is not reported, the incidence of lipo sarcoma in Caucasians and African- Americans is similar (5.5/million in Caucasian males, 5.9/million in African- American males, 3.5/million in Caucasian females, 3.0/million in African-American females) (SEER 1983-1987).! i9 Clark et al. describe the fusion o f EWS to CHN in skeletal myxoid chondrosarcoma the result of a t(9;22Xq22-31 ;ql 1-12). 121 CHN (Chondrosarcoma) is a member of the steroid/thyroid receptor family with two zinc fingers allowing DNA-binding. The entire gene is translocated to N-terminal domain of EWS. The 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. breakpoint in the EWS gene occurred in intron 7. Abnormalities of the EWS region and a translocation t(9;22) are frequent findings in chondrosarcomas. The tumor sample used in the study came from a 66 year old male in a London hospital, but no ethnicity information was p r o v i d e d . ^ 2 Table 2: EWS in non-Ewing's sarcoma tumors and ethnicity. Gene Product Disease Ethnicity of tumor samples Reference EWS/FLIl olfactory neuroblastoma Caucasian (one cell line unknown) Sorensen et al. (1996) EWS/FLIl Biphenotypic tumors Caucasian Sorensen et al. (1995) EWS/WT1 Desmoplastic small round-cell tumor Seen in a variety of ethnicities Gerald et al. (1998) EWS/CHN Myxoid chondrosarcoma No ethnic information presented Clark et al. (1996) Gill et al. (1995) EWS/CHOP Myxoid lipo sarcoma No ethnic information, assume Caucasian Panagopoul os et al. (1996) EWS/ATF-1 Malignant melanoma of soft parts Caucasian in one case, others unknown Zucman et al. (1993) Mack TM (1995) found incidence of lipo sarcoma and malignant melanoma of soft tissue similar in Caucasians and African-Americans (SEER, 1983-1987). If the EWS gene alone is responsible for the ethnic incidence pattern in Ewing's sarcoma, then one can postulate that the tumors also containing EWS fused to other DNA binding proteins would show a similar ethnic incidence pattern. To date, whenever the ethnic information on a case is provided, it has been in Caucasian individuals. However, cytogenetic observations show the presence o f translocations 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. involving EWS and a DNA binding protein in a proportion of these tumors and the ethnic specific incidence for two o f the tumors, malignant melanoma o f soft parts and lipo sarcoma, shows no ethnic difference in incidence. Furthermore, several reports come from Caucasian areas and bias towards inclusion o f only Caucasians may be present. Larger multiethnic studies would be needed to conclusively determine the ethnic distribution of EWS in these tumors. Alternatively, the FLIl or ERG portion of the fusion gene may be responsible for the ethnic specificity of Ewing's sarcoma. However, this would imply that two separate genes, which are also found in other animal species, underwent separate mutational events only in Caucasians, which produced the same ethnic specific results. This is a more unlikely event, but one that must also be investigated further. IV.B. Transformation of Cell Lines bv Ewing’s Sarcoma Fusion Proteins There is no animal model for Ewing’s sarcoma (no animal naturally presents with a Ewing’s sarcoma tumor) and transfection o f EWS/FLIl in to human cell lines has only recently been accomplished (Dr. Triche, personal com m unication).^ EWS/FLIl is able to transform NIH3T3 cells (mouse fibroblast cell lines) and thus this model is most often used by investigators. In comparison, it is not able to transform all rodent fibroblast cell lines. *23 a subclone o f NIH3T3 cells, YAL7, is not transformed by EWS/FLI-1 even though Western analysis shows equivalent levels of the fusion protein. This implies that EWS/FLIl is modulating a different set of genes in this cell line in comparison with NTH3T3 cells. 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IV.C. Target Genes of the Ewing’s Sarcoma Fusion Protein If the translocation process or the fusion protein itself is not responsible for the disproportionate incidence of Ewing’s sarcoma across ethnicities, then perhaps upstream or downstream events are involved. Any gene differentially regulated by EWS/FLIl would be a candidate. However, the gene and its protein product would have to be involved in the majority of cases and be vital to the transformation of cells by the EWS/ets fusion protein. In order to also explain the ethnically distinct incidence pattern, this gene would have to show a genetic difference between Caucasian and non-Caucasians. Braun et al. used representational difference analysis to identify target genes specifically up-regulated or down-regulated by the EWS/FLIl fusion p r o t e i n . *24 cDNA was prepared from NIH3T3 cells expressing the EWS/FLIl fusion protein and NIH3T3 cells expressing FLIl and digested with DpnIL One set of adapters is ligated to the digest products and PCR amplification is carried o u t These adapter molecules are removed from the EWS/FLIl cDNA and new adapter molecules are applied to these cDNA fragments only. The two populations o f cDNA molecules are hybridized (excess o f FLIl cDNA) and PCR is carried out that only amplifies the new adapter molecules. Homodimers representing the specific products of EWS/FLIl are amplified whereas heterodimers (old and new adapter molecules) and homodimers of the old adapter molecules are not amplified. Thus, genes that are specific to EWS/FLIl fusion protein are preferentially amplified. In order to determine genes that are down-regulated, the opposite procedure is used (FLIl minus 58 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. EWS/FLIl). Results from this investigation identified 10 target genes of the EWS/FLIl fusion protein. Three of the genes were identified by sequence analysis. These genes were a cytochrome P-450 (similar to human CYP4F2 and CYP4F3), cytokeratin 15 (an intermediate filament), and stromelysin I (a metalloproteinase family member that digests extra-cellular matrix proteins). The role of cytokeratin 15 in the tumorogenesis of Ewing’s sarcoma is unknown. The role of the P-450 o > - hydroxylase (normally involved in the activation of inflammatory mediators) may be the modulation of immune function or by interfering in signal transduction pathways. Stromelysin I acts during tissue repair and embyrogenesis. However, during tumorogenesis these proteins have been implicated in metastasis. Unidentified genes that were up-regulated were named EWS/FLIl-activated transcript (EAT). Unidentified genes that were down-regulated were named EWS/FLIl-repressed transcript (ERT). EAT-2 was further characterized by Thompson et al. *25 This protein contains a src-homology2 (SH2) domain that selectively binds phosphotyrosine. During signal transduction, SH2 domains function to bring proteins in close proximity such that they can enzymatically activate one another and other s u b s t r a t e s . 126 a related sequence in the human genome has been identified and mapped to chromosome lq22. EAT-2 is upregulated in NIH3T3 cells transformed with EWS/FLIl and transcripts appear after 4 hours, reaching steady state levels within 24 hours after induction (inducible metallothionein promoter using Zn+ 2 as inducer). EAT-2 is also upregulated by NIH3T3 cells transformed with EWS/ERG 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and TLS/FLI1 (TLS(FUS) is structurally similar to EWS and fused with CHOP in myxoid liposarcoma). YAL7 cell lines are not transformed by EWS/FLI1, and EAT- 2 is also nbt up regulated in this cell line. EAT-2 is normally expressed in murine spleen and lung and is able to bind to phosphotyrosine. The human homolog o f EAT-2 is detectable via RT-PCR in Ewing's sarcoma cell lines and in normal thymus and peripheral blood lymphocytes. However, EAT-2 was also detected in some cell lines that do not express the EWS/FLI1 fusion protein (fibrosarcoma and cervical carcinoma). This indicates that EAT-2 may not be unique to Ewing’s sarcoma. Also, EAT-2 can not by itself transform NIH3T3 cells as overexpression of EAT-2 alone does not induce transformation. However, EAT-2 is likely to be one component of the transformation process. May et al. used representational difference analysis with a different restriction enzyme digest (H infl instead of Dpnll) and detected up regulation of manic fringe inNIH3T3 cells transfected with E W S /F L I1 .1 2 7 Manic fringe is a member of the fringe family and involved in somatic development. Fringe proteins may function as secreted glycosyl transferases and may stimulate cell growth in an autocrine and paracrine fa sh io n . 1 2 7 ,1 2 8 Manic fringe transcripts were detected in Ewing’ s/pPNET cell lines and are able to promote tumorogenesis in SCID mice. Like EAT-2, manic fringe transfected NIH3T3 cells are unable to form colonies on soft agar. Thus, manic fringe is only one part o f the transformation process o f EWS/FLI1. 6 0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. EAT-2 and manic fringe are candidate genes that could be responsible for the ethnic specific incidence of Ewing's sarcoma. However, the consistent involvement of these genes in Ewing's sarcoma has not been documented. IV.D. Mutations in Tumnr Suppressor Genes and Oncogenes Radig et a l investigated possible expression and mutations in the p5 3 gene (exons 4-8) and the ras-genes (H-ras, N-ras, K-ras; codons 12 and 13 of exon 1 and codon 61 of exon 2) in Ewing’s sarcoma. 129 in addition, expression of the negative regulator of p53, MDM2 was investigated. Twenty-four cases were analyzed and 3 were found to express p53 (12.5%) with one carrying a T — > A transversion in codon 238 of exon 7 (Cys to Ser). 5/24 expressed MDM2 while none of the cases expressed any of the ras genes nor were any mutations found in the ras genes in the areas screened. The authors state that p53 mutations are a rare event in Ewing’s sarcoma since their work and work of others found mutations in only 10/106 (9.43%) Ewing’s sarcoma cases. Stahl et a l found mutations in the ras-genes in 5 of 24 Ewing’s sarcoma cases and 0 of 6 pPNET cases studied. 130 Burchill et a l found evidence to suggest that ras may be activated and play a role in behavior of Ewing’s sarcoma and pPNET cells. 131 One cell line of each type o f tumor was studied and found to express ras in an activated state despite any mutations in the ras gene itself. The authors believe that the absence of neurofibromin, a negative regulator of activated ras, is responsible for the activation of ras in these two cell lines. de Alva et aL note that aberrant expression o f p53 may be an indicator of poorer p r o g n o s i s . *32 They found 6 o f 53 cases (11%) to have p53 immunoreactivity 61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in greater than 20% of nuclei Although only 11% of patients had aberrant p53 expression they had significantly poorer outcomes (p=0.001, RR = 9) than patients without aberrant expression. 1V.E. Insulin-like Growth Factor^! Pathway The insulin-like growth factor-I (IGF-I) pathway is perhaps the best candidate for the genetic factor besides the EWS gene. Evidence o f IGF-I pathway expression is found in primary tumor samples, NIH3T3 cells transfected with EWS/FLI1, and tumor cell lines. Furthermore, the incidence pattern o f Ewing's sarcoma which increases during puberty, is parallel to the increase in IGF-I levels during this time period. Thus higher levels o f IGF-I during adolescence may be responsible for the increase in incidence of Ewing's sarcoma. IV.E.1. The Role of IGF-I IV.E.1.L IGF-I and Us Receptor Growth hormone, which is released by the pituitary, positively stimulates the release of IGF-I by the liver and local sites. IGF-I is involved in a long loop negative feedback system on the production of GH. IGF-I is involved in growth and differentiation of several tissues. In the serum, there are at least six binding proteins which bind IGF-I and may mediate some of IGF-I effects. *33 The IGF-I gene is 713 amino acids long and comprised o f 4 exons that span over 45 kb on chromosome 12.134 The IGF-IR gene is encoded by 21 exons and spans approximately 100 kb on chromosome 15.*33 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Dore et a l discuss the role of IGF-I in the central and peripheral nervous system (CNS and PNS) by promoting growth and survival. 136 IGF-I modulates effects of neurotransmitters and may affect neurotransmission and learning in the CNS. It increases myelination and promotes survival PNS Schwann cells. Anlar et a l report that IGF-I is involved in the development of the CNS. 13 " 7 Specifically, it acts on the growth cones of neurons and affects motility and chemotaxis. By protecting cells from apoptosis, it is involved in the development and survival of selected neurons. IGF-I is involved in the promotion of neuronal survival acting via two separate protein k i n a s e s . 138 i g f activates phosphoinositide 3-kinase (PI3K) which in turn activates both the serine-threonine Akt kinase and another kinase, the p70 ribosomal protein S6 kinase (pTO56*). The pathway via Akt plays a major role in neuronal survival. Baserga et a l reviewed the role o f the IGF-I receptor in cell growth, transformation and apoptosis. 133 The IGF-IR and insulin receptor (IR) are 70% homologous. Although it is similar to the insulin receptor, the p subunit of the IGF- IR is ten times more mitogenic than that of the IR The IGF-IR contains a and P subunits linked by disulfide bonds. The a subunit is extra cellular and responsible for ligand binding, whereas the p subunit spans the cell membrane containing both extracellular and intracellular domains. The major substrates of the IGF-IR are IRS1, IRS2, She, and protein 14.3.3. IRS1 is involved in the mitogenicity of both 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the IGF-IR and the IR. IRS I is encoded by a single exon on chromosome 2q36-37 and comprises a 1242 amino acid p r o t e i n . 139,140 After interactions with these downstream proteins, the biological role of the IGF-IR is mediated via various pathways. One major pathway is the ras pathway which involves PI3 kinase, Grb2, Sos, and other molecules. Other growth factors can induce this pathway, although the IGF-IR is the major activator of this pathway. The ras pathway is not the only pathway activated through the IGF-IR, and a ras independent pathway is postulated to exist. The IGF-I receptor is involved in cell mitogenicity, transformation (formation of colonies on soft agar or form tumors in mice), and anti-apoptotic signaling. IRS1 is involved in both anti-apoptosis and transformation. IV.E. 1. iL IGF-l in Bone Remodeling Ewing’s sarcoma occurs in and around the area of the bone. There is evidence to suggest that the IGF-I pathway is also involved in bone biology or formation. Canal is et al. describe several factors involved in the regulation of bone remodeling including transforming growth factor p, bone-derived growth factor (or beta2 microglobulin), insulin-like growth factor, and platelet-derived growth factor. 141 The authors note that IGF-I is expressed by bone and cartilage tissues and it is the local expression o f IGF-I that is most important when it comes to growth of these tissues. IGF-I is capable of increasing the number of osteoblasts in culture and also has a stimulatory role on the function o f these cells. Several authors have shown a relationship between polymorphisms in IGF-I and bone mineral d e n s i t y . 1*2,143 64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IV.E.2. Insulin-like Growth Factor-I and Puberty Ewing’s sarcoma incidence increases in childhood and peaks during adolescence. IGF-I levels follow a similar pattern. Juul et al. investigated the serum levels o f IGF-I in 1,430 healthy individuals, from all ages, in order to determine the normal variation in levels as one ages. 144 infant subjects were sampled for blood four times until 9 months of age whereas all other study subjects only submitted one blood sample. Children under the age o f 5 were concurrently being treated for hernias while all other individuals had no chronic diseases and were not on any medications. Free IGF-I increased from childhood and peaked during puberty. The free IGF-I levels in females increased 1 to 2 years earlier than that of males. Results for total IGF-I were similar to free IGF-I. This pattern mimics the incidence curve of Ewing's sarcoma which increases and peaks during puberty. IV.E.3. IGF-I pathway and Ewing's Sarcoma Several investigators have found evidence o f an autocrine loop involved in Ewing’s sarcoma (Table 3 below). Blockage o f this pathway has led to decrease growth in cell lines and mice. Yee et al. describe a difference in the expression of insulin-like growth factor-I (IGF-I) and insulin-like growth factor-II (IGF-H) between Ewing’s sarcoma/neuroectodermal and neuroblastoma. 145 Nine of the ten neuroectodermal cell lines (Ewing’s sarcoma, pPNET, and esthesioneuroblastoma) containing an EWS/FLI1 fusion transcript expressed mRNA for IGF-I whereas 0 of the IS neuroblastoma or pPNET tumors without the EWS/FLI1 fusion protein expressed 65 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IGF-I. In comparison, all twelve o f the neuroblastoma cell lines expressed IGF-II mRNA. All cell lines investigated, including those without the t(l 1;22) translocation, expressed mRNA for the insulin-like growth factor-I receptor (IGF- IR). Furthermore, in an Ewing’s sarcoma cell line that expressed IGFBP-2 (a binding protein that can regulate the effects of IGF-I), antibody to the IGF-IR significantly decreased cell growth compared to baseline. This process was reversible by introduction of excess IGF-I. The authors hypothesize that IGF-I may be involved in an autocrine growth loop in which it is secreted by the Ewing’s tumor cells and binds to IGF-I receptors expressed by these same cells, stimulating growth and proliferation. 145 Also interesting to note is that this same group rarely found IGF-I in tumors of epithelial origin, which is the same embryonic origin of Ewing's sarcoma. 146 In comparison to the IGF-I autocrine growth loop in Ewing’s sarcoma, neuroblastoma cell line proliferation has been shown to be under control of IGF-II, either in an autocrine or paracrine f a s h i o n . 147 IGF-II was shown to increase proliferation of four neuroblastoma cell lines. However, IGF-II mRNA was found in only 6 of 22 neuroblastoma cell lines and 5 of 21 tumor samples. Adjacent cells in the tumor samples were found to express IGF-II in all 21 tumor samples. Furthermore, fetal adrenal cortical cells express IGF-II while no IGF-II was detected in cortical cells o f children aged l- l 1 months. The authors hypothesize that the spontaneous regression of neuroblastoma that occurs in some newborns may be due in part to the reduced levels o f IGF-II. 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Toretsky et al. determined that the presence of insulin-like growth factor-1 receptor (IGF-IR) was necessary for transformation of mouse fibroblast cells by the EWS/FLI1 fusion p r o t e i n . 148 Using fibroblast cell lines derived from IGF-IR knockout and wild type mice, transfection of the fusion protein only produced colonies on soft agar in the wild type colonies although RT-PCR determined that the fusion protein was expressed in both clones. Soft agar is used to determine anchorage independent growth, an indication of transformation. The nature of the interaction of the IGF-IR and the EWS/FLI1 fusion protein was sought. No evidence for an increase in the synthesis or activation of the IGF-IR protein was found. However, distal signaling o f the IGF pathway showed increased levels o f phosphorylation of the IRS1 protein in clones containing the fusion protein and IGF- IR compared to clones containing only the IGF-IR after IGF induction. This increase in phosphorylation was due to a lower basal phosphorylation level in the EWS/FLI1 containing cells. The authors note that most oncogenes do require the presence of IGF-IR and there may be no significance to the finding that EWS/FLI1 requires this receptor or signaling pathway in cell culture. The exact mechanism of interaction between the EWS/FLI1 fusion protein or its downstream proteins and the IGF-I pathway remains unknown. There may be reason to investigate the interaction of the two pathways considering the epidemiologic evidence for a peak in the incidence of Ewing’s sarcoma coinciding with the peak of IGF levels during puberty. Furthermore, IGF-I acting through its receptor is believed to interact with other growth factors in controlling parasympathetic neural survival and development in the 67 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. chicken e m b r y o 149. Thus, the IGF-I pathway is involved the hypothesized precursor cell of Ewing’s sarcoma providing further evidence of the significant role o f this pathway. van Valen et aL described the function of IGF-I receptors in Ewing’s sarcoma cells. 1^0 IGF-I binding to these cells increased glucose uptake, increased glycogen synthesis, and increased synthesis o f DNA. A strong argument for the involvement of the IGF-I pathway in the malignant behavior of Ewing’s sarcoma comes from treatment related experiments. The IGF-I autocrine loop has been hypothesized to be a potential therapeutic target in Ewing's cases by Scotlandi et al. 1 ^ 1 In order for intervention to be successful in patients, it must first be shown that the autocrine loop is functional in most if not all Ewing's sarcoma/pPNET tumors. These investigators found that the IGF-I and its receptor were present in all 6 cell lines and all 8 clinical samples studied. Furthermore, the presence of the t(l 1:22) or the t(21:22) was confirmed in all tissue samples by RT- PCR. Other growth factors and corresponding receptors studied included TGFP(transforming growth factor-beta), NGF(nerve growth factor), EGF(epidermal growth factor), TGFa, and FGF(fibroblast growth factor). However, none displayed consistent expression except TGFp (found in all samples) and the NGF receptor (7/8 clinical samples). In vitro, the addition of an antibody (aIR3) that specifically binds IGF-IR resulted in a decrease in Ewing VpPNET cell line proliferation, increase in apoptosis, loss of chemotaxis response, and loss of ability to grow in soft agar. 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Furthermore, the loss of ability to form colonies in soft agar or migrate in response to a chemotaxic signal when exposed to <xIR3 indicates a role for the IGF-IR in the malignant behavior of these cells. Additionally, in 1998, Scotlandi et al. presented further in vivo evidence that inhibition of the IGF-I autocrine loop could inhibit the growth and metastasis of Ewing’s sarcoma c e l l s . 152 Subcutaneous injection o f Ewing’s sarcoma cells into athymic mice was followed by either treatment with the aIR3 antibody or suramin, a non-specific inhibitor of growth factor mediated autocrine loops including IGF-I. Five of the nine mice treated with aIR3 displayed complete regression o f all tumors. Further evidence o f the effect of inhibition of the IGF-I loop is absence of any lung metastasis in any of the mice treated with aIR3 compared to 43% of the mice in the control group that developed lung metastases. In comparison, suramin was much less effective in producing complete regression (only lo f 5 at the highest dose tested). Only 10% of the mice on suramin developed lung metastases compared to 40% in the control group. Suramin also decreased the number of bone metastases in the treatment group compared to the control group. The authors conclude that the IGF-I autocrine loop is a consistent feature o f Ewing’s sarcoma and may play a role in pathogenesis and that the pathway may be a target for future therapeutic approaches. Kamiel et al. investigated the interaction between the IGF-IR promoter and a fusion protein that combinedthe EWS protein to the Wilm’s tumor 1 protein found in desmopiastic small round cell tumor. 153 The authors chose this line of 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. investigation because WT1 has been shown to bind to the promoter region o f IGF-IR and decrease expression of this gene. However, the EWS/WT1 fusion protein increased expression of the IGF-IR promoter. The authors also investigated the action of the whole EWS gene and found that it reduced expression of the IGF-IR promoter by 58% compared to control levels by an unknown mechanism. This may indicate that loss of one normal copy of EWS could lead to increased IGF-I receptor levels that could provide a growth advantage. Toretsky et al. described the effects of intervention o f downstream molecules on apoptosis in Ewing’s sarcoma cells. 154 Phosphoinositide 3-hydroxide kinase (PI3K) and Akt (a serine-threonine kinase), both downstream targets o f the IGF-IR pathway, were shown to play a role in the inhibition o f apoptosis. PI3K was inhibited by wortmannin and LY294002 and subsequent apoptosis was able to proceed more readily than in non-inhibited cells. Akt is a direct downstream target of PI3K and overexpression of Akt leads to inhibition o f doxorubicin induced apoptosis. These findings suggest that Ewing’s sarcoma cells may exhibit a growth advantage by activation of the IGF-IR signaling pathway thereby reducing the role of apoptosis and promoting cell survival. 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. fashion in the presence supraphysiologic levels of insulin which acts through the IGF-IR. While not every clone was able to form colonies in soft agar and aggregates to the same extent, there was relationship between level of IGF-IR expression and ability to display both of these features of neoplastic change. In order to demonstrate that the in vivo level of IGF-IR expression was sufficient to induce tumor formation, the transfected NIH3T3 cell line was injected into nude mice with subsequent tumor formation observed. The rat and human fibroblast cell lines were unable to form colonies on soft agar (lack of transformation), although they displayed phenotypic changes after addition o f 5 |ig/ml of insulin. IV.F. Nerve Growth Factor and Receptors Another explanation for the occurrence of Ewing's sarcoma in the vicinity of the bone may be the expression o f nerve growth factor (NGF) from bone tissue. IV.F.1. Role of Nerve Growth Factor NGF is a trophic factor located on chromosome lpl3.1 composed o f 2 exons spanning 7.7 kb and is involved in both the prevention of cell death and also apoptosis via two receptors, the high affinity and low affinity receptors. * 56,157 xia et al. have shown that NGF may regulate whether a neuronal cell undergoes apoptosis or continues to survive via two separate MAP (mitogen activated protein) kinase family pathways. 158 NGF, acting via the ERK (extracellular signal-regulated kinase) pathway promotes proliferation, differentiation and cell survival while inhibiting apoptosis. Withdrawal of NGF results in suppression of the ERK pathway and an increase in the JNK (c-JUN NH2-terminal protein kinase) and p38 pathway 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. leading to apoptosis. Yao and Cooper have shown that PI3K (phosphatidylinositol-3 kinase) is involved in the prevention o f cell death when NGF acts on the Trk receptor. 159 Greene and Kaplan discuss the properties o f the two receptors for NGF; Trk receptor (tyrosine receptor kinase) and the p75 receptor. 160 The trk (trkA - high affinity receptor) pathway activates downstream cascade proteins such as ras and SNT leading to the promotion o f cell survival and differentiation. 161 The role o f the p75 or low affinity receptor is less understood, but it is believed to induce cell death and also affect the function o f the trk receptor. IV.F.2. Expression o f Nerve Growth Factor Receptors in Ewing’s Sarcoma Ewing’s sarcoma cells express the high affinity receptor (trkA) for NGF, which plays a role in the prevention o f cell death. * 62,163 Thomson et al. used c-fbs expression induced by exposure o f NGF to indirectly determine the existence o f Junctional high-affinity (trkA) receptors for NGF in four Ewing’s sarcoma cell l i n e s . 162 Nogueira et al. detected expression o f Trk receptors in paraffin-embedded tumor samples using an antibody to the tyrosine kinase r e c e p t o r . 163 Using this approach, trkA was detected in 7/10 cases o f Ewing’s sarcoma. Five o f the ten Ewing’s sarcoma cases came from tumors with features o f neural differentiation and all five expressed trkA. Although this is a small sample size, the authors contend that differentiated Ewing’s sarcoma expresses trkA; however, this needs to be confirmed with at larger number o f samples. Because this work was done in primary 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. tumor samples, not cell lines, there is evidence to suggest that the nerve growth factor/trk receptor pathway could be involved in Ewing’s sarcoma. Sugimoto et al. also found expression o f high affinity (trkA) NGF receptors on 6 / 8 Ewing’s sarcoma cell lines without evidence o f a neural phenotype. ^ 6 O f these six cell lines, only one was functional as determined by c-fos induction. Another explanation for the predilection o f Ewing’s sarcoma in bone may involve vitamin D induced expression o f NGF and trophic effects o f this factor on Ewing’s sarcoma cells. Veenstra et al. reported on the mechanism o f vitamin D induced expression o f NGF in rat osteosarcoma cell l i n e . 164 Using a hGH reporter gene construct, the authors show that the AP-1 site found in the first intron o f the NGF gene is necessary for vitamin D induced NGF expression from osteoblast cells. The vitamin D receptor did not bind this site directly and vitamin D did not increase expression o f c-fos and c-jun, which are transcription factors known to bind to AP-l sites. In light o f these findings, the authors concluded that vitamin D increased the binding affinity o f this site indirectly. Comet et al. report that 1,25- dihydroxyvitamin D3 increases the expression o f nerve growth factor (NGF) mRNA in Schwann cells in vitro, peripheral nervous system cells also derived from the neural c r e s t . 165 If vitamin D levels influence the effects o f NGF on Ewing's sarcoma cells, the fact that 1,25-dihydroxy vitamin D levels are highest during the peak stages o f puberty might also help explain the increase in incidence o f Ewing's sarcoma during adolescence. 166 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V. Clinical Prognostic Factors The prognosis for individuals diagnosed with Ewing’s sarcoma is not optimistic. For children, the survival rate is between 20 and 70% (approximately 50% overall) depending upon whether or not the disease is metastatic or localized at d i a g n o s i s . ^ 6 * 7 Baldini et al. reported that adults with Ewing’s sarcoma may have an unfavorable outcome in comparison to their younger co u n terp arts.^ Based on 37 adult cases treated at Dana-Farber Cancer Institute between 1979 to 1996, the overall 5-year survival rate was 37% (49% for localized disease and 0% for metastatic disease). The authors note that other studies o f childhood Ewing’s sarcoma cases have similar rates. However, the adult cases seem to remain on the lower ends o f the 5-year survival rates. V.A. Translocation Heterogeneity and Prognosis Although heterogeneity exists in the breakpoint region, clinical data suggest that the location of the breakpoint may be an important prognostic factor. In a small preliminary study (N=28) patients with the type 2 (EWS exon 7 and FLIl exon 5) fusion transcript had more metastatic disease compared to the type 1 (EWS exon 7 to FLIl exon 6 ) fusion transcript. The type 2 transcript tended to be found in the central axis while the type 1 transcript tended to be localized and found in the extremities.^! The same group used a multi-center study to analyze the EWS/FLI1 and EWS/ERG breakpoint regions involved in 147 cases o f Ewing’s s a r c o m a . ^2 No correlation between the any breakpoint region and age, sex, tumor location, volume or extension was found. Comparing only patients with localized disease, those with 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. a type 1 (N=31) fusion transcript had a significantly better relapse free survival than individuals with other types o f fusion transcripts (N=24, p=0.04). The reasoning behind this investigation was the fact that the variable location o f the breakpoint may impact the hinge region (helix loop helix domain) which could be structurally and functionally different depending on the amount o f contribution from EWS and FLIl or ERG. However, Ginsberg et al. found no difference in clinical behavior o f the two major fusion products, E W S / F L I 1 and E W S / E R G . 1 6 9 Thirty cases o f E W S - E R G and 1 0 6 cases o f EWS-FLI1 were selected from multiple institutional sites and analysis showed no difference in event-free survival and overall survival between the two groups. Later, work by De Alva et al. has confirmed the findings o f Zoubek et al. in a separate multi-center cohort o f p a t i e n t s . 93 The location o f the EWS/FLI1 breakpoints in 99 patients was found to be associated with prognosis. Only patients with an EWS/FLI 1 fusion transcript were eligible. In a multivariate analysis, both metastasis at diagnosis (RR = 2.6, p=0.008) and type 1 fusion protein (RR=0.37, p=0.014) were found to be prognostic factors for overall survival. Among patients with localized disease (N=74), those with a type 1 fusion protein had better survival than individuals without a type 1 transcript (RR=0.32, p=0.034). The authors found no correlation between fusion transcript type and tumor volume. In conclusion, it appears that the ‘hinge’ o r ‘spacer’ region between the N- terminal domain o f EWS and the DNA binding domain o f FLIl may serve a functional role in the tumorogenicity o f Ewing’s sarcoma. 