University of Southern California Dissertations and Theses

Randomized comparison of oral mercaptopurine versus oral thioguanine for standard risk acute lymphoblastic leukemia

(USC Thesis Other)

Randomized comparison of oral mercaptopurine versus oral thioguanine for standard risk acute lymphoblastic leukemia

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content RANDOMIZED COMPARISON OF ORAL MERCAPTOPURINE
VERSUS ORAL THIOGUANINE FOR STANDARD RISK
ACUTE LYMPHOBLASTIC LEUKEMIA
Copyright 2004
by
Hongfang Dong
A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(APPLIED BIOSTATISTICS/EPIDEMIOLOGY)
December 2004
Hongfang Dong
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
UMI Number: 1424243
INFORMATION TO USERS
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 bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
UMI
UMI Microform 1424243
Copyright 2005 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
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
ACKNOWLEDGMENTS
I would like to express my sincerest appreciation to Dr. Harland Sather, chairman
of my thesis committee, for his guidance and support throughout the development of
this thesis. Thanks also go to Dr. Susan Groshen and Dr. Todd Alonzo for their
serving on my thesis committee and most importantly, for their valuable suggestions.
Also thank CCG-1952 study group for providing the dataset.
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
TABLE OF CONTENTS
Acknowledgments ii
List of Tables iv
List of figures v
Abstract vi
Introduction 1
Methods 9
Results 19
Discussion 33
Reference 39
Appendix 43
iii
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
LIST OF TABLES
1. Demographic characteristics of study subjects (n=2027) 19
2. Life table analysis of VODLS incidence for TG patients (n=1017) 21
3. Univariate Cox regression analysis for risk of event in the whole
cohort (n=2027) 27
4. Univariate Cox regression analysis for risk of event in TG
patients (n=1017) 30
5. Multivariate Cox regression analysis for risk of event in TG
patients (n= 1017) 31
6. Cox regression analysis of treatment effect comparisons (n=2027) 32
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
LIST OF FIGURES
1. Incidence of VODLS on patients with TG treatment (n=1017) 22
2. Incidence of VODLS Comparison of Gender (n= 1017) 22
3. Incidence of VODLS Comparison of Cohort (n=1017) 23
4. Incidence of VODLS Comparison of Gender: Cohort 1
(Randomized before 01/23/98) (n=440) 23
5. Incidence of VODLS Comparison of Cohort: Males (n=577) 24
6. Incidence of VODLS Comparison of CNS Disease status at
diagnosis (n=1017) 24
Al. Metabolic process of MP and TG converting into thioguanine
nucleotide 43
A2. Treatment Schema 44
A3. CCG-1952 Event-Free Survival from Randomization Comparison
of Thiopurine Regimens 45
A4. CCG-1952 Event-Free Survival from Randomization Comparison
of CNS Prophylaxis Regimens 46
A5. CCG-1952 Event-Free Survival from Randomization Comparison
of Thiopurine Regimens: Cohort 1 (Randomized before 01/23/98) 47
A6. CCG-1952 Event-Free Survival from Randomization Comparison
of Thiopurine Regimens:Cohort 2 (Randomized after 01/23/98) 48
V
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
ABSTRACT
The CCG-1952 study is a large, randomized clinical trial to treat children with
newly diagnosed standard risk acute lymphoblastic leukemia (ALL). One of the main
aims of this study is to find out whether treatment with oral 6-thioguanine (TG) in
ALL patients improves event free survival compared to treatment with oral
6-mercaptopurine (MP). Cox regression analyses on 2027 randomized patients with
cumulative TG dose as a time-dependent variable revealed that, patients who received
higher cumulative TG dose (above median) had only half the risk of MP patients
(HR=0.54; 95%CI=0.38-0.77). Despite the better effect of TG therapy over MP
therapy, 206 of 1017 TG patients (20%) experienced episode of hepatic
veno-occlusive disease-like syndrome during the therapy although most toxicities
were mild and reversible. Thus, further research is needed to identify those who could
receive TG therapy safely and derive the greatest benefit in terms of improved
event-free survival.
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
INTRODUCTION
Acute lymphoblastic leukemia (ALL) is the most common form of childhood
cancer, representing one fourth of all pediatric cancers (Ries et al, 2003). In the United
States, about 2,400 children and adolescents below 20 years o f age are diagnosed with
ALL each year (Smith et al, 1999). ALL is characterized by the emergence of a vast
amount of abnormally immature white blood cells in the blood and bone marrow
(Sawyers et al, 1991). It has been noticed that boys are slightly more likely to be affected
by ALL than girls (Brincker, 1982) and the whites are more susceptible to ALL than the
African-Americans (Brincker, 1982; Sandler, 1992). Few factors have been identified to
be associated with increased risk of developing ALL to this day (Pullen et al, 1999). To a
large extent, there is still little understanding of the etiology of ALL.
Despite the unknown causes, treatment strategies of combining chemotherapy or
intrathecal chemotherapy have successfully increased five-year event free survival (EFS)
to 75% - 80% of children with ALL (Pui et al, 1994; Schrappe et al, 2000; Gaynon et al,
2000; Harms et al, 2000). However, numerous important therapeutic issues still remain to
be studied for a significant further improvement in treatment outcome. The systematic
method of studying these issues and identifying excellent curative therapy requires large,
randomized clinical trials that are carefully designed (Bleyer, 1997).
The CCG-1952 study is a large clinical trial carried out by the Children’s Oncology
Group to treat children with newly diagnosed “standard risk” (SR) acute lymphoblastic
leukemia (ALL). Standard risk was defined as ages 1 through 9 years, inclusive, and
white blood cell (WBC) less than 50,000/ul, using the criteria of the Risk Classification
l
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Workshop held in 1993 (Smith et al, 1996). The study was designed as a 2x2 factorial
experiment to compare event-free survival (EFS), freedom from central nervous system
(CNS) relapse, and overall survival of patients treated with either mercaptopurine (MP)
or thioguanine (TG) as the maintenance therapy and of patients treated with either
intrathecal methotrexate (ITM) or IT “triples” (ITT) of methotrexate, cytarabine and
hydrocortisone for presymptomatic CNS therapy.
One of the main aims of this study is to find out whether treatment with oral 6-
thioguanine (TG) in ALL patients improves event free survival (EFS) compared to
treatment with oral 6-mercaptopurine (MP). Investigators of this study hypothesized a
possible 7.5% higher 4-year event free survival achieved by treatment with oral TG than
that achieved by treatment with oral MP. The rationale underlying this hypothesis is
grounded in potential therapeutic benefits of TG over MP in the treatment of ALL
suggested by both theories and laboratory studies.
MP and TG were both synthesized in the early 1950's. MP was approved for use
in clinics in 1953 because of its easy preparation, while research into TG continued for 10
more years before its use in clinics (Zimm, 1988; Elion et al, 1962). In the 1950’s, the
application of MP in patients with ALL raised one-year survival from 29% to 52%. No
particular investigation was done on TG except that a cooperative study conducted in the
1960’s indicated an equal remission rate for TG and MP in childhood ALL (Brubaker et
al., 1964). Hence, the fact that MP forms the basis of the continuing therapy o f childhood
ALL across the world has resulted from historical factors, not based on a clear advantage
over TG.
2
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
TG and MP are closely related, both in structure and in metabolism. The thio
compound of these two DNA antimetabolites becomes cytotoxic only after its conversion
into a nucleotide. Thioguanine nucleotides (TGNs) generate cytotoxicity mainly by two
mechanisms: (1). being incorporated into DNA and RNA as thioguanosine, which
prevent the resulting nucleic acids from directing proper protein synthesis; (2). competing
for enzyme with naturally occurring purine base, which in turn inhibits purine
biosynthesis (Zimm, 1988; Nelson et al, 1975; Bostrom et al, 1993). TG is converted
into an active form of 6-thioguanylate (TGMP) in a more straight-forward way than MP,
with only one enzymatic step in contrast with three enzymatic steps of MP. (See
Appendix A l reproduced from the CCG-1952 study protocol).
Parts of the intermediates produced during MP conversion are susceptible to
thiopurine methyltransferase. This methylation, in turn, could decrease the amount of
TGNs derived from MP. The more TGNs are incorporated into DNA, the more
cytotoxicity occurs (Zimm, 1988; LePage, 1963; Nelson et al, 1975; Bostrom et al, 1993).
Therefore, simpler conversion into an active form of TGNs implies stronger cytotoxicity
of TG over MP. Results from several in vitro studies showed consistency with the above
theoretical assumptions. For instance, TG was proved to be significantly more cytotoxic
at equimolar concentrations than MP in human ALL cell lines in a study published in
1994 (Adamson et al, 1994). Some in vivo data also demonstrated that TG might have a
potential cytotoxic advantage over MP in the treatment of ALL. In a study conducted by
Lennard et al to compare the pharmacokinetics of both drugs in a small group of children
with ALL, the investigators found “At 7 days, 6-TGN concentrations ranged from 959 to
3
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
2361 pmol 8 x 10(-8) RBCs, in all cases significantly higher (P = 0.002) than those
produced by the same patients on 6-MP.” (Lennard et al, 1993).
The CCG-1952 study was started in May, 1996, and was closed to patient entry
on February 1, 2000 after meeting accrual goals. During this period, a total of 2185