7 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V.B. Other Cytogenetic Changes and Prognosis Kullendorff et al. analyzed cytogenetic aberrations other than the t(l 1 ;22) or t(21t22) translocations in 21 “short-term cultured tumor s a m p l e s ” . *70 Trisomy 8 was the most common finding (8/21) followed by trisomy 12 (5/21), trisomy 2, 5, 9, 15 (3 patients each). Also, three patients showed loss of some o f the short arm o f chromosome I . Ten patients bad died at last follow-up. Gain o f material on the long arm o f chromosome 1 and chromosome number greater than 50 were found only in patients that were not alive (3 in each case out o f 10 that expired). These findings may represent tumors with a poor prognosis or tumors that were diagnosed at a later stage and thus more likely to be metastatic and lethal. Mugneret et al. found secondary chromosomal changes in 75 o f 82 Ewing’s sarcoma specimens. ^ Most changes resulted in an increase in chromosome number and trisomy 8 was the most common, found in half o f these tumors. A t(l ;16) translocation was observed in 15 cases, unbalanced in 11 o f these cases. The authors were unable to correlate any information with tumor volume or outcome. Also, while the authors made an attempt to use short term cultures for cytologic study, some specimens came from long term culture and chromosomal changes may reflect in vitro changes that occurred in culture. VI. Conclusions Ewing's sarcoma is a rare, but relatively well studied childhood tumor. It is characterized by a number o f inconsistent epidemiologic findings, except for its ethnically disproportionate incidence pattern and its increase in incidence during 77 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. puberty. The ethnically distinct incidence pattern points to the involvement o f an underlying genetic factor. The molecular genetics o f Ewing's sarcoma has been extensively investigated including its characteristic translocation event, yet the presence o f the underlying genetic factor responsible for the incidence pattern remains unexplained. Given its ubiquitous presence in the disease and involvement in pathogenesis, one has to consider the EWS gene as a prime candidate. Yet at the same time, several lines o f evidence suggest that EWS or the translocation may be a random event and therefore, some other genetic fector(s) may explain the incidence pattern of Ewing’s sarcoma. The insulin growth factor-I pathway and nerve growth factor pathway are potential candidate regions due to their postulated roles in the biology o f Ewing’s sarcoma. Our conclusion is that whatever explains the ethnic specific incidence o f Ewing's sarcoma will also explain the disease itself. 78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 2: Analysis of Case Family Control Study o f Ewing’s Sarcoma VII. Introduction Besides rendering the study o f Ewing's sarcoma difficult, the rarity o f this cancer necessitates investigation o f the disease by means o f a case-control study design. In this type o f study, cases are identified by means o f a reporting agency after their diagnosis. Selection o f a suitable control group is critical when using a case-control study design to investigate an association between Ewing's sarcoma and genetic polymorphic markers. Population controls could have been used; however, a disadvantage to this study design would be the risk o f false associations between the marker and disease due to ethnic confounding (as discussed below). Thus, for the purposes o f this investigation, family controls, matched on ethnicity, were used. Essentially, the frequency o f a polymorphic marker in the case was compared to the family control group. Analysis o f this type o f study is still evolving as will be discussed below, and attempt was made to use the best and most thorough use o f the available data from the study. Ewing's sarcoma/pPNET patients were identified retrospectively and prospectively from several sources (Childrens Hospital Los Angeles (CHL A) tumor registry, Cancer Survelleince Program (CSP) o f Los Angeles County, the California Cancer Registry (CCR) and our study web site). We contacted the participants via mail and telephone correspondence. Cases consulted with family members to decide whether or not they were interested in participating. The case/family controls that agreed to paridcpants either submitted a biological sample (blood or mouthwash) via 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the mail or visited Childrens Hospital Los Angeles to donate a sample. Archived tissue taken at surgery or biopsy was used for deceased patients. This combination o f sample procurement, along with the use o f registries covering a wide geographical area and period o f time (1985 to present), proved to be an effective and efficient method to enroll a sufficient number o f participants in a timely manner. VIII. Available Data for Analysis After approximately two years o f enrollment, 97 cases and respective family controls (total o f378 individuals) were enrolled. A breakdown o f the family structure by living status o f the case is found in Table 2 below. O f the 97 cases, 55 (56.7%) were male and 42 (43.3%) were male. Seventy-seven were Caucasian (79.4%), 19 (19.6%) were Hispanic/Latino, and 1 (1%) was other/unspecified ethnic background. Average age at diagnosis was 19.04. Eighty-seven cases were Ewing’s sarcoma and 10 (10.3%) were pPNET tumors. Fifty-three cases (54.6%) came from the CCR, 25 cases (25.8%) from the Los Angeles CSP, 17 cases (17.5) from CHLA, and 2 cases (2.1%) from our study web site. 1 o f the seventeen deceased patients, 10 had normal tissue, 5 had tumor tissue, and 2 had mixed normal and tumor tissue available for analysis. For cases that only had one parent available, 22 had mothers (8 8 %) available and only 3 had fathers (12%) available. An additional 20 half siblings were also enrolled on to the study sharing 13 parents with the case (9 mothers and 4 fathers). 80 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table I: Available Families for Analysis: Living Patients Family Structure Number Breakdown Parents (Both) 1 2 Parent (Only One) No sibs 5 Both Parents, Sibs 46 1 sibling: 24 2 siblings: 19 3 siblings: 2 4 siblings: 1 One Parent and Siblings 1 2 1 sibling: 7 2 siblings: 3 3 siblings: 2 Siblings only 5 1 sibling: 1 2 siblings: 4 Totals 80 patients 135 parents 1 0 0 siblings Deceased Patients Family Structure Number Breakdown Parents (Both) 2 Parent (Only One) No sibs 1 Patient, Parents, Sibs 8 1 sibling: 5 2 siblings: 2 3 siblings: 1 One Parent and Siblings 6 1 sibling: 5 2 siblings: 1 Totals 17 patients 27 parents 19 siblings Overall Total (378 individuals) 97 patients 162 parents 119 siblings Several methods exist for analyzing familial based study designs and what follows is a discussion o f which o f these methods is best suited for analyzing the data Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. available from this study. Much work has been done on the theory o f familial control study design and the exiting methods are still evolving to analyze the data from such a study. An attempt will be made to analyze the above data in the most appropriate and robust way possible. Emphasis will be placed on sources o f bias and the use o f all available data. Two terms are often used in the literature to describe the relationship between a marker allele and the disease causing allele: linkage and linkage disequilibrium. Linkage will be used to describe the association o f the marker allele and disease causing allele within nuclear families. Linkage disequilibrium will be used to describe the association o f the marker allele and disease causing allele in a population, Le. across many nuclear families. The term association will be used to describe the presence o f both linkage (within families) and linkage disequilibrium (across families). In addition, a disease causing allele is the allele responsible for the genetic determinate o f the disease process (or partially responsible for if a polygenic disease process exists). A marker allele is found at a polymorphic site within hypothetical vicinity o f disease causing allele on the gene o f interest. Vm.A. Statistics o f Parental Control Studies (both parents available! Khoury discussed the assumptions and strengths/weaknesses o f a case- parental control d e s i g n . 173 As with any epidemiologic study, the choice o f a control group is essential to the validity o f any study results. Ethnicity is an important confounder in the study o f candidate genes. The use o f a ‘convenient’ population- based comparison group may lead to spurious findings as case and control groups are 82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. likely to have different ethnic backgrounds and thus different allele frequencies independent o f disease status. In genetic studies, genetic markers are used to identify a genetic region that is thought to be in linkage disequilibrium with the true disease causing allele. However, by measuring a marker some distance from the disease causing allele but still in linkage disequilibrium with the disease causing allele, misclassification occurs. This misclassification is nondifferential and leads to underestimation o f the odds ratio (towards unity). Since the true underlying disease causing allelic mutation is unknown in this study, this misclassification cannot be avoided. However, if significant association is found between a marker and the disease causing allele, use o f further markers upstream or downstream from the original marker can be used to further narrow down the region o f the true underlying disease causing allele. Khoury states that the use o f multiple markers leads to an increased possibility o f finding a small p-value based on a spurious association. An assumption for case-parental control study is that the marker genotypes follow Mendelian inheritance. An example o f non-Mendelian inheritance would be fragile X disease in which allele repeat length can increase in the next generation. All markers used in the present study follow Mendelian inheritance. Another limitation o f the study design is that the control group (parental genotypes) may not be representative o f the underlying population at risk, especially if certain parental genotypes are associated with reproduction. In order to compensate for this limitation, we have enrolled living siblings and will use this control group to determine if there is an influence on reproductive ability for a given parental 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. genotype. Analysis o f a case-parental control study is a matched case-control design with controls being the genotypes the case could have inherited from the parents, but did not. Khoury sets up a 2 x 2 table to compare presence or absence o f an allele or genotype in case and control subjects as below (Table 2): Table 2: 2 x 2 Table for case-parental control study Case allele status Control allele status (based on nontransmitted parental alleles) Present Absent Present To U 0 Absent Vo Wo Odds ratio Uo/Vo Khoury expands the table to analyze the effect o f an environmental factor on a genotype by stratifying on the factor and forming the following table (Table 3): Table 3: Stratification o f exposure in case-parental control study Case allele status Control allele status (based on nontransmitted parental alleles) Present Absent Exposure Absent Present T0 U0 Absent Vo W o Odds ratio (among unexposed) 1 IW o Exposure Present Present T t u , Absent V, W! Odds ratio (among exposed) 1 u ,/v , 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This does not allow the assessment o f the effect o f the environment alone. However, one can assess the effect o f an exposure on a person with a given genotype or carrying a given allele on a multiplicative scale (Le., gene environment interaction). Self et al. described a likelihood based method for estimating the association between a disease and allele. In the study o f rare diseases, the best approach is to use a case-control study design. However selection o f a control group has always been a challenging task for genetic association studies due to confounding effects o f ethnicity. The authors present a study design that uses parental alleles to construct a set of hypothetical siblings for use as an ethnically matched control group. Under Mendelian laws, there are four possible sibling combinations from parental alleles. Under the null hypothesis, these four combinations are equally likely to be inherited by the case and any deviation would indicate an association between the allele and the disease. Since sibling data is not always available, the authors note that their procedure is a viable alternative to using siblings as a control group. This is also true for Ewing’s sarcoma/pPNET tumors in which the peak in incidence occurs in adolescence and parents are likely to be available. Spielman et al. described a method o f testing association using the transmission test for linkage disequilibrium or transmission/disequilibrium test (TDT).175 The XDT uses data from families with one or more affected individuals and at least one parent heterozygous at the marker locus. If interested in one allele (for example allele 1 versus allele 2 for a biallelic locus), the tests compares the 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. number o f times the heterozygous parent passed allele 1 to the affected child to the number o f times the heterozygous parent passed the alternative allele (allele 2 ) to the affected child. We would expect the heterozygous parent to pass the specific allele to the affected patient 50% o f the time, with any deviation indicating linkage/association between the marker and disease. The TDT is computed using a variation o f McNemar’s test for matched data using the data set up in the table below with a x2 = (b-c) 2 / (b+c) where a-d equal the number in each genetic category (Table 4): Table 4: Transmission disequilibrium test Non-transmitt ed Allele Transmitted Allele allele 1 allele 2 Total allele 1 a b a + b allele 2 c d c + d Total a + c b + d The authors note that the TDT test uses only heterozygous parents and that it tests for linkage between disease and marker only in the presence o f population association (linkage disequilibrium). The test can be extended to include families with more than one affected child and for markers with more than two alleles. Schaid and Sommer presented two methods for analyzing case-parental control studies. The first method is appropriate to use when Hardy-Weinberg equilibrium (HWE) is present (a likelihood conditional on HWE) while the second method can be used in the case that Hardy-Weinberg equilibrium does not hold for the study population (a likelihood conditional on parental genotypes (CPG)). Both methods use likelihood methods and are based on two genetic relative risks: ( 1 ) a 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relative risk for homozygote carriers o f allele 1 versus homozygotes without allele 1, and (2 ) a relative risk o f heterozygote carriers o f allele 1 versus homozygotes without allele 1. The authors argue that in the case o f a missing parent, there is still information present to allow sufficient analysis. Using available information for the case and one parent, the contribution o f possible genotypes o f the missing parent can be summed and added to the likelihood. Comparison o f the HWE-likelihood and the CPG-likelihood indicate that the HWE is more efficient for higher attributable risks o f the candidate gene (if the disease risk is due to one or a few alleles in the population), but that the CPG method is better for higher relative risks (high risk o f disease if individual from the population carries the allele). The HWE-likelihood is thus better suited for the situation in which relative risks are small and attributable risk is high. However, the authors note that the CPG method is more robust, especially when parents come from different ethnic backgrounds. Sample size estimates for a 25% attributable risk and relative risk (RR) o f 2 to 4 are between 80 and 215 families for a dominant model and 60 to 2 0 0 families for a recessive model. For studies in which a large number o f candidate loci are investigated, the authors suggest a Bonferoni adjustment such that the p-value must be less than a = ax/m where a j is the ‘desired experiment-wise error rate’ and m is the number o f candidate genes under study. If applied to this study, the type I error rate would become 0.05/6 = 0.008. A ‘multi-tiered study’ is suggested for investigating candidate genes. Sibling data can be included in the analysis only if the presence o f the disease is dependent upon the genotype o f the candidate gene and other genes 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. are not involved in the disease. If this is the case, then siblings are independent, conditional on the parental genotypes. The involvement o f one or more genes may or may not be the case in the present study. Schaid and Sommer compared a number o f proposed statistics used for case- parental control association studies while proposing two new statistical methods.177 The matched analysis o f the TDT is compared to an unmatched analysis. The authors conclude that the use o f an unmatched analysis o f matched data can lead to a ‘conservative statistical test’ with subsequent loss o f power in the presence o f a strong confounder (ethnicity). The authors propose the use o f their likelihood-based approach, which allows for the modeling o f dominant and recessive genetic relative risks using simple logistic regression. Comparison o f power for a dominant, recessive, and additive model show that the likelihood methods for dominant or recessive methods were most powerful if the true underlying disease models were dominant or recessive, respectively. The TDT method was most powerful for an additive model o f disease. Schaid introduced the use o f a score statistic that is both robust and able to utilize information from a multi-allelic locus. *78 Score statistics are derived from the conditional likelihood methods o f previous investigators (174.176-177) given by the following equation: L =ITVi exp(X, B) S g e G i exp(X« p) N = all triplets o f cases and parents Xi= coded vector o f the case 88 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Xa= coded vector o f one o f the four possible marker genotypes that the parents could have produced G= the four possible marker genotypes that the parents could have produced (3= log-odds o f disease for K-l o f the component alleles at the locus with K alleles, compared to the arbitrarily selected remaining allele. The likelihood is the same as a conditional logistic regression model with three pseudosibling controls (three genotypes the case could have inherited from the parents, but did not). Score statistics are used to test Ho: P=0, no association between genetic marker and disease (marker allele frequencies were equally likely). Genotypes and alleles are coded as shown in table 5: Table 5: Example o f coding alleles versus genotypes for K = 3 alleles GEN GDOM GREC GTDT Coding o f Coding o f Coding o f Coding o f Genotype Allele Allele Allele Genotype A B c A B c A B c AA 0 0 0 0 0 1 0 0 1 0 0 2 0 0 AB I 0 0 0 0 1 1 0 0 0 0 1 1 0 AC 0 1 0 0 0 1 0 1 0 0 0 1 0 1 BB 0 0 1 0 0 0 1 0 0 1 0 0 2 0 BC 0 0 0 1 0 0 1 1 0 0 0 0 1 1 CC 0 0 0 0 1 0 0 1 0 0 1 0 0 2 The general (GEN) model codes individual genotypes and can lead to a large number o f degrees o f freedom for a locus with many alleles. The dominant (GDOM) model assigns the value o f the jth element o f the vector X as X j = 1 only if that individual has one or two alleles o f type j and 0 otherwise. The recessive (GREC) model assigns the value x ,— 1 only if that individual has both alleles o f type j. The 89 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. additive (GTDT) model, which generalizes the TDT, assigns the value xj = to 0,1, or 2 depending upon whether the individual carries 0 ,1 , or 2 alleles o f type j. Schaid shows that if the underlying disease model is not known, then the GTDT (generalized transmission disequilibrium test) is the most powerful to find an association between a marker and disease. 1 Sample size calculations are given for the required number o f case-parent triplets to have 80% power at the 5% significance level for detecting a candidate locus with the indicated number o f alleles and relative and attributable risks in table 6 : Table 6 : Sample size estimates for a case-parental control study design adapted from __________________ Genetic Model_________________ Schaid (1996) Dominant_________________ Recessive_______ Attributable Risk__________ Attributable Risk______ # o f Relative Alleles Risk 5% 10% 20% 5% 10% 20% 2 n=736* 404 260 1,254 542 271 4 439 228 125 741 304 133 1 0 413 2 2 1 1 1 0 707 279 114 2 969 533 343 1,652 714 356 4 578 300 164 976 400 176 1 0 545 278 145 931 367 150 * The required number o f case-parent triplets to have 80% power at the 5% significance level for detecting a candidate locus with the indicated number o f alleles and relative and attributable risks. Attributable risk is defined as the “percentage o f disease in the population that is attributable to the marker alleles”. A value o f 10% is assigned to a complex disease. The relative risk (RR) is defined as the risk o f disease for a given genotype (or allele) compared to the arbitrarily chosen baseline genotype (or allele). RR depends on both linkage disequilibrium and recombination. Schaid also 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. not allele 2 and the case/parent set would be included. Thus, the frequency o f allele 1 and allele 2 in the population will influence the genotype o f the missing parent and affect the inclusion o f case-available parent in the test statistic. The authors use the example o f an extremely rare allele 2. If an available allele 1/allele 2 parent is considered for inclusion in the study, then most likely the missing parent will be allele 1/allele I. Thus, allele I will be passed from the missing parent and half the time allele 1 from the available parent (allele 1 /allele 1 case - included in analysis) and half the time allele 2 (allele 1/allele 2 case - discarded). This will lead to accumulation o f evidence for association between allele I and disease, leading to bias. The authors conclude that for a bi-allelic locus, all case-single parent pairings must be discarded. This does not hold for multi-allelic markers in which a heterozygous case and a parent with a different heterozygous genotype can be included. Schaid and Li present a method by which missing parental data is inferred from the affected and unaffected siblings o f the case. ^ a likelihood ratio statistic that incorporates the probabilities o f the missing parental alleles is used to test for association. Curtis later points out that the use o f the affected sib to imply the missing parental allele may also introduce bias in a similar manner. 1 ^ 2 Sun el al. propose another test statistic termed the 1-TDT which avoids the bias noted by Curtis and Sham.1^ Two test statistics are evaluated, Tt and T2. Ti = bi-Cj/ sqrt V where V = bi+ci and bi=Aoi+Au and ci = A10 + A21 from the following table for a bi-allelic marker with alleles N and M (Table 7): 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 7: 1-TDT for missing parental controls Parental Genotype Case Genotype NN(0) NM(1) MM(2) NN(0) A o o Aoi 0 NM(1) Aio A „ A |i MM(2) 0 A2i A22 Ti is a valid test statistic only if two assumptions are true. First, for a given genotype, males and females must have the same mating preference. Secondly, the probability o f missing a father or mother is 50% in a nuclear family missing one parent. In our study, there is no reason to believe that the first assumption is not true. O f the families with only one available parent, 8 8 % are mothers and 12% are fathers. Thus, the second assumption does not hold true for this study population. When the two assumptions do not hold true, the authors propose a second statistic T2. This statistic takes in to account the number o f families with either the father or mother available. In order to use all available family situations in an analysis, the authors provide a suggestion for analyzing the case o f a mixed case/parental/sibling control study, by breaking the data up in to the following four groups: ( 1) Case and both parents, (2) Case and one parent - no siblings, (3) Case and siblings - no parents, (4) Case and one parent and siblings. For the first group, the TDT can be used. For group 2, the 1-TDT is appropriate. For group 3, the S-TDT is used (see below). For group 4, the 1-TDT or the S-TDT can be used, but the authors note that the S-TDT is usually more powerful. Knapp developed another approach to reconstruct missing parental alleles with an adjustment to correct for any biases introduced during the r e c o n s t r u c t i o n . 184 93 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The procedure proposed by the author is termed the reconstruction combined TDT (RC-TDT). The TDT and S-TDT (see below for further discussion o f this sibling based test) based procedure reconstructs missing parental alleles if possible, corrects for the reconstruction to avoid bias, and uses available sibling data if parental reconstruction is not possible. However, only modest gains in power over the S- TDT test were observed for the RC-TDT. Clayton describes a new transmission/disequilibrium test statistic that can be applied when parental information is missing. *85 The approach is based on the TDT and involves the use o f a partial likelihood. It uses any available parental information along with additional affected and unaffected siblings. Previously, Cervino et a l have shown that the TRANSMIT software developed from this approach is very robust in handling different data s e t s J 8 6 Lake et a l point out that the method relies on variables based on complex population genetics theories that are difficult to validity and thus assign values for use in an a n a l y s i s . 187 V m .C . Statistics for SihKng Control Studies Curtis described the use o f siblings as controls in genetic association studies in comparison to traditional population based controls. 182 The author explains that the TDT is not the optimal design in late onset diseases because parental controls will be largely unavailable. Although methods for implying the parental genotypes are available by utilizing unaffected sibs, the author argues that this approach may not produce the missing genotypes and may introduce bias. Therefore, the author proposes the use o f an unaffected sibling as an ethnically matched control. This 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. approach is argued to be valid; testing for both linkage and association as affected and unaffected siblings are equally likely to inherit a parental allele under the null hypothesis. In a qualitative manner, the author notes that the power o f a sibling control design is much lower compared to population control design because siblings are more likely to carry the same alleles as the case. However, one does not have to worry about false positive associations due to population stratification in the sibling control design. In order to increase the power o f the test, the author proposes sampling multiple unaffected siblings and using the most genetically distinct sibling as the control for the case (i.e., if case is allele 1/allele 1 and siblings are allele 1/allele 1 , allele l/allele 2 , and allele 2 , allele 2 then the allele 2 /allele 2 sibling would be used), which does not introduce bias according to the author. However, information is lost when more than one unaffected sibling is found since other siblings may have been used as controls (Le., several allele 2 /allele 2 and allele l/allele 2 siblings would have been discarded from the data set). Speilman and Ewens proposed the use o f the sib-TDT (S-TDT) when parental data is unavailable, but unaffected siblings are a v a i l a b l e . 1*8 The S-TDT tests whether or not a given allele frequency among affected cases differs from the frequency o f the allele in the case’s unaffected sibling(s). The authors propose the combination o f the TDT for families with parental data and the S-TDT for families with sibling data into one z score test statistic. In the case that parental and sibling data are available from a family, the authors propose analyzing only the parental data using the TDT since it is more powerful than the S-TDT. Alternatively, if a family 9 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. has missing parental genotype information with available unaffected siblings, the S- TDT should be used. Monks et cd. proposed another test based on the S-TDT and compared this test statistic to previous sibling test statistics. 1 89 The authors note that in the presence o f larger sibships containing more than one affected or unaffected sibling, all o f the sibling tests are only valid as a test o f linkage. Curtis’ approach to this problem was to shrink the larger sibship by randomly selecting only one affected and then selecting the most genetically different unaffected sibling. *82 The proposed test also has discards information if larger sibships are potentially available, but allows addition o f parental information. The argument refering to the inappropriate use o f sibling controls is based on conditional logistic regression (CLR) based approach (i.e. S-TDT and other variations o f this test statistic). One assumption o f the CLR model is that siblings are independent. The score statistic used in the CLR model is based on a numerator (observed minus expected values) and denominator (some measurement o f variance) as are many statistics. Linkage between the disease causing allele and a marker allele will cause affected siblings to carry similar marker alleles. Thus, there will be an underestimate o f the variance (or variability in marker data compared to completely independent controls, Le., population controls) in the sibship which will affect the test statistic (score test). Use o f sibling pairs (one affected and one unaffected) minimizes this bias by decreasing the number o f available for the variance estimate and in the case o f the Curtis’ sibling test, using the most 96 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. genetically distinct unaffected sibling (creates more variance in the data if it truly exists). Horvath and Laird present an additional sibling control test statistic, the SDT (sibship disequilibrium test). ^ The test allows for the inclusion o f markers with multiple alleles and can add families with parental information into a single test statistic. The authors note that the S-TDT test proposed by Spielman and Ewens is a valid test for linkage disequilibrium only if sibling pairs are used or the assumption o f no linkage is made because the test requires siblings to be independent. The proposed SDT allows for the use o f data from all affected and unaffected siblings while still testing for both linkage and linkage disequilibrium. The SDT does not account for the correlation between siblings and thus is a better test than previous sibling control statistics. The SDT is the nonparametric sign test o f the difference in the mean number o f alleles o f interest in the affected and unaffected siblings (Le, the mean number o f type 1 alleles in the cases minus the mean number o f type 2 alleles in the unaffected siblings). Exact p-vahies can be easily calculated and the test can be extended to test multiple alleles at a locus, thus the variance is not estimated avoiding the problem encountered by previous investigators. The TDT and SDT can be combined in studies containing both sibling and parental data. Schaid and Rowland present a general approach to using a multivariate score statistic for study designs using siblings. 191 A score statistic for sibling control analysis is presented which is the same as a score statistic for conditional logistic regression (CLR) model stratified on sibships. 97 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Siegmund et al. note that the previously described test employing a conditional logistic regression analysis for sibling data proposed by Schaid and Rowland is not always valid as a test o f linkage disequilibrium in the presence o f linkage. 1 92 7 ^ variance o f the score test is underestimated when there is linkage between the true disease causing allele and a marker allele. This violates the assumption o f the conditional likelihood, that the sibling’s marker and disease status are independent. The use o f multivariate regression is proposed that utilizes a Wald test with a robust variance estimate for correlated outcome data (Le., a better estimate o f variance in the data compared to the variance estimated in the score statistic). This test statistic has the benefit o f using procedures that can be coded using SAS® or other statistical analysis software. The test uses all the marker data from affected and unaffected siblings. Different genetic models such as dominant, recessive, or additive, can be coded. The authors show that although linkage between marker and disease causing allele can increase the false positive rate, this increase in minimal (<6 % for sibship o f size four) when the genetic odds ratio is low. In addition, the authors note that the standard Wald tests for conditional logistic regression may perform adequately under most situations. In comparison to the SDT test, the use o f the robust Wald variance estimate had more power to detect an association. V m .D . Comparison o f Statistics for Family-based Study Design* Schaid examines the history along with the strengths and weaknesses o f the TDT. 193 The first study design using familial controls used the non-transmitted 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. alleles o f the parents o f the case (only one pseudosibling control) with the test statistic termed the “haplotype relative risk” (HRR). ^ Analysis o f such a study involves the use o f traditional case control methods, yet there are limitations in the ability o f the HRR to detect linkage and linkage disequilibrium. The TDT is an improvement on the HRR in that it allows for ‘testing linkage in the presence o f association’ or linkage disequilibrium. The following formula is used to calculate the TDT: TDT = (a - b)2 /(a + b) where a is equal to the number o f times allele 1 is transmitted and b is equal to the number o f times allele 2 is transmitted for a bi-allelic marker For both methods, the case and parents o f the case are needed. The TDT can also be applied to multi-allelic markers by computing a separate TDT for each allele o f the marker. However, correction for multiple testing is suggested. Additional methods have been proposed to perform a global test statistic for the multi-allelic locus (testing for an association between the disease and the entire locus instead o f individual alleles). Schaid comments on the appropriateness o f the global test statistic versus the corrected TDT method by noting that the corrected TDT is more powerful when only one allele is associated with disease status, but that the global test is more powerful when more than one allele o f a multi-allele locus is associated with disease status. 1^3 Furthermore, the TDT is best suited for the situation in which the allelic effect is multiplicative Le. relative risk for carrying one copy o f the allele is r and for two copies is r2 (Le. RR for heterozygote is 3 and homozygote is 9). 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Schaid notes that one can estimate r as a/b where a and b are from the above formula for the TDT. One limitation in the power o f the TDT is from the number o f parents heterozygous at the marker locus. For multi-allelic markers, there also needs to be enough parents heterozygous for an at risk allele in order to detect an association. Thus, if a small number o f parents are heterozygous for the correct allele then there will be a diminished likelihood o f detecting any association with that allele. Other factors involved in decreasing the ability to detect an association are the distance between the marker locus o f choice and the disease susceptibility locus, the amount o f linkage disequilibrium between the two loci, and the genotype relative risk o f the disease allele. Schaid discusses the ability o f the TDT to measure different genetic effects such as parent-of-origin effects and allelic h e t e r o g e n e i t y ^ . However, he notes that in order to do so, the data must be subsetted, likely leading to sparse numbers in some instances. One remedy to this is to apply conditional logistic regression to model (CLR) the different factors and even interactions. In order to set up the model, the case is compared to three pseudosiblings comprised o f the genotypes the case could have inherited from the parents but did not. Gene- environmental interactions can also be modeled by CLR. Schaid and Rowland present a general approach to using a multivariate score statistic for study designs using various controls including parental, sibling, or even unrelated controls. 191 A score statistic for sibling control analysis is presented which is the same as a score statistic for conditional logistic regression (CLR) model stratified on sibships. This is coded in a similar manner to the CLR method for use 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with parental controls (see above). In this scheme, the allele counts are coded as 0, 1 , or 2 depending upon the number o f alleles o f a specified type that an individual carries (additive model)- Environmental factors can be added to the model along with gene-environment interactions. As Schaid has proposed previously, pseudosibling controls can be modeled in similar manner. 193 However, environmental risk factors cannot be evaluated, only the interaction between an environmental factor and marker genotype can be assessed as discussed by K h o u r y . 1 7 3 Unrelated controls can be coded into the model in a similar manner as sibling controls, although confounding due to ethnicity is a limitation o f this approach. In addition, confounders such as ethnic background can be stratified on and thus allows some control for this factor. Power comparisons illustrate several key points o f family-based controls. In general, parental controls can give similar but lesser power to detect an association than unrelated controls. Sibling controls have less power to detect an association than parental and unrelated controls. Additional unaffected siblings for a single affected case increases the power, but with a diminishing effect after approximately 4 siblings (follows formula M/M+l where M = number o f unaffected sibs). Thus, when combining family-based control designs with both sibling and parental genotypes, the parents should be used as controls for the case. Cervino and Hill propose the use o f a likelihood ratio association test (LRAT) for studying family-based control designs and compare this test to others in the situation for which at feast one parent is missing. The LRAT is based on a 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. classical likelihood approach to the study o f family based designs. It assumes that the missing parent does not depend upon the transmitted allele, Le., missing at random. The authors compare several statistical approaches including the approach by Clayton to reconstruct missing parental alleles based on an algorithm and the approach by Knapp that uses either reconstructed parental alleles or sibling data itself in the S-TDT. 184,185 The authors conclude that incomplete families should be incorporated in the analysis since information is lost and power decreased when families with missing parents are discarded. The authors note that the approach used by Knapp is favored, since both parents (even if one is missing) or sibling data is used. 184 Alternatively, the approach o f Clayton, which is based on the TDT and includes reconstruction o f missing parental genotype, is useful, but sibling data alone is not used. ^85 Lake et al. describe a method to test a broad range o f association based statistics that test for linkage disequilibrium in the presence o f linkage. 187 The authors note that in testing for association, one can develop two null hypotheses. Specifically, if 6 is the recombination parameter and 5 the measure o f allelic association, the first null hypothesis can be stated as Ho: 5=0 and 9 = V z (test o f association) while the second null hypothesis can be stated as Ho: 5=0 and 8 < ‘ A (test o f association in the presence o f linkage). Both have the same alternative hypothesis (Ha: 5>0 and 9 < l A). However, sibling marker genotypes are correlated and thus, problems arise when addressing the second null hypothesis. According to 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the authors, this can compromise the level o f significance, a . In order to adjust for the correlation o f sibling genotypes, the authors propose using the Rabino witz-Laird (RL) algorithm, which is a more robust estimate o f the variance, termed the empirical variance-covariance estimator (EV-FBAT option in F B A T ) . * 9 5 The authors note that the method o f Siegmund et al. is restricted to discordant sibships (does not use parental information) and thus, does not produce a test statistic as large as that from the EV-FBAT method which incorporates the entire family structure in to the test s t a t i s t i c . *92 This method has been incorporated in to the F B A T program and is utilized by the - e option o f the f b a t command (see below). V m .L Statistics for Family Based Controls Studies In November o f2000, Laird et al. described a unified statistical method to analyze any family-based association study using a single test statistic. 1 ^ This approach is robust in that it allows for different types o f controls (both siblings and parents o f the case simultaneously) and allows for testing an additive, dominant, or recessive m odel The authors approach the analysis o f family-based tests by conditioning on only the traits (disease status) and parental genotypes. Distribution o f parental genotypes is based on Mendelian inheritance or complex algorithms if parental information is missing. 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In this approach, the authors define the following test statistic for N nuclear families with ni offspring per family: S = £ ijT ijX ij Where Xij = function o f the ij* offspring at the tested locus (value depends upon genetic model and can be a vector to represent multiple alleles together) and Ty= 1 if the offspring is affected and = 0 if unaffected. The authors comment that depending upon how Xg and Tjj are coded, one arrives at the previously described tests such as the TDT (Tjj - 1 for affected and Xij = number o f type 1 alleles o f the ij* offspring, Le., 0,1, or 2 for an additive model or 0,1 for a dominant model) and conditional logistic regression. As stated previously, the distribution o f S is constructed from only the observed traits and parental genotypes. Since the statistic is based on independent transmission o f a marker allele from parent to offspring (an independent event for each offspring), the test avoids previous critiques o f lack o f sibling marker independence. When parental genotypes are known, Mendel’s laws are used to determine the distribution o f the offspring genotypes. If parental genotypes are missing, the appropriate offspring distribution is derived from the observed parental genotype (if only one parent is missing) and any available offspring (see tables in Rabinowitz and Laird, 2 0 0 0 ). *95 In order to implement this unified approach, the authors developed a software package termed FBAT (Family Based Association Test) which is available at http://www.biostat.harvard.edu/~fbat/default.htmL 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Rabinowitz and Laird present details o f this unified approach. Using the same test statistic framework, the authors are able to test different genetic models, different sampling schemes, different patterns o f missing alleles, different null hypotheses, addition o f covariates, and arbitrary phenotypes. The approach is based on functional and classical methods o f conditioning on the minimal sufficient statistics for the null hypothesis (based on traits or disease status and parental genotypes). For example, if both parents’ genotypes are available then the minimal sufficient statistic is the parental alleles and the measured trait, and the conditional distribution is based on Mendelian transmission probabilities. When parental alleles are missing, the authors develop an algorithm to derive the minimal sufficient statistic and the conditional distribution o f the observed data. Essentially, they have reproduced previous tests, provided a way to discriminate which method to use and allowed for the use o f both parental and sibling data simultaneously thus not losing information. Their approach is applicable and valid for two different approaches, one being a genome wide scan for association and the other being the use o f association to fine map a candidate gene in an area that has already shown linkage. The FBAT program developed from this approach allows one to use all the information from a family pedigree in order to test for association. FBAT allows the user to read in a file o f data containing information on pedigree structure, affection status, and genotype for study participants. SAS® can be used to format the data to the specified FBAT format. If appropriate, additional phenotype information can be read in as a separate file allowing for the inclusion o f this data in the analysis. The 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. program allows for several different genetic models to be tested including additive, dominant, and recessive (alleles coded as per Shaid’s p a r a m e t e r s ) . ^ Furthermore, for a multi-allelic locus, one can test the alleles separately or all alleles simultaneously (global test). If parental information is missing, the program uses the genotypes o f all unaffected members o f the pedigree to determine the distribution o f the test statistic. One can also perform the SDT on the markers o f interest, using only the siblings. We will attempt to code half siblings into the model. This program will be used to test for an association in the data collected for this study. The data will be analyzed in several different approaches. VIILF. Considerations in Use o f Younger Siblings Siblings that have reached the age o f diagnosis o f the case and are disease free at the time o f sampling can be considered as controls for the study. A problem encountered in using sibling controls arises when the sibling is younger than the case when he/she was diagnosed. It is possible that since the sibling has not reached the age o f diagnosis o f the case, that the sibling could become a case and be incorrectly included in the study as a control. However, Ewing's sarcoma is extremely rare in siblings (only four reported cases in literature) despite the implications o f involvement o f a genetic factors). Thus, one could argue that younger siblings be included in the analysis because the likelihood o f future diagnosis as a case prior to the age their sibling was diagnosed is extremely small. 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IX. Implementation o f Case/Parent/Sibling Study IX. A. Sample Size Estimates No sample size estimates exist for a case-family control study involving parents and siblings so we based our sample size on estimates for a case-parental controls study, (see Schaid table o f sample sizes above) IX.B. Sources o f Participants In order to enroll our projected sample size, we started to enroll patients from several sources after attaining Institutional Review Board (IRB) approval from Childrens Hospital Los Angeles (CHLA). The first source o f patients was from CHLA (both incident and prevalent cases) diagnosed from 1985 onwards. After also obtaining a listing from the Cancer Surveillance Program (CSP) o f Los Angeles County o f cases diagnosed since 1985, we began to contact physicians and patients. Difficulties in locating patients were encountered and it became clear that additional sources o f patients would be needed. On the suggestion o f a member o f the CSP, we contacted the State run cancer registry, the California Cancer Registry (CCR). Subsequently, we obtained a listing o f cases from the CCR diagnosed since 1988. One region o f the CCR, Orange County/San Diego County/Imperial County required a separate IRB approval from their regional center (UCI). Once this additional IRB approval was obtained, we were able to proceed in contacting a list o f patients diagnosed since 1988 from this region. More recently, we have begun a web based recruiting effort to enroll patients/family member that visit a web site we set up (http://www.jmavtlab-chla-usc.com/). We have successfully enrolled 5 patients (2 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. families have turned in samples for analysis) in the first 3 months using this approach. The diagnosis for these web-based patients must be verified prior to inclusion in the analysis o f the study. IX.C. Participation/Enrollment Issues and Rates One o f the largest problems in undertaking a retrospective study was the ascertainment o f cases. As stated previously, the rarity o f Ewing’s sarcoma made it difficult to attain the projected sample size. In order to overcome this problem we utilized several cancer registries and cases diagnosed as for back in time as 198S. As a result, we had difficulty in locating the patient and/or family members. In general, we were most successful in locating recently diagnosed individuals. However, on several occasions, recently diagnosed patients/fomily members did not want to participate due to emotional adjustment to their disease, recent death from the disease, or unwillingness to deal with the healthcare/research system. Despite these difficulties, we were able to enroll 97 patients and 281 family members (phis an additional 19 half siblings) in a relatively short period o f time; approximately 2 years. Our overall participation rate for contacted cases was approximately 70%. IX.D. Sources o f DNA for the Study One o f problems with ascertainment from a wide geographical area was the ability to obtain DNA for genotyping in an economical and timely fashion. In order to accomplish this, we used a method o f obtaining mouthwash samples described by Lum et al. 197 DNA was isolated from buccal cells obtained from a mouthwash sample by study participants. Written instructions, sampling vials and study 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. documents were sent study participants and all sampling was done at the home o f the study participant. This proved to be a reliable method and since three samples were obtained per participant, we had only one occasion where an individual needed to be recontacted because no DNA was attainable from any o f the three samples. We were able to enroll participants from across the United States and other international sites such as Mexico, Ireland and Israel using this method. Blood samples were used for patients or family o f a patient from CHLA. In addition, several non-CHLA families also elected to visit CHLA and donated blood samples. On other occasions, individuals were mailed a package including an empty blood tube that was drawn by their local medical facility and then returned via a mailed package kit for blood transport. We were provided with the baby tooth (molar) o f a patient that died o f Ewing’s sarcoma. Adapting a method o f obtaining DNA for forensic studies from teeth, we successfully extracted DNA from this specimen and included the genomic DNA from the patient.^® In brief, the tooth was wrapped in aluminum foil, submersed in liquid nitrogen for several minutes and pulverized. The crushed material was subjected to proteinase digestion followed by phenolxhloroform DNA extraction as followed for the mouthwash procedure. Archival tissue specimens were re-examined by one o f two staff pathologists at CHLA. Slides were classified as normal, tumor, or mixed by the pathologist. Only slides containing all normal tissue or areas o f easily separable normal tissue were classified as normal for the purpose o f our analysis. Tissue was scraped from 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the slide using a separate disposable scalpel per patient. The tissue was proteinase digested overnight prior to PCR. X. Analysis of Current Study The FBAT program allows for a robust analysis o f a broad range o f data from family-based association studies. Standard pedigree files are read in to the stand alone software program that is freely available on the web in Mac, Windows, and Solaris/Sparc operating systems. Documentation is also available for operating the software. The interactive program is run by a set o f easily learned commands. All available information from a family (parents, case, siblings) is used to test the null hypothesis o f no linkage and no association. The affection status can be either dichotomous, measured or time-to-onset. Bi- and multi-allelic tests can be performed. The bi-allelic mode tests for association between an individual allele and the disease. Multi-allelic mode tests for association between the gene (one allele or a combination o f alleles at a particular loci) and the disease. Additional degrees o f freedom in the multi-allelic mode may decrease power such that an association between a single allele is found for the bi-allelic mode and no association is found in the multi-allelic mode (df=k-l, k = # o f marker alleles for additive model). For an additive model with two marker alleles, the multi-allelic and bi-allelic tests are the same; the test statistic for the m ulti-allelic mode is the square o f either o f the bi- allelic tests. The additive model is the default model; however, dominant and recessive models can be tested. Schaid has shown that the additive model is the most robust model when the underlying mode o f inheritance is unknown 178. For a bi- 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. allelic marker, the test statistic for the dominant model for a given allele results in the same value as the test statistic for recessive model for the other allele. This can be seen in the example o f the EWS gene (C and T alleles). For the C allele, the dominant model compares C/C and C/T to T/T. Likewise, the recessive model for the T allele is the same model, but comparing T/T to C/C and C/T. Using the wrong model will lead to a decrease in power for the alternative hypothesis, but does not invalidate the test under the null hypothesis. An additional command makes it possible to perform the SDT using only sibling data. Adjustment for multiple tests must also be considered. When conducting multiple tests, especially with multi-allelic markers, a conservative approach would be to use a Bonferoni adjustment o f the p-value. Since this is an initial evaluation and each locus has some biologically plausible association with the disease, one can reasonably argue that adjustment is not desirable and thus, we will not use the Bonferoni adjustment. However, the significance o f any association should take this in to account. X.A. Teat Using All Available Participants The data set will include living and deceased cases along with their respective parents and full siblings as family controls. This is the fullest data set and will be used to compare with various subsets o f this data in the following sections. The following table (Table 8) displays the results o f the bi-allelic mode for different genetic models (additive, dominant, recessive). The informative family number is the number o f families with informative matings (at least one heterozygous parent or 111 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sibling with different genotype than case). Results will be discussed in Section X.F below on page 122. Significant or marginally significant tests o f association will be bolded. Table 8: Tests o f association using all available study participants M arker Allele Inform ative Family # P-Value Add Dom Rec Genetic Model Additive Dom inant Recessive NGF N allele 38 37 13 0.608 0.452 0.821 n allele 38 13 37 0.608 0.821 0.452 IGF CA 188 bp 6 6 0 — 1.000 — 190 bp 15 15 I 0.617 0.515 — 192 bp 54 24 43 0.120 0.104 0.395 194 bp 40 36 9 0.128 0.280 0.165 196 bp 22 22 2 1.000 0.829 — 198 bp 2 2 0 — — — 186 bp I I 0 — — — 178 bp 0 0 0 — — — IGF-IR 90 bp 51 35 28 0.933 0.596 0.640 93 bp 50 27 35 0.966 0.495 0.596 96 bp I 1 0 — — — IRS-1 132 bp 7 7 0 — 0.186 — 134 bp 41 25 24 0.165 0.739 0.016 136 bp 43 28 23 0.955 0.286 0.271 138 bp 2 2 0 — — — 140 bp 10 10 0 0.527 0.527 — 142 bp 2 2 0 — — — IGF-IR M allele 58 44 26 0.309 0.362 0.560 m allele 58 26 44 0.309 0.560 0.362 E S I1841 C allele 34 9 32 0.019 0.003 0.187 T allele 34 32 9 0.019 0.187 0.003 The multi-allelic mode o f the recessive model for the IRS 1 gene resulted in a p-value o f 0.0239 for the locus. For an additive model o f the IRS 1 gene in the multi- 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. allelic mode a p-value o f0.099 was obtained. For the recessive model, the - e option resulted in a p-value o f 0.021 for the IRS1 134 bp allele. The multi-allelic mode (tests all alleles together) o f the recessive and dominant models for the EWS gene resulted in a p-value o f 0.009 for the locus. (The multi-allelic mode for a bi-allelic marker results in the same p-value as the bi-allelic test). For the additive model, the use o f the - e option (see above for discussion o f this command) resulted in a less significant p-value for the EWS gene o f 0.031. For the dominant and recessive models, the use o f the - e option resulted in a p-value o f 0.029 for the EWS gene. X.B. Test Using Living Patients Only Inclusion o f archival tissue from deceased cases is controversial and therefore, analysis o f living patients only with parents and full siblings will be carried out separately for comparison. Results are displayed in a similar manner (Table 9) as above and results are discussed below in Section X.F below on page 122. 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 9: Tests o f association using living patients only Marker Allele Informative Fam ily# P-Value Add Dom Rec Genetic Model Additive Dominant Recessive NGF N allele 35 34 13 0.923 0.803 0.821 n allele 35 13 34 0.923 0.821 0.803 IGF CA 188 bp 6 6 0 — 1.000 — 190 bp 15 15 1 0.617 0.515 — 192 bp 46 19 36 0.147 0.117 0.453 194 bp 36 32 8 0.076 0.217 0.099 196 bp 17 17 1 0.637 0.714 — 198 bp I I 0 — ------ — 186 bp 1 1 0 — ------ — 178 bp 0 0 0 — ------ — IGF-IR 90 bp 45 31 25 0.826 0.638 0.859 93 bp 44 24 31 0.928 0.690 0.638 96 bp 1 1 0 — — — IRS-1 132 bp 7 7 0 — 0.186 — 134 bp 36 23 21 0.309 0.726 0.056 136 bp 39 26 21 0.817 0.482 0.247 138 bp 1 1 0 — — — 140 bp 10 10 0 0.527 0.527 — 142 bp 2 2 0 — — — IGF-IR M allele 53 40 23 0.480 0.610 0.576 m allele 53 23 40 0.480 0.576 0.610 ESI 1841 C allele 27 6 27 0.062 0.039 0.204 T allele 27 27 6 0.062 0.204 0.039 The multi-allelic mode o f the recessive model for the IRS gene resulted in a p-value o f0.067. The multi-allelic mode o f the recessive and dominant models for the EWS gene resulted in a p-value o f 0.078 for the locus. For the additive model, the use o f the - e option resulted in a less significant p-value for the EWS gene o f 0.086. For the dominant and recessive models, the use o f the - e option resulted in a p-value o f 0.137 and 0.187 for the C and T alleles o f the EWS gene respectively. 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X.C._____ Full Versus H alf Siblings Half sibs are coded in to the program as the following example (where pid=pedigree id, id=unique participant id mid=mother id fid=father id): Ewing’s Case Half-Sibling pid=201 pid=201 mid=O02 mid=002 fid=001 fid=003 A different father id (fid) is used to code for the father’s id o f the half-sibling, but the same mother id is used as the affected half-sibling case. A comparison o f the results from data o f only the full siblings was compared to the results from data o f both the full and half siblings. No difference in test statistic values was noted. It is not apparent whether the lack o f difference is from the lack o f informative data from the 19 half siblings or whether the FBAT program cannot correctly account for this family structure. Since the half siblings are not in a listing o f a specified pedigree, it is quite possible that although the FBAT program is capable o f incorporating a number o f family structures, it is incapable o f handling data from half siblings. We are currently in communication with the group that developed the FBAT program and are attempting to resolve this issue. Figure 2: Coding o f half siblings in to the model Father Mother Step-Father pid=201 pid=201 pid=201 id=001 id=002 id=003 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X .P. Test Using Parents Only and Siblings Only It is possible that an allele may be associated with a survival/growth advantage, making the allele more likely to be found in living offspring. Thus, a case may preferentially inherit the allele, not because it is associated with the disease, but because it is associated with survival/growth. Living unaffected siblings would also preferentially inherit the allele and thus show no difference when compared to the case. Therefore, we will compare the results obtained from living unaffected siblings only (SDT) and results obtained from using only parents as controls (hypothetical siblings)(Table 10 & 11). Living patients were used for this analysis. 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 10: Tests o f association using parental controls only Marker Allele Informative Family # P-Value Add DomRec Genetic Model Additive Dominant Recessive NGF N allele 31 30 12 1.000 0.774 0.622 n allele 31 12 30 1.000 0.622 0.774 IGF CA 188 bp 5 5 0 0.655 0.655 — 190 bp 14 14 1 0.796 0.686 ------ 192 bp 40 17 32 0.116 0.051 0.521 194 bp 32 29 7 0.046 0.186 0.041 196 bp 17 17 1 0.637 0.714 ------ 198 bp 0 0 0 — — ------ 186 bp I 1 0 — — ------ 178 bp 0 0 0 — — ------ IGF-IR 90 bp 37 29 19 0.386 0.626 0.385 93 bp 36 18 29 0.466 0.522 0.626 96 bp 1 1 0 — — — IRS-1 132 bp 5 5 0 0.180 0.180 — 134 bp 30 17 19 0.505 0.310 0.056 136 bp 32 23 16 0.423 0.588 0.047 138 bp 1 1 0 — — — 140 bp 9 9 0 0.317 0.317 — 142 bp 2 2 0 ------ - — — IGF-IR M allele 43 33 18 0.889 0.857 1.000 m allele 43 18 33 0.889 1.000 0.857 E S I1841 C allele 25 4 25 0.041 0.021 0.153 T allele 25 25 4 0.041 0.153 0.021 The multi-allelic mode o f the additive, dominant and recessive models for the IGF-I gene resulted in p-values o f 0.362, 0.291,0.108. For the dominant model, the use o f the - e option resulted in a p-value o f0.079 for the 192 bp allele o f IGF-I CA repeat. For the recessive model, the use o f the - e option resulted in a p-value of 0.114 for the 194 bp allele o f IGF-I CA repeat. 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The multi-allelic mode o f the additive, dominant and recessive models for the IRS1 gene resulted in p-values o f 0.092,0.450,0.017 respectively. For the recessive model, the use o f the - e option resulted in p-values o f0.063 and 0.051 for the IRS1 134 and 136 bp alleles respectively. The multi-allelic mode o f the recessive and dominant models for the EWS gene resulted in a p-value o f0.035 for the locus. For the additive model, the use o f the - e option resulted in a less significant p-value for the EWS gene o f 0.056. For the dominant and recessive models, the use o f the - e option resulted in a p-value o f 0.131 and 0.136 for the C and T alleles o f the EWS gene respectively. 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 11: Tests o f association using sibling controls only Marker Allele P-Value (Exact) Living & Deceased Cases Living Only NGF N allele 1.000 0.814 n allele 1.000 0.814 IGF CA 188 bp — — 190 bp 1.000 1.000 192 bp 0.281 0.424 194 bp 0.383 0.238 196 bp 0.344 1.000 198 bp — — 186 bp — — 178 bp — — IGF-IR 90 bp 0.860 1.000 93 bp 0.720 0.851 96 bp — — IRS-1 132 bp — — 134 bp 0.359 0.804 136 bp 1.000 0.814 138 bp — — 140 bp — — 142 bp — — IGF-IR M allele 0.392 0.720 m allele 0.392 0.720 ES11841 C allele 0.134 0.454 T allele 0.134 0.454 If deceased patients are included, the p-value for the EWS gene becomes 0.134 for the bi-allelic and 0.088 for the multi-allelic test. X.E. Test Using Caucasian Cases and Hispanic/Latino Separately The incidence is highest in individuals o f Caucasian descent and intermediate in individuals o f Hispanic/Latino descent. While this may be due to admixture o f European genes in the Hispanic/Latino population, there may be separate genes 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. influencing the risk o f disease in each population. Therefore, association between markers and disease will be carried out for these two groups (Tables 12 & 13). Living patients, parents, and foil siblings will be included in the analysis. Table 12: Tests o f association for Caucasian cases Marker Allele Informative Family # P-Value Add DomRec Genetic Model Additive Dominant Recessive NGF N allele 29 28 11 0.833 0.939 0.559 n allele 29 11 28 0.833 0.559 0.939 IGF CA 188 bp 6 6 0 — 1.000 — 190 bp 12 12 1 0.782 0.662 — 192 bp 35 15 27 0.290 0.353 0.486 194 bp 31 27 8 0.214 0.556 0.099 196 bp 13 13 0 0.782 0.782 — 198 bp 1 1 0 — ------ — 186 bp 1 1 0 — ------ — 178 bp 0 0 0 — ------ — IGF-IR 90 bp 36 24 19 0.684 0.458 0.840 93 bp 35 18 24 0.797 0.648 0.458 96 bp 1 1 0 — — — IRS-1 132 bp 7 7 0 — 0.186 — 134 bp 29 16 20 0.147 0.981 0.039 136 bp 32 25 14 0.704 0.413 0.658 138 bp 0 0 0 — — — 140 bp 9 9 0 — 0.739 — 142 bp 2 2 0 — — — IGF-IR M allele 42 30 20 0.380 0.478 0.551 m allele 42 20 30 0.380 0.551 0.478 E S I1841 C allele 21 4 21 0.557 0.334 0.776 T allele 21 21 4 0.557 0.776 0.334 The multi-allelic mode o f the recessive model for the IRS gene resulted in a p-value o f0.097. For the recessive model, the use o f the - e option resulted in a p- value o f 0.053 for the 134 bp allele o f the IRS 1 gene. 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 13: Tests o f association for Hispanic cases Marker Allele Informative Fam ily# P-Value Add Dom Rec Genetic Model Additive Dominant Recessive NGF N allele 6 6 2 0.480 0.670 — n allele 6 2 6 0.480 ------ 0.670 IGF CA 188 bp 0 0 0 ------ --- — 190 bp 3 3 0 --- --- — 192 bp 11 4 9 0.281 --- 0.770 194 bp 5 5 0 0.075 0.075 — 196 bp 4 4 1 — — — 198 bp 0 0 0 — — — 186 bp 0 0 0 — — — 178 bp 0 0 0 — — — IGF-IR 90 bp 9 7 6 0.782 0.683 1.000 93 bp 9 6 7 0.782 1.000 0.683 96 bp 0 0 0 — — — IRS-1 132 bp 0 0 0 — — — 134 bp 6 6 1 0.257 0.297 _ 136 bp 6 1 6 0.059 — 0.061 138 bp 1 1 0 — — — 140 bp 1 I 0 — — — 142 bp 0 0 0 — — — IGF-IR M allele 10 9 3 0.732 0.683 — m allele 10 3 9 0.732 — 0.683 E S I1841 C allele 6 2 6 0.007 _ 0.029 T allele 6 6 2 0.007 0.029 — - For the additive model, the use o f the - e option resulted in a p-value o f 0.025 for the 136 bp allele o f IRS1 CA repeat. For the recessive model, the use o f the - e option resulted in a p-value o f0.049 for the 136 bp allele o f IRS1 CA repeat. The multi-allelic mode o f the recessive and dominant models for the EWS gene resulted in a p-value o f0.020 for the locus. For the additive model, the use o f 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the - e option resulted in a less significant p-value for the EWS gene o f 0.021. For the dominant and recessive models, the use o f the - e option resulted in a more significant p-value for the EWS gene o f 0.017. X.F. Discussion o f Pertinent Results A discussion o f the pertinent results will focus on consistencies and inconsistencies within the results obtained. Attention will be drawn on any conclusions that can be drawn and the strengths and weaknesses o f these conclusions. Each gene and its polymorphic marker will be discussed individually. Several factors limit the conclusions one can draw from the current data. Estimates o f the length o f linkage disequilibrium surrounding a marker are approximately 60 kb in Northern European p o p u latio n s.