patients were recruited into the study. Some patients with unfavorable cytogenetic
features (t(4 ;ll) and t(9;22) translocations; hypodiploidy < 44 chromosomes) or with
poor response at day 14 of standard induction phase were not eligible to be randomized,
and were treated on a more intensive chemotherapy regimen than the basic treatments to
be studied in the 2x2 design. In addition, since residual leukemic blasts in bone marrow
(BM) aspirates have been shown to have significant effect on event free survival, only
those subjects whose day 28 BM aspirate was M l status (<5% blast in a BM aspirate) at
the end of standard induction phase were eligible for randomization. By these criteria,
2027 eligible patients were randomized to one of four treatment regimens for the purpose
of testing the above mentioned hypothesis and other research hypotheses:
1. Regimen A l: oral MP and IT MTX, 509 patients;
2. Regimen A2: oral MP and ITT, 501 patients;
3. Regimen B 1: oral TG and IT MTX, 509 patients;
4. Regimen B2: oral TG and ITT, 508 patients.
The study schema reproduced from the CCG-1952 study protocol is given in Appendix
A2. There were 94 non-randomized patients who were assigned to a more intensive
regimen called Augmented BFM and denoted in the schema as Regimen C.
4
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
After the study was initiated, it was gradually noted that a “hepatic veno-
occlusive disease-like syndrome (VODLS)” was found to develop in some patients
receiving TG regimens. This was characterized by acute onset congestive hepatopathy
and consumptive thrombocytopenia, but often was a mild and reversible form of VOD,
and. VODLS patients usually resumed chemotherapy after their recovery, but then had
MP substituted, rather than TG, as the thiopurine. VODLS did not recur in any patient
switched to the use of MP. From May 1, 1996 through January 23, 1998, about 30
patients had VODLS in those receiving TG. While most of these toxicities were transient
episodes that could be managed effectively by switching to MP without long lasting
effects, there were a small percentage of patients with more serious complications. The
2 2
original 100% protocol dose of TG was then reduced from 60mg/M /d to 50 mg/M /d by
an urgent advisory issued on January 23, 1998. Following this dose reduction, the study
chair (Dr. Linda Stork) reported that while episodes of VODLS continued to occur, they
were generally milder and the incidence decreased by about 1/3. Because occasional
severe effects still occurred in certain patients, the study was modified again on April 27,
2001 by having all patients on the TG regimen discontinue TG and substitute MP on or
before June 1, 2001. No episode of VODLS has occurred since then.
Among the VODLS patients, about half of them required hospitalization for
several days, 40% required platelets and 60% received peripheral RBC transfusions
(pRBCs). In addition, one patient spent more than 3 weeks in the ICU with VODLS and
other problems before his full recovery, and three patients underwent acute hepatic
failure. Although two of them recovered soon, one patient died of presumed post
5
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
transplant lymphoproliferative disorder more than 1 year after a successful liver
transplant. However, there has been no significant liver toxicity developed on treatment
following resolution o f VODLS and no fatalities directly attributable to TG.
The investigators did not expect such a high proportion (20%) o f VODLS among
patients taking TG because VODLS was not considered to be a common side effect of
TG and excessive toxicity of this type had never occurred in previous studies. Neither
did they expect the severity of VODLS (about 4% of the patients taking TG had severe
VODLS), as very few reports, and none in children, had mentioned it in previous studies.
Basic life table analyses of event free survival (EFS) outcome of TG versus MP and
of ITM versus ITT have been done during the study. According to a recent update of the
study, there were 140 events in 1017 TG patients and 188 events in 1010 MP patients. The
EFS comparison shows a highly significant difference (p=0.004) with a relative hazard rate
of 0.74 for TG compared to MP (Appendix A3). The EFS in these two different treatment
regimens was approximately the same for the first year, then TG gradually showed more
treatment benefit relative to MP, with most of the difference occurring after 3 years of
follow-up. But we need to note that during the study, about 1/3 of the patients on TG
regimens switched to MP regimens because of VODLS or as a result of study amendments
changing the dosage approach. For the comparison of CNS prophylaxis regimens, no
significant difference in EFS outcome occurred between ITM and ITT (Appendix A4).
For one analysis of interest, the 1017 TG patients were also divided into 2 cohorts
based on the initial daily TG target dose at the time the patient first entered the study: cohort
1 (56.7% of the total TG patients) at 60 mg/M2 and cohort 2 (43.3% of the total TG patients)
6
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
at 50 mg/M2. EFS comparison of patients in TG cohort 1 versus patients on MP regimens
enrolled during the same time period shows a significant benefit of TG over MP (p=0.005,
RHR=0.63) (Appendix A5), yet such benefit could not be found in TG cohort 2 when
compared to patients in MP regimens enrolled during the same time period (p=0.20,
RHR=0.82) (Appendix A6). When the analyses were stratified by gender, the estimated 6-
year EFS is significantly higher among boys on TG over that among boys on MP in both
cohorts (cohortl: p=0.01, RHR=0.60; cohort2: p=0.04, RHR=0.67), while it is not
significantly different between girls on TG or MP in either cohorts (cohortl: p=0.21,
RHR=0.70; cohort2: p=0.77, RHR=T.07). This gender effect is of interest because the
duration of therapy for males in this clinical trial is one year longer than for females, and
during this extra year period of maintenance therapy the randomized thiopurine (TG or MP)
continues to be received. Typically in childhood ALL, the EFS outcome for females is
eventually better than that for males. In this study the females do have a better outcome, but
only by about 5% at 5 years (79% versus 84%, log rank p = 0.01).
What has been analyzed thus far was based only on original assigned daily TG and
MP dose intake at the time of study entry. However, the actual cumulative TG doses that
patients received varies greatly, depending on when they enrolled into the study, whether
they were male or female, and whether they required dose reductions based on blood counts.
The study began in May 1996, and the last patients accmed onto the study completed
therapy in approximately either April 2002 (females) or April 2003 (males). As one
example, a TG patient who entered the study before December 26, 1997 (thus, he/she was
randomized before January 23, 1998) was originally scheduled to receive a 100% TG dose
7
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
as 60mg/M2 /d, which was then reduced to 50mg/M2 /d after January 23, 1998 if he/she was
still on study receiving therapy. In another case, a TG patient who entered the study after
December 26, 1997 was scheduled initially to receive a 100% TG dose o f 50mg/M2 /d. If
this patient was still receiving protocol therapy after April 27, 2001, the treatment
switched to MP from TG. Furthermore, boys received one more year therapy in
maintenance phase than girls per the protocol treatment plan, and patients who had an
episode of VODLS during the study were switched from TG to MP at that time.
Since the cumulative TG dose received by ALL patients changes in value over the
whole study, it might be more appropriate statistically to treat it as a time dependent
variable in survival analysis. As a result, the statistical analysis in the present paper will
focus on using Cox regression to assess the predictive influence of cumulative TG dose
for treatment regimens, with adjustment for gender, WBC and other prognostic factors.
This master’s thesis is aimed at developing a better answer to the question
regarding which patients had the most and least benefit to the use o f TG, and better
clarify its role in this study. Some of these results may be helpful in addressing issues
related to the possible use of TG in the design o f future studies in childhood ALL.
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
METHODS
Subjects
The subjects in this study were newly diagnosed standard risk ALL patients, who
did not receive prior treatment (Stork et al, 2001). The diagnosis of ALL was made by
bone marrow aspiration or bone morrow biopsy. ALL was restricted to bone marrow
infiltrated with more than 25% LI or L2 lymphoblasts, according to the French-
American-British (FAB) Cooperative Group Proposal for the Classification of Acute
Leukemia (Bennett et al, 1976). Patients with massive lymphadenopathy, massive
splenomegaly, and/or large mediastinal mass at diagnosis were eligible for this study.
Patients with CNS or testicular leukemia at diagnosis also qualified for the study (Stork et
al, 2001).
All the eligible patients received Standard Induction therapy initially. On the basis
of cytogenetic findings and their day 7, day 14 and day 28 bone marrow (BM) status
response to Induction therapy, patients were assigned to either randomized regimens or
non-randomized intensive therapy. The analysis in this paper will be limited to the data of
those assigned to randomized regimens. These subjects included 2027 ALL patients
whose day 28 BM was M l status. They were assigned to one of the four treatment
regimens and received oral TG or oral MP as maintenance thiopurine depending on
which regimen they were in.
Baseline Evaluation o f Patients
According to the CCG-1952 study protocol, baseline evaluations were performed
on all study subjects. These included demographic data collection and physical
9
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