^ However, the amount and strength o f linkage disequilibrium surrounding a marker may vary and thus limit the ability to detect association. Therefore, even though a result is non-significant, the gene may still be involved in the disease. Sample size and linkage disequilibrium play a role in our ability to detect an association. Currently we have enrolled a selected population o f Ewing’ sarcoma cases. Any significant result may be a spurious association in the sample size obtained or may actually be associated with the disease. X.F.1. NGFGene None o f the results show a significant association between the marker used for the NGF gene and Ewing’s sarcoma at this point. This non-association can represent a true non-association or the efiect o f an insufficient sample size and inadequate power to detect an association. In addition, a low am ount o f linkage 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. disequilibrium between the true underlying disease causing allele and a marker allele may necessitate the use o f additional markers that are in closer proximity and thus more likely to be in linkage disequilibrium. X.F.2. IGF-I Gene Tests o f association using living and deceased cases as well as parents and full siblings, do not show any association between the IGF-I gene and Ewing’s sarcoma. In addition, tests o f association using the living cases only display similar non-significant results. When only the parents are included in the analysis, a marginally significant association for the 192 bp (dominant model, p=0.051) and the 194 bp (additive and recessive model, p=0.046 and 0.041 respectively) and Ewing’s sarcoma is noted. This may represent a true association, but is likely due to spurious association. Additionally, since there is no association for the sibling only model and the strength o f the association is not present when the siblings are added to the model, this may be a case o f allele association with survival/growth advantage. In this situation, the 192 and 194 bp alleles may provide a growth advantage or other developmental advantage and thus carriers are more likely to survive. X .FJ. IRS1 Gene Tests o f association using living and deceased cases as well as parents and full siblings, show a significant association between the IRS1 gene and Ewing’s sarcoma for the recessive model involving the 134 bp allele (p=0.016). The strength o f this association becomes marginally significant when the deceased cases are removed from the analysis (p=0.056). When only the parents are included in the 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. analysis, a marginally significant association for the 134 bp (recessive model, p=0.056) and the 136 bp (recessive model, p=0.047) and Ewing’s sarcoma is noted. There is no association for the sibling only model. This also may be a case o f allele association with survival/growth advantage or lower power in the sibling only m odel Tests of association in the Caucasian and Hispanic/Latino populations show similar p-values for an association with the IRS1 134 bp allele in the recessive model (p-value 0.039 and 0.061). Allele frequency o f the 134 bp allele in cases is 52.6%, while it is 50.5% in living siblings and 48.6% in pseudosiblings (both parents available). For the 134 bp allele, 24.7% o f cases were homozygous while 20.4% o f siblings and 18.4% o f pseudo siblings were homozygous carriers o f the same genotype. Thus it is possible that there is an association between the homozygous presence o f the 134 bp allele and Ewing’s sarcoma. However, the strength o f this association is not present in allele data sets tested. X.F.4. IGF-IR Gene None o f the results show a significant association between either o f the two markers used for the IGF gene (promoter region and exon 16) gene and Ewing’s sarcoma. This non-association can represent a true non-association or the effect o f an insufficient sample size and not enough power to detect an association. In addition, a low amount o f linkage disequilibrium between the true underlying disease causing allele and a marker allele may necessitate the use o f additional markers that are in closer proximity and thus more likely to be in linkage disequilibrium. 124 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. However, the use o f two markers on separate ends o f the gene decreases the likelihood o f this possibility. X.F.5. EWS Gene Tests o f association using living and deceased cases as well as parents and full siblings, show a significant association between the EWS gene and Ewing's sarcoma for the additive model (p=0.019). For the dominant and recessive models overall tests o f association resulted in p-values o f0.009. Testing each allele separately revealed that the C and T allele are significantly associated in the dominant and recessive model with Ewing’s sarcoma (p=0.003). In the model using the living patients only, the strength o f the association is reduced. For the additive model, the p-value is 0.062 and for the dominant and recessive models, the p-value is 0.039 for the C and T allele respectively. Intriguingly, there is no significant association between the EWS gene and Ewing's sarcoma for the Caucasian cases whereas there is a strong association when the Hispanic cases are used (p=0.007) however, this is based on only 6 informative families. The model using parents only shows significant association for the additive model (p=0.041) and the dominant and recessive models for the C and T alleles respectively (p=0.021). No association was found between the EWS gene and Ewing's sarcoma in the sibling only m odel Explanation for this discrepancy could be decreased power in the sibling only model or growth/survival advantage o f an allele. Allele frequency o f the T allele in cases is 23.6%, while it is 17.9% in living siblings and 14.8% in pseudosiblings (both parents available). For the T allele, 9.0% o f cases were homozygous while 5.0% o f siblings 125 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and 1.1% o f pseudo siblings were homozygous carriers o f the same genotype. Although only a limited number o f families were informative for this locus, significant association was found between homozygous carriers o f the T allele and Ewing’s sarcoma. X.F.6. Interaction between the EWS gene and IRS1 gene The FBAT program offers the advantage o f being able to analyze all the information obtained from family based control studies. However, it only tests for an association and is unable to determine the genetic relative risk. Work is being undertaken at USC (Peter Kraft, personal communication) to allow for this.200 Currently interaction terms for gene-gene interactions are not part o f the FBAT software. Any gene-gene interaction can be modeled using the previously described logistic regression model proposed by Schaid. 178 This only allows the separate analysis o f case and parents or case and siblings. Addition o f an interaction term for EWS T/T and IRS1 134 bp/134 bp did not result in significant interaction on a multiplicative scale for models using pseudosiblings and living siblings (p-value o f 0.101 and 0.196 respectively). 126 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 3: Grant Proposal for a Case/Sibling/Parental Control Study of Ewing’s Sarcoma I. Abstract In the United States, Ewing’s sarcoma has an incidence o f approximately 2.8 per million in children under the age o f fifteen o f all ethnicities combined and is the second most common bone tumor after osteosarcoma. Worldwide there is a well- established difference in incidence o f this disease geographically and ethnically and in some populations there is a virtual absence o f this cancer. The lack o f a migration pattern or change in incidence as a given population moves to a different geographic location suggests involvement o f a genetic factor. Although this observation has been known for some time, and much is known about the molecular genetics o f this disease, the responsible genetic factorfs) have not been identified. Ewing’s sarcoma is most common during adolescence, with the highest age specific incidence rates occurring at about thirteen years o f age in the United States. Although there is an overall slight male preponderance o f the disease, the age specific incidence is higher in females until the start o f the second decade after which the incidence for males increases and remains higher, longer than in females o f the same age. The peak in incidence o f Ewing’s sarcoma during adolescence is parallel to the adolescent growth spurt. Thus, biologic changes occurring during puberty may be affecting the incidence o f Ewing’s sarcoma. These biologic changes may also be related to the genetic factor responsible for the ethnic specific incidence o f Ewing’s sarcoma. 127 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ewing* s sarcoma family o f tumors is characterized by a reciprocal translocation between the EWS gene on chromosome 22ql2 and members o f the ets- transcription factor family o f genes on either chromosome 1 lq24 (FLI1) or 21q21 (ERG). In ninety eight percent o f Ewing’s sarcoma cases there is evidence o f such a translocation and introduction o f this translocation in to cell lines leads to tumor transformation. Thus, this translocation is specific to Ewing’s sarcoma and likely to be related to the etiology o f this disease. Insulin-like growth factor-I (IGF-I), part o f the growth hormone (GH)/IGF-I axis that is upregulated during puberty, has been shown to be a distinguishing marker for Ewing’s sarcoma. Expression o f the insulin like growth factor-I receptor (IGF-IR) is required in vitro for transformation o f cell lines containing the EWS/FLI1 fusion protein. A down stream target o f the IGF-IR, insulin receptor substrate 1 (IRS1) is hyperphosphorylated in cell lines with the fusion protein compared to those without the fusion protein, indicating that it may “communicate” with the fusion protein pathway. Further downstream targets o f the IGF pathway are thought to modulate the tumor cells ability to fend o ff apoptosis, thus leading to cancer cell survival. In addition, Ewing’s sarcoma cells express high affinity receptors for nerve growth factor. Vitamin D induced expression o f nerve growth factor by osteoblasts may account for the predilection o f Ewing’s sarcoma to bone tissue and this pathway may be related to disease etiology. We hypothesize that the genetic factor(s) responsible for the ethnic specific incidence o f Ewing’s sarcoma may be involved in the translocation event, the insulin-like growth factor pathway, and/or the nerve growth factor pathway. In order 128 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to test our hypothesis, we plan enroll at least 125 cases with Ewing’s sarcoma identified through Childrens Hospital Los Angeles and the California Cancer Registry along with both their parents and siblings as ethnically matched controls. We will investigate any difference in the distribution o f polymorphic markers among cases and compare the distribution to that obtained by constructing hypothetical siblings based on the parental genotypes and compared to living sibling controls. Under the null hypothesis, we would expect that any living sibling or any hypothetical sibling genotype will equally likely to be transmitted as any case genotype. Any deviation would indicate an association between the disease and the genotype. This study design will allow us to control for the possible confounding effects o f ethnicity. n . Background and Purpose In the United States, Ewing’s sarcoma (ES) has an incidence o f approximately 2.8 per million in children under the age o f fifteen, all ethnicities combined, and is the cause o f about 2% o f all childhood cancers. * >201 Worldwide there is a well documented difference in incidence geographically and ethnically, and in some populations there is virtual absence o f disease (see Figure 1 b e l o w ) .2*10 Interestingly, the incidence o f Ewing’s sarcoma in African individuals and African- Americans is similarly low, pointing to a genetic etiology. Recent work by our group has also shown that the incidence in Asian-Americans is similar to the incidence rate o f Ewing’s sarcoma in Asia.^1 This is in contrast to cancers associated with an environmental etiology which show a change in incidence upon 129 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. immigration into another geographic site and/or p o p u l a t i o n . 2 0 2 por the combined tumor registries in the United States (Delaware Valley, Los Angeles, New York and the SEER registries), the incidence in African-Americans is 0.35 per million ages 0- 1 4 5,6,10 fo r three combined tumor registries from Africa (Uganda, Nigeria, and Zimbabwe) in which population at risk was known, the incidence rate was 0.22 per million ages 0-14, and based on only one case. 10 In Asian populations, the incidence rates per million ages 0-14 for China and Japan were 0.4 and 0.51 respectively while the incidence rate in Los Angeles County was 0.2 and 0.5 per million for males and females ages 0 to 35 r e s p e c t i v e l y . ^ 1 In contrast, several populations have a much higher incidence rate. The age standardized incidence rates for Caucasian population is around 2 per million in children ages 0-14.10 In addition, in Bombay, India and Israeli Jews, the incidence rates are 2.1 and 2.0 respectively. 1 ® Incidence in Hispanic/Latino populations appears to be intermediate in comparison to Caucasian populations (between 1.1 per million in Cuba to 1.5 per million in Puerto Rico and Los Angeles) Other studies find a similar pattern Le. the incidence of Ewing’s sarcoma is very low compared to the incidence in non-African or Asian p o p u l a t i o n s . ^ " ^ Therefore, we can categorize low risk populations (Asian and African), high risk populations (Caucasian), and intermediate-high risk populations (Hispanic/Latino). The most commonly proposed hypothesis addressing this unique ethnic specific incidence pattern is unidentified genetic factors. 130 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 1: Worldwide Incidence Rates of Ewing's ■ = - Sarcoma 6 2.5 ------------------------- O f t Figure 1: Worldwide incidence rates o f Ewing’s sarcoma. (Adapted from Parkin et a l, 1993.) In the mid 1980s, further advances in the understanding o f the pathology o f Ewing’s sarcoma led to new tools to aid in the diagnosis o f this disease. This led to the discovery that other tumors shared the same genetic makeup as Ewing’s sarcoma with reclassification o f these tumors in to what is now known as the ‘Ewing’s sarcoma family o f tumors’ (ESFT). Ewing’s sarcoma constitutes 87% o f all the tumors classified as ESFT. Besides bone tumors (87%), this group includes extraosseous Ewing’s (8%) and peripheral primitive neuroectodermal tumors (pPNETs) (5%).* Histopathologically Ewing’s sarcoma family o f tumors is characterized as small round blue cell tumors, with evidence o f neural markers and differentiation and a characteristic genetic change, a translocation, t(l 1 £2). The distinction between osseous and extraosseous Ewing’s sarcoma is often difficult because tumors primary to the bone can contain soft tissue components and tumors in the soft tissues may invade the bone. Therefore, for the purposes o f this proposal, 131 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. we will include both types of tumors when referring to Ewing’s sarcoma. Furthermore, it has become clear that the pPNET tumors are essentially the same tumor as Ewing’s sarcoma with classification o f each together by many pathologists. Thus, we will include the pPNET tumors in our reference to Ewing’s sarcoma for this proposal. Ewing's sarcoma does not occur in animals, however there exists a cancer model with an ethnically distinct incidence pattern that is similar to the model of Ewing’s sarcoma in humans, i.e. prostate cancer in men. African-American men have a three-fold risk of developing prostate cancer compared to Caucasian men of the same age and same socioeconomic background. The risk for developing prostate cancer is associated, individually and in combination, with the polymorphisms found in the vitamin D receptor (VDR) and the androgen (AR) genes, i.e. the risk is associated with (1) the length of the poly-A microsatellite on the 3’ untranslated region (3’ UTR) o f the VDR gene and with (2) the length of the CAG micro satellite repeats o f the AR gene.203-208 These polymorphisms show ethnically distinct distributions, i.e. African-American men have a higher prevalence o f at least one long (Aig to A22) VDR poly-A allele and a higher prevalence of at least one AR allele with fewer than 20 CAG repeats. The higher prevalence o f these alleles may explain the higher incidence of prostate cancer in African-American men. Ewing’s sarcoma has an ethnically and geographically distinct incidence pattern. We will test for an association between Ewing’s sarcoma and several candidate loci. 132 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Similarly to prostate cancer, the candidate loci under study may explain the different risk of Ewing’s sarcoma in different ethnic groups. Ewing’s sarcoma most often occurs in the vicinity o f bone, however, several lines o f investigation support the hypothesis that Ewing’s sarcoma is derived from cells o f the neural crest. In support of this hypothesis, Ewing’s sarcoma cell lines can develop neural differentiation when exposed to such agents as serum-depleted medium, tetradecanoylphorbol-13 acetate, retinoic acid, nerve growth factor, and dibutyryl cyclic adenosine m o n o p h o s p h a t e . 5 8 > 2 0 9 - 2 1 2 Ewing’s cell lines also possess receptors for neuropeptide Y and dopamine D-l receptors along with expression of c h o l e c y t o k i n i n . 5 4 , 5 7 , 2 1 3 Neuropeptide Y is a parasympathetic regulatory protein and presence o f cholecytokinin suggests a postganglionic origin.58 Cells o f the neural crest develop into a number of different cell types, namely Schwann cells, melanocytes, neuroendocrine cells, sympathetic neurons, and parasympathetic neurons. A pluripotent neural crest cell is hypothesized to give rise to each of these differentiated cells. Therefore, it can be hypothesized that Ewing’s sarcoma represents a disruption in the differentiation process of parasympathetic neurons. 58 In ninety eight percent of Ewing’s sarcoma cases, a recurring translocation occurs between the EWS gene on chromosome 22ql2 and two members of the ets- transcription factor family, FLI1 (1 lq24) or ERG (21q21), or on rare occasions other members of the ets-gene f a m ily .^ ^ ^ transcription factor family share 1 3 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. homology in a 3 ’ D N A binding d o m a i n . 7 1 The chimeras result in the joining of the amino terminal end of EWS, which has transcriptional activation activity, with the carboxy terminal DNA binding regions of FLI1 or ERG.68»?0 The breakpoints involved in producing the chimeric products differ and a large number of heterogeneous translocation products r e s u lt.? !5 The breakpoint region in the EWS gene on chromosome 22 covers an area o f 2-3 kb while the breakpoint region of FLI1 on chromosome 11 covers a much larger area o f30-40 kb.^2 a glutamine-rich region of the EWS domain is the region where translocation occurs. The EWS gene is always involved in the translocation event, making it an obvious candidate for the genetic factor responsible for the ethnic specificity of Ewing’s sarcoma. We will investigate this possibility by using a polymorphic marker (C/T transition) in the vicinity o f the EWS gene. We should find an association between this marker and the disease if this marker is in linkage disequilibrium with the true disease causing allele. Due to the postulated genetic factor involved in Ewing’s sarcoma and the presence an almost ubiquitous translocation, Zucman-Rossi et al. has investigated possible interethnic variation in the region o f the breakpoint in the EWS gene.!®? They found a rare ethnic specific deletion of several Alu repeats in the intron 6 region that exclusively occurred in individuals of African descent (49/576 alleles studied, only 2 homozygous for mutation; not found in any of the 81 Caucasians studied). This group hypothesizes that this mutant allele could account for some of the genetic protection found in Africans and African-Americans. However, more 1 3 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. individuals of African descent were studied than Caucasians, and the mutation has a prevalence o f only 8% in the African population. Hence it could only protect a small number of individuals and is therefore unlikely to account for the large discrepancy o f Ewing’s sarcoma across ethnicity. Furthermore, Kovar points out that the majority o f Ewing’s sarcoma translocation rearrangements take place in intron 7 not intron 6, although the Alu repeats could be involved in chromosomal structure and stabilization. 103 if the intron 6 polymorphism plays a role, it most likely does not by itself account for the large ethnic difference in Ewing’s sarcoma cases. Thus other (genetic) elements must account for the ethnically disproportionate incidence of Ewing’s sarcoma. If one ethnic population is predisposed to the translocation event, one could hypothesize that the translocation would occur at specific sites within each gene. Instead, the translocation breakpoints span relatively large regions in each gene involved, as discussed above. Furthermore, Zucman-Rossi et al. found no sequences known to be involved in site specific translocation events in the region surrounding the t r a n s l o c a t i o n . 9 5 Taken together, the large region involved in the translocation and the lack of a specific signal sequence points to a random event which could happen as frequently in Caucasian and non-Caucasians. Instead, other genes downstream of EWS/FLIlmay differentially interact with the translocation event in high and low risk populations. Further evidence of the randomness o f the translocation event comes from indirect evidence supporting the idea that the long arm o f chromosome 22 where 135 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. E W S is located is “highly r e c o m b i n o g e n i c ” . ^ Budarf et al. report that chromosome 22q contains the breakpoints for a constitutional t(l 1;22) translocation found proximal to the X light chain locus (involved in the t(8;22) of variant Burkitt lymphoma) and the BCR1 region (involved in the t(9;22) o f acute lymphocytic leukemia) are all proximal to the breakpoint region involved in Ewing's sarcoma.^ However, none of the other 22q translocations show the same ethnic specificity as Ewing’s sarcoma. Ewing’s sarcoma is most common during adolescence, with the highest age specific incidence rates occurring around thirteen years o f age in the United States A * 7,216 The pattern of incidence for Ewing’s sarcoma is parallel to the adolescent growth spurt. 17,216 The age specific incidence o f the pubertal growth spurt for girls precedes that of boys by 12 to 18 months, similar to the incidence of Ewing’s sarcoma of bone. 1*216 Ample evidence links Ewing’s sarcoma to the physiological changes occurring during puberty. The onset of puberty is marked by an increase in the amplitude of luteinizing hormone (LH) and follicle-stimulating hormone (FSH) which leads to an increased levels of gonadal sex steroids. 133 The latter change leads to the development of secondary sexual characteristics, skeletal changes, and muscle mass d e v e l o p m e n t . 133 Simultaneously, changes occur in the growth hormone (GH)/insulin-like growth factor-I (IGF-I) pathway. 133 GH positively influences the release of IGF-I by the liver and local sites and IGF-I is involved in a 136 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. long loop negative feedback system on the production of GH. Both GH and IGF are involved in many o f the body’s growth processes including bone growth. 141 Insulin-like growth factor-I, for which serum levels peak during puberty, is expressed along with its receptor by Ewing’s sarcoma cell lines containing the t(l 1:22) translocation. 144.145 Moreover, the IGF-I receptor (IGF-IR) is necessary for transformation of cell lines containing the EWS/FLI-1 fusion protein. 148 A downstream target of IGF-IR, insulin receptor substrate-1 (IRS 1) shows increased phosphorylation following stimulation with IGF-I in the presence of the EWS/FLI-1 fusion protein compared to clones without the fusion protein. *48 The exact mechanism by which the fusion protein and the IGF-I pathway interact is currently unknown. However, this interaction between the IGF-I pathway and the t(l 1 ;22) translocation protein are intriguing in light o f the role of IGF-I during puberty and the peak incidence of Ewing’s sarcoma during this period. Thus, IGF-I, IGF-IR, or IRS-1, because they provide a link between changes occurring during puberty and the translocation product are potential candidate genes worth investigating. Furthermore, IGF-I acting through its receptor is believed to interact with other growth factors in controlling parasympathetic neural survival and development in the chicken embryo. 149 Thus, the IGF-I pathway is involved in the hypothesized precursor cell o f Ewing’s sarcoma providing further evidence of the significant role of this system. An explanation for the predilection o f Ewing’s sarcoma in bone may involve the interaction o f vitamin D induced expression o f NGF and trophic effects of this 137 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. factor on Ewing’s sarcoma cells. NGF is a trophic factor involved in both the prevention of cell death and also apoptosis via two r e c e p t o r s . ^ 6 Ewing’s sarcoma cells show evidence of expressing the high affinity receptor (trkA) for NGF which plays arole in preventing cell death. 162,163 Although the vitamin D induced expression of NGF in Ewing’s sarcoma is unknown, in a cell line derived from a related cell o f the neural crest lineage, Schwann cells, vitamin D increases the expression of VDR m R N A and nerve growth factor (NGF) m R N A . 1 6 5 Interestingly, in osteoblasts, vitamin D also increases the expression NGF; however, the role of NGF secretion in the bone is not u n d e r s t o o d . 164 If vitamin D levels influence the effects of NGF on Ewing’ s sarcoma cells, the fact that 1,25-dihydroxy vitamin D levels are highest during the peak stages of puberty might also help explain the increase in incidence o f Ewing's sarcoma during adolescence. 166 III. Hypothesis & Specific Aims of Research Ewing’s sarcoma is the second most common childhood bone related cancer. There is a well known ethnic specific incidence of Ewing’s sarcoma with an overwhelming majority of patients of Caucasian descent (96%). This characteristic incidence pattern is hypothesized to be due to genetic factors. The incidence of Ewing’s sarcoma peaks during the second decade of life coinciding with the adolescent growth spurt Ewing’s sarcoma also has a virtually ubiquitous translocation between the EWS gene on chromosome 22ql2 and two members of the ets-transcription factor family, FLI1 (1 lq24) or ERG (21q21). 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. III.A. Hypothesis We hypothesize that the well-known ethnic specific incidence o f Ewing’s sarcoma points to an underlying genetic cause of this disease. We propose to identify the underlying genetic factor(s) by searching for candidate genes involved in this disease. By using a fam ily based control population, we will control for the confounding effects of ethnicity that have plagued past studies o f this type. m.B. Specific Aims m .B .l. Primary Aims In order to test our hypothesis, we will conduct a case-parental/sibling control analysis of the fo U o wing genes: 1. published polymorphic markers of the (1) insulin growth factor-I; (2) the insulin growth factor-I receptor; (3) the insulin receptor substrate-1; and (4) nerve growth factor genes. 2. a single nucleotide polymorphism located upstream of the EWS gene. These markers (1) are activated during the pubertal growth spurt; (2) may interact with the t(l 1 ;22) fusion protein of the Ewing's sarcoma transbcation; (3) alternatively are involved in the EWS/ets gene downstream pathways; and/or (4) are involved in bone growth and development. HLB.2. Secondary Aims 1. If an associatfon is identified between a polymorphic marker and Ewing’s sarcoma, we will include additional flanking polymorphic markers or direct 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sequencing of the region in a search for additional markers closer to the causal allele or that allele itself. 2. If an apparent disease-causing polymorphism^) is identified, we will compare the prevalence of this/these polymorphism^) among Caucasian and non- Caucasian cases and controls in order to determine if the polymorphism^) explain the ethnic specific incidence of Ewing’s sarcoma. 3. In the event that our investigation does not show an association between any of the candidate genes chosen, we will analyze single nucleotide polymorphisms located throughout the human genome in a later study. To accomplish our primary aims, we will conduct a case-parental/sibling control study enrolling 125 cases o f Ewing’s sarcoma and their respective parental/sibling controls (total N approximately 450). The selection o f a control group in association studies of this type is especially challenging because of the confounding effects of ethnicity. If we were to use a conventional case-control study with controls selected at random from the general population and found an association between Ewing’s sarcoma and a particular genetic polymorphism, we would have two possible explanations. First, the genetic polymorphism may be truly associated with Ewing’s sarcoma. Second, the ethnic background may be different between our case group and control group and this difference may be responsible for the difference in polymorphic allele frequencies. Therefore, in order to eliminate the second possibility o f different 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ethnic backgrounds of our case and control group, we must match on this factor. This is best accomplished by comparing the case to a family control (either a living sibling or set of hypothetical sibling controls, constructed from the parents' genotypes). Both living and hypothetical siblings come from the same parents and thus have the same ethnic background. The design is relatively simple; genotypes are determined for the case, living siblings, and parents and the parents’ genotype is used to construct hypothetical siblings made up from the alleles the parents did not pass to their affected child (see figure 2 below). Figure 2; Genotypes used for case family control study Parental Controls A/a A/a A/A A/a a/a A/a A/a a/a Ewing’s Sarcoma Sibling Constructed from Case Controls parents’ genotype ( Pseudosibling Controls) Under the null hypothesis, we would expect that any living sibling or any hypothetical sibling genotype will equally likely to be transmitted as any case genotype. Any deviation would indicate an association between the disease and the genotype. 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IV. K xnerim ental Design and M ethods IV. A. Study Design IV.A.1. Case-Parental/Sibling Control The initial study design will enroll 125 cases and respective parental/sibling controls. Cases will be selected from the tumor registry of Childrens Hospital Los Angeles (CHLA), the Cancer Surveillance Program (CSP) of Los Angeles County, and the California Cancer Registry (CCR). Cases will be needed from these tumor registries due to the rarity of Ewing’s sarcoma and possible nonparticipation of contacted controls. The population based CSP program, which is organized by the University of Southern California, and California Cancer Registry (CCR), began collecting information on all cases of cancer in Los Angeles and the state of California in 1970 and 1985 respectively.217 Information on cases is obtained from hospitals and pathology laboratories within each county.217 Data is collected concerning surname, first name, ethnicity, birthplace, birth date, gender, religion, social security number, date of diagnosis, pathologic diagnosis, tumor histology, staging, and address at time o f diagnosis.^!? The tumor registry o f CHLA collects similar information on each case and must report their cases to the CSP which then reports to the California Cancer Registry. Fresh frozen or archived tumor sections will be obtained for each case in which the family controls agree to participate. Blood or mouthwash samples will be obtained from living cases and controls. DNA from these samples will be extracted for later PCR based genotyping. Genotyping will include the following genes and methods (see Tablet below): 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 1: Genetic markers and method of genotyping Gene Polymorphism Method Section Insulin Growth Factor-I (IGF-I) C A repeat y3 3 P ATP end labeling of PCR primer see VLB. 1. Insulin Growth Factor-I Receptor (IGF-IR) AGG repeat y3 3 P ATP end labeling of PCR primer see VI.B.2. Insulin Growth Factor-I Receptor (IGF-IR) M nll restriction site polymorphism PCR followed by restriction enzyme digest see VI.B.3. Insulin Receptor Substrate 1 (IRS I) CA repeat y3 3 P ATP end labeling of PCR primer see VI.B.4. Nerve Growth Factor (NGF) BgUI restriction site polymorphism PCR followed by restriction enzyme digest see VI.B.5. EWS M nll restriction site polymorphism PCR followed by restriction enzyme digest see VI.B.6. The use of parental/sibling controls avoids the problem o f ethnic confounding, but has the limitation that overmatching on genotype may occur. 174,176,177 for this reason, a slightly larger sample size is needed to obtain the same relative efficacy as a traditional case-control study in a case-parental controls d e s i g n . *76,178 Although the software and study analysis will permit us to use the entire family as a control group, the availability of both hypothetical and living sibling controls will allow us to do a separate analysis for comparison of cases with only hypothetical siblings and cases compared to living siblings. We will compare the results of the two analyses in order to determine if there is a difference in the final results depending upon the source o f the control selected, (see discussion in X. Advantages and Limitations section). 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V. Data Collection V.A. Basic Demographic and Clinical D ata This information will be collected on all 125 cases (~147-167 cases will be contacted with expected 75%-85% enrollment). It will be limited to items that can be collected from information available via hospital records, such as surname, first name, ethnicity, birthplace, birth date, gender, religion, social security number, date of diagnosis, pathologic diagnosis, tumor histology, staging, and address at time of diagnosis. This information is collected by, and available from, the participating cancer registries. V.B._____ Contact Procedures and Specimen Collection V .B .l. Family Controls and Cases (Prevalent & Incident) (Figure 3) Families and cases (living and deceased) will be contacted in the following manner: a.) Cases will be identified via the Childrens Hospital Los Angeles (CHLA) tumor registry, the Cancer Surveillance Program (CSP) of Los Angeles County or the California Cancer Registry (CCR), and other registries that agree to participate in the future. b.) Upon identification of a prevalent case and after waiting a period of at least eight weeks for incident cases, a letter will be sent to the diagnosing physician (if name is available), notifying the physician of our intent to contact their patient and inquiring about the life status of the prevalent case. If the physician does not have a major objection to the contact of the patient for this study, we will proceed to the next step (step d). 144 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. c.) In the event that we do not receive any reply from the physician two weeks after the initial letter, we will contact the case and his/her family directly without prior approval from the treating physician. d.) We will write to the family of the case to further explain the study, how we obtained the cases name and diagnosis, and a questionnaire to collect information about the family members interested in participating in the study. If any case/family refuses to participate, no further attempt will be made to contact that case/family. Family members o f deceased cases will be given a separate letter that has been adapted in a way to address the sensitivity o f this situation. For the family of deceased cases, we will request permission to obtain biopsy or surgical tissue obtained from the case. We will use genetic material from these samples for the purposes of this study. e.) Included in the initial letter will be a self addressed stamped envelop to allow the case and family the chance to respond as to their interest in participating in the study and also provide us with the family information questionnaire. Two weeks later, the letter will be followed up with a telephone call to provide an opportunity for parents and cases to ask questions and increase the chance o f enrollment. f.) If no telephone contact is made, we will re-verify our data and a second letter will be sent to the subject and a second telephone conversation will be attempted two weeks later. g.) If the subjects express written or verbal lack o f interest in participation at any time, attempts at contact will stop there. 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. h.) However, if no contact is made, we will attempt to locate the subjects through census, credit bureau or post office data and a third letter and phone contact two weeks later will be attempted for each possible address. L) If the family and case (if under the age o f eighteen) or the case (if eighteen years o f age or older) agree to participate in the study, we will request consent to obtain a blood or mouthwash sample from the case, both parents and living siblings. If the case is willing to participate, we will obtain consent directly from the case if eighteen years of age or older together with the parent’s consent to participate in the study. If the case or any sibling is under the age of eighteen, we will require that the parent or legal guardian also sign the consent form. j.) Alternatively, a study web site has been published on the world wide web and patients/family members can visit the web site and if interested, can submit similar family information forms for participation. All other contact procedures will be followed as for participants identified through other sources. Emphasis will be placed on verifying the diagnosis by obtaining a pathology report for each case identified in this manner. V .B J. Specimen collection We will obtain blood or mouthwash samples from the study participants. Mouthwash samples will be obtained via a mailed package with instructions for obtaining a mouthwash sample and returning the sample to CHLA. Blood samples will be obtained as an 'add-on' blood sample to the routine check up blood work done during a follow-up visit to the oncology clinic at Childrens Hospital Los Angeles. 1 4 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Alternatively, blood sample mailing kits can be mailed to the case/family member with instructions to have the blood drawn at their current place of follow-up care (treating oncologist (case) or primary care physician (family member)). Blood samples obtained from outside sites will be promptly mailed or delivered to Childrens Hospital Los Angeles. Postage will be provided to all study participants. Blood samples will consist o f one 10 cc EDTA tube to be drawn for each participant via regular phlebotomy protocol. Instructions for mouthwash samples will consist of brushing ones teeth and waiting for I hour afterwards while forgoing food. Samples contain lOcc of Listerine® mouthwash solution or a generic brand thereof in a 50mL centrifuge tube, which is swished in the participants mouth for 30 seconds to 1 minute and subsequently returned to the provided 50mL t u b e . * 97 Three samples are obtained on three separate days and all sam ples are returned via the mail to CHLA. For each type of samples, subsequent isolation of genetic material to be carried out in Dr. Van Tomout's laboratory. We will obtain tumor tissue when available for each case. Banked or archived tumor tissue will be obtained from the hospital that obtained the sample via a previous biopsy or surgery. 147 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3: Enrollment of Participants Wait at least 8 weeks alter ^ diagnosis (incident cases only) Proceed to contact Permission Denied Permission Granted Stop any further rontarf No response from family (2 weeks later) Not Interested No phone response from subject Patient/ Family \ a g r e e to participate Patient/Family agree to participate Verbal or Written Informed Consent Denied Patient/ Family agree to participate Proper written consent obtained No response from parent Study subject agrees to participate No contact made Verbal or Written Informed Consent Denied No response in two weeks Cases identified through Web site Cases identified through CHLA, CSP, CCR, or other registry Attempt to contact by phone to verify receipt o f study letter Letter sent to explain purpose o f study and request consent to participate Verify diagnosis and treatment. Analyze samples fix* genetic polymorphisms Obtain family information, subsequently consent parents, siblings and living cases. Verify contact information and attempt to re-contact (2nd letter and phone call 2 weeks later) Obtain mouthwash and blood samples from patient/family, obtain pathology report and archived samples from cases. Contact diagnosing physician and inform o f intent to contacted family and case (if no diagnosing physician, contact patient/family directlv) Use alternate sources o f contact information (CSP data, US Census data, Postal records, credit bureau) (3rd letter and phone call 2 weeks later) 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V T - laboratory Procedures V1.A. Biologic Samples For all living study participants, a blood or mouthwash sample will be obtained. For the deceased cases, preferably a normal tissue specimen will be obtained. Alternatively, in the event that a normal tissue sample is not available, a tumor/mixed tissue sample will be obtained. VLB. Genetic Analysis All genetic analyses are optimized for use with DNA obtained from blood, mouthwash, or archived tissue. Hence, all amplified segments are less than 300 base pairs. Publshed methods have been used and optimized when necessary. VI.B.1. Specific Methods for Genotyping o f the Insulin-like Growth Factor-I Gene CA Repeat Polymorphism A CA repeat in the promoter region of the insulin-like growth factor gene has been shown to correlate with IGF-I serum levels. The repeat is approximately 1 kb upstream from the transcription start s i t e . 142*218 -j^e primers, forward 5’-GCT AGC CAG CTG GTG TTA TT-3’ and reverse 5’-ACC ACT CTG GGA GAA GGG TA-3\ will be used to amplify 188-198 bp products after initially labeling the reverse primer (1.5 pmol) with in a reaction that included T4 kinase (0.5 U) and y3 3 ? ATP (1.5|iCi/pL). The PCR reaction is carried out using 1 U Taq polymerase, 1.5 pmol each primer, 1.5mM MgCh, IX reaction buffer, and 0.2mM each dNTP. Initial denaturation is carried out at 95°C for 4 min followed by 35 cycles of denaturation for 30 seconds at 94°C, annealing for 30 seconds at 60°C, and extension 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. for 40 seconds at 72°C. One volume of PCR reaction is mixed with 2 volumes formamide loading dye and heated to 7S°C for 2-3 minutes immediately before loading on to a 6% acrylamide geL The gel is run at 1300 V for 2-3 hours. The gel is dried and exposed for 3-6 hours on radiography film. In order to score genotypes, samples are run with a known standard. VT.B.2. Specific Methods for Genotyping of the Insulin-like Growth Factor-I Receptor Gene AGG Repeat Polymorphism In order to determine if there is an association between the IGF-IR locus and Ewing’s sarcoma, we plan to study a microsatellite repeat found within the promoter region of this l o c u s . 2 19,220 Xhe primers, 5’-GCT GAG GGA GGA GGC GGC-3’ and 5’-GGC GAG GGG CAG AAA CGC-3’, amplify a trinucleotide (AGG)„ repeat in the IGF-IR locus with products that rang in size from 90 to 96 bp. The reverse primer (1.5pmol) is initially labeled with y3 3 ? in a reaction that included T4 kinase (0.5 U) and y^P ATP (1.5p.Ci/nL) prior to PCR amplification. The PCR reaction is carried out using 1 U Taq polymerase, 1.5 pmol each primer, 1.5mM MgCfe, IX reaction buffer, 0.2mM each dNTP and 10% DMSO. Initial denaturation is carried out at 95°C for 4 min followed by 35 cycles of denaturation for 30 seconds at 94°C, annealing for 30 seconds at 66°C, and extension for 40 seconds at 72°C. One volume of PCR reaction is mixed with 2 volumes formamide loading dye and heated to 75°C for 2-3 minutes immediately before loading on to a 6% acrylamide gel The gel is run at 1300 V for 2-3 hours. The gel is dried and exposed for 3-6 hours on 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. radiography film. In order to score genotypes, samples are run with a known standard. V I.B3. Specific Methods for Genotyping o f the Insulin-like growth factor I receptor exon 16 M nll polymorphism In order to determine if there is an association between the IGF-IR locus and Ewing's sarcoma, we also plan to study a G to A transition at nucleotide position 3174 of exon 16.221 The silent polymorphism codes for glutamate in the ATP binding domain of the gene. The primers, forward 5’-TGC ITT AAT TAC GGT TTC TTC -3 ’ and reverse 5’-GCT TTT CAG GAA CTT TCT CTT-3’ amplify a 280 bp section surrounding exon 16. The PCR reaction is carried out using 0.3 U Taq polymerase, 8.0 pmol each primer, l.5mM MgCk, IX reaction buffer, and 0.5mM each dNTP. Initial denaturation is carried out at 95°C for 4 min followed by 35 cycles of denaturation for 30 seconds at 94°C, annealing for 30 seconds at 62°C, and extension for 40 seconds at 72°C. PCR products are digested at 37°C overnight with 0.5 U o f M nll and visualized by ethidium bromide staining after electrophoresis on a 3.0% agarose gel (1:2 NuSieve agarose/regular agarose). The polymorphism is detected by the presence of an M nll restriction site when the G allele is present. Two internal M nll restriction sites also exist and the scoring o f alleles has previously been outlined by Abu-Amero et al. with fragment sizes o f40, 132,20, and 88 when the G allele is present.221 Alternatively, fragment sizes o f40, 132, and 108 are obtained when the A allele is present. The M nll recognition site is CCTCNNNNNNN". 151 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VI.B.4. Specific Methods for Genotyping o f the Insulin Receptor Substrate I Gene CA Repeat Polymorphism In order to determine if there is an association between the IRS1 locus and Ewing’s sarcoma, we plan to study microsatellite repeats found within a 12.S kb region of the gene. 139,140.£22 The primers, forward S’-GTT CAT TAA TAT TGT TCA ACT GTG G-3’ and reverse 5’-AAT TAA TTT GAA ACC CGT TTG ATG G- 3’, amplify a (CA)„ repeat in the IRS1 locus that ranges in size from 134 to 142 bp. The reverse primer (1.5pmol) is initially labeled with y3 3 P in a reaction that included T4 kinase (0.5 U) and y^P ATP (1.5pCi/pL) prior to PCR amplification. The PCR reaction is carried out using 1 U Taq polymerase, 1.5 pmol each primer, 1.5mM MgCfe, IX reaction buffer, and 0.2mM each dNTP. Initial denaturation is carried out at 95°C for 4 min followed by 35 cycles o f denaturation for 30 seconds at 94°C, annealing for 30 seconds at 58°C, and extension for 40 seconds at 72°C. One volume of PCR reaction is mixed with 2 volumes formamide loading dye and heated to 75°C for 2-3 minutes immediately before loading on to a 6% acrylamide gel. The gel is run at 1300 V for 2-3 hours. The gel is dried and exposed for 3-6 hours on radiography film. In order to score genotypes, samples are run with a known standard. VI.B.5. Specific Methods for Genotyping o f the Nerve Growth Factor B g lll Restriction Enzyme Polymorphism In order to determine if there is an association between the NGF locus and Ewing’s sarcoma, we plan to study a restriction enzyme polymorphism found in the 152 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. first intron of the g e n e . 2 2 3 The primers, forward 5’-ACA GCG TCA GTC AGA CC-3’ and reverse 5’-GTA ATT GCA TCA CAG GTA C -3\ amplify a product that is 193 bp in size. The PCR reaction is carried out using 0.3 U Taq polymerase, 8.0 pmol each primer, 1.5mM MgCfe, IX reaction buffer, and 0.5mM each dNTP. Initial denaturation is carried out at 95°C for 4 min followed by 35 cycles of denaturation for 30 seconds at 94°C, annealing for 30 seconds at 54°C, and extension for 40 seconds at 72°C. PCR products are digested at 37°C overnight with 0.5 U B glll and visualized by ethidium bromide staining after electrophoresis on a 3.0% agarose gel (1:2 NuSieve agarose/regular agarose). Cutting of the 193 bp product by B glll results in fragments of 126 and 67 bp. The B glll recognition site is A'GATCT. VI.B.6. Specific Methods for Genotyping o f the EWS gene M nll Restriction Enzyme Polymorphism In order to determine if there is an association between the EWS locus and Ewing’s sarcoma, we plan to study a restriction enzyme polymorphism (C/T transition) found 4 3 4 2 bp upstream of the EWS g e n e . 2 2 4 , 2 2 5 The primers, forward 5’- G C A A A G G A G C T G C A G G A A G -3’ and reverse 5’- T G G C A G T G C T G C T G G C A G -3’, amplify a product that is 2 0 3 bp in size. The PCR reaction is carried out using 0 .3 U Taq polymerase, 8 . 0 pmol each primer, 1.5mM M g C h , IX reaction buffer, and 0.5mM each dNTP. Initial denaturation is carried out at 9 5 ° C for 4 min followed by 35 cycles of denaturation for 3 0 seconds at 9 4 ° C , annealing for 3 0 seconds at 6 1 ° C , and extension for 4 0 seconds at 7 2 ° C . PCR products are digested 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. at 37°C overnight with 0.5 U of M nll and visualized by ethidium bromide staining after electrophoresis on a 3.0% agarose gel (1:2 NuSieve agarose/regular agarose). Cutting of the 203 bp product when the C is present by M nll results in fragments of 114 and 89 bp. The M nll recognition site is CCTCNNNNNNN". VII. Quality Controls PCR contamination precautions include separate areas for pre-PCR preparation, PCR-amplification reactions, and visualization of PCR products. In addition, each set of assays will include non-nucleic acid PCR reactions as negative controls. Aerosol resistant pipette tips will be used in all PCR amplifications to avoid contamination between reactions. Furthermore, ultraviolet lighting is used after preparation of each PCR assay to reduce the chance of contamination. PCR will be carried out in separate area from post PCR gel visualization. Standard methods for preparation of high molecular weight DNA as described by Almoguera et al. will be used as a guideline to prepare the paraffin embedded sections for PCR a m p l i f i c a t i o n . 2 2 6 Similar methods will be used to isolate DNA from fresh frozen samples and mouth wash s a m p l e s . 1 9 7 , 2 2 7 a s indicated, we will optimize each method for application in using archival tissue. Extended fixation and years of storage diminish the ability to amplify larger DNA fragments due to a decay over time in nucleic acid s e q u e n e e . 2 2 8 However, our preliminary data from studies using archival tissue and the current literature suggest that DNA strands of less than 350 bp can routinely be amplified in archival t i s s u e . 2 2 9 154 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V m . Prelim inary D ata We have obtained Institutional Review Board approval from Children's Hospital Los Angeles (CCI #99-028) and have enrolled 97 patients along with available family members (N=378). Currently, we are continuing to enroll incident patients from California Cancer Registry and its affiliated CSP of Los Angeles County in addition to any incidence cases that present to CHLA. We also continue to enroll patients that are identified via our study web site (http://www.jmavtlab- chla-usc.com/) and verifying the diagnosis of these cases via pathology reports. Current work focuses on obtaining samples from all interested participants. For all patients, we are continuing to contact and obtain tissue samples/pathology reports from individual hospitals. To date, we have analyzed the CA repeat polymorphism o f IGF-I, the AGG repeat and M nll polymorphism o f IGF-I receptor, the CA repeat of IRS 1, the B glll polymorphisms of NGF, and the M nll polymorphism of EWS in a subset (cases=97) o f our study population with the following results: (See Chapter 2 for discussion of preliminary results). Although promising results are found for an association between the EWS gene and the I RSI gene, no conclusions can be drawn at this time until a larger sample size is enrolled. 155 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IX. Statistical Considerations IX.A. Case Parental/Sibling Control Design Much work has been done on the theory of familial control study design and different methods are evolving for analyzing the data from such a study. Attempt will be made to analyze the data obtained horn this study in the most appropriate and robust way possible. Emphasis will be placed on sources of bias and use o f all available data. Two terms are used often in the literature to describe the relationship between a marker and disease allele: linkage and linkage disequilibrium. Linkage will be used to describe the association of both the marker and disease allele within nuclear families. Linkage disequilibrium will be used to describe the association of the marker and disease causing allele in a population i.e. across many nuclear families. Association will be used to describe the presence o f both linkage (within families) and linkage disequilibrium (across families). Previous authors have implemented various test statistics for analyzing case- parental data, case-single parent data, and case sibling data. In addition, methods exist for use of bi-allelic markers and/or multi-allelic markers. Recently, Laird et al. described a unified statistical method using a single test statistic that allowed analysis o f many different family structures. 196 This approach is robust in that it allows for different types o f controls (both siblings and parents simultaneously) and allows for testing an additive, dominant, or recessive model. The authors approach the analysis o f family-based tests by separating the test statistic from the distribution 1 5 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of this test statistic. The authors define the following test statistic for N nuclear families with ni offspring per family: S=£ijTijXij where Xij = function of the ij* offspring at the tested locus (value depends upon genetic model and can be a vector to represent multiple alleles together) T jj= 1 if the offspring is affected and = 0 if unaffected. The authors comment that depending upon how Xjj and Ty are coded, one arrives at the previously described tests such as the transmission/disequilibrium test (TDT) and conditional logistic regression (i.e., Ty = 1 for affected (case) and Xy = number o f type 1 alleles or 0,1, or 2 for an additive model or 0,1 for a dominant model). In order to construct the distribution of S, the authors condition only on all observed traits and parental genotypes. When parental genotypes are known, Mendel’s laws are used to determine the distribution of the offspring genotypes. If parental genotypes are missing, the appropriate offspring distribution is derived from the observed parental genotype and any available offspring (see tables in Rabinowitz and Laird, 2 0 0 0 ). 195 in order to implement this unified approach, the authors developed a software package termed FBAT (Family Based Association Test) which is available at http://www.biostat.harvard.edu/~fbat/default.html (see section below for more discussion o f this software package). Rabinowitz and Laird present details of this unified approach. 195 Using the same test statistic framework noted above, the authors are able to test different genetic models, different sampling schemes, different patterns of missing alleles, 1 5 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. different null hypotheses, addition of covariates, and arbitrary phenotypes. The approach is based on functional and classical methods of conditioning on the minimal sufficient statistics for the null hypothesis. For example, if both parents’ genotypes are available then the minimal sufficient statistic is the parental alleles and the measured trait and the conditional distribution is based on Mendelian transmission probabilities. When parental alleles are missing, the authors develop an algorithm too derive the minimal sufficient statistic and the conditional distribution of the observed data (see tables in Rabinowitz and Laird, 2 0 0 0 ) J 95 Most importantly, all family members (parents and siblings) are used as controls instead of previous statistics, which separated the analysis and discarded information when both parents and siblings were available. The test framework is applicable and valid for two different approaches, one being a genomic scan for association, and the other being the use of association to fine map a candidate gene in an area that has already shown linkage. Adjustment for multiple tests must be considered. When conducting multiple tests, especially with multi-allelic markers, a conservative approach would be to use a Bonferoni adjustment o f the p-value. 1 since this is an initial evaluation and each locus has some biologically plausible association with the disease, one can reasonably argue that adjustment is not desirable and thus, we will not use the Bonferoni adjustment. However, the significance o f any association should take this in to account. 158 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IX.A.1. The Family Based Association T est (FBAT) software The FBAT program allows one to use all the information from a family pedigree in order to test for a s s o c i a t i o n . 1^6 FBAT allows the user to read in a file of data containing information on pedigree structure, affection status, and genotype for study participants. SAS® will be used to format the data to the specified FBAT format. If appropriate, additional phenotype information can be read in as a separate file allowing for the inclusion of this data in the analysis. The bi-allelic mode tests for association between an individual allele and the disease. A multi-allelic mode tests for association between the gene (one allele or a combination of alleles at a particular loci) and the disease. The additive model is the default model, however dominant, and recessive models can be tested. Schaid has shown that the additive model is the most robust model when the underlying mode of inheritance is u n k n o w n . 178 Using the wrong model will lead to a decrease in power for the alternative hypothesis, but does not invalidate the test under the null hypothesis. If parental information is missing, the program uses the genotypes of all unaffected members of the pedigree to determine the distribution of the test statistic. One can also perform the SDT on the markers of interest, using only the siblings. We will attempt to code half siblings into the model. The FBAT program will be used to test for an association in the in data collected for this study. The data will be analyzed in several different approaches. First, all data from living and deceased cases, parents and siblings will be used. Second, to determine if the use of tumor tissue may invalidate any association found 1 5 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in the initial analysis, deceased cases will be removed from the model In order to determine if alleles may be associated with a growth or survival advantage, analysis of parental and sibling controls will be undertaken separately and results compared. In the event that different alleles confer susceptibility in the two major ethnic groups (Caucasian and Hispanic/Latino), analysis will be undertaken on each group separately and results compared. DCA.1.L Limitations of the FBAT program The FBAT program offers the advantage of being able to analyze all the information obtained from family based control studies. However, it only tests for an association and is unable to determine the genetic relative risk. Work is being undertaken at USC (Peter Krafr, personal communication) to allow for t h i s . 2 0 0 Currently interaction terms for gene-gene interactions are not part o f the FBAT software. Any gene-gene interaction can be modeled in a previously described conditional logistic regression (CLR) model proposed by Schaid. 178 The CLR model only allows for the separate analysis of case and parents, or case and siblings, but attempts will be made to add interaction terms to the basic model to determine any significant interaction. IX.A.2. Sample Size Estimates Sample size estimates are not available for a case-parent/sibling study. Instead, the following table (Table 2 below) which is adapted from Schaid provides an estimate o f the sample size needed for a case-parental control design.*?® We will base our sample size estimates on this table. Sample sizes are given in order to have 160 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80% power at the 5% significance level for detecting an association between the disease and a locus with one high-risk allele responsible for the stated relative and attributable risks (see Table 2 below).* From the information provided in the table, a sample size of 60 to 195 cases for a dominant model (56 to 180 for recessive model) would give us adequate power to detect a relative risk between 2 and 4 provided the allele explains 30 to 40% or more of the disease risk for a marker with 4 alleles. Table 2: Sample size estimates (adapted from Schaid DJ, 1996): ____________________ Genetic Model_______________________ Dominant Recessive Attributable Risk Attributable Risk if of Relative Alleles Risk 10% 20% 30% 40% 10% 20% 30% 40% 4 2 404* 260 195 150 542 271 180 123 4 228 125 90 60 304 133 86 56 10 221 110 80 57 279 114 72 46 8 2 533 343 258 198 714 356 237 161 4 300 164 119 87 400 176 113 74 10 278 145 104 75 367 150 94 60 * The required number of case-parent triplets to have 80% power at the 5% significance level for detecting a candidate locus with the indicated number of alleles and relative and attributable risks. Given a high likelihood that a genetic element is responsible for Ewing’s sarcoma, we expect that an associated allele will explain a high percentage of Ewing’s sarcoma cases, perhaps higher than 40%. The attributableness o f a given locus will depend on whether or not other loci are also involved in the disease process. Thus, if one locus is involved, we expect to find that it is responsible for a majority if not all cases of Ewing’s sarcoma. However, if there is a number of loci involved, each may be responsible for a certain percentage o f Ewing’s sarcoma 161 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cases. The relative risk will depend on several factors. First, Ewing’s sarcoma is a rare tumor with an incidence around 2.8 per million under the age of fifteen in the U.S A If there is an association between the disease and a common polymorphism, we expect to find a low relative risk, indicating that many individuals may carry the at risk allele, yet very few ever develop the disease. Alternately, if an association is found between a rare polymorphism and the disease, we would expect to find a higher relative risk, indicating that carriers of the ‘at risk’ allele would be more likely to develop the disease. Our attributable risk is likely to be high given the high possibility of a genetic etiology of this disease and our relative risk will likely be low due to the rarity of this tumor. For a sample size of 125 cases, we will have adequate power to detea an association between a given allele and Ewing’s sarcoma for relative risks that are in the range of 2 to 4 and attributable risks in the range o f 40% or greater. These estimates seem reasonable and are worth investigating. IX.AJ. Survival Advantage versus Disease Association There has been some debate over the interpretation of a positive result from a case-parental control design. Any association between a particular allele and the disease can have two altremative explanations. The first is that the allele is associated with the disease in our case series. The second explanation is that the allele may confer a growth or survival advantage in our living cases compared to the non-living hypothetical sibling controls derived from the parents’ genotype. We would therefore find the allele more often in our living case series, as it is associated with an increased chance for survival. This second explanation is less likely, but 162 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. since we are investigating factors involved in growth and differentiation of tissues, we are concerned that this explanation could decrease the validity o f any positive findings. Thus, one way to investigate the second explanation would be to perform a separate analysis using unaffected siblings of the case. If our positive findings still hold true for the comparison between the living case and the living siblings, then we can argue that the association is between the allele and the disease. FBAT allows one to specify a sibling disequilibrium (SDT) which test for associations between marker and disease in unaffected siblings o f the case. Compared to the parental controls, the use of only siblings as controls is less efficient and an increase in sample size may be needed to have the same power to detect an association. We do not know of any study that has shown an association between an allele and a survival advantage in humans, however this hypothetical situation may exist in light of the polymorphisms under investigation. IX.A.4. Use of Siblings as Controls Siblings that have reached the age of diagnosis of the case and are disease free at the time o f sampling can be considered as controls for the study. A problem encountered in using sibling controls arises when the sibling is younger than the age at diagnosis o f the case. It is possible that this sibling could become a case and be incorrectly included in the study as a control However, Ewing's sarcoma is extremely rare in siblings (only four reported cases in literature) despite the implications o f involvement of a genetic f a c t o r ( s ) . 