examination (lymph node size, spleen size, liver size, testes size, height, weight, and
body surface area), and chest X-ray examination. Bone marrow (BM) aspiration or
biopsy samples were also obtained for BM studies. Bone marrow status was determined
morphologically by counting the number of blasts and normal cells on aspirate smears.
M l status BM was defined as less than 5% blasts in a BM regardless of the proportion of
mature lymphocytes, M3 status as more than 25% blasts in a BM aspirate and M2 status
lying in between.
Immunophenotypic analysis of BM studies included monoclonal antibodies
against CD 19, CD 10 (B-precursor phenotype) and CD2, CD7 (T-cell phenotype). A
positive phenotype was defined as more than 30% of the blast cells staining with
antibody. Additionally, blood studies were carried out by complete blood count: initial
WBC count, initial platelet count, and initial hemoglobin. Initial WBC was the first WBC
count at the local CCG institution, no matter how the count changed after hydration or
transfusion.
Regarding VOD-like syndrome, episodes of hepatomegaly and/or ascites and/or
VOD were all required to be reported. Any of the three conditions was considered as an
episode of VODLS.
TG Treatment fo r patients randomized to TG regimens
As indicated in the CCG-1952 study protocol, there were 7 phases o f the treatment plan:
1. Induction (28 days)
No thiopurine therapy was given in this phase.
10
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
2. Consolidation (28 days)
Consolidation began at the 28th day of Induction, or when peripheral counts
recovered with ANC > 1000/ul and platelet count > 100,000/ul, whichever occurred last.
From day 1 to day 27 (27 days), 60mg/M2 /day TG was given before the January 23, 1998
advisory; daily oral TG dose was reduced to 50mg/M after that date.
3. Interim Maintenance (IM) #1 (56 days)
Interim Maintenance #1 began at the 28th day of Consolidation or when peripheral
counts recovered with ANC > 1000/ul and platelet count >100,000/ul, whichever
'j
occurred last. From day 0 to day 49 (50 days), 60mg/M /day TG was given before
January 23,1998; daily oral TG dose was reduced to 50mg/M2 after that date.
4. Delayed Intensification (DI) #1 (56 days)
Delayed Intensification #1 began at the 56th day of IM #1, or when peripheral counts
recovered with ANC > 1000/ul and platelet count > 100,000/ul, whichever occurred last.
• y
From day 28 to day 41, patients received 60mg/M /day TG for 14 days. The 100% dose
was not reduced after January 23, 1998 because the 60mg/M daily TG dose has been a
traditional dosage used in this phase in past studies without toxicity problems.
5. Interim Maintenance (IM) #2 (56 days)
Interim Maintenance #2 began at the 56th day of DT #1 or when peripheral counts
recovered with ANC > 1000/ul and platelet count > 100,000/ul, whichever occurred
last. From day 0 to day 49 (50 days), 60mg/M2 /day TG was given before January 23,
1998; daily oral TG dose was reduced to 50mg/M2 after that date.
1 1
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
6. Delayed Intensification (DI) #2 (56 days)
Delayed Intensification #2 began at the 56th day of IM #2, or when peripheral counts
recovered with ANC > 1000/ul and platelet count > 100,000/ul, whichever occurred last.
From day 28 to day 41, patients received 60mg/M2/day TG for 14 days. Same as in DI #1,
the 100% daily TG dose was not reduced after January 23,1998.
7. Maintenance
Maintenance began at the 56th day of DI #2 or when peripheral counts recovered with
ANC > 1000/ul and platelet count > 100,000/ul, whichever occurred last. Measured from
the beginning of IM #1, the duration of therapy was 2 years for girls and 3 years for boys.
Eight-four days was the length of a course in the maintenance phase. When the end of
therapy was reached, the current course was stopped no matter whether the course was
completed or not. Daily oral TG of 60mg/M2 was given before January 23, 1998, then
was reduced to 50mg/M2 thereafter. After April 17, 2001, all the TG patients on protocol
were changed to MP.
During each of the above phases, patients with an episode of VODLS were
switched from TG to MP at that time. Throughout therapy, the target dose was also
continually adjusted, based on weekly or monthly blood counts, in an attempt to prevent
dangerous reductions in blood counts.
Statistical analyses
Incidence rates of hepatic veno-occlusive disease-like syndrome (VODLS) were
calculated using Kaplan-Meier life table methods (Kaplan and Meier, 1958). The
standard deviation of the life table estimate was calculated using Peto’s method (Breslow,
12
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
1975). Log-rank tests were applied to the test for VODLS incidence differences among
different genders group, WBC count groups and CNS status groups (Mantel and
Haenszel, 1959). In a two-group comparison, the statistic for the log rank test follows a
chi-square distribution under the null hypothesis with one degree of freedom. In order to
clarify the influence of initial daily TG target dose on VODLS incidence, the TG patients
were divided into 2 cohorts. Patients that had been randomized into TG regimens before
January 23, 1998 and received daily TG target dose of 60mg/M2 were considered to be
cohort 1. The other patients that had been randomized into TG regimens after January 23,
1998 and initially received a daily TG target dose of 50mg/M were grouped into cohort
2. The log-rank test was also employed to test for VODLS incidence differences between
these two cohorts. Relative hazard rates (RHRs) for comparisons were obtained by the
following formula:
OH El
O ltEl
where O j is the observed number of events, and E j is the expected number o f eventsin
group j calculated assuming Ho.
As mentioned in the Introduction, TG doses received by patients changed their
value during the observation period. Thus we examined total cumulative TG dose as a
time dependent variable and used Cox proportional hazards regression to assess its
predictive influence for treatment regimens.
Cox regression permits a multivariate comparison of life table outcome (Cox,
1972). The Cox model incorporates the covariate vector z=(zi, Z 2, ..., Zk), together with
the response variable of interest T in the following model structure:
13
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
A (t I z) = A0 (t) exp /J z
where A0 (t) is an unspecified baseline hazard function. Hence, the hypothesis testing
procedure allows for a completely general baseline hazard rate. Hypothesis tests for a
candidate prognostic factor typically examine the relative risk of the independent
variables (or covariates), and the baseline hazard function becomes a nuisance function
that cancels out for the purpose of the hypothesis test. The procedure uses a Partial
Likelihood estimation method:
The conditional probability that a subject j has an event at time t{, given the subjects at
risk set, R, at that time is:
L = /U Q expQffz) = exp (j?z)
3 Z^oCOexpC^z) ^ e x p (^ z )'
je R je R
J
The partial likelihood is a product of the likelihood for all the events: PL = Y \ L j , where
;=i
J is the total number of distinct event risk sets and Lj is the likelihood contribution for
each risk set. This results in
J 6X 0 ^ 6^ zl
PL = — t— v Using an iterative estimation procedure to maximize this function
7=1 / JeXP\PZ)
js R
with respect to/?, we obtain the maximum likelihood estimates of the /? parameters. By
exponentiating the term /?(Z j2-Zji), we then get the maximum likelihood estimate of the
hazard ratio for two levels of a particular covariate, Z j. To test the hypothesis that the
hazard ratio(s) equals one, we can obtain the usual likelihood ratio test statistic by
subtracting the log likelihood with the covariate(s) included from the one without
14
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
covariate(s), and multiply it by -2 (e.g. -2(LnL(0)-LnL(/?))). The statistic asymptotically
follows a chi-square distribution under the null hypothesis.
One o f the advantages of Cox regression is that it can incorporate time dependent
covariates. When time-dependent variables exist, Cox model can be extended as
A (t | z(t)) = T0 (t) exp p z (t) .
One or more o f the covariates in vector z could be time-dependent. The coefficient
estimating procedure is still based on the partial likelihood function.
There are two major types of time-dependent covariates in Cox regression:
external and internal covariates (Kalbfleisch and Prentice, 1980). An external covariate is
usually involved with study or environmental factors. Its value taken in time does not
necessarily require subjects to be observed at that particular time, and the value usually is
applicable to all subjects. The ozone (O3) and nitrogen dioxide (NO2) levels in
atmosphere at different time in certain communities where the observed subjects live is
an example of external covariate. On the other hand, an internal covariate is one that
requires subjects to be under investigation up to that time and its value is subject-specific.
For instance, the cumulative dosage of TG received by individaul patients in this study at
any time, t, is an internal time-dependent variable.
The approximate cumulative dose of TG was calculated at the end of every
therapy phase for each patient, using different candidate dosage values for non-VODLS
• • 9
patients and VODLS patients. The theoretical maximum (based on 60 or 50 mg/M ) was
used for non-VODLS patients if they had received the actual protocol schedule. The
candidate dose used for VODLS patients was based on an average dose determined from
15
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
a sample of 75 VODLS patients. The cumulative TG doses that VODLS patients received
during Consolidation, Interim Maintenance #1, Delayed Intensification #1 and Delayed
Intensification #2 phases were quite close to the target doses, but the doses received in
Interim Maintenance #2 and Maintenance phases were only about 88% and 70% of the
target doses, respectively. One of the causes for VODLS patients to receive reduced dose
might be the toxicity that VODLS patients were experiencing. Since precise
computerized data were not collected in this study on the actual dose received of
chemotherapy for each patient, it is necessary to use these types of projected doses
mentioned above.
In Cox regression analysis of this study, when an EFS event occurred for a
particular subject in a certain phase, the cumulative TG doses received for all patients in
the corresponding risk set at this time point were calculated from the sum of the projected
TG doses received at previously finished therapy phases, plus the projected amount they
received from the beginning of the current phase until the time the event occurred.
The primary endpoint examined in the Cox regression analysis was event-free
survival. It was determined as the interval from time of randomization (which occurred
on approximately day 27 or day 28 of the Induction phase if a patient had no treatment
delays during induction therapy) until the first significant event. An event in this study
was defined as any o f the following cases: induction death, induction failure (non
response), marrow relapse, CNS relapse, testicular relapse, “other” leukemia relapse,
second malignant neoplasm (SMN), or death in remission.
16
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Cumulative TG dose received by TG patients was tested both as a continuous
variable and as a categorical variable. As a categorical variable, it was divided both into
three groups (low, medium and high) according to its 33% and 67% percentile point, and
two groups (low, high) according to its median. Univariate Cox regression analysis was
also performed on other prognostic factors: gender, race/ethnicity, age at diagnosis, WBC
count, CNS status, liver rating, spleen rating, lymphadenopathy rating, CD7 and CD 10.
Age at diagnosis was treated as a dichotomous variable: 12-23 months or more than 71
months versus 24-71 months. WBC count was grouped into two categories: less than
20,000/ul and equal to or more than 20,000/ul. CNS status was coded as three groups: no
CNS disease, CNS disease, and CNS disease at diagnosis. Liver rating, spleen rating and
lymphadenopathy rating were grouped into two categories: normal and enlarged
(combining moderate enlarged and markedly enlarged). CD7 and CD 10 were coded as
binary variables: less than 30% cells expressing the antigen versus equal to or more than
30. All these factors were also entered into Cox regression model with cumulative TG
dose one at a time to assess their influence on the predictive effect of TG.
All the above analyses were restricted to patients receiving oral TG. In order to
assess the effect of TG with adjustment for the treatment regimens (TG vs MP), the
analysis was also performed in all the 2027 randomized patients with the treatment
regimen variable included. A log-likelihood ratio test was applied to determine whether
cumulative TG dose modified the impact of treatment regimens on event-free survival.
Throughout the analyses, a p-value smaller than 0.05 was considered statistically
significant. The SAS statistical software program LIFETEST was used for obtaining life
17
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
table estimates and performing the log rank test, and program PHREG was used for Cox
regression analysis (SAS version 8.2; SAS Institute Inc., Cary, NC, USA). The Epilog
statistical software was used to crate the event-free survival graphs (Epicenter Software,
Pasadena, CA, USA).
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
RESULTS
Descriptive Analyses
Table 1 presents some demographic characteristics of the randomized subjects. A
total of 2027 patients were randomized to either MP regimens or TG regimens.
Approximately 56 percent of them were boys and 44 percent were girls. The whites
formed the largest ethnic group, making up two-thirds of the patients, followed by
Hispanics, accounting for 22 percent. African-Americans were 3.6 percent of the
randomized patients. Only about 8 percent of the patients were younger than 2 years old,
and about 70 percent were between 2 - 6 years old. These are typical percentages that
could be expected in a clinical trial involving children with standard risk ALL.
Table 1. Demographic characteristics of study subjects (n=2027)
Variable N (% )
Sex Male 1129 (55.7%)
Female 898 (44.3%)
Race/Ethnicity White 1376 (67.9%)
Hispanic 444 (21.9%)
African-American 72 (3.6%)
Asian 29(1.4)
Other 106 (5.2%)
Age at diagnosis (months)* 12-23 165 (8.1%)
24-71 1402 (69.2%)
72-119 460 (22.7%)
Treatment regimens MP 1010(49.8%)
TG 1017 (50.2%)
* Infants less than 12 months of age at diagnosis were not eligible for entry on this study. Also,
children ten years of age or older at diagnosis were not eligible since both of these age categories
are not considered standard risk ALL.
19
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Life Table Analyses o f the Incidence o f VODLS
There were 216 patients with a first episode of VODLS in 1017 TG patients. The
earliest occurrence of VODLS was on the day that the patient was randomized, and the
latest was 1070 days after randomization. For those with an episode, median time to
occurrence was 415 days following randomization (mean = 392 days). One hundred and
fifty-six out of 216 patients (72%) had the first episode of VODLS in the maintenance
phase. In contrast, only 10 percent of the patients had the first episode o f VODLS in the
early post-randomization phase of consolidation and interim maintenance # 1.
The estimated VODLS incidence rates were 9.3%, 20.6%, 22.3% at the first,
second and third years of follow-up, respectively (Figure 1). Among these VODLS
patients, 135 were males and 81 were females. The incidence comparison showed a
marginal significant difference for females compared to males (p=0.07; relative risk=0.77
times lower in females) (Figure 2) (Table 2). After dividing TG patients into two cohorts
based on whether they were randomized before or after January 23, 1998 (the day on
which TG dose was reduced from 60mg/M2 /day to 50mg/M2 /day), we found that there
were 104 VODLS occurrences in 440 patients of cohort 1 (randomized before January 23,
1998) and 112 VODLS occurrences in 577 patients of cohort 2 (randomized after January
23, 1998). In the life table comparison of cohort 2 versus cohort 1 for incidence, the
comparison was not significantly different, although it is higher in cohort 1 (p=0.13;
relative risk=0.81) (Figure 3) (Table 2).
In cohort 1, 247 males had 70 VODLS occurrences and 193 females had 34
VODLS occurrences; and in cohort 2, altogether 330 males had 65 VODLS occurrences
20
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
and 247 females had 47 VODLS occurrences. Incidence comparison of gender in cohort
1 showed a significant difference (p=0.01) with a relative risk of 0.61 for females
compared to males (Figure 4), yet no such difference could be found in cohort 2 (p=0.80;
RR=0.96). When the analyses were stratified by gender, the incidence comparison
showed significant difference between cohort 2 and cohort 1 among boys (p=0.02;
RR=0.67) (Figure 5), but no difference among girls (p=0.71; RR=0.93). Incidence of
VODLS for patients with WBC greater than 20,000/ul was not significantly different
from those with WBC less than 20,000/ul (p=0.28). Neither did the incidence differ
among patients by CNS disease status (p=0.61) (Figure 6) (Table 2).
Table 2. Life table analysis of VODLS incidence for TG patients (n=1017)
Variable Relative Risk p-value£
Sex Male
Female 0.77 0.07
Cohort Cohortl
Cohort 2 0.81 0.13
Males Cohortl
Cohort2 0.67 0.02
Females Cohortl
Cohort2 0.93 0.71
Cohort 1 Male
Female 0.61 0.01
Cohort2 Male
Female 0.96 0.80
WBC < 20,000/ul
> 20,000/ul 0.96 0.28
CNS disease
£___i • i
No CNS
CNS disease
CNS at diagnosis
1.07
1.03 0.61
£ p-value is for log-rank test
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Figure 1. Incidence o f VODLS on patients with TG treatment (n=1017)
.50
.40
.30
.20
.10
7 4 5 6 1 2 3
Y ears from Randomization
Figure 2. Incidence of VODLS Comparison of Gender (n=1017)
.50
.40
FEMALE
MALE
.30
.20
.10
1 2 3 4 5 7 6
Years from Randomization
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Figure 3. Incidence o f VODLS Comparison o f Cohort (n=1017)
.50
.40
COHORT 2
COHORT 1
.30
.20
.10
1 2 3 4 5 6 7
Years from Randomization
Figure 4. Incidence of VODLS Comparison of Gender: Cohort 1 (Randomized before
01/23/98) (n=440)
.50
FEMALE
.40
MALE
.30
.20
.10
1 2 3 4 5 6 7
Years from Randomization
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Figure 5. Incidence o f VODLS Comparison o f Cohort: Males (n=577)
.50
COHORT 2
.40
COHORT 1
.30
.20
.10
7 3 4 5 6 1 2
Years from Randomization
Figure 6. Incidence of VODLS Comparison of CNS Disease status at diagnosis (n=1017)
.50
CNS D isease at Diag
CNS Disease
No CNS D isease
.40
.30
.20
.10
1 3 2 4 5 6 7
Y e a r s fro m R a n d o m iz a tio n
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Descriptive Analyses o f Cumulative TG Dose
The theoretical duration of therapy in this study for males was 3 years from the
start of the Interim Maintenance #1 phase, and 2 years from the start of that phase for
females. Since thiopurine therapy begins at the start of the Consolidation phase (i.e.
following Induction), the duration of TG therapy was approximately 3 years plus 28 days
for males, and 2 years plus 28 days for females. Two hundred and fourteen patients
switched from TG treatment to MP treatment because they had VODLS during
chemotherapy. Two hundred and twenty-one out of 577 males (38.3%) and 98 out of 440
(22.3%) females also terminated TG treatment and changed to MP treatment by June 1,
2001 according to the Urgent Advisory made in April, 2001.
Projected cumulative doses of TG administered to the 1017 randomized patients
ranged from 0 mg/m2 to 55810 mg/M2, with a median dose of 31630 mg/M2 (mean =
29519 mg/M ). Among 9 patients who did not receive any TG doses, 4 patients withdrew
by parental or physician choice, 3 refused to follow the randomized assignment, 1 had