1 3 - 1 6 Thus, one could argue that 163 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. younger siblings be included in the analysis because the likelihood of future diagnosis as a case prior to the age their sibling was diagnosed is extremely small. IX.A.5. Future Studies In the event that an association is found between a marker and the disease, we will include additional flanking polymorphic markers or direct sequencing in a search for markers closer to the causal allele or that allele itself. X. Advantages and Limitations of the Study Prior to undertaking this study, we must first consider all possible alternatives in order to determine if the one postulated is plausible. The first alternative would be that the true causal factor for the difference in incidence by each ethnicity is not genetic: the true underlying causal factor involved in Ewing's sarcoma may be environmental in nature. Environmental factors have been investigated in several studies with a lack o f evidence for a role of any such factor in the etiology of Ewing’s sarcoma.^^-2 1 (reviewed in 230) jjjg ethnic specific incidence pattern is consistently found in the literature, the magnitude of the difference is large (11:1 in the U.S.), and it does not show any migration pattern that would e expected if an environmental factor were i n v o l v e d . 2 ^ 3 , 5 - 1 0 The translocation t(ll;22)(q24;ql2) (EWS/FLI1) may be the genetic element responsible for the disproportionate incidence o f Ewing’s sarcoma. The Caucasian and Hispanic populations may be genetically more at risk for this translocation as a somatic alteration or allow the product of the translocation to transform normal cells. 164 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. We will address this point by including a study o f the EWS gene which is present in every translocation event. Alternatively, the FLI1 or ERG portion of the fusion gene may be responsible for the ethnic specificity of Ewing's sarcoma. However, this would imply that two separate genes underwent separate mutational events only in Caucasians, which produced the same ethnic specific results. This is an unlikely event, but one that must also be investigated further in future studies. It may be that all ethnicities form the translocation at same rate, but the specific genetic element allowing cellular transformation may be present at a much higher frequency in susceptible ethnic groups. The insulin-like growth factor pathway displays evidence for an interaction with the fusion protein. By including polymorphic genes from this pathway, we are attempting to address this hypothesis. Due to the rarity of Ewing's sarcoma, it is possible that the disease may result from a rare genetic defect present only in the Caucasian population and the relatively common polymorphisms under investigation, if associated with the disease, would account for a higher prevalence o f the disease. Again, the marker alleles under study are expected to be in linkage disequilibrium with the disease allele. A common marker allele may be in linkage disequilibrium with a rare disease allele in Ewing’s sarcoma cases. Our study design will allow us to determine if the cases are more likely to carry this more common marker allele than would be expected by chance. The case-parental/sibling control study design will allow us to test for an association between the selected marker alleles and Ewing’s sarcoma while 165 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. eliminating the confounding effects of ethnicity. The relative efficiency o f this design approaches the traditional case-population control design. 178 Because of the nature of the relationship between case and family controls, we expect enrollment to be higher than in population based control studies. The markers under investigation are in non-functional areas of the candidate gene and any association between the disease and a specific marker allele indicates that the marker may be in linkage disequilibrium with the true causal allele. Since the true disease causing allele is not measured directly, use of marker alleles will result in non-differential misclassification with a bias in the estimate o f the risk estimate towards the null. 173 However, given that we do not know the true disease causing allele and have only non-fonctional marker alleles at our disposal, we will follow up any significant results with a search for mutations/polymorphisms in the candidate region. The likelihood of finding a marker that is in linkage disequilibrium is limited by the region of linkage disequilibrium surrounding the marker allele. Thus, it is possible that we may be using a marker that is in the close vicinity o f the true disease causing allele, but that the two alleles are not in linkage disequilibrium and we would therefore be unable to detect any association with this region. This is a valid limitation o f any candidate gene search, however, recent work by Retch et al. have shown that the region of linkage disequilibrium could extend up to 60 kb in Northern European individuals. ^ This is a relatively large region and we can comfortably conclude that on average, a region this large can be excluded if we do not find an association with a given marker allele. Furthermore, a region this 166 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. large should be adequate to search for linkage disequilibrium. For example, the EWS gene is approximately 40 kb in size, therefore a marker in the vicinity of this gene should theoretically uncover an association between a true disease causing allele. For the IGF-IR gene which spans more than 100 kb, we have chosen to use an additional marker located in the 3’ end of the gene (exon 16) in order to adequately search for association. Although we run the risk of finding a spurious association between the disease and a marker allele with increasing number of candidate genes under investigation, we will set the level of significance at 0.05. This level of significance is chosen because at this time there are a limited number of candidate loci under investigation with some biologic reason for choosing each one. If a marker allele is found to be associated with the disease, we must also consider the notion that the reason the marker allele is found more often in the cases is that it confers a “survival advantage”. If this were the case, we would expect to find the marker allele being passed on to the living cases, rather than the hypothetical sibling controls. This possibility can be assessed by comparing the distribution of marker alleles in the cases with the distribution of alleles obtained from any living siblings enrolled in the study. We plan to enroll families even when the case is deceased. This is controversial because if we are hypothesizing that these loci are involved in pathogenesis, then they may be altered during oncogenesis o f Ewing’s sarcoma. If we do find a difference between normal and tumor tissue, this finding would still be 167 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. important new information. However, because we are intending to study the etiology, not the pathology of this disease, we will endeavor, whenever possible, to use normal tissue from the material provided to us. There is no prior evidence to suggest that these loci would be altered in tumor tissue. In order to verify that the genetic loci under investigation are not altered as a result o f tumorogenesis, we will compare the genotype o f incident cases taken from tumor tissue to the genotype from peripheral blood samples. This will allow us to determine the frequency of alterations in genetic tissue during tumor development i.e. loss of heterozygosity. Once genotype data is collected on all study subjects, we will compare the parental genotypes with the case’s genotype to ensure agreement with Mendelian inheritance. Any inconsistencies will be repeated. Incompatible data could reflect changes in the tumor tissue, however it is also likely to indicate infidelity. In some cases, we have found it possible to locate normal tissue from prior surgeries (appendectomies, tonsillectomies) and in one case, DNA has been provided from the baby teeth saved by the family. While the ethnic frequency of the polymorphism is important and we would expect it to be if the marker loci under investigation were in fact the disease causing allele, the allele frequency o f the marker does not need to follow the ethnic frequency of Ewing’s sarcoma. There may in fact be no relationship between marker allele frequency and ethnic frequency of the disease under investigation. As stated previously, we expect that the marker alleles are in linkage disequilibrium with the true causal allele. The amount of linkage disequilibrium can vary by ethnicity.206 168 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Moreover, since we are looking at marker loci, we believe that the frequency of the marker in relation to population specificity will largely depend upon when the disease causing allele became present in the human population (see Figs. 4 & 5 below). For figure 4, the marker and disease causing allele become linked before the populations separate and there is preferential inheritance of the disease causing allele along with maker allele 1 in population 1. In this case, we would expect the marker frequency to be ethnic specific. In figure 5, the disease causing mutation occurs within population 1 after separating. In this case, the marker allele frequency would not be ethnic specific. One would expect to find a smaller proportion of marker 1 alleles since individuals with Ewing's sarcoma likely died before being able to pass the allele on. However, since Ewing's is such a rare disease, this decrease would be minimal. Less than 100% disequilibrium, population admixture, multiple disease causing alleles, etc. only complicate this picture and lead to the argument that the marker frequency may or may not be important. Under the null hypothesis, we would expect that any hypothetical sibling genotype will be equally likely to be transmitted as any case genotype. Any deviation would indicate an association between the disease and the genotype. 169 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 4: Example of marker frequency that is population specific Beginning Population M -------------D----- m------------ d preferential inheritance ^ / \ of M D pG pU jation i Population 2 Population 3 M D . m--------------------- m------------ d_, m------------ d , m------------ d , m d . m d_, m------------ d J m------------ d_, m------------ d m------------ d m------------ d 25% M 0% M 0% M Figure 5: Example of marker frequency that is not population specific Beginning Population M— d m— d * J r \ Population 1 Population 2 Population 3 rn_ d , m------------ d .----- m------------d_, m d m d m------------d l^ j— — — d ^ f4r;---------d M n---------- d D M — - d M----------- d 50% M X 50% M 50% M new mutation Legend M - Marker allele 1 m - marker allele 2 D - disease allele d - non-disease allele Given 100% linkage disequilibrium between marker and disease causing allele 170 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. If the genetic factors) responsible for the etiology of Ewing’s sarcoma differ in different ethnicities, Le. different genetic cause o f the disease in Caucasian versus Hispanic/Latino ethnicities, we v/ill be diminishing the ability to find an association between a marker and the disease by including all ethnicities in the analysis. We will therefore undertake a separate analysis for the two major ethnic groups (Caucasian and Hispanic/Latino) to determine if any association can be found within each ethnic group separately. Although the possibility o f different genetic factors responsible for the disease in each ethnic group is plausible, we believe the same genetic factor is more likely to be inherited via a common ancient relative from the European/Western Asia region with incidence dependent upon admixture. Thus, the use of all available participants irrespective of ethnicity is still valid. Separate analysis will decrease the effective sample size, necessitating additional accrual of participants of each respective ethnic group to approach the study sample size previously discussed. An alternate approach to studying previously determined polymorphisms would be to sequence regions o f the insulin growth factor I gene and receptor, insulin receptor substrate 1, nerve growth factor and EWS gene in cases with Ewing’s sarcoma prior to starting case-control studies. This would allow identification of new and specific polymorphisms. However, polymorphisms have been described in these genes and assays have been development for archival tissue. Therefore, this investigation will be conducted in a case-familial control manner using existing polymorphic markers with the possibility o f later sequencing selected regions of the 1 7 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. same genes for new polymorphisms should an association be found between the marker and the disease. It is possible that we are investigating the wrong candidate genes, and that other receptors, hormones, environmental factors, etc. may be involved. In the event that we do not find an association between the mentioned genes and Ewing’s sarcoma, we must consider this possibility. However, given the level o f current understanding of the disease, we believe that our choice of candidate loci is a good starting point and the candidate loci under investigation are the most likely to be involved given our current knowledge of the disease. With the current study design, further modification and inclusion o f other genes is possible, all that is needed is the development of a PCR based approach to identify the polymorphic locus. Furthermore, new technology is providing the use of genome wide scanning to search for candidate polymorphic markers. This technology can be adapted to use in the current study since it is based on PCR amplification of multiple short segments ofDNA. XI. Significance o f the Study Although the ethnic specific incidence data indicates that genetic factors) play a role in the etiology o f Ewing’s sarcoma, to date, this phenomenon has not been fully explained. The identification o f the genetic factor(s) involved will be a significant step towards understanding the tumor biology of this disease. Essentially, the genetic factorfs) are critical to the pathogenesis of this disease that afflicts mostly children and young adults. Therefore, it is critical that the genetic factor(s) be 172 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. explained in order to fully understand why individuals develop the disease and perhaps provide a critical step in the transformation process for which targeted therapy could be used to increase survival. Although this is a rare disease, it primarily afflicts the young, therefore any information that might lead to an increase in survival will thus have a large effect on the number of years o f life lived. XTI. Human Subjects Discussion o f the six points pertaining to studies with human subjects: Human subjects will be involved in the work outlined by donating tissue samples taken at the time of surgical biopsy when they were diagnosed with Ewing’s sarcoma or during a follow up visit. Parental controls, siblings, and living cases will be involved by donating venous blood taken via cutaneous needle stick or buccal cells obtained via a mouthwash sample. We will obtain information on ethnic background, sex, age, and case/control status on each individual enrolled on to the study. This is the extent of the involvement of human subjects in this proposed investigation. The characteristics of the cases are anticipated to be as follows: 125 cases of Ewing’s sarcoma in children and adolescents ranging in age from 1 to 40. All cases have been clinically diagnosed with Ewing’s sarcoma and health status will vary depending on the stage of disease at the time of diagnosis and the result of treatment. A biopsy is routinely taken to confirm the diagnosis or follow up on disease progression o f this cancer and extra tissue from this biopsy will be used for the work done in this proposal Ewing’s sarcoma has a disproportionate incidence by ethnicity with 95% of the cases being Caucasian, 1.8% being African descent and 173 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2% being of other ethnic background. 1 Thus, by virtue of the nature of this disease, we anticipate that a majority o f the cases will be Caucasian and we will therefore enroll primarily Caucasian subjects. However, all cases of Ewing’s sarcoma/pPNET regardless o f ethnicity, will be enrolled in the study if they choose to do so. We anticipate an equal number of both sexes to be enrolled because Ewing’s sarcoma has an equal distribution across both sexes (slight male preponderance in some studies).216 The proposed work is intended to fund preliminary data in what we anticipate to be 125 cases. Children are selected as cases because Ewing’s sarcoma is extremely rare in adults and is primarily considered a childhood cancer. Controls will be selected based on their parental or sibling relationship to the case. We anticipate the enrollment o f250 controls (125 fathers and 125 mothers) ranging in age from 20 to 80 years o f age. We will also enroll any available siblings. We anticipate the enrollment of 75 to 100 siblings. We anticipate that the health status of the controls to be similar to the general population. Enrollment of minority controls will depend on enrollment of minority cases as outlined above. After permission is granted, a blocd sample or mouthwash sample will voluntarily be obtained and used for analysis. Women will be selected as controls due to their maternal/sibling relationship to the case. Institutional Review Board approval has been obtained from Childrens Hospital Los Angeles (CCI#99- 028). Research material will be obtained from tissue biopsies (cases) taken at the time of diagnosis or follow up and blood draws or mouthwashings taken voluntarily 174 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (parental controls, siblings, and living cases). The biopsy material is routinely taken to confirm the diagnosis o f Ewing’s sarco ma/pPNET and extra tissue from this biopsy will be used for the work done in this proposal. This tissue will be taken from existing tissue stored at Childrens Hospital Los Angeles or the Cancer Surveillance Program of Los Angeles County and California Cancer Registry (community hospitals). Blood and mouthwash samples taken from the parents of the case, siblings, and living cases will be obtained specifically for research purposes. Existing records of the case will be used to obtain the age at diagnosis, sex, ethnicity, and contact information for the parents. The controls will be asked to submit the same information. All participants will provide informed consent along with release of the patient’s medical information and tissue samples. Cases will be selected from the tumor bank of Childrens Hospital of Los Angeles. All prospective controls will be selected based on their parental or sibling relationship to the case. Controls will be contacted in the following manner. Contact information will be obtained from Childrens Hospital Los Angeles and from the Cancer Surveillance Program of Los Angeles or the California Cancer Registry, which keeps records of every cancer case in Los Angeles County and in the state of California. The diagnosing physician of the case will be contacted and notified of our intent to contact the case and/or family o f the case. The patient/family will initially be contacted via a letter explaining the nature and importance of the study. After this, a follow-up phone conversation will be undertaken to further facilitate contact and enrollment of controls. Documentation o f consent will be collected in 175 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the form of a signed letter of consent form each parent. Due to the nature of the relationship with the case, we expect high enrollment rates for controls. Despite this, we anticipate 15-25% of attempted contacts to not participate. Therefore, we will contact approximately 147 - 166 cases to enroll a projected 125 case and parental control pairs. Parents, siblings and living cases will be able to donate blood samples at Childrens Hospital o f Los Angeles (if in the vicinity of Los Angeles) or submit mouthwash samples via U.S. Postal service. DNA isolation will be carried out in Dr. VanTomout’s laboratory. Living cases, siblings (and some parental controls) may undergo some psychological distress regarding a biood draw. This is a common concern for children and may be compounded in cases by the repeated blood work and other diagnostic tests undertaken during treatment for the disease. If this proves to be an overwhelming psychological distress among the incident cases, we can revert to using mouthwash samples. Cases and controls may undergo some psychological distress due to the recollection of their (child’s/sibling’s) diagnosis, treatment and outcome. We do not believe that this may be a serious risk however it may limit enrollment of some living cases, parents and siblings. Also, we anticipate that a chance to contribute some insight into the disease of their child will outweigh the psychological stress and may help some cases/parents/siblings to feel they helped future victims of this disease. We do not anticipate or foresee any social, legal or other risks to the cases or controls. 176 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The potential psychological risk to the living cases, parental and sibling controls will be minimized by careful, sensitive means of contacting the study subjects. We feel that by first contacting the diagnosing physician who is likely to know the family o f the case and make sound judgment as to the mourning status of the parents and risk o f psychological stress, we will minimize the risk to those individuals most likely to be affected. By first approaching the parents with a letter explaining the nature of the study and by follow-up via phone conversation, we will carefully approach the subject causing the least amount o f psychological stress. The potential psychological risk to the living cases will be minimized by careful, sensitive means of obtaining a blood or mouthwash sample. The risk of confidentiality will be minimized by giving all case and control samples a unique number code (with code kept by Dr. Van Tornout) and analysis will be carried out in an anonymous fashion. We anticipate that the knowledge gained will outweigh the risk of psychological stress for the following reasons. First, we recognize that controls and living cases may undergo some psychological distress due to the recollection o f their (child’s/sibling’s) diagnosis, treatment and outcome. However, we do not believe that this will be a serious risk that would limit enrollment to a great extent. We anticipate that the chance to contribute some insight into the disease o f their child/sibling or their own disease will outweigh the psychological stress and may help some living cases, parents, and siblings to feel they helped future victims o f this disease. Also, some parents, siblings and cases may feel that this is a way to give 177 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. back to the medical/scientific community for the help and care they gave to their family during the disease process. Furthermore, the use of parental controls is an effective means for answering the questions proposed in this work. 178 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References (Numerical) 1. Horowitz, M.E., Malawer, M.M., Woo, S.Y. & Hicks, M.J. Ewing's sarcoma family o f Tumors: Ewing's sarcoma of bone and soft tissue and the peripheral primitive neuroectodermal tumors, in Principles and Practice o f Pediatric Oncology (eds. Pizzo, P.A. & Poplack, D.G.) 831-863. (Lippincott-Raven Publishers., Philadelphia, PA, 1997). 2. Fraumeni, J.F. & Glass, A.G. Rarity o f Ewing's sarcoma among U.S. Negro children. Lancet 1, 366-367 (1970). 3. Jensen, R.D. & Drake, R.M. Rarity o f Ewing's tumour in Negroes. Lancet 1, 777(1970). 4. Linden, G. & Dunn, J.E. Ewing's sarcoma in Negroes. Lancet. 1,1171 (1970). 5. Polednak, A.P. Primary bone cancer incidence in Black and White residents of New York State. Cancer 55, 2883-2888 (1985). 6. Gumey, J.G., Severson, R.K., Davis, S. & Robison, L.L. Incidence of cancer in children in the United States - Sex-, race-, and 1-year age-specific rates by histologic type. Cancer 75, 2186-2195 (1995). 7. Edington, G.M., Bohrer, S.P. & Midlemiss, J.H. Ewing's sarcoma in Negroes. Lancet 1, 1171-1172(1970). 8. Oyemade, G.A.A. & Abioye, A. A. Primary malignant tumors of bone: incidence in ladan, Nigeria. J Natl Med Assoc 74,65-68 (1982). 9. Li, F.P., Tu, J.T., Liu, F.S. & Shiang, E.L. Rarity of Ewing's sarcoma in China. Lancet 1, 1255 (1980). 10. Parkin, D.M., Stiller, C.A. & Nectoux, J. International variations in the incidence of childhood bone tumors. IntJC an 53, 371-376 (1993). 11. Parra, E.J. et al. Estimating African American admixture proportions by use of population-specific alleles. Am JH um Genet 63, 1839-1851 (1998). 12. Chakraborty, R., Kamboh, M.I., Nwankwo, M. & Ferrell, R.E. Caucasian genes in American Blacks: new data. Am JH um Genet 5 0 ,145-155 (1992). 179 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13. Zamora, P. et al. Ewing's tumor in brothers. Am J Clin Oncol 9,358-360 (1986). 14. Huntington, R.W., SheffeL, D.J., Iger, M. & Henkelmann, C. Malignant bone tumors in siblings. J Bone Joint Surg 42-A, 1065-1075 (1960). 15. Joyce, M.J. et al. Ewing's sarcoma in female siblings: a clinical report and review of the literature. Cancer 53, 1959-1962 (1984). 16. Hutter, R.V.P., Francis, K.C. & Foote, F.W. Ewing's sarcoma in siblings: report of the second known ooccurrence. Am J Surg 107, 598-603 (1964). 17. Fraumeni, J.F. Stature and malignant tumors of bone in childhood and adolescence. Cancer 20, 967-973 (1967). 18. Pendergrass, T.W., Foulkes, M.A., Robison, L.L. & Nesbit, M.E. Stature and Ewing's sarcoma in childhood. Am JP ed Hem One 6, 33-39 (1984). 19. Buckley, J.D. et al. Epidemiology o f osteosarcoma and Ewing's sarcoma in childhood: A study o f305 cases from the Children's Cancer Group. Cancer 83, 1440-1448 (1998). 20. Holly, E.A., Aston, D.A., Ahn, D.K. & Kristiansen, J.J. Ewing's bone sarcoma, paternal occupational exposure and other factors. Am JEpidem 135, 122-129 (1992). 21. Winn, D.M. et al. A case-control study of the etiology of Ewing's sarcoma. Can Epi Bio Prev 1, 525-532 (1992). 22. Pui, C.-H., Dodge, R.K., George, S.L. & Green, A.A. Height at diagnosis of malignancies. Arch Dis Child 62,495-499 (1987). 23. Hum, L., Kreiger, N. & Finkelstein, M.M. The relationship between parental occupation and bone cancer risk in offspring. International Journal o f Epidemiology 27, 766-771 (1998). 24. Holman, C.D., Reynolds, P.M., Byms, M.J., Trooter, J.M. & Armstrong, B.K. Possible infectious etiology o f six cases of Ewing's sarcoma in Western Australia. Cancer 52,1974-1976 (1983). 25. Larrson, S.-E. & Lorentzon, R. The geographic variation of the incidence of malignant primary bone tumors in Sweden. J Bone Joint Surg 56-A, 592-600 (1974). 180 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26. Tilly, H., Bastard, C., Chevallier, B., Halkin, E. & Monconduit, M. Chromosomal abnormalities in secondary Ewing's sarcoma. Lancet I, 812 (1984). 27. Me Keen, E.A., Hanson, M.R., Mulvihill, J.J. & Glaubiger, D.L. Birth defects with Ewing's sarcoma. NEJM 309, 1522 (1983). 28. Cope, J.U., Tsokos, M., Hleman, L.J., Gridley, G. & Tucker, M.A. Inguinal hernia in patients with Ewing sarcoma: a clue to etiology. Med Ped One 34. 195-199 (2000). 29. Narod, S.A., Hawkins, M.M., Robertson, C.M. & Stiller, C.A. Congenital anomalies and childhood cancer in Great Britian. Am J Hum Genet 60,474- 485 (1997). 30. Fein-Levy, C., Gorlick, R., Meyers, P.A., Healey, J. & Huvos, A.G. Ewing's sarcoma in a patient with congenital optic atrophy. J Ped Hem/One 20, 577- 579(1998). 31. Nakissa, N., Constine, L.S., Rubin, P. & Strohl, R. Birth defects in three common pediatric malignancies; Wilm's tumor, neuroblastoma and Ewing's sarcoma. Oncol 42, 358-363 (1985). 32. Novakovic, B., Goldstein, A.M., Wexler, L.H. & Tucker, M.A. Increased risk of neuroectodermal tumors and stomach cancer in relatives of patients with Ewing's sarcoma family of tumors. J Natl Can Inst 86, 1702-1706 (1994). 33. Haas, O.A., Argyriou-Tirita, A. & Lion, T. Parental origin of chromosomes involved in the translocation t(9;22). Nature 3 5 9 ,414-416 (1992). 34. Hartley, A.L., Birch, J.M., Marsden, H.B., Harris, M. & Blair, V. Malignant disease in the mothers o f children with Ewing's tomour. M ed Ped One 16,95- 97 (1988). 35. Hartley, A.L. et al. Cancer incidence in the families of children with Ewing's tumor. JN atl Can Institute 83, 955-956 (1991). 36. McDougall, A. Malignant tumour at site of bone plating. J Bone Joint Surg 38B, 709-713 (1956). 37. Tayton, K.J.J. Ewing's sarcoma at the site o f a metal plate. Cancer 4 5 ,413- 415 (1980). 38. Austin, D.F., Nelson, V.E. & Johnson, L.F. Epidemiologic characteristis of childhood cancer. Front Radiat Ther One 16, 9-17 (1982). 181 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39. McWhirter, W.R., Dobson, C. & Ring, I. Childhood cancer incidence in Australia, 1982-1991. Int J Cancer 65, 34-38 (1996). 40. Ross, J.A. et al. Seasonal variations in the diagnosis o f childhood cancer in the United States. Brit j Can 81,549-553 (1999). 41. Chan, K.W., Rosen, G. & Pollack, M.S. Distribution of HLA antigens in patients with Ewing's sarcoma. Tiss Antigen 16, 314-316 (1980). 42. Lipinski, M. et al. Neuroectoderm-associated antigens on Ewing's sarcoma cell lines. Can Res 47, 183-187 (1987). 43. Sanchez-Prieto, R. et al. An association between viral genes and human oncogenic alterations: the adenovirus El A induces the Ewing tumor fusion transcript EWS-FLI1. Nat M ed 5, 1076-1079 (1999). 44. Me lot, T. & Delattre, O. El A and the Ewing tumor translocation. Nat M ed 5, 1331 (1999). 45. Kovar, H. El A and the Ewing tumor translocation. Nat Med 5, 1331 (1999). 46. Kovar, H. et al. Adenovirus El A does not induce the Ewing tumor-associated gene fusion EWS-FLI 1. Can Res 60, 1557-1560 (2000). 47. de Alva, E., Sanchez-Prieto, R. & Ramon y Cajal, S. Adenovirus El A and Ewing tumors. Nat Med 6 ,4 (2000). 48. Meric, F., Liao, Y., Lee, W.-P., Pollock, R.E. & Hung, M.-C. Adenovirus 5 early region 1A does not induce expression of the Ewing sarcoma fusion product EWS-FLI 1 in breast and ovarian cancer cell lines. Clin Can Res 6, 3832-6 (2000). 49. Cope, J.U. A viral etiology for Ewing's sarcoma. Med Hypoth 55, 369-372 (2000). 50. Istas, A.S., Demmler, G.J., Dobbins, J.G. & Stewart, J.A. Surveillance for congenital cytomegalovirus disease: a report from the National Congenital Cytomegalovirus Disease Registry. Clin Irtfect Dis 20,665-670 (1995). 51. Dayton, J.P., Cozen, W., Deapen, D. & Van Tomout, J.M. Ethnic specific incidence pattern of Ewing's sarcoma in Los Angeles County from 1985 to 1998. in preparation. 182 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52. Boppana, S.B., Rivera, L.B., Fowler, K.B., Mach, M. & Britt, W.J. Intrauterine transmission o f cytomegalovirus to infants of women with preconceptional immunity. NEJM 3 4 4 ,1366-1371 (2001). 53. Ewing, J. Diffuse endothelioma ofbone. Proc N Y Pathol Soc 21, 17 (1921). 54. van Valen, F., Winklemann, W. & Jurgens, H. Expression of functional Y1 receptors for neuropeptide Y in human Ewing's sarcoma cell lines. J Can Res Clin Oncol 118, 529-536 (1992). 55. Lee, C.S. et al. EWS/FLI1 fusion transcript detection and MIC2 immunohistochemical staining in the diagnosis of Ewing's sarcoma. Ped Pathol & Lab Medicine 16, 379-392 (1996). 56. Sugimoto, T., Umezawa, A. & Hata, J. Neurogenic potential of Ewing's sarcoma cells. Virchows Arch 430,41-46 (1997). 57. Navarro, S., Gonzalez-Devesa, M., Ferrandez-Izquierdo, A., Triche, T.J. & Llombart-Bosch, A. Scanning electron microscopic evidence of neural differentiation in Ewing's sarcoma cell lines. Virch Arch A Path Anat 416, 383-391 (1990). 58. Thiele, C. Pediatric peripheral neuroectodermal tumors, oncogenes, and differentiation. Can Invest 8, 629-639 (1990). 59. Aurias, A., Rimbaut, C., Buffe, D., Dubousset, J. & Mazabraud, A. Chromosomal translocations in Ewing's sarcoma. NEJM 309, 496-497 (1983). 60. Turc-CareL, C., Philip, I., Berger, M.-P., Philip, T. & Lenoir, G.M. Chromosomal translocations in Ewing's sarcoma. NEJM 8, 497-498 (1983). 61. Whang-Peng, J. et a l Chromosome translocation in peripherla neuroepithelioma. NEJM 311, 584-585 (1984). 62. Zucman, J. et a l Cloning and characterization of the Ewing's sarcoma and peripheral neuroepithelioma t(l 1;22) translocation breakpoints. Gen Chrom Can 5, 271-277 (1992). 63. PlougasteL, B., Zucman, J., Peter, M., Thomas, G. & Delattre, O. Genomic structure of the EWS gene and its relationship to EWSR1, a site of tumor- associated chromosome translocation. Genomics 18, 609-615 (1993). 183 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64. Mao, X., Miesfeldt, S., Yang, H., Leiden, J.M. & Thompson, C.B. The FLI-1 and chimeric EWS-FLI-1 oncoproteins display similar DNA binding specificities. J Biol Chem 2 6 9 ,18216-18222 (1994). 