VODLS at time of randomization, and 1 had significant toxicities during the induction
period and went off study soon after randomization without receiving any post-induction
CCG-1952 therapy. For the patients who received cumulative TG dose far below the
average (below the 5th percentile), 60% of them had VODLS shortly after they were
randomized. Most of the other 40% patients either had events early on the study or
withdrew from protocol therapy, resulting in a lower cumulative TG dose.
Projected median cumulative TG dose was 31230 mg/M2 (mean = 31813 mg/M2 )
for males, and was 33305 mg/M2 (mean = 26510 mg/M2 ) for females. Males received
25
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
average 5303 mg/M2 more cumulative TG dose than females because generally they have
a longer treatment duration. However, more males (38.3%) switched treatment from TG
to MP than females (22.3%) because of the urgent advisory of terminated TG treatment
towards the end o f the study for patients still receiving treatment.
Cox Proportional Hazard Regression Analyses fo r Risk o f Events
Univariate Cox proportional hazard regression analysis (without time-dependent
variable) was first performed in the whole cohort. Patients randomized to TG regimens
had significantly lower risk of event than MP patients with a relative hazard rate of 0.73
(95%CI=0.59-0.91) (Table 3). Gender, age at diagnosis, WBC, liver rating, spleen rating
and CNS disease status were highly significantly associated with EFS in the whole cohort.
Gender was an important prognostic indicator, with females having a risk only 0.75
(95%CT=0.60-0.93) times that of males. CNS disease status also was a significant
prognostic indicator in the whole cohort: a patient with CNS disease had much higher
hazard than the one without (HR=2.02; 95%CI=1.34-2.91). Race/ethnicity and CD7 were
marginally associated with EFS, while CD 10 and lymphadenopathy rating did not
statistically relate to the risk of an EFS event.
When the analysis was restricted to TG patients, the Cox regression analysis with
TG cumulative dose as a time-dependent covariate revealed that patients receiving high
cumulative TG dose (above median) had a significantly lower hazard rate than those
receiving low cumulative TG dose (below median) (HR=0.62; 95%CI=0.40-0.96; p=0.03)
(Table 4). When simplistically dividing TG patients into three groups according to the
cumulative TG doses they ultimately received, the medium group had a slightly higher
26
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
risk of developing an event compared with the low group (HR=1.15; 95%CI=0.74-1.80),
and the high group had the lowest risk (HR=0.62; 95%CI=0.36,1.08). This three group
comparison was borderline statistically significant (p=0.06).
Table 3. Univariate Cox regression analysis for risk of event in the whole cohort (n=2027)______
Variable N HR (95% Cl) p-value1
Regimen MP 1010
TG 1017 0.73 (0.59, 0.91) 0.004
Sex Male 1129
Female 898 0.75 (0.60, 0.93) 0.01
Race/Ethnicity White 699
Hispanic 223 1.28 (0.99, 1.66)
African-American 37 1.31 (0.76, 2.24)
Asian 17 1.40 (0.62, 3.15)
Other 41 0.93 (0.55, 1.58) 0.33
Age at diagnosis 12-23 or 72+ 625
(months) 24-71 1402 0.61 (0.49, 0.77) <0.0001
WBC < 20,000/ul 1660
> 20,000/ul 367 1.69 (1.32, 2.16) <0.0001
CD7 <30 1863
> 30 164 1.37 (0.96, 1.96) 0.08
CD10 <30 255
> 30 1772 0.94 (0.68, 1.30) 0.70
Liver rating Normal 1032
Enlarged 995 1.54(1.24, 1.92) 0.0001
Spleen rating Normal 1108
Enlarged 919 1.54(1.24, 1.92) 0.0001
Lymphadenopathy rating Normal 1035
Enlarged 992 1.13 (0.91, 1.40) 0.28
CNS disease No CNS 1828
CNS disease 113 2.02(1.34, 2.91)
CNS at diagnosis 30 1.26 (0.56, 2.82) 0.003
£ p-value is for likelihood ratio test
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
In the univariate Cox models performed on TG patients, the hazard was higher for
patients with enlarged liver (HR=2.08; 95%CI=1.45-2.97), enlarged spleen (HR=2.05;
95%CI=1.45-2.89), and for patients with WBC greater than 20,000/ul (HR=1.85;
95%CI=0.41-0.81) (Table 4). Age at diagnosis also had important influence upon EFS.
For patients of ages between 24-71 months at diagnosis, the hazard rate was highly
significantly lower than patients who were diagnosed before 24 months or after 71
months (HR=0.58; 95%CI=0.41-0.81). On the other hand, the risk of event was not
associated with gender, race/ethnicity, CD7, CD10, lymphadenopathy rating or CNS
disease status among TG patients (Table 4).
In the multivariate models performed on TG patients only, gender and all the
significant prognostic factors identified through univariate analysis were tested with
cumulative TG dose separately. None of them had an important modifying influence on
the effect of cumulative TG dose upon EFS (Table 5). Neither were significant
interactions between these prognostic factors and cumulative TG dose found, although
CNS disease at diagnosis achieved a moderately significant interaction effect (p=0.06).
Table 6 presents the detailed treatment comparisons among the whole randomized
cohort of TG and MP patients. At a low cumulative TG dose, TG did not show significant
benefit over MP, although the hazard for patients who received a low cumulative TG
dose was lower than patients who received MP (HR=0.85; 95%CI=0.66-1.08). At a high
cumulative TG dose, TG reduced the risk by almost half compared with MP (HR=0.54;
95%CI=0.38-0.77). When TG patients were divided into three groups, patients in low and
medium cumulative dose groups had relatively lower, but non-significant hazard rate;
28
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
while patients in the high cumulative dose group had a significantly lower risk with a
relative hazard rate o f 0.47 (95%CI=0.30-0.74) compared with patients in MP group.
These treatment effects, however, significantly differed by gender (Table 6). In
the tests for interaction between treatments and gender, likelihood ratio tests revealed
significant p values: p=0.03 and p=0.02 for MP with two-group TG patients and for MP
with three-group TG patients, respectively. In contrast with the results from the whole
cohort, low cumulative TG dose showed significant benefit over MP in males (HR=0.71;
95%CI=0.51-0.97), but not in females (HR=1.09; 95%CI=0.49-1.32). For three-group
TG patients, the risk was significantly decreased with increasing cumulative TG dose
received in males. In females, on the other hand, the risk was slightly higher in the
medium TG group than the low group, and the high TG group had a highly significant
lower risk (HR=0.22; 95%CI=0.05-0.91) (Table 6).
According to the study chair (Linda Stork, M.D.), many non-VODLS patients
also had dose reduction for oral TG because of toxicities they experienced, just as many
VODLS patients had dose reduction (i.e. cumulative TG doses received by VODLS
patients in Interim Maintenance #2 and Maintenance phases were about 88% and 70% of
the target doses, respectively). As one type of Cox regression analysis, she suggested that
we should use the same adjustment to calculate the cumulative doses received by non-
VODLS patients as we did for VODLS patients. We did the analysis again using the
adjusted dose and found that the results were similar to those summarized above (data not
shown).
29
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Table 4. Univariate Cox regression analysis for risk of event in TG patients (n=1017)
Variable N HR (95% Cl) p-value£
Cumulative TG dose Linear trend 1017 0.23
Cumulative TG dose Low 509
High 508 0.62 (0.40, 0.96) 0.03
Cumulative TG dose Low 339
Medium 344 1.15 (0.74, 1.80)
High 334 0.62 (0.36, 1.08) 0.06
Sex Male 577
Female 440 0.90 (0.64, 1.26) 0.55
Race/Ethnicity White 1378
Hispanic 444 1.36 (0.93,2.01)
African-American 72 1.09 (0.44,2.68)
Asian 29 2.02 (0.74, 5.49)
Other 106 0.79 (0.29, 2.16) 0.40
Age at diagnosis 12-23 or 72+ 326
(months) 24-71 691 0.58(0.41,0.81) 0.001
WBC < 20,000/ul 839
> 20,000/ul 178 1.85 (1.28,2.69) 0.001
CD7 <30 928
> 30 89 1.29 (0.76, 2.21) 0.35
CD10 <30 137
> 30 880 0.83 (0.52, 1.32) 0.42
Liver rating Normal 484
Enlarged 533 2.08(1.45,2.97) <0.0001
Spleen rating Normal 537
Enlarged 480 2.05 (1.45,2.89) <0.0001
Lymphadenopathy Normal 496
rating Enlarged 521 0.96 (0.69, 1.33) 0.80
CNS disease No CNS 928
CNS disease 47 1.64(0.86,3.13)
CNS at diagnosis 14 1.04 (0.26, 4.21) 0.37
£ p-value is for likelihood ratio test
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Table 5. Multivariate Cox regression analysis for risk of event in TG patients (n=1017)
Variable HR (95%CI) p-value£ §
TG dose Low
High 0.62 (0.40, 0.96) 0.03
Sex Male
Female 0.92 (0.65, 1.29) 0.62
TG*Sex 0.35
TG dose Low
High 0.64 (0.42, 0.99) 0.04
Age at diagnosis 12-23 or 72+
(months) 24-71 0.59 (0.42, 0.82) 0.002
TG*Age 0.14
TG dose Low
High 0.61 (0.40, 0.94) 0.03
WBC count <20,000/ul
>20,000/ul 1.87(1.29,2.72) 0.001
TG*WBC 0.11
TG dose Low
High 0.62 (0.40, 0.95) 0.03
Liver size Normal
Enlarged 2.08 (1.45,2.97) <0.0001
TG*Liver 0.86
TG dose Low
High 0.62 (0.40, 0.96) 0.03
Spleen size Normal
Enlarged 2.05 (1.45,2.89) <0.0001
TG* Spleen 0.80
TG dose Low
High 0.61 (0.40, 0.95) 0.03
CNS disease No CNS disease
CNS disease 1.61 (0.84, 3.06)
CNS at diagnosis 1.06 (0.26, 4.30) 0.15
TG*CNS 0.06
£ p-value is for likelihood ratio test; p-value for main effect test is with the present of another
main effect in the model
§ p-value for interaction test is with both main effects in the model
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Table 6. Cox regression analysis of treatment effect comparisons (n=2027)
HR (95% CD
N All Patients N Males N Females (n=898)
MP 1010 552 458
TG 1017 1.00(1.00, 1.00) 577 1.00(1.00, 1.00) 440 1.00(1.00, 1.00)
p-value§ 0.002 0.002 0.24
p-value for interaction of treatment and gender§ £ : 0.85
MP 1010 552 458
TG Low 509 0.85(0.66,1.08) 296 0.71(0.51,0.97) 213 1.09(0.49,1.32)
TG High 508 0.54(0.38,0.77) 281 0.51(0.33,0.80) 227 0.60(0.34,1.07)
p-value§ 0.003 0.002 0.15
p-value for interaction of treatments and gender§ £ : 0.03
MP 1010 552 458
TG Low 339 0.77(0.58,1.02) 214 0.72(0.50,1.03) 125 0.80(0.49,1.32)
TG Medium 344 0.88(0.64,1.21) 105 0.70(0.42,1.18) 239 1.24(0.80,1.93)
TG High 334 0.47(0.30,0.74) 258 0.47(0.29,0.77) 76 0.22(0.05,0.91)
p-value§ 0.002 0.005 0.02
p-value for interaction of treatments and gender§£: 0.02
§ p-value is for likelihood ratio test;
£ p-value for interaction test is with both main effects in the model
3 2
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
DISCUSSION
Both TG and MP were synthesized a half-century ago and have similarities in
structure and metabolism, but MP has been used as one of the main chemotherapeutic
agents in the maintenance therapy of ALL since then, while TG has been largely ignored.
In the most recent 10 years, interest in the use of TG has increased because of its
metabolic pattern and the possibility that it may be a more effective anti-leukemic agent.
It was observed that children with ALL having a low level of erythrocyte TGN seemed to
be at increased risk of disease relapse (Lilleyman & Lennard, 1994). TG can be converted
directly to TGN by hypoxanthine-guanine phosphoribosyl transferase (HGPRT), thus
avoiding the formation of MMPR. A higher level of TGN produced by TG as contrasted
with MP TGN level has been observed in children with leukemia (Lancaster DL et al,
1998). It was also found in this study that TG was well tolerated. In addition, in vitro data
has suggested that lymphoblasts are more sensitive to TG than MP (Evans & Relling,
1994; Adamson et al, 1994). Thus, for these reasons, patients on TG treatment might
have a lower risk of events than those on MP treatment.
The activity of thiopurine methyltransferase (TPMT) affects thiopurine cytotoxity
by methylating the initial metabolite of MP, thus reducing erythrocyte TGN level. TPMT
activity is determined by genes and varies among individuals (Weinshilboum & Sladek,
1980). A study conducted on a White population indicated that 89% of subjects inherited
high TPMT activity (Weinshilboum & Sladek, 1980). Lennard et al observed lower
concentration of TGN in patients with high TPMT activity (Lennard et al, 1990). It is
33
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
possible that TG has advantage over MP especially in patients with high TPMT activity
and no response to MP.
In this large-scale, randomized clinical study of comparisons of oral
mercaptopurine (MP) versus oral thioguanine (TG) among 2027 patients, the researchers
hypothesized that substitution of oral TG for oral MP during maintenance therapy would
improve event free survival for patients with standard risk ALL. We observed 188 events
in 1010 MP patients and 140 events in 1017 TG patients. Life table analysis revealed an
increase in event free survival for patients on TG treatment, compared with patients on
MP treatment. O f interest, most of this difference in EFS begins to occur relatively late
after therapy would be completed for the majority of patients.
When treating cumulative TG dose as a time-dependent variable, patients who
received lower cumulative TG doses (below the median) had a 15 percent lower risk of
events than patients who received MP treatment, while patients who received higher
cumulative TG doses (above median) had only half the risk of MP patients. Although no
significant linear trend was observed in which risk decreased as cumulative TG dose
increased, patients who received the highest level of cumulative TG dose (above the 67th
percentile) had 40 percent lower risk of events than patients at the lowest level (below the
34th percentile), and this highest group had less than half the risk of patients who
received MP. Thus, the results of the CCG-1952 clinical trial lend support to the
hypothesis by showing significant decreased risk in patients given higher cumulative TG
dose in this large randomized study.
34
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
The effect of high cumulative TG dose was most notable in females. Compared
with girls who received MP, girls who received cumulative TG dose above the 67th
percentile had an almost 80 percent lower risk of developing an event. This suggests that
female patients might benefit more from TG treatment if their treatment duration could
extend longer, e.g. as long as what male patients received.
For patients receiving TG treatment, higher cumulative TG dose (above median)
was associated with a decreased risk of event. This was also the case when gender, age at
diagnosis, WBC, liver rating, spleen rating and CNS disease status were each included
separately in the model as a possible confounder, which indicates TG operates
independently of other prognostic factors on ALL. The association between TG and risk
was not modified by those prognostic factors either (i.e., no statistical interaction effect
with TG).
In addition to cumulative TG dose, age at diagnosis, WBC, liver rating, spleen
rating, CNS disease status are factors that influenced event free survival in this study.
These results are reasonably consistent to those observed in other studies (Smith et al,
1996; Reiter et al, 1994; Chessells et al, 1995; Mahmoud et al, 1993; Gilchrist et al, 1994;
Pui et al, 1999). In the current study, gender is not significantly associated with risk of
event among TG patients. A plausible explanation is that one year less chemotherapy
received by girls than boys cancels out superior prognosis for girls. Interestingly, gender
is still an important prognostic indicator for MP patients. Girls receiving MP had a
significantly lower hazard rate than boys (relative risk = 0.63), even though they also
received one year less chemotherapy than boys.
35
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
In contrast to the decrease in risk for highest cumulative TG dose, we found a
slight increase in risk for the group receiving intermediate cumulative levels of TG. This
result is unexpected and difficult to explain.
Thus, the statistical analysis of this study indicates a better effect of TG therapy
compared to MP therapy in preventing the development of treatment failure in children
with ALL. However, a large proportion randomized to TG treatment had to be switched
to MP treatment because of the hepatic toxicity. A relatively high incidence of hepatic
veno-occlusive disease-like syndrome (VODLS) was observed in patients who received
TG treatment. True hepatic veno-oclusive disease (VOD) is thought to be a rare
complication of thiopurine therapy (Larrey et al, 1988), but it has been the most
commonly reported complication of TG therapy. Most of these problems are observed in
the combined therapy of TG and other chemotherapeutic agents (Griner et al, 1976;
Krivoy et al. 1982; Satti et al. 1982; D ’Cruz et al. 1983). Most VODLS cases in our
study were mild and reversible after terminating TG. Thus, the toxicities were generally
transient episodes without long lasting effects, and no patients had VODLS recur after
they switched to MP treatment.
In the current study, we found that the overall incidence difference of VODLS on
TG between boys and girls only achieved borderline significance. But after TG patients
were divided into two cohorts according to their randomization date (before or after
January 23, 1998, the day that the daily TG dose was reduced from 60mg/M2 to
50mg/M2 ), we found that boys had a significantly higher incidence rate than girls among
patients whose initial TG dose was 60mg/M2, but no such difference was found among
36
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
patients whose initial TG dose was 50mg/M2. The gender predilection in this study
suggests a sex-linked genetic polymorphism (Stoneham et al, 2003) or other gender-
related factors (e.g. hormones, etc) may be involved in the metabolic pathways of TG.
However, the mechanism of hepatic toxicity remains uncharacterized (Lilleyman et al,
1984). The results also suggest that this male preponderance seems to occur at a higher
daily TG dose level. Furthermore, patients randomized to TG treatments before January
23, 1998 had a significantly higher incidence rate than those given TG after January 23,
1998 among boys, but no such difference was found among girls. This fact indicates that
males are more sensitive to the daily TG dose level. Thus, it also suggests that the
complex biological pathways might be influenced by unknown gender-specific
mechanisms.
We should note that a limitation of this study is the lack of accurate computerized
measurements of cumulative TG doses. We had to estimate them from the theoretical
maximum daily dose. The actual doses patients received are generally less than
theoretical doses because of different degrees of toxicity occurring for the patient
population. We made dose adjustments for VODLS patients, but not for non-VODLS
patients since those data are not currently available. Another limitation is that the
cumulative dose probably dose not tell the whole story. A lower dose level given for a
longer time, or a higher dose level for a shorter time might have totally different
biological effects. The optimal dose level and the duration of the therapy whereby
patients would obtain the biggest benefit cannot be determined from the current single
clinical trial. Possibly that is an area worthy of future study.
37
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
In summary, we found that higher cumulative dose TG therapy resulted in lower
risk than MP therapy, but a higher incidence of VODLS was also observed in patients
who received TG therapy. So it would be prudent to monitor liver enzymes in patients
who receive TG therapy (Rulyak et al, 2003). Further research is needed to clarify the
mechanism of hepatic toxicity in order to better identify those who could safely receive
TG therapy and presumably derive the greatest benefit in terms of improved event-free
survival.
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
REFERRENCE
Adamson P.C., Poplack D.G., Balis F.M., et al. (1994). The cytotoxicity of thioguanine
vs mercaptopurine in acute lymphoblastic leukemia. Leukemia Res, 18(11), 805-10