65. Ben-David, Y„ Giddens, E.B., Letwin, K. & Berstein, A. Erythroleukemia induction by Friend murine leukemia virus: insertional activation of a new membe of the ets gene family, Fli-1, closely linked to c-ets-1. Genes & Develop 5,908-918 (1991). 66. McKeon, C. et al. Indistinguishable patterns o f protooncogene expression in two distinct but closely related tumors: Ewing's sarcoma and neuroepithelioma. Can Res 4 8 ,4307-4311 (1988). 67. Dunn, T., Praissman, L., Hagag, N. & Viola, M.V. ERG gene is translocated in an Ewing's sarcoma cell line. Can Genet Cytogenet 76, 19-22 (1994). 68. Sorensen, P.H.B. et al. A second Ewing's sarcoma translocation, t(21;22), fuses the EWS gene to another ETS-family transcription factor, ERG. Nat Genet 6, 146-151 (1994). 69. Ishida, S. et al. The geneomic breakpoint and chimeric transcripts in the EWSR1-ETV4/E1AF gene fusion in Ewing's sarcoma. Cytogen Cell Genet 82, 278-283 (1998). 70. Delattre, O. et al. The Ewing family of tumors - a subgroup of small-round cell tumors defined by specific chineric transcripts. NEJM331,294-299 (1994). 71. Delattre, O. et al. Gene fusion with an ETS DNA-binding domain caused by chromosome translocation in human tumors. Nature 359, 162-165 (1992). 72. Stolow, D.T. & Haynes, S.R. Cabeza, a Drosophilia gene encoding a novel RNA binding protein, shares homology with EWS and TLS, two genes involved in human sarcoma formation. Nucl Acid Res 23, 835-843 (1995). 73. Aman, P. et al. Expression patterns fo the human sarcoma-associated genes FUS and EWS and the genomic struction of FUS. Genomics 37, 1-8 (1996). 74. Bertolotti, A. et al. EWS, but not EWS-FLI1, is associated with both TFIID and RNA polymerase II: interactions between two members of the TET family, EWS and hTAFII68, and subunits of TFIID and RNA polymerase II complexes. Mol Cell Bio 18, 1489-1497 (1998). 184 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75. Bertolotti, A., Lutz, Y., Heard, D.J., Chambon, P. & Tora, L. hTAFII68 a novel RNA/SSDNA-binding protein with homology to the prooncoproteins TLS/FUS and EWS is associated with both TFIID and RNA polymersase II. EM BO 1 5 ,5022-5031 (1996). 76. Petermann, R. et al. Oncogenic EWS-Flil interacts with hsRPB7, a subunit of human RNA polymerase II. Oncogene 17,603-610 (1998). 77. Rossow, K.L. & Janknecht, R. The Ewing's sarcoma gene product functions as atranscritptional activator. Can Res 61, 2690-2695 (2001). 78. Bhat, N.K., Fisher, R.J., Fujiwara, S., Ascione, R. & Papas, T.S. Temporal and tissue-specific expression of mouse ets genes. Proc Natl Acad Sci 84, 3161-3165(1987). 79. Selleri, L. et al. Cloning of the entire FLI1 gene, disrupted by the Ewing's sarcoma translocation breakpoint on 1 lq24, in a yeast artificial chromosome. Cytogenet Cell Genet 67, 129-136 (1994). 80. Prasad, D.D.K., Rao, V.N. & Reddy, S.P. Structure and expression of human Fli-1 gene. Can Res 52, 5833-5837 (1992). 81. Rao, V.N., Ohno, T., Prasad, D.D.K., Bhattacharya, G. & Reddy, E.S.P. Analysis o f the DNA-binding and transcritional activation functions of human Fli-l protein. Oncogene 8, 2167-2173 (1993). 82. Siddique, H.R., Rao, V.N., Lee, L. & Reddy, E.S.P. Characterization of the DNA binding and transcriptional activation domains of the erg protein. Oncogene 8, 1751-1755 (1993). 83. Yi, H. et al. Inhibition o f apoptosis by normal and aberrant Fli-1 and erg proteins involved in human solid tumors and leukemias. Oncogene 14, 1259- 1268 (1997). 84. Troung, A.H.L. & Ben-David, Y. The role of Fli-1 in normal cell function and malignant transformation. Oncogene 19, 6482-6489 (2000). 85. Mager, A.M. et al. The avian fli gene is specifically expressed during embryogenesis in a subset of neural crest cells giving rise to mesenchyme. Int JD ev Biol 42, 561-572 (1998). 86. Ohno, T., Rao, V.N. & Reddy, E.S.P. EWS/Fli-1 chimeric protein is a transcriptional activator. Can Res 53, 5859-5863 (1993). 185 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87. Bailly, R.-A. et al. DNA-binding ans transcriptional activation properties of the EWS-FLI 1 fusion protein resulting from the t(l 1 ;22) translocation in Ewing’ s sarcoma. Mol Cell Bio 14, 3230-3241 (1994). 88. Lessnick, S.L., Braun, B.S., Denny, C.T. & May, W.A. Multiple domains mediate transformation by the Ewing's sarcoma EWS/FLI-1 fusion gene. Oncogene 10,423-431 (1995). 89. Knezevich, S.R. et al. Absence of detectable EWS/FLIl expression after therapy-induced neural differentiation in Ewing sarcoma. Hum Pathol 29. 289-294 (1998). 90. Olsen, R.J. & Hinrichs, S.H. Phosphorylation of the EWS IQ domain regulates transcriptional activity. Oncogene 20, 1756-1764 (2001). 91. Zoubek, A. et al. Variability of EWS-chimeric transcripts in Ewing tumours: A comparison of clinical and molecular data. B rJ Can 70,908-913 (1994). 92. Zoubek, A. et al. Does expression of different EWS chimeric transcripts define clinically distinct risk groups of Ewing tumor patients? J Clin Ocol 14, 1245-1251 (1996). 93. de Alava, E. et al. EWS-FLI I fusion transcript structure is an independent determinant of prognosis in Ewing's sarcoma. J Clin Oncol 16, 1248-55 (1998). 94. Desmaze, C. et al. Multiple chromosomal mechanisms generate an EWS/FLIl or an EWS/ERG fusion gene in Ewings tumors. Can Genet Cyto 97, 12-19(1997). 95. Zucman-Rossi, J., Legoix, P., Victor, J.-M., Lopez, B. & Thomas, G. Chromosome translocation based on illegitimate recombination in human tumors. Proc Natl Acad Sci 95,11786-11791 (1998). 96. Kovar, H. et al. Cryptic exons as a source of increased diversity of Ewing tumor-associated EWS-FLI1 chimeric products. Genomics 60, 371-374 (1999). 97. Obata, K., Hiraga, H., Nojima, T., Yoshida, M.C. & Abe, S. Molecular characterization o f the genomic breakpoint junction in a t(l 1 ;22) translocation in Ewing sarcoma. Genes Chrom Can 25,6-15 (1999). 186 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98. Budarf, M., Sellinger, B., Griffin, C. & Emanuel, B.S. Comparative mapping of the constitutional and tumor-associated 11;22 translocations. Am J Hum Genet A S, 128-139(1989). 99. Zackai, E.H. & Emanuel, B.S. Site-specific reciprocal translocation, t(l 1 ;22), in several unrelated families with 3 :1 meiotic disjunction. Am J Med Genet 7, 507-521 (1980). 100. Fraccaro, M., Lindsten, J., Ford, C.E. & Iselius, L. The llq;22q translocation: A European collaborative analysis of 43 cases. Hum Genet 56. 21-51 (1980). 101. Groffen, J. et al. Philadelphia chromosomal breakpoints are clustered within a limited region, bcr, on chromosome 22. Cell 36,93-99 (1994). 102. Zucman-Rossi, J., Batzer, M.A., Stoneking, M., Delatrre, O. & Thomas, G. Interethnic polymorphism of EWS intron 6: genome plasticity mediated by Alu retroposition and recombination. Hum Genet 9 9 ,357-363 (1997). 103. Kovar, H. Ewing's sarcoma and peripheral primitive neuroectodermal tumors after their genetic union. Curr Opin Oncol 10, 334-342 (1998). 104. Whang-Peng, J., Freter, C.E., Knutsen, T., Nanfro, J.J. & Gazdar, A. Translocation t( 11 ;22) in esthesioneuroblastoma. Can Genet Cyto 29. 155- 157 (1987). 105. Sorensen, P.H.B. et al. Olfactory neuroblastoma is a peripheral primitive neuroectodermal tumor related to Ewing's sarcoma. Proc Natl Acad Sci 93, 1038-1043 (1996). 106. Nelson, R.S., Perlman, E.J. & Askin, F.B. Is esthesioneuroblastoma a peripheral neuroectodermal tumor? Hum Path 26, 639-641 (1995). 107. Argani, P. et al. Olfactory neuroblastoma is not related to the Ewing family of tumors. Am JSurgP ath 22, 391-398 (1998). 108. Mezzelani, A. et al. Esthesioneuroblastoma is not a member of the primitive peripheral neuroectodermal tumour-Ewing's group. Brit J Can 81, 586-591 (1999). 109. Sorensen, P.H. et al. Biphenotypic sarcomas with myogenic and neural differentiation express the Ewing’ s sarcoma EWS/FLI1 fusion gene. Can Res 55, 1385-1392 (1995). 187 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110. Triche, T J. personal communication. (1999). 111. Thomer, P. et al. Is the EWS/FLI1 fusion transcript specific for Ewing sarcoma and peripheral primitive neuroectodermal tumor? Am J Path 148, 1125-1138(1996). 112. Burchill, S.A., Wheeldon, J., Cullinane, C. & Lewis, I.J. EWS-FLI1 fusion transcripts in patients with typical neuroblastoma. E u rJ Can 33,239-243 (1997). 113. Hawkins, J.M., Craig, J.M., Seeker-Walker, L.M., Prentice, H.G. & Mehta, A.B. Ewing's sarcoma t(l 1;22) in a case of acute nonlymphocytic leukemia. Can Genet Cytogenet 55, 157-161 (1991). 114. Gerald, W.L. et al. Clinical, pathologic, and molecular spectrum of tumors associated with t(l I;22)(pl3;ql2): Desmoplatic small round-cell tumor and its variants. J Clin Oncol 16, 3028-3036 (1998). 115. Gerald, W.L. personal communication - electronic mail. (1999). 116. Zucman, J. et al. EWS and ATF-1 gene fusion induced by t(12;22) translocation in malignant melanoma of soft parts. Nature Genetics 4. 341- 345 (1993). 117. Epstein, A.L., Martin, AO. & Kempson, R. Use of a newly established human cell line (SU-CCS-1) to demonstrate the relationship of clear cell sarcoma to malignant melanoma. Cancer Research 44, 1265-1274 (1984). 118. Stenman, G., Kindblom, L.-G. & Angervall, L. Reciprocal translocatin t(12;22)(ql3;ql3) in clear-cell sarcoma of tendons and aponeuroses. Genes, Chrom & Cancer 4 ,122-127 (1992). 119. Mack, T.M. Sarcomas and other malignancies o f soft tissue, retro peritoneum, peritoneum, pleura, heart, mediastinum, and spleen. Cancer 75, 211-244 (1995). 120. Panagopoulos, I. et al. Fusion of EWS and CHOP genes in myxoid liposarcoma. Oncogene 12,489-494 (1996). 121. Clark, J. et al. Fusion of the EWS gene to CHN, a memeber of the steroid/thyroid receptor gene superfamily, in a human myxoid chondrosarcoma. Oncogene 12, 229-235 (1996). 188 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122. Gill, S. et al. Fusion of the EWS gene to a DNA segment from 9q22-31 in human myxoid chondrosarcoma. Genes, Chromosomes & Cancer 12, 307- 310(1995). 123. May, W.A. et al. The Ewing's sarcoma EWS/FLI-1 fusion gene encodes a more potent transcriptional activator and is a more powerful transforming gene than FLI-1. Mol Cell Biol 13, 7393-7398 (1993). 124. Braun, B.S., Frieden, R., Lessnick, SX., May, W.A. & Denny, C.T. Identification of target genes for the Ewqing's sarcoma EWS/FLI fusion protein by representational difference analysis. Mol Cell Biol 15,4623-4630 (1995). 125. Thompson, A.D. et al. EAT-2 is a novel SH2 domain containing protein that is up regulated by Ewing's sarcoma EWS/FLI 1 fusion gene. Oncogene 13, 2649-2658 (1996). 126. Pawson, T. & Gish, G. SH2 and SH3 domains: from structure to function. Cell 71, 359-362 (1992). 127. May, W.A. et al. EWS/FLI 1-induced manic fringe renders NIH3T3 cells tumorigenic. Nat Genet 17,495-497 (1997). 128. Yuan, Y.P., Schultz, J., Mlodzik, M. & Bork, P. Secreted fringe-like signaling molecules may be glycosyl transferases. Cell 88, 9-11 (1997). 129. Radig, K. et al. p53 and ras mutations in Ewing's sarcoma. Pathol Res Pract 194, 157-162(1998). 130. Stahl, U. et al. Ras protoonkogen aktivierung in tumoren der Ewing-gruppe. Verh Dtsch Ges Path 81, 571 (1997). 131. BurchilL, S.A., Berry, P.A., Bradbury, F.M. & Lewis, I.J. Contrasting levels of p2Iras activation and expression of neurofibromin in peripheral primitive neuroectodermal tumour and neuroblastoma cells, and their response to retinoic acid. JNeuro Sci 157, 129-137 (1998). 132. de Alva, E. et al. Prognostic impact of p53 status in Ewing sarcoma. Cancer 89, 783-792 (2000). 133. Baserga, R., Hongo, A., Rubini, M., Prisco, M. & Valentinis, B. The IGF-I receptor in cell growth, transformation and apoptosis. Biochim BiophysActa 1332, F105-F126 (1997). 189 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134. de Pagter-Holthuizen, P. et al. Organization o f the human genes for insulin- like growth factors I and II. FEBS Lett 179, 243-246 (1985). 135. Abbott, A.M., Bueno, R., Pedrini, M.T., Murray, J.M. & Smith, R.J. Insulin like growth factor I receptor gene structure. J Biol Chem 267, 10759-10763 (1992). 136. Dore, S., Kar, S. & Quirion, R. Rediscovering an old friend, IGF-I:potential use in the treatemnt of neurodegenerative diseases. TINS 20, 326-331 (1997). 137. Anlar, B., Sullivan, K.A. & Feldman, E.L. Insulin-like growth factor-I and central nervous system development. Horm Metab Res 31, 120-125 (1999). 138. Dudek, H. et al. Regulation of neuronal survival by the serine-threonine protein kinase Akt. Science 275,661-668 (1997). 139. Stoffel, M. et al. Human insulin receptor substrate-1 gene (IRSl):chromosomal localization to 2q35-q36.1 and identification of a simple tandem repeat DNA polymorphism. Diabetologia 36, 335-337 (1993). 140. Araki, E. et al. Human skeletal muscle insulin receptor substrate-1, characterization of the cDNA, gene, and chromosomal localization. Diabetes 42, 1041-1054(1993). 141. Canalis, E., McCarthy, T. & Centrella, M. Growth factros and the regulation of bone remodeling. JC lin Invest 81, 277-281 (1988). 142. Rosen, C.J. et al. Association between serum insulin growth factor-I (IGF-I) and a simple sequence repeat in IGF-I gene: implications for genetic studies of bone mineral density. J Clin Endo Metab 83,2286-2290 (1998). 143. Miyao, M. et al. Polymorphism o f insulin growth factor I gene and bone mineral density. C alcif Tissue Int 63, 306-311 (1998). 144. JuuL, A. et al. Free insulin-like growth factor I serum levels in 1430 healthy children and adults, and its diagnostic value in patients suspected of growth hormone deficiency. J Clin Endo Metab 82, 2497-2502 (1997). 145. Yee, D. et al. Insulin-like growth factor I expression by tumors of neuroectodermal origin with the t(l 1:22) chromosomal translocation: a potent autocrine growth factor.J C lin Invest 86,1806-1814 (1990). 190 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146. Yee, D. et al. Analysis o f IGF-I gene expression in malignancy, evidence for a paracrine role in human breast cancer. Mol Endocrinol 3, 509-517 (1989). 147. El-Badry, O.M. et al. Insulin-like growth factor-II mediated proliferation of human neuroblastoma. J Clin Invest 8 7 ,648-657 (1991). 148. Toretsky, J.A., Kalebic, T., Blakesley, V., LeRoith, D. & Helman, L.J. The insulin-like growth factor-I receptor is required foe EWS/FLI-1 transformation of fibroblasts. JB io l Chem 272, 30822-30827 (1997). 149. Crouch, M.F. & Hendry, I. A. Co-activation of insulin-like growth factor-I receptors and protein kinase C results in parasympathetic neural survival. J Neuroscience Res 28, 115-120 (1991). 150. van Valen, F., Winkelmann, W. & Jurgens, H. Type I and II insulin-like growth factor receptors and their function in human Ewing's sarcoma cells. J Can Res Clin Oncol 118, 269-275 (1992). 151. Scotlandi, K. et al. Insulin-like growth factor I receptor-mediated circuit in Ewing's sarcoma/peripheral neuroectodermal tumor: a possible therapeutic target. Can Res 56,4570-4574 (1996). 152. Scotlandi, K. et al. Blockage of insulin-like growth factor-I receptor inhibits the growth o f Ewing's sarcoma in athymic mice. Can Res 58,4127-4131 (1998). 153. Karniel, E., Werner, H., Rauscher, F.J., Benjamin, L.E. & LeRoith, D. The IGF-I receptor gene promoter is a molecular target for the Ewing's sarcoma- Wilm's tumor 1 fusion protein. JB io l Chem 271, 19304-19309 (1996). 154. Toretsky, J.A., Thakar, M., Eskenazi, A.E. & Frantz, C.N. Phosphoinositide 3-hydroxide kinase blockade enhances apoptosis in the Ewing's sacoma family of tumors. Can Res 59, 5745-5750 (1999). 155. Kaleko, M., Rutter, W.J. & Miller, A.D. Overexpression of the human insulinlike growth factor I receptor promotes ligand-dependent neoplastic transformation. Mol Cell Biol 10,464-473 (1990). 156. Frade, J.M. & Barde, Y.A. Nerve growth factor: two receptors, multiple fimtions. Bio Essays 20, 137-145 (1998). 157. Ullrich, A., Gray, A., Berman, C. & Dull, T.J. Human beta-nerve growth factor gene sequence highly homologous to that of mouse. Nature 303, 821- 825 (1983). 191 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158. Xia, Z., Dickens, M., Raingeaud, J., Davis, R.J. & Greenberg, M.E. Opposing effects fo ERK and JNK-p38 MAP kinases on apoptosis. Science 270, 1326- 1331 (1995). 159. Yao, R. & Cooper, G.M. Requirement for phosphatidylinositol-3 kinase in the prevention of apoptosis by nerve growth factor. Science 267, 2003-2006 (1995). 160. Greene, L.A. & Kaplan, D.R. Early events in neurothophin signalling via Trk and p75 receptors. Curr Opin Neurobio 5, 579-587 (1995). 161. Kaplan, D.R. & Stephens, R.M. Neurotrophin signal transduction by the trk receptor. J Neurobiology 25, 1404-1417 (1994). 162. Thomson, T.M., Pellicer, A. & Greene, L.A. Functional receptors for nerve growth factor in Ewing's sarcoma and Wilm's tumor cells. J o f Cell Phys 141, 60-64(1989). 163. Nogueira, E., Navarro, S., Pellin, A. & Llombart-Bosch, A. Activation of TRK genes in Ewing's sarcoma. Trk A receptor expression linked to neural differentiation. Diagn Mol Path 6, 10-16 (1997). 164. Veenstra, T.D., Fahnestock, M. & Kumar, R. An AP-1 site in the nerve growth factor promoter is essential for 1,25-dihydroxyvitamin D3-mediated nerve growth factor expression in osteoblasts. Biochem 37, 5988-5994 (1998). 165. Comet, A. et al. 1,25-dihydroxyvitamin D3 regulates the expression of VDR and NGF gene in Schwann cells in vitro. JNeurosci Res 53, 742-746 (1998). 166. Illich, J.Z. et al. Calcitriol and bone mass accumulation in females during puberty. CalcifTiss Int 61, 104-109 (1997). 167. Arndt, C.A.S. & Crist, W.M. Common musculoskeletal tumors of childhood and adolescence. NEJM341, 342-352 (1999). 168. Baldini, E.H. et al. Adults with Ewing's sarcoma/primive neuroectodermal tumor-Adverse effect of older age and primary extraosseous disease on outcome. Ann Surg 230, 79-86 (1999). 169. Ginsberg, J.P. et al. EWS-FLI1 and EWS-ERG gene fusions are associated with similar clinical phenotypes in Ewing's sarcoma. JC lin One 1 7 ,1809- 1814 (1999). 192 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 170. Kullendorff, C.-M. et al. Cytogenetic aberrations in Ewing's sarcoma: are secondary changes associated with clinical outcome? Med Ped One 32, 79-83 (1999). 171. Mugneret, F., Lizard, S., Aurias, A. & Turc-Carel, C. Chromosomes in Ewing's sreoma. II. Nonrandom additional changes, trisomy 8 and der(16)t(l;16). Can Genet Cyto 32, 239-245 (1988). 172. Available on the world wide web at http://www.jmavtlab-chla-usc.com/.. 173. Khoury, M.J. Genetic Epidemiology, 609-621 (Lippincott-Raven, Philidelphia, 1998). 174. Self, S.G., Longton, G., Kopecky, K.J. & Liang, K.Y. On estimating HLA/disease association with application to a study of aplastic anemia. Biometrics 47, 53-61 (1991). 175. Spielman, R.S., McGinnis, R.E. & Ewens, W.J. Transmission test for linkage disequilibrium: the insulin gene region and insulin-dependent diabetes mellhus (IDDM). Am J Hum Genet 52, 506-516 (1993). 176. Schaid, D.J. & Sommer, S.S. Genotype relative risks: methods for design and analysis of candidate-gene associations. Am J Hum Genet 5 3 ,1114-1126 (1993). 177. Schaid, D.J. & Sommer, S.S. Comparison of statistics for candidate-gene association studies using cases and parents. Am J Hum Genet 55,402-409 (1994). 178. Schaid, D.J. General score tests for associations of genetic markers with disease using cases and their parents. Genet Epi 13,423-449 (1996). 179. Schaid, D.J. Likelihoods and the TDT for the case-parents design. Genet Epidemiology 16,250-260 (1999). 180. Curtis, D. & Sham, P.C. A note on the application of the transmission disequilibrium test when a parent is missing. Am JH um Genet 56, 811-812 (1995). 181. Schaid, D.J. & Li, H. Genotype relative-risks and association tests for nuclear families with missing parental data. Genet Epidemiology 14,1113-1118 (1997). 193 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. Curtis, D. Use of siblings as controls in case-control association studies. Ann Hum Genet 61, 319-333 (1997). Sun, F., Flanders, W.D., Yang, Q. & Khoury, M.J. Transmission disequilibrium test (TDT) when only one parent is available: the 1-TDT. Am J Epidemiology 150, 97-104 (1999). Knapp, M. The transmission/disequilibrium test and parental-genotype reconstruction: the reconstruction-combined transmission/disequilibrium test. Am J Hum Genet 64, 862-870 (1999). Clayton, D. A generalization of the transmission/disequilibrium test for incertain-haplotype transmission. Am J Hum Genet 65, 1170-1177 (1999). Cervino, A.C.L. & Hill, A.V.S. Comparison of tests for association and linkage in incomplete families. Am J Hum Genet 67, 120-132 (2000). Lake, S.L., Blacker, D. & Laird, N.M. Family-based tests of association in the presence of linkage. Am JH um Genet 67, 1515-1525 (2000). Spielman, R.S. & Ewens, W.J. A sibship test for linkage in the presence of association: the sib transmission/disequilibrium test. Am J Hum Genet 62, 450-458 (1998). Monks, S.A., Kaplan, N.L. & Weir, B.S. A comparative study of the sibship tests of linkage and/or association. Am J Hum Genet 63, 1507-1516 (1998). Horvath, S. & Laird, N.M. A discordant-sibship test for disequilibrium and linkage: no need for parental data. Am JH um Genet 63, 1886-1897 (1998). Schaid, D J. & Rowland, C. Use of parents, sibs, and unrelated controls for detection of associations between genetic markers and disease. Am JH um Genet 63, 1492-1506(1998). Siegmund, K.D., Langholtz, B., Kraft, P. & Thomas, D.C. Testing linkage disequilibrium in sibships. Am JH um Genet 67, 244-248 (2000). Schaid, D.J. Transmission disequilibrium, family controls, and great expectations. Am JH um Genet 63,935-941 (1998). Falk, C.T. & Rubenstein, P. Haplotype relative risks: an easy reliable way to contruct a proper control sample for risk calculations. Ann Hum Genet 51, 227-233 (1987). 194 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 195. Rabinowitz, D. & Laird, N. A unified approach to adjusting association tests for population admixture with arbitrary pedigree structure and arbitrary missing marker information. Hum Hered 50, 211-223 (2000). 196. Laird, N.M., Horvath, S. & Xu, X Implementing a unified approach to family-based tests of association. Gen Epi 19, S36-S42 (2000). 197. Lum, A. & Le Marc hand, L. A simple mouthwash method for obtaining genomic DNA in molecular epidemiologic studies. Can Epid Biomark Prev 7, 719-24(1998). 198. Sweet, D. & Hildebrand, D. Recovery of DNA from human teeth by cryogenic grinding. J Forensic Sci 43, 1199-1202 (1998). 199. Retch, D.E. et al. Linkage disequilibrium in the human genome. Nature 411, 199-204 (2001). 200. Kraft, P. personal communication. (2001). 201. Christ, W.M. & Kun, L.E. Common solid tumors of childhood. NEJM 324, 461-465 (1991). 202. Ziegler, R.G. et al. Migration patterns and breast cancer risk in Asian- American women. JN a tl Can Inst 85, 1819-1827 (1993). 203. Taylor, J.A. et al. Association of prostate cancer with vitamin D receptor gene polymorphism. Cancer Res 56,4108-4110 (1996). 204. Hakami, J.M., Schoenberg, M.P. & Rondinelli, R.H. Androgen receptor CAG (glutamine) and GGC (glycine) repeat lengths as potential risk factors for prostate cancer. Am Assoc Cancer Res 37,258-262 (1996). 205. Ingles, S.A. et al. Association of prostate cancer risk with genetic polymorphisms in vitamin D receptor and androgen receptor. JN a tl Cancer Inst 89,166-170 (1997). 206. Ingles, S .A. et al. Strength of linkage disequilibrium between two vitamin D receptor markers in five ethnic groups: implications for association studies. Cancer Epidemiol Biomarkers Prev 6, 93-98 (1997). 207. Irvine, R.A., Yu, M.C., Ross, R.K. & Coetzee, G.A. The CAG and GGC microsatellites of the androgen receptor gene are in linkage disequilibrium in men with prostate cancer. Cancer Res 55,1937-1940 (1995). 195 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 208. Coetzee, G.A. & Ross, R.K. Prostate and the androgen receptor (letter). JNCI 86, 872-873 (1994). 209. Noguera, R., Triche, T.J., Navarro, S., Tsokos, M. & Llombart-Bosch, A. Dynamic model o f differentiation in Ewing's sarcoma cells: comparative analysis fo morphologic, immunocytochemical, and oncogene expression parameters. Lab Invest 66,143-151 (1992). 210. Hara, S., Adachi, Y., Kaneko, Y., Fujimoto, J. & Hata, J. Evidence for heterogeneous groups o f neuronal differentiation of Ewing's sarcoma. B rJ Cancer 64, 1025-1030 (1991). 211. CRegan, S., Diebler, M.F., Meunier, FJM. & Vyas, S. A Ewing's sarcoma cell line showing some, but not all, of the traits of a cholinergic neuron. J Neurochem 64, 69-76 (1995). 212. Lizard-NacoL S., Volk, C., Lizard, G. & Turc-Carel, C. Abnormal expression of neurofilament proteins in Ewing's sarcoma cell cultures. Tumor Biol 13, 36-43 (1992). 213. van Valen, F. & Keck, E. Induction of glycogenolysis in cultured Ewing's sarcoma cells by dopamine and beta-adrenergic agonists. J Can Res Clin One 114, 266-272 (1988). 214. Denny, C.T. Gene rearrangements in Ewing’ s sarcoma. Can Inv 14, 83-88 (1996). 215. Zucman, J. et al. Combinatorial generation of variable fusion proteins in Ewing family tumors. E M B O J12, 4481-4487 (1993). 216. Miller, R.W., Boice, J.D. & Curtis, R.E. Bone Cancer, in Cancer Epidemiology and Prevention (eds. Shottenfeld, D. & Fraumeni, J.F.) 971- 983 (Oxford Univ Press, New York, 1996). 217. Parkin, D.M. et al. (eds.). Cancer Incidence in Five Continents, 301,326 (International Agency for Research on Cancer, Lyon, 1992). 218. Weber, J.L. & May, P.E. Abundant class of human DNA polymorphisms which can be typed using the polymerase chain reaction. Am JH um Genet 44, 188-196(1989). 219. Meloni, R., Fougerousse, F., Roudaut, C. & Beckmann, J.S. Trinucleotide repeat polymorphism at the human insulin-like growth factor I receptor gene (IGF1R). Nucleic Acids Res 20, 1427 (1992). 196 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 220. Dib, C. et al. A comprehensive genetic map o f the human genome based on 5,264 microsatellites. Nature 380, 152-154 (1996). 221. Abu-amero, S., Preece, M., Wakeling, E., Moore, G. & Stanier, P. A common polymorphism in exon 16 of the human insulin-like growth factor-1 receptor gene (IFG1R). Mol Cell Probes 11, 381-383 (1997). 222. Vigouroux, C. et al. Genetic exclusion of 14 candidate genes in lipoatropic diabetes using linkage analysis in 10 consanguineous families. JC lin Endo Metab 82, 3438-3444 (1997). 223. Breakefield, X.O. et al. Structural gene for beta-nerve growth factor not defective in familial dysautonomia. Proc Natl Acad Sci 81,4213-6 (1984). 224. Mullikin, J.C. et al. An SNP map of human chromosome 22. Nature 407, 516-520 (2000). 225. Altshuler, D . et al. An SNP map o f the human genome generated by reduced repressentation shotgun sequencing. Nature 407, 513-520 (2000). 226. Almoguera, C. et al. Most humans carcinomas of the exocrine pancreas contain mutant c-K-ras genes. Cell 53, 549-554 (1988). 227. Turbett, G.R. & Sellner, L.N. The use of optimal cutting temperature compound can inhibit amplification by polymerase chain reaction. Diagn Mol Pathol 6, 298-303 (1997). 228. Goelz, S.E., Hamilton, S.R. & Vogelstein, B. Purification of DNA from formaldehyde fixed and paraffin embedded tissue. Biochem Biophys Res Commun 130,118-126 (1985). 229. Crisan, D. & Mattson, J.C. Amplification of intermediate-size DNA sequences from formalin and B-5 fixed tissue by polymerase chain reaction. Clin Biochem 25,1099-1103 (1992). 230. Dayton, J.P., Buckley, J.D. & Van Tomout, J.M. A review of the epidemiology of Ewing's sarcoma in relation to the known tumor biology, in preparation . 197 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References (Alphabetical) 1. Abbott, A.M., Bueno, R., Pedrini, M.T., Murray, J.M. & Smith, R.J. Insulin like growth factor I receptor gene structure. JB io l Chem 2 67,10759-10763 (1992). 2. Abu-amero, S., Preece, M., Wakeling, E., Moore, G. & Stanier, P. A common polymorphism in exon 16 of the human insulin-like growth factor-1 receptor gene (IFG1R). Mol Cell Probes 11, 381-383 (1997). 3. Almoguera, C. et al. Most humans carcinomas of the exocrine pancreas contain mutant c-K-ras genes. Cell 53, 549-554 (1988). 4. Altshuler, D. et al. An SNP map of the human genome generated by reduced repressentation shotgun sequencing. Nature 407, 513-520 (2000). 5. Aman, P. et al. Expression patterns fo the human sarcoma-associated genes FUS and EWS and the genomic struction of FUS. Genomics 37, 1-8 (1996). 6. Anlar, B., Sullivan, K.A. & Feldman, E.L. Insulin-like growth factor-I and central nervous system development. Horm Metab Res 31, 120-125 (1999). 7. Araki, E. et al. Human skeletal muscle insulin receptor substrate-1, characterization of the cDNA, gene, and chromosomal localization. Diabetes 42, 1041-1054(1993). 8. Argani, P. et al. Olfactory neuroblastoma is not related to the Ewing family of tumors. Am JSurg Path 22, 391-398 (1998). 9. Arndt, C.A.S. & Crist, W.M. Common musculoskeletal tumors of childhood and adolescence. NEJM341, 342-352 (1999). 10. Aurias, A., Rimbaut, C., Bufife, D., Dubousset, J. & Mazabraud, A. Chromosomal translocations in Ewing's sarcoma. NEJM 309,496-497 (1983). 11. Austin, D.F., Nelson, V.E. & Johnson, L.F. Epidemiologic characteristis of childhood cancer. Front Radiat Ther One 16, 9-17 (1982). 12. Available on the world wide web at http://www.jmavtlab-chla-usc.com/.. 13. fiailly, R-A. et aL DNA-binding ans transcriptional activation properties of the EWS-FLI1 fusion protein resulting from the t(l 1;22) translocation in Ewing’ s sarcoma. Mol Cell Bio 14, 3230-3241 (1994). 198 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14. Baldini, E.H. et al. Adults with Ewing's sarcoma/primive neuroectodermal tumor-Adverse effect o f older age and primary extraosseous disease on outcome. Ann Surg 230, 79-86 (1999). 15. Baserga, R., Hongo, A., Rubini, M., Prisco, M. & Valentinis, B. The IGF-I receptor in cell growth, transformation and apoptosis. Biochim Biophys Acta 1332, F105-F126 (1997). 16. Ben-David, Y., Giddens, E.B., Letwin, K. & Berstein, A. Erythroleukemia induction by Friend murine leukemia virus: insertional activation of a new membe of the ets gene family, Fli-1, closely linked to c-ets-1. Genes & Develop 5, 908-918 (1991). 17. Bertolotti, A. et al. EWS, but not EWS-FLIl, is associated with both TFIID and RNA polymerase II: interactions between two members of the TET family, EWS and hTAFII68, and subunits of TFIID and RNA polymerase II complexes. Mol Cell Bio 18, 1489-1497 (1998). 18. Bertolotti, A., Lutz, Y., Heard, D.J., Chambon, P. & Tora, L. hTAFII68 a novel RNA/SSDNA-binding protein with homology to the prooncoproteins TLS/FUS and EWS is associated with both TFIID and RNA polymersase II. EMBO 15, 5022-5031 (1996). 19. Bhat, N.K., Fisher, R.J., Fujiwara, S., Ascione, R. & Papas, T.S. Temporal and tissue-specific expression of mouse ets genes. Proc Natl Acad Sci 84, 3161-3165(1987). 20. Boppana, S.B., Rivera, L.B.. Fowler, K.B., Mach, M. & Britt, W.J. Intrauterine transmission o f cytomegalovirus to infants of women with preconceptional immunity. NEJM 344,1366-1371 (2001). 21. Braun, B.S., Frieden, R., Lessnick, S.L., May, W.A. & Denny, C.T. Identification of target genes for the Ewqing's sarcoma EWS/FLI fusion protein by representational difference analysis. Mol Cell Biol 15,4623-4630 (1995). 22. Breakefield, X.O. et al. Structural gene for beta-nerve growth factor not defective in familial dysautonomia. Proc Natl Acad Sci 81,4213-6 (1984). 23. Buckley, J.D. et al. Epidemiology of osteosarcoma and Ewing's sarcoma in childhood: A study o f 305 cases from the Children's Cancer Group. Cancer 8 3 ,1440-1448 (1998). 199 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24. Budarf M., Sellinger, B., Griffin, C. & Emanuel, B.S. Comparative mapping of the constitutional and tumor-associated ll;22 translocations. Am J Hum Genet AS, 128-139(1989). 25. Burchill, S.A., Berry, P.A., Bradbury, F.M. & Lewis, I.J. Contrasting levels of p2Iras activation and expression of neurofibromin in peripheral primitive neuroectodermal tumour and neuroblastoma cells, and their response to retinoic acid. JNeuro Sci 157,129-137 (1998). 26. Burchill, S.A., Wheeldon, J., Cullinane, C. & Lewis, I.J. EWS-FLIl fusion transcripts in patients with typical neuroblastoma. E u rJ Can 33, 239-243 (1997). 27. Canalis, E., McCarthy, T. & Centrella, M. Growth factros and the regulation of bone remodeling. JC lin Invest 81, 277-281 (1988). 28. Cervino, A.C.L. & Hill, A.V.S. Comparison of tests for association and linkage in incomplete families. Am JH um Genet 67, 120-132 (2000). 29. Chakraborty, R., Kamboh, M.I., N wank wo, M. & Ferrell, R.E. Caucasian genes in American Blacks: new data. Am JH um Genet 5 0 ,145-155 (1992). 30. Chan, K.W., Rosen, G. & Pollack, M.S. Distribution of HLA antigens in patients with Ewing's sarcoma. Tiss Antigen 16, 314-316 (1980). 31. Christ, W.M. & Kun, L.E. Common solid tumors of childhood. NEJM 324, 461-465 (1991). 32. Clark, J. et al. Fusion of the EWS gene to CHN, a memeber of the steroid/thyroid receptor gene superfamily, in a human myxoid chondrosarcoma. Oncogene 12, 229-235 (1996). 33. Clayton, D. A generalization of the transmission/disequilibrium test for incertain-haplotype transmission. Am J Hum Genet 6 5 ,1170-1177 (1999). 34. Coetzee, G.A. & Ross, R.K. Prostate and the androgen receptor (letter). JNCI 86, 872-873 (1994). 35. Cope, J.U. A viral etiology for Ewing's sarcoma. M ed Hypoth 55, 369-372 (2000). 36. Cope, J.U., Tsokos, M., Hleman, L.J., Gridley, G. & Tucker, M.A. Inguinal hernia in patients with Ewing sarcoma: a clue to etiology. M ed Ped One 34, 195-199 (2000). 200 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37. Comet, A. et al. 1,25-dihydroxyvitamin D3 regulates the expression of VDR and NGF gene in Schwann cells in vitro. JNeurosci Res 53, 742-746 (1998). 