Bleyer W.A. (1997). The U.S. pediatric cancer clinical trials programmes: international
implications and the way forward. Eur J Cancer, 33(9), 1439-47
Bostrom B., Erdmann G. (1993). Cellular pharmacology of 6-MP in ALL. Am J Pediatr
Hematol Oncol, 15, 80-6
Breslow N. (1975). Analysis of survival data under the proportional hazards model. Int.
Stat. Review, 1, 45-58
Brincker H. (1982). Population-based age- and sex-specific incidence rates in the four
main types o f leukaemia. Scand J Haematol, 29, 241-249
Brubaker C.C., Kontras E., Leikin S., et al. (1964). Comparative evaluation of 6-MP, TG
and thioguanosine in the treatment of acute leukemia in children: results of a cooperative
study. Proc Int Soc Hematol, 9, 577
Chessells J.M., Bailey C., Richards S.M. (1995). Intensification of treatment and survival
in all children with lymphoblastic leukaemia: results of UK Medical Research Council
trial UKALL X. Medical Research Council Working Party on Childhood Leukaemia.
Lancet, 345 (8943), 143-8
Cox D. (1972). Regression models and life tables. J R Stat Soc, 34, 187-220
D'Cruz C.A., Wimmer R.S., Harcke H.T., et al. (1983). Veno-occlusive disease of the
liver in children following chemotherapy for acute myelocytic leukemia. Cancer, 52,
1803-7
Elion G.B., Callahan S.W., Hitchings G.H., et al. (1962). Experimental, clinical and
metabolic studies of thiopurine. Cancer Chemother Rep, 16, 197
Gaynon P.S., Stork L., Sather H., et al. (1999). Leukemic progenitor cell content of pre-
and post-induction chemotherapy bone marrow specimens from children with newly
diagnosed or relapsed acute lymphoblastic leukemia (ALL). Proceedings o f the American
Society o f Clinical Oncology, 18: A-2187, 567a
Gaynon P.S., Trigg M.E., Heerema N.A., et al. (2000). Children's Cancer Group trials in
childhood acute lymphoblastic leukemia: 1983-1995. Leukemia, 14, 2223-33
39
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Gehan E. (1965). A generalized Wilcoxon test for comparison arbitrarily singly-censored
samples. Biometrica, 52, 203-223
Gilchrist G.S., Tubergen D.G., Sather H.N., et al. (1994). Low numbers o f CSF blasts at
diagnosis do not predict for the development of CNS leukemia in children with
intermediate-risk acute lymphoblastic leukemia: a Childrens Cancer Group report. J Clin
Oncol, 12 (12), 2594-600
Griner P.F., Elbadawi A., Packman C.H. (1976). Veno-occlusive disease o f the liver after
chemotherapy of acute leukemia. Report of two cases. Ann Intern M ed 1976, 55,578-82
Harms D.O., Janka-Schaub G.E. (2000). Co-operative study group for childhood acute
lymphoblastic leukemia (COALL): long-term follow-up of trials 82, 85, 89 and 92.
Leukemia, 14(12), 2234-9
Kaplan E., Maier P. (1958). Nonparametric estimation from incomplete observation. J
Am Stat Assoc, 53, 457-481
Krivoy N., Raz R., Carter A., et al. (1982). Reversible hepatic veno-occlusive disease and
6-thioguanine. Ann Intern Med, 96, 788
Lancaster D.L., Lennard L., Rowland K., Vora A.J., Lilleyman J.S. (1998). Thioguanine
versus mercaptopurine for therapy of childhood lymphoblastic leukaemia: a comparison
of haematological toxicity and drug metabolite concentrations. British Journal o f
Haematology, 102(2), 439-43
Lennard L., Davies H.A., Lilleyman J.S. (1993). Is 6-thioguanine more appropriate than
6-mercaptopurine for children with acute lymphoblastic leukaemia? British Journal o f
Cancer, 68(1), 186-90
LePage G.A. (1963). Basic biochemical effects and mechanism of action of 6-
thioguanine. Cancer Res, 23, 1202
Lilleyman, J.S., Lennard, L. (1994). Mercaptopurine metabolism and risk of relapse in
childhood lymphoblastic leukaemia. Lancet, 343, 1188-1190
Mahmoud H.H., Rivera G.K., Hancock M.L., et al. (1993). Low leukocyte counts with
blast cells in cerebrospinal fluid of children with newly diagnosed acute lymphoblastic
leukemia. N Engl J Med, 329 (5), 314-9
Nelson J.A., Carpenter J.W., Rose L.M. (1975). Mechanisms of action of 6-TG, 6-MP,
and 8-azaguanine. Cancer Res, 315, 2872
4 0
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Pui C.H., Boyett J.M., Relling M.V., et al. (1999). Sex differences in prognosis for
children with acute lymphoblastic leukemia. J Clin Oncol, 17 (3), 818-24
Pui C.H., Crist W.M. (1994). Biology and treatment of acute lymphoblastic leukemia. J
Pediatr, 124, 491-503
Pullen J., Shuster J.J., Link M., et al. (1999). Significance of commonly used prognostic
factors differs for children with T cell acute lymphocytic leukemia (ALL), as compared
to those with B-precursor ALL. A Pediatric Oncology Group (POG) study. Leukemia, 13
(11), 1696-707
Reiter A., Schrappe M., Ludwig W.D., et al. (1994). Chemotherapy in 998 unselected
childhood acute lymphoblastic leukemia patients. Results and conclusions of the
multicenter trial ALL-BFM 86. Blood, 84 (9), 3122-33
Ries L.A., Eisner M.P., Kosary C.L., et al., eds. (2003). SEER Cancer Statistics Review,
1975-2000. National Cancer Institute. Bethesda, MD.
Rulyak S.J., Saunders, M., Lee, S.D. (2003). Hepatotoxicity Associated With 6-
Thioguanine Therapy for Crohn's Disease. Journal o f Clinical Gastroenterology, 36(3),
234-237
Sandler D.P. (1992). Epidemiology and etiology of acute leukemia: An update. Leukemia,
6(4), 3-5
Satti M.B., Weinbren K., Gordon-Smith E.C. (1982). 6-thioguanine as a cause of toxic
veno-occlusive disease o f the liver. J Clin Pathol, 35, 1086-91
Sawyers C.L., Denny C.T., Witte O.N. (1991). Leukemia and the disruption of normal
hematopoiesis. Cell, 64, 337-350
Schrappe M., Reiter A., Ludwig W.D., et al. (2000). Improved outcome in childhood
acute lymphoblastic leukemia despite reduced use of anthracyclines and cranial
radiotherapy: results of trial ALL-BFM 90. Blood, 95, 3310-22
Smith M., Arthur D., Camitta B., et al. (1996). Uniform approach to risk classification
and treatment assignment for children with acute lymphoblastic leukemia. J Clin Oncol,
14(1), 18-24
Smith M.A., Ries L.A., Gurney J.G., et al. (1999). Leukemia. In: Ries L.A., Smith M.A.,
Gurney JG, et al., eds. Cancer incidence and survival among children and adolescents.
United States SEER Program 1975-1995. National Cancer Institute. Bethesda, MD
41
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Stoneham S., Lennard L., Coen P., Lilleyman J., Saha V. (2003). Veno-occlusive disease
in patients receiving thiopurines during maintenance therapy for childhood acute
lymphoblastic leukaemia. British Journal o f Haematology, 123(1), 100-2
Stork L.C., Erdmann G., Adamson P., et al. (1998). Oral 6-thioguanine causes relatively
mild and reversible hepatic veno-occlusive disease(VOD). J Pediatr Hematol Oncol, 20,
400a
Stork L.C., Sather H., Yanofsky R., et al. (2001). Hyperdiploidy with trisomy 10 and
TEL-AML 1 expression among children with standard risk acute lymphoblastic leukemia
(SR-ALL): a CCG-1952 report. Proceedings o f the American Society o f Clinical
Oncology, 20, A -1476
Weinshilboum R.M., Sladek S.L. (1980). Mercaptopurine pharmacogenetics: Monogenic
inheritance of erythrocyte thiopurine methyltransferase activity. Am J Hum Genet, 32,
651-62
Zar J. (1984). Biostatistical Analysis. Prentice-Hall, Englewood Cliffs
Zimm S. (1988). Thiopurines in cancer chemotherapy. Wellcome Oncology Monograph,
Research Triangle Park, NC
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
APPENDIX
Figure A l. Metabolic process of MP and TG converting into thioguanine nucleotide
SH
SH
V^~N
6-Mercaptopurine
PRPP
A .
HGPRT
©
SH
IHOH
TIMP
IMP Dehydrogenase
0 = P - 0 —CH,
OH OH
TXMP
< S >
ATP G uanylate Synthetase
SH
A
6-Thioguanine
PRPP
HGPRT
©
n h 2-
OH OH
6-TGMP
Incorporation into DNA
43
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Figure A2. Treatment Schema
Standard 3-Drug Induction
IT Ara-C, VCR, PDN, L-Asp, IT MTX
M1/M2 <-
i
I
Day 7 BM
I
Day 14
BM
i
M3
1
Off Protocol
Therapy
• Day 28 BM -
I
M1
i
Randomize
Reg A1
IT MTX
Reg A2 Reg B2
MP TG
ITT ITT
Reg B1
TG
IT MTX
Consolidation
VCR, PDN.M PorTG,
IT MTX or ITT
Interim Maintenance #1
VCR, PDN, MTX po, MP or TG
Delayed Intensification #1
VCR, L-Asp, DXM, DOX, CPM, TG.
Ara-C, IT MTX or ITT
Interim Maintenance #2
VCR, PDN, MTX po, MP or TG
Delayed Intensification #2
VCR, L-Asp, DXM, DOX, CPM, TG,
Ara-C, IT MTX or ITT
Maintenance
VCR, PDN, MTX po, MP or TG,
IT MTX or ITT
Unfavorable
C ytogenetics
-» M3
Continue Reg C
Standard Induction
(Augm ented BFM)
Dauno, PDN, VCR,
L-Asp, IT MTX
Day 35 BM » M3
M1/M2
Continue
Off Protocol
Therapy
Consolidation
CPM. Ara-C, MP. VCR, IT MTX.
PEG
Interim M aintenance#1
MTX IV, VCR, PEG
Delayed Intensification #1
VCR. PEG, DXM, DOX, CPM, TG,
Ara-C, IT MTX
Interim M aintenance #2
MTX IV, VCR, PEG , IT MTX
Delayed Intensification #2
VCR, PEG, DXM, DOX, CPM, TG,
Ara-C, IT MTX
M aintenance
VCR. PON, MTX po, MP, IT MTX
4 4
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Figure A3. CCG-1952 Event-Free Survival from Randomization
Comparison o f Thiopurine Regimens
1.00
— MP
------ TG
.90
.80
Log rank p = .004
RHR for TG vs. MP = 0.72
.70
.60
1017 992 939 861 597 374 136
2 4 6
Y ears
Years from Randomization
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Figure A4. CCG-1952 Event-Free Survival from Randomization
Comparison o f CNS Prophylaxis Regimens
1.00
Triple IT
IT MTX
.90
.80
.70
Log rank p = .30
RHR for Triple IT vs. IT MTX =
.60
i t.’ ce
1018 985 926 856 595 129 369
2 4 6
Y ears
Years from Randomization
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Figure A5. CCG-1952 Event-Free Survival from Randomization Comparison o f
Thiopurine Regimens:Cohort 1 (Randomized before 01/23/98)
1.00
MP
TG
.90
.80
Log rank p = .005
RHR for TG vs. MP = 0.63
.70
.60
402 414 384 365 336 298 136
2 4 6
Years from Randomization
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Figure A6. CCG-1952 Event-Free Survival from Randomization Comparison o f
Thiopurine Regimens:Cohort 2 (Randomized after 01/23/98)
1.00
MP
TG
.90
.80
Log rank p = .20
RHR for TG vs. MP = 0.82
.70
.60
555 496 261 76 603 590
2 4
Years from Randomization
6
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Recreational physical activity and risk of breast cancer: The California Teachers Study

PDF

Polymorphisms in genes involved in steroid hormone metabolism and mammographic density changes in women randomized to menopausal estrogen and progesterone therapy

PDF

A case-control study of passive smoking and bladder cancer risk in Los Angeles

PDF

Validation of serum cotinine as a biomarker of environmental tobacco smoke exposure: Validation with self-report and association with subclinical atherosclerosis in non-smokers

PDF

Selective laser trabeculoplasty for the treatment of glaucoma

PDF

The association between recreational physical activity and mammographic density

PDF

The influence of family structures on adolescent smoking among multicultural adolescents in Hawaii

PDF

Progression of carotid intima-media thickness and plasma antioxidants: The Los Angeles Atherosclerosis Study

PDF

Cigarettes and alcohol in relation to colorectal cancer within the Singapore Chinese Health Study

PDF

Association between body mass and benign prostatic hyperplasia in Hispanics: Role of steroid 5-alpha reductase type 2 (SRD5A2) gene

PDF

Lymphedema: Impact on breast cancer survivors

PDF

Predictive value of CT coronary artery calcium scanning for coronary heart disease in asymptomatic subjects with diabetes mellitus

PDF

Comparisons of metabolic factors among gestational diabetes mellitus probands, siblings and cousins

PDF

Predictors of mammography use among California teachers

PDF

BRCA1 mutations and polymorphisms in African American women with a family history of breast cancer identified through high throughput sequencing

PDF

Family history, hormone replacement therapy and breast cancer risk on Hispanic and non-Hispanic women, The New Mexico Women's Health Study

PDF

Metabolic effects of magnesium supplementation in women with a history of gestational diabetes mellitus

PDF

Insulin-like growth factor 1 genotype, phenotype and breast cancer risk, by racial/ethnic group

PDF

Effect of hormone therapy on the progression of carotid-artery atherosclerosis in postmenopausal women with and without established coronary artery disease

PDF

beta3-adrenergic receptor gene Trp64Arg polymorphism and obesity-related characteristics among African American women with breast cancer: An analysis of USC HEAL Study

Asset Metadata

Creator Dong, Hongfang (author)

Core Title Randomized comparison of oral mercaptopurine versus oral thioguanine for standard risk acute lymphoblastic leukemia

Degree Master of Science

Degree Program Applied Biostatistics and Epidemiology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag biophysics, medical,health sciences, oncology,Health Sciences, Pharmacology,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-321460

Unique identifier UC11329223

Identifier 1424243.pdf (filename),usctheses-c16-321460 (legacy record id)

Legacy Identifier 1424243.pdf

Dmrecord 321460

Document Type Thesis

Rights Dong, Hongfang

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