38. Crisan, D. & Mattson, J.C. Amplification of intermediate-size DNA sequences from formalin and B-5 fixed tissue by polymerase chain reaction. Clin Biochem 25, 1099-1103 (1992). 39. Crouch, M.F. & Hendry, I.A. Co-activation of insulin-like growth factor-I receptors and protein kinase C results in parasympathetic neural survival. J Neuroscience Res 28, 115-120 (1991). 40. Curtis, D. & Sham, P.C. A note on the application of the transmission disequilibruim test when a parent is missing. Am JH um Genet 56, 811-812 (1995). 41. Curtis, D. Use of siblings as controls in case-control association studies. Ann Hum Genet 61, 319-333 (1997). 42. Dayton, J.P., Buckley, J.D. & Van Tomout, J.M. A review of the epidemiology o f Ewing's sarcoma in relation to the known tumor biology, in preparation. 43. Dayton, J.P., Cozen, W., Deapen, D. & Van Tomout, J.M. Ethnic specific incidence pattern of Ewing's sarcoma in Los Angeles County from 1985 to 1998. in preparation. 44. de Alava, E. et al. EWS-FLI1 fusion transcript structure is an independent determinant of prognosis in Ewing's sarcoma. JC lin Oncol 16, 1248-55 (1998). 45. de Alva, E. et a l Prognostic impact of p53 status in Ewing sarcoma. Cancer 89, 783-792 (2000). 46. de Alva, E., Sanchez-Prieto, R. & Ramon y Cajal, S. Adenovirus El A and Ewing tumors. Nat Med 6, 4 (2000). 47. de Pagter-Holthuizen, P. et al. Organization of the human genes for insulin like growth factors I and II. FEBS Lett 179, 243-246 (1985). 48. Delattre, O. et al. Gene fusion with an ETS DNA-binding domain caused by chromosome translocation in human tumors. Nature 359, 162-165 (1992). 201 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49. Delattre, O. et al. The Ewing family o f tumors - a subgroup o f small-round cell tumors defined by specific chineric transcripts. NEJM 331,294-299 (1994). 50. Denny, C.T. Gene rearrangements in Ewing's sarcoma. Can Inv 14, 83-88 (1996). 51. Desmaze, C. et al. Multiple chromosomal mechanisms generate an EWS/FLI 1 or an EWS/ERG fusion gene in Ewings tumors. Can Genet Cyto 97, 12-19 (1997). 52. Dib, C. et al. A comprehensive genetic map of the human genome based on 5,264 microsatellites. Nature 380, 152-154 (1996). 53. Dore, S., Kar, S. & Quirion, R. Rediscovering an old friend, IGF-I rpotential use in the treatemnt of neurodegenerative diseases. TINS 20, 326-331 (1997). 54. Dudek, H. et al. Regulation o f neuronal survival by the serine-threonine protein kinase Akt. Science 275,661-668 (1997). 55. Dunn, T., Praissman, L., Hagag, N. & Viola, M.V. ERG gene is translocated in an Ewing's sarcoma cell line. Can Genet Cytogenet 76, 19-22 (1994). 56. Edington, G.M., Bohrer, S.P. & Midlemiss, J.H. Ewing's sarcoma in Negroes. Lancet 1, 1171-1172(1970). 57. El-Badry, O.M. et al. Insulin-like growth factor-II mediated proliferation of human neuroblastoma. JC lin Invest 87,648-657 (1991). 58. Epstein, A.L., Martin, A.O. & Kempson, R. Use of a newly established human cell line (SU-CCS-1) to demonstrate the relationship o f clear cell sarcoma to malignant melanoma. Cancer Research 4 4 ,1265-1274 (1984). 59. Ewing, J. Diffuse endothelioma ofbone. Proc N Y Pathol Soc 21, 17 (1921). 60. Falk, C.T. & Rubenstein, P. Haplotype relative risks: an easy reliable way to contruct a proper control sample for risk calculations. Ann Hum Genet 51, 227-233 (1987). 61. Fein-Levy, C., Gorlick, R., Meyers, PA., Healey, J. & Huvos, AG. Ewing's sarcoma in a patient with congenital optic atrophy. JP ed Hem/One 20, 577- 579 (1998). 202 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62. Fraccaro, M., Lindsten, J., Ford, C.E. & Iselius, L. The 1 lq;22q translocation: A European collaborative analysis of 43 cases. Hum Genet 56, 21-51 (1980). 63. Frade, J.M. & Barde, Y.A. Nerve growth factor: two receptors, multiple funtions. Bio Essays 20, 137-145 (1998). 64. Fraumeni, J.F. & Glass, A.G. Rarity of Ewing's sarcoma among U.S. Negro children. Lancet 1, 366-367 (1970). 65. Fraumeni, J.F. Stature and malignant tumors of bone in childhood and adolescence. Cancer 20, 967-973 (1967). 66. Gerald, W.L. et al. Clinical, pathologic, and molecular spectrum o f tumors associated with t(l I;22)(pl3;ql2): Desmoplatic small round-cell tumor and its variants. J Clin Oncol 16, 3028-3036 (1998). 67. Gerald, W.L. personal communication - electronic mail. (1999). 68. Gill, S. et al. Fusion of the EWS gene to a DNA segment from 9q22-31 in human myxoid chondrosarcoma. Genes, Chromosomes & Cancer 12, 307- 310(1995). 69. Ginsberg, J.P. et al. EWS-FLI1 and EWS-ERG gene fusions are associated with similar clinical phenotypes in Ewing's sarcoma. JC lin One 17, 1809- 1814(1999). 70. Goelz, S.E., Hamilton, S.R. & Vogelstein, B. Purification of DNA from formaldehyde fixed and paraffin embedded tissue. Biochem Biophys Res Commun 130, 118-126(1985). 71. Greene, L.A. & Kaplan, D.R. Early events in neurothophin signalling via Trk and p75 receptors. Curr Opin Neurobio 5, 579-587 (1995). 72. Groffen, J. et al. Philadelphia chromosomal breakpoints are clustered within a limited region, bcr, on chromosome 22. Cell 36, 93-99 (1994). 73. Gurney, J.G., Severson, R.K., Davis, S. & Robison, L.L. Incidence o f cancer in children in the United States - Sex-, race-, and 1-year age-specific rates by histologic type. Cancer 75,2186-2195 (1995). 74. Haas, O.A., Argyriou-T irita, A. & Lion, T. Parental origin of chromosomes involved in the translocation t(9;22). Nature 359,414-416 (1992). 203 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75. Hakami, J.M., Schoenberg, M.P. & Rondinelli, R.H. Androgen receptor CAG (glutamine) and GGC (glycine) repeat lengths as potential risk factors for prostate cancer. Am Assoc Cancer Res 37,258-262 (1996). 76. Hara, S., Adachi, Y., Kaneko, Y., Fujimoto, J. & Hata, J. Evidence for heterogeneous groups of neuronal differentiation o f Ewing's sarcoma. B rJ Cancer 64, 1025-1030 (1991). 77. Hartley, A.L. et al. Cancer incidence in the families o f children with Ewing's tumor. JN a tl Can Institute 83, 955-956 (1991). 78. Hartley, A.L., Birch, J.M., Marsden, H.B., Harris, M. & Blair, V. Malignant disease in the mothers o f children with Ewing's tomour. Med Ped One 16,95- 97 (1988). 79. Hawkins, J.M., Craig, J.M., Seeker-Walker, L.M., Prentice, H.G. & Mehta, A.B. Ewing's sarcoma t(l 1;22) in a case of acute nonlymphocytic leukemia. Can Genet Cytogenet 55, 157-161 (1991). 80. Holly, E.A., Aston, D.A., Ahn, D.K. & Kristiansen, J.J. Ewing's bone sarcoma, paternal occupational exposure and other factors. Am JEpidem 135, 122-129 (1992). 81. Holman, C.D., Reynolds, P.M., Byms, M.J., Trooter, J.M. & Armstrong, B.K. Possible infectious etiology of six cases of Ewing's sarcoma in Western Australia. Cancer 52, 1974-1976 (1983). 82. Horowitz, M.E., Malawer, M.M., Woo, S.Y. & Hicks, M.J. Ewing's sarcoma family o f Tumors: Ewing's sarcoma of bone and soft tissue and the peripheral primitive neuroectodermal tumors, in Principles and Practice of Pediatric Oncology (eds. Pizzo, P.A. & Poplack, D.G.) 831-863. (Lippincott-Raven Publishers., Philadelphia, PA, 1997). 83. Horvath, S. & Laird, N.M. A discordant-sibship test for disequilibrium and linkage: no need for parental data. Am JH um Genet 63, 1886-1897 (1998). 84. Hum, L., Kreiger, N. & Finkelstein, M.M. The relationship between parental occupation and bone cancer risk in offspring. International Journal o f Epidemiology 27, 766-771 (1998). 85. Huntington, R.W., SheffeL, D.J., Iger, M. & Henkelmann, C. Malignant bone tumors in siblings. J Bone Joint Surg 42-A, 1065-1075 (1960). 204 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86. Hutter, R.V.P., Francis, K.C. & Foote, F.W. Ewing's sarcoma in siblings: report o f the second known ooccurrence. Am JSurg 107, 598-603 (1964). 87. Illich, J.Z. et al. Calcitriol and bone mass accumulation in females during puberty. CalcifTiss Int 61, 104-109 (1997). 88. Ingles, S.A. et al. Association o f prostate cancer risk with genetic polymorphisms in vitamin D receptor and androgen receptor. JN atl Cancer Inst 8 9 ,166-170 (1997). 89. Ingles, S. A. et al. Strength of linkage disequilibrium between two vitamin D receptor markers in five ethnic groups: implications for association studies. Cancer Epidemiol Biomarkers Prev 6, 93-98 (1997). 90. Irvine, R.A., Yu, M.C., Ross, R.K. & Coetzee, G.A. The CAG and GGC microsatellites of the androgen receptor gene are in linkage disequilibrium in men with prostate cancer. Cancer Res 55,1937-1940 (1995). 91. Ishida, S. et al. The geneomic breakpoint and chimeric transcripts in the EWSR1-ETV4/E1AF gene fusion in Ewing's sarcoma. Cytogen Cell Genet 82, 278-283 (1998). 92. Istas, A.S., Demmler, G.J., Dobbins, J.G. & Stewart, J.A. Surveillance for congenital cytomegalovirus disease: a report from the National Congenital Cytomegalovirus Disease Registry. Clin Infect Dis 20, 665-670 (1995). 93. Jensen, R.D. & Drake, R.M. Rarity of Ewing's tumour in Negroes. Lancet 1, 777(1970). 94. Joyce, M.J. et al. Ewing's sarcoma in female siblings: a clinical report and review of the literature. Cancer 53, 1959-1962 (1984). 95. Juul, A. et al. Free insulin-like growth factor I serum levels in 1430 healthy children and adults, and its diagnostic value in patients suspected of growth hormone deficiency. JC lin Endo Metab 82, 2497-2502 (1997). 96. Kaleko, M., Rutter, W.J. & Miller, A.D. Overexpression of the human insulinlike growth factor I receptor promotes ligand-dependent neoplastic transformation. Mol Cell Biol 10,464-473 (1990). 97. Kaplan, D.R. & Stephens, R.M. Neurotrophin signal transduction by the trk receptor. J Neurobiology 25, 1404-1417 (1994). 205 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98. KarnieL, E., Wemer, H., Rauscher, F.J., Benjamin, L.E. & LeRoith, D. The IGF-I receptor gene promoter is a molecular target for the Ewing's sarcoma- Wilm's tumor 1 fusion protein. J Biol Chem 271, 19304-19309 (1996). 99. Khoury, M.J. Genetic Epidemiology, 609-621 (Lippincott-Raven, Philidelphia, 1998). 100. Knapp, M. The transmission/disequilibrium test and parental-genotype reconstruction: the reconstruction-combined transmission/disequilibrium test. Am J Hum Genet 64, 862-870 (1999). 101. Knezevich, S.R. et aL Absence of detectable EWS/FLI 1 expression after therapy-induced neural differentiation in Ewing sarcoma. Hum Pathol 29, 289-294 (1998). 102. Kovar, H. El A and the Ewing tumor translocation. Nat Med 5, 1331 (1999). 103. Kovar, H. et al. Adenovirus El A does not induce the Ewing tumor-associated gene fusion EWS-FLI1. Can Res 60, 1557-1560 (2000). 104. Kovar, H. et al. Cryptic exons as a source of increased diversity of Ewing tumor-associated EWS-FLI1 chimeric products. Genomics 60, 371-374 (1999). 105. Kovar, H. Ewing's sarcoma and peripheral primitive neuroectodermal tumors after their genetic union. Curr Opin Oncol 10, 334-342 (1998). 106. Kraft, P. personal communication. (2001). 107. Kullendorff C.-M. et al. Cytogenetic aberrations in Ewing's sarcoma: are secondary changes associated with clinical outcome? Med Ped One 32, 79-83 (1999). 108. Laird, N.M., Horvath, S. & Xu, X. Implementing a unified approach to family-based tests of association. Gen Epi 19, S36-S42 (2000). 109. Lake, S.L., Blacker, D. & Laird, N.M. Family-based tests of association in the presence o f linkage. Am JH um Genet 67, 1515-1525 (2000). 110. Larrson, S.-E. & Lorentzon, R. The geographic variation of the incidence of malignant primary bone tumors in Sweden. J Bone Joint Surg 56-A, 592-600 (1974). 206 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 111. Lee, C.S. et al. EWS/FLI 1 fusion transcript detection and MIC2 immunohistochemical staining in the diagnosis o f Ewing's sarcoma. Ped Pathol & Lab Medicine 16, 379-392 (1996). 112. Lessnick, S.L., Braun, B.S., Denny, C.T. & May, W.A. Multiple domains mediate transformation by the Ewing's sarcoma EWS/FLI-1 fusion gene. Oncogene 10,423-431 (1995). 113. Li, F.P., Tu, J.T., Liu, F.S. & Shiang, E.L. Rarity of Ewing's sarcoma in China. Lancet 1, 1255 (1980). 114. Linden, G. & Dunn, J.E. Ewing's sarcoma in Negroes. Lancet 1,1171 (1970). 115. Lipinski, M. et al. Neuroectoderm-associated antigens on Ewing's sarcoma cell lines. Can Res 47, 183-187 (1987). 116. Lizard-Nacol, S., Volk, C., Lizard, G. & Turc-Carel, C. Abnormal expression of neurofilament proteins in Ewing’ s sarcoma cell cultures. Tumor Biol 13, 36-43 (1992). 117. Lum, A. & Le Marchand, L. A simple mouthwash method for obtaining genomic DNA in molecular epidemiologic studies. Can Epid Biomark Prev 7, 719-24 (1998). 118. Mack, T.M. Sarcomas and other malignancies of soft tissue, retro peritoneum, peritoneum, pleura, heart, mediastinum, and spleen. Cancer 75, 211-244 (1995). 119. Mager, A.M. et al. The avian fli gene is specifically expressed during embryogenesis in a subset o f neural crest cells giving rise to mesenchyme. Int J Dev Biol 42, 561-572 (1998). 120. Mao, X., Miesfeldt, S., Yang, H., Leiden, J.M. & Thompson, C.B. The FLI-1 and chimeric EWS-FLI-1 oncoproteins display similar DNA binding specificities. JB io l Chem 269,18216-18222 (1994). 121. May, W.A. et al. EWS/FLI 1-induced manic fringe renders NIH3T3 cells tumorigenic. Nat Genet 17,495-497 (1997). 122. May, W.A. et aL The Ewing's sarcoma EWS/FLI-l fusion gene encodes a more potent transcriptional activator and is a more powerful transforming gene than FLI-1. Mol Cell Biol 13, 7393-7398 (1993). 207 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. McDougall, A. Malignant tumour at site of bone plating. J Bone Joint Surg 38B, 709-713 (1956). McKeen, E.A., Hanson, M.R., Mulvihill, J.J. & Glaubiger, D.L. Birth defects with Ewing's sarcoma. NEJM 309, 1522 (1983). McKeon, C. et al. Indistinguishable patterns of protooncogene expression in two distinct but closely related tumors: Ewing's sarcoma and neuroepithelioma. Can Res 48,4307-4311 (1988). McWhirter, W.R., Dobson, C. & Ring, I. Childhood cancer incidence in Australia, 1982-1991. Int J Cancer 65, 34-38 (1996). Meloni, R., Fougerousse, F., Roudaut, C. & Beckmann, J.S. Trinucleotide repeat polymorphism at the human insulin-like growth factor I receptor gene (IGF1R). Nucleic Acids Res 20, 1427 (1992). Melot, T. & Delattre, O. El A and the Ewing tumor translocation. Nature Med 5, 1331 (1999). Meric, F., Liao, Y., Lee, W.-P., Pollock, R.E. & Hung, M.-C. Adenovirus 5 early region 1A does not induce expression of the Ewing sarcoma fusion product EWS-FLI1 in breast and ovarian cancer cell lines. Clin Can Res 6, 3832-6 (2000). Mezzelani, A. et aL Esthesioneuroblastoma is not a member of the primitive peripheral neuroectodermal tumour-Ewing's group. Brit J Can 81, 586-591 (1999). Miller, R.W., Boice, J.D. & Curtis, R.E. Bone Cancer, in Cancer Epidemiology and Prevention (eds. Shottenfeld, D. & Fraumeni, J.F.) 971- 983 (Oxford Univ Press, New York, 1996). Miyao, M. et al. Polymorphism of insulin growth factor I gene and bone mineral density. C alcif Tissue Int 63, 306-311 (1998). Monks, S.A., Kaplan, N.L. & Weir, B.S. A comparative study of the sibship tests of linkage and/or association. Am JH um Genet 63, 1507-1516 (1998). Mugneret, F., Lizard, S., Aurias, A. & Turc-Carel, C. Chromosomes in Ewing's srcoma. II. Nonrandom additional changes, trisomy 8 and der(16)t(l;16). Can Genet Cyto 32,239-245 (1988). 208 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135. Mullikin, J.C. et al. An SNP map of human chromosome 22. Nature 407, 516-520 (2000). 136. Nakissa, N., Constine, L.S., Rubin, P. & Strohl, R. Birth defects in three common pediatric malignancies; Wilm's tumor, neuroblastoma and Ewing's sarcoma. Oncol 42,358-363 (1985). 137. Narod, S.A., Hawkins, M.M., Robertson, C.M. & Stiller, C.A. Congenital anomalies and childhood cancer in Great Britian. Am J Hum Genet 60,474- 485 (1997). 138. Navarro, S., Gonzalez-Devesa, M., Ferrandez-Izquierdo, A., Triche, T.J. & Llombart-Bosch, A. Scanning electron microscopic evidence o f neural differentiation in Ewing's sarcoma cell lines. Virch Arch A Path Anat 416, 383-391 (1990). 139. Nelson, R.S., Perlman, E.J. & Askin, F.B. Is esthesioneuroblastoma a peripheral neuroectodermal tumor? Hum Path 26,639-641 (1995). 140. Nogueira, E., Navarro, S., Pellin, A. & Llombart-Bosch, A. Activation of TRK genes in Ewing's sarcoma. Trk A receptor expression linked to neural differentiation. Diagn Mol Path 6, 10-16 (1997). 141. Noguera, R., Triche, T.J., Navarro, S., Tsokos, M. & Llombart-Bosch, A. Dynamic model of differentiation in Ewing's sarcoma cells: comparative analysis fo morphologic, immunocytochemical, and oncogene expression parameters. Lab Invest 66,143-151 (1992). 142. Novakovic, B., Goldstein, A.M., Wexler, L.H. & Tucker, M.A. Increased risk of neuroectodermal tumors and stomach cancer in relatives of patients with Ewing's sarcoma family of tumors. JN a tl Can Inst 86,1702-1706 (1994). 143. Obata, K., Hiraga, H., Nojima, T., Yoshida, M.C. & Abe, S. Molecular characterization of the genomic breakpoint junction in a t( 11 ;22) translocation in Ewing sarcoma. Genes Chrom Can 25, 6-15 (1999). 144. Ohno, T., Rao, V.N. & Reddy, E.S.P. EWS/Fli-1 chimeric protein is a transcriptional activator. Can Res 53, 5859-5863 (1993). 145. Olsen, R.J. & Hinrichs, S.H. Phosphorylation of the EWS IQ domain regulates transcriptional activity. Oncogene 20, 1756-1764 (2001). 209 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146. O'Regan, S., Diebler, M.F., Meunier, F.M. & Vyas, S. A Ewing's sarcoma cell line showing some, but not all, of the traits o f a cholinergic neuron. J Neurochem 64, 69-76 (1995). 147. Oyemade, G.A.A. & Abioye, A. A. Primary malignant tumors of bone: incidence in Iadan, Nigeria. JN a tl Med Assoc 74, 65-68 (1982). 148. Panagopoulos, I. et al. Fusion o f EWS and CHOP genes in myxoid liposarcoma. Oncogene 12, 489-494 (1996). 149. Parkin, D.M. et al. (eds.). Cancer Incidence in Five Continents, 301, 326 (International Agency for Research on Cancer, Lyon, 1992). 150. Parkin, D.M., Stiller, C.A. & Nectoux, J. International variations in the incidence o f childhood bone tumors. Int J Can 53, 371-376 (1993). 151. Parra, E.J. et al. Estimating African American admixture proportions by use of population-specific alleles. Am J Hum Genet 63, 1839-1851 (1998). 152. Pawson, T. & Gish. G. SH2 and SID domains: from structure to function. Cell 71, 359-362 (1992). 153. Pendergrass, T.W., Foulkes, M.A., Robison, L.L. & Nesbit, M.E. Stature and Ewing's sarcoma in childhood. Am JP ed Hem One 6, 33-39 (1984). 154. Petermann, R. et al. Oncogenic EWS-Flil interacts with hsRPB7, a subunit of human RNA polymerase II. Oncogene 17,603-610 (1998). 155. Plougastel, B., Zucman, J., Peter, M., Thomas, G. & Delattre, O. Genomic structure of the EWS gene and its relationship to EWSR1, a site o f tumor- associated chromosome translocation. Genomics 18,609-615 (1993). 156. Polednak, A.P. Primary bone cancer incidence in Black and White residents ofNew York State. Cancer 55, 2883-2888 (1985). 157. Prasad, D.D.K., Rao, V.N. & Reddy, S.P. Structure and expression of human Fli-1 gene. Can Res 52, 5833-5837 (1992). 158. Pui, C.-H., Dodge, R.K., George, S.L. & Green, A.A. Height at diagnosis of malignancies. Arch Dis Child 62, 495-499 (1987). 159. Rabinowitz, D. & Land, N. A unified approach to adjusting association tests for population admixture with arbitrary pedigree structure and arbitrary missing marker information. Hum Hered 50, 211-223 (2000). 210 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160. Radig, K. et al. p53 and ras mutations in Ewing's sarcoma. Pathol Res Pract 194, 157-162 (1998). 161. Rao, V.N., Ohno, T., Prasad, D.D.K., Bhattacharya, G. & Reddy, E.S.P. Analysis o f the DNA-binding and transcritional activation functions of human Fli-1 protein. Oncogene 8,2167-2173 (1993). 162. Retch, D.E. et al. Linkage disequilibrium in the human genome. Nature 411, 199-204 (2001). 163. Rosen, C J. et al. Association between serum insulin growth factor-I (IGF-I) and a simple sequence repeat in IGF-I gene: implications for genetic studies of bone mineral density. J Clin Endo Metab 83, 2286-2290 (1998). 164. Ross, J.A. et al. Seasonal variations in the diagnosis o f childhood cancer in the United States. Brit j Can 81, 549-553 (1999). 165. Rossow, K.L. & Janknecht, R. The Ewing's sarcoma gene product functions as atranscritptional activator. Can Res 61, 2690-2695 (2001). 166. Sanchez-Prieto, R. et al. An association between viral genes and human oncogenic alterations: the adenovirus El A induces the Ewing tumor fusion transcript EWS-FLI1. Nat Med 5,1076-1079 (1999). 167. Schaid, D.J. & Li, H. Genotype relative-risks and association tests for nuclear families with missing parental data. Genet Epidemiology 1 4 ,1113-1118 (1997). 168. Schaid, D.J. & Rowland, C. Use of parents, sibs, and unrelated controls for detection of associations between genetic markers and disease. Am JH um Genet 63, 1492-1506 (1998). 169. Schaid, D.J. & Sommer, S.S. Comparison of statistics for candidate-gene association studies using cases and parents. Am JH um Genet 55,402-409 (1994). 170. Schaid, D.J. & Sommer, S.S. Genotype relative risks: methods for design and analysis of candidate-gene associations. Am JH um Genet 53, 1114-1126 (1993). 171. Schaid, D.J. General score tests for associations o f genetic markers with disease using cases and their parents. Genet Epi 13,423-449 (1996). 211 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172. Schaid, D J. Likelihoods and the TDT for the case-parents design. Genet Epidemiology 16,250-260 (1999). 173. Schaid, D.J. Transmission disequilibrium, family controls, and great expectations. Am JH um Genet 63,935-941 (1998). 174. Scotland!, K. et a l Blockage o f insulin-like growth factor-I receptor inhibits the growth of Ewing's sarcoma in athymic mice. Can Res 58, 4127-4131 (1998). 175. Scotlandi, K. et al. Insulin-like growth factor I receptor-mediated circuit in Ewing's sarcoma/peripheral neuroectodermal iumor: a possible therapeutic target Can Res 56,4570-4574 (1996). 176. Self, S.G., Longton, G., Kopecky, K.J. & Liang, K.Y. On estimating HL A/disease association with application to a study of aplastic anemia. Biometrics 47, 53-61 (1991). 177. Selleri, L. et al. Cloning of the entire FLI1 gene, disrupted by the Ewing's sarcoma translocation breakpoint on 1 lq24, in a yeast artificial chromosome. Cytogenet Cell Genet 67, 129-136 (1994). 178. Siddique, H.R., Rao, V.N., Lee, L. & Reddy, E.S.P. Characterization of the DNA binding and transcriptional activation domains of the erg protein. Oncogene 8, 1751-1755 (1993). 179. Siegmund, K.D., Langholtz, B., Kraft, P. & Thomas, D.C. Testing linkage disequilibrium in sibships. Am JH um Genet 67, 244-248 (2000). 180. Sorensen, P.H. et al. Biphenotypic sarcomas with myogenic and neural differentiation express the Ewing's sarcoma EWS/FLI1 fusion gene. Can Res 55, 1385-1392(1995). 181. Sorensen, P.H.B. et aL A second Ewing's sarcoma translocation, t(21;22), fuses the EWS gene to another ETS-family transcription factor, ERG. Nat Genet 6,146-151 (1994). 182. Sorensen, P.H.B. et aL Olfactory neuroblastoma is a peripheral primitive neuroectodermal tumor related to Ewing's sarcoma. Proc Natl Acad Sci 93, 1038-1043 (1996). 183. Spielman, R.S. & Ewens, W.J. A sibship test for linkage in the presence of association: the sib transmission/disequilibrium test. Am JH um Genet 62, 450-458 (1998). 212 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 184. Spielman, R.S., McGinnis, R.E. & Ewens, W.J. Transmission test for linkage disequilibrium: the insulin gene region and insulin-dependent diabetes mellitus (IDDM). Am JH um Genet 52, 506-516 (1993). 185. Stahl, U. et al. Ras protoonkogen aktivierung in tumoren der Ewing-gruppe. Verh Dtsch Ges Path 81, 571 (1997). 186. Stenman, G., Kindblom, L.-G. & Angervall, L. Reciprocal translocatin t(12;22)(ql3;ql3) in clear-cell sarcoma of tendons and aponeuroses. Genes, Chrom & Cancer 4 , 122-127 (1992). 187. Stoflel, M. et al. Human insulin receptor substrate-1 gene (IRSl)xhromosotnal localization to 2q35-q36.1 and identification of a simple tandem repeat DNA polymorphism. Diabetologia 36, 335-337 (1993). 188. Stolow, D.T. & Haynes, S.R. Cabeza, a Drosophilia gene encoding a novel RNA binding protein, shares homology with EWS and TLS, two genes involved in human sarcoma formation. Nucl Acid Res 23, 835-843 (1995). 189. Sugimoto, T., Umezawa, A. & Hata, J. Neurogenic potential of Ewing's sarcoma cells. Virchows Arch 430, 41-46 (1997). 190. Sun, F., Flanders, W.D., Yang, Q. & Khoury, M.J. Transmission disequilibrium test (TDT) when only one parent is available: the 1-TDT. Am J Epidemiology 150, 97-104 (1999). 191. Sweet, D. & Hildebrand, D. Recovery o f DNA from human teeth by cryogenic grinding. J Forensic Sci 43, 1199-1202 (1998). 192. Taylor, J.A. et al. Association of prostate cancer with vitamin D receptor gene polymorphism. Cancer Res 56, 4108-4110 (1996). 193. Tayton, K.J.J. Ewing's sarcoma at the site of a metal plate. Cancer 45,413- 415 (1980). 194. Thiele, C. Pediatric peripheral neuroectodermal tumors, oncogenes, and differentiation. Can Invest 8, 629-639 (1990). 195. Thompson, A.D. et aL EAT-2 is a novel SH2 domain containing protein that is up regulated by Ewing's sarcoma EWS/FLI1 fusion gene. Oncogene 13, 2649-2658 (1996). 213 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 196. Thomson, T.M., Pellicer, A. & Greene, L.A. Functional receptors for nerve growth factor in Ewing's sarcoma and Wilm's tumor cells. J o f Cell Phys 141, 60-64 (1989). 197. Thomer, P. et al. Is the EWS/FLI1 fusion transcript specific for Ewing sarcoma and peripheral primitive neuroectodermal tumor? Am J Path 148, 1125-1138 (1996). 198. Tilly, H., Bastard, C., Chevallier, B., Halkin, E. & Monconduit, M. Chromosomal abnormalities in secondary Ewing's sarcoma. Lancet I, 812 (1984). 199. Toretsky, J.A., Kalebic, T., Blakesley, V., LeRoith, D. & Helman, L.J. The insulin-like growth factor-I receptor is required foe EWS/FLI-1 transformation o f fibroblasts. J Biol Chem 272, 30822-30827 (1997). 200. Toretsky, J.A., Thakar, M., Eskenazi, A.E. & Frantz, C.N. Phosphoinositide 3-hydroxide kinase blockade enhances apoptosis in the Ewing's sacoma family o f tumors. Can Res 59, 5745-5750 (1999). 201. Triche, T J. personal communication. (1999). 202. Troung, A.H.L. & Ben-David, Y. The role of Fli-lin normal cell function and malignant transformation. Oncogene 19, 6482-6489 (2000). 203. Turbett, G.R. & Sellner, L.N. The use of optimal cutting temperature compound can inhibit amplification by polymerase chain reaction. Diagn Mol Pathol 6, 298-303(1997). 204. Turc-Carel, C., Philip, I., Berger, M.-P., Philip, T. & Lenoir, G.M. Chromosomal translocations in Ewing's sarcoma. NEJM S, 497-498 (1983). 205. Ullrich, A., Gray, A., Berman, C. & Dull, T.J. Human beta-nerve growth factor gene sequence highly homologous to that of mouse. Nature 303, 821- 825 (1983). 206. van Valen, F. & Keck, E. Induction o f glycogenolysis in cultured Ewing's sarcoma cells by dopamine and beta-adrenergic agonists. J Can Res Clin One 114, 266-272 (1988). 207. van Valen, F., Winkelmann, W. & Jurgens, H. Type I and II insulin-like growth factor receptors and their function in human Ewing's sarcoma cells. J Can Res Clin Oncol 118,269-275 (1992). 214 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 208. van Valen, F., Winklemann, W. & Jurgens, H. Expression o f functional Y1 receptors for neuropeptide Y in human Ewing's sarcoma cell lines. J Can Res Clin Oncol 118, 529-536 (1992). 209. Veenstra, T.D., Fahnestock, M. & Kumar, R. An AP-1 site in the nerve growth factor promoter is essential for 1,25-dihydroxyvitamin D3-mediated nerve growth factor expression in osteoblasts. Biochem 37, 5988-5994 (1998). 210. Vigouroux, C. et al. Genetic exclusion of 14 candidate genes in lipoatropic diabetes using linkage analysis in 10 consanguineous families. J Clin Endo Metab 82,3438-3444 (1997). 211. Weber, J.L. & May, P.E. Abundant class o f human DNA polymorphisms which can be typed using the polymerase chain reaction. Am J Hum Genet 44, 188-196(1989). 212. Whang-Peng, J. et al. Chromosome translocation in peripherla neuroepithelioma. NEJM 311, 584-585 (1984). 213. Whang-Peng, J., Freter, C.E., Knutsen, T., Nanfro, J.J. & Gazdar, A. Translocation t(l 1;22) in esthesioneuroblastoma. Can Genet Cyto 29, 155- 157(1987). 214. Winn, D.M. et al. A case-control study of the etiology of Ewing’ s sarcoma. Can Epi Bio Prev 1, 525-532 (1992). 215. Xia, Z., Dickens, M., Raingeaud, J., Davis, R.J. & Greenberg, M.E. Opposing effects fo ERK and JNK-p38 MAP kinases on apoptosis. Science 270, 1326- 1331 (1995). 216. Yao, R. & Cooper, G.M. Requirement for phosphatidylinositol-3 kinase in the prevention o f apoptosis by nerve growth factor. Science 267,2003-2006 (1995). 217. Yee, D. et al. Analysis of IGF-I gene expression in malignancy, evidence for a paracrine role in human breast cancer. Mol Endocrinol 3,509-517 (1989). 218. Yee, D. et aL Insulin-like growth factor I expression by tumors of neuroectodermal origin with the t(l 1:22) chromosomal translocation: a potent autocrine growth factor. JC lin Invest 86,1806-1814 (1990). 215 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 219. Yi, H. et al. Inhibition of apoptosis by normal and aberrant Fli-1 and erg proteins involved in human solid tumors and leukemias. Oncogene 14, 1259- 1268 (1997). 220. Yuan, Y.P., Schultz, J., Mlodzik, M. & Bork, P. Secreted fringe-like signaling molecules may be glycosyl transferases. Cell 88, 9-11 (1997). 221. Zackai, E.H. & Emanuel, B.S. Site-specific reciprocal translocation, t(l 1 ;22), in several unrelated families with 3 :1 meiotic disjunction. Am J Med Genet 7, 507-521 (1980). 222. Zamora, P. et al. Ewing's tumor in brothers. Am J Clin Oncol 9 , 358-360 (1986). 223. Ziegler, R.G. et al. Migration patterns and breast cancer risk in Asian- American women. JN a tl Can Inst 85, 1819-1827 (1993). 224. Zoubek, A. et al. Does expression o f different EWS chimeric transcripts define clinically distinct risk groups of Ewing tumor patients? J Clin Ocol 14, 1245-1251 (1996). 225. Zoubek, A. et al. Variablility of EWS-chimeric transcripts in Ewing tumours: A comparison of clinical and molecular data. B r J Can 70,908-913 (1994). 226. Zucman, J. et al. Cloning and characterization of the Ewing's sarcoma and peripheral neuroepithelioma t(l 1;22) translocation breakpoints. Gen Chrom Can 5, 271-277 (1992). 227. Zucman, J. et al. Combinatorial generation of variable fusion proteins in Ewing family tumors. E M B O J12, 4481-4487 (1993). 228. Zucman, J. et al. EWS and ATF-1 gene fusion induced by t(12;22) translocation in malignant melanoma o f soft parts. Nature Genetics 4, 341- 345 (1993). 229. Zucman-Rossi, J., Batzer, M.A., Stoneking, M., Delatrre, O. & Thomas, G. Interethnic polymorphism o f EWS intron 6: genome plasticity mediated by Alu retroposition and recombination. Hum Genet 9 9 , 357-363 (1997). 230. Zucman-Rossi, J., Legoix, P., Victor, J.-M., Lopez, B. & Thomas, G. Chromosome translocation based on illegitimate recombination in human tumors. Proc Natl Acad Sci 9 5 , 11786-11791 (1998). 216 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Cluster analysis of p53 mutational spectra
PDF
CYP17 polymorphism and risk for colorectal adenomas
PDF
Association between body mass and benign prostatic hyperplasia in Hispanics: Role of steroid 5-alpha reductase type 2 (SRD5A2) gene
PDF
Colorectal cancer risks in Singapore Chinese: Polymorphisms in the insulin-like growth factor-1 and the vitamin D receptor
PDF
Biopsychosocial factors in major depressive disorder
PDF
Effect of genetic factors in the development of childhood lymphocytic leukemia (ALL)
PDF
beta3-adrenergic receptor gene Trp64Arg polymorphism and obesity-related characteristics among African American women with breast cancer: An analysis of USC HEAL Study
PDF
Assessment of fatigue as a late effect of therapy among survivors of childhood leukemia
PDF
Genetic risk factors in breast cancer susceptibility: The multiethnic cohort
PDF
Extent, prevalence and progression of coronary calcium in four ethnic groups
PDF
A case-control study of passive smoking and bladder cancer risk in Los Angeles
PDF
Androgens and breast cancer
PDF
Association of vitamin D receptor gene polymorphisms with colorectal adenoma
PDF
Comparisons of metabolic factors among gestational diabetes mellitus probands, siblings and cousins
PDF
BRCA1 mutations and polymorphisms in African American women with a family history of breast cancer identified through high throughput sequencing
PDF
Association between latchkey status and smoking behavior in middle school children
PDF
A descriptive analysis of medication use by asthmatics in the Children's Health Study, 1993
PDF
Family history, hormone replacement therapy and breast cancer risk on Hispanic and non-Hispanic women, The New Mexico Women's Health Study
PDF
A linear model for measurement errors in oligonucleotide microarray experiment
PDF
Dietary fiber intake and atherosclerosis progression: The Los Angeles Atherosclerosis Study
Asset Metadata
Creator
Dayton, Jeffrey Paul
(author)
Core Title
A case/parental/sibling control study of Ewing's sarcoma/peripheral primitive neuroectodermal tumor (pPNET)
School
Graduate School
Degree
Doctor of Philosophy
Degree Program
Epidemiology
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
biology, biostatistics,health sciences, oncology,health sciences, public health,OAI-PMH Harvest
Language
English
Contributor
Digitized by ProQuest
(provenance)
Advisor
Buckley, Jonathan D. (
committee chair
), [illegible] (
committee member
), Haile, Robert W. (
committee member
)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c16-216930
Unique identifier
UC11339210
Identifier
3073769.pdf (filename),usctheses-c16-216930 (legacy record id)
Legacy Identifier
3073769.pdf
Dmrecord
216930
Document Type
Dissertation
Rights
Dayton, Jeffrey Paul
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA
Tags
biology, biostatistics
health sciences, oncology
health sciences, public health