University of Southern California Dissertations and Theses

Body wall primary tumors in non-rhabdomyosarcoma soft tissue sarcoma

(USC Thesis Other)

Body wall primary tumors in non-rhabdomyosarcoma soft tissue sarcoma

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 1

Body wall primary tumors in Non-Rhabdomyosarcoma Soft Tissue Sarcoma

By Qian Qian

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN
CALIFORNIA
In Partial Fulfillment of the Requirements for the Degree
MASTER OF SCIENCE
BIOSTATISTICS

December 2018

2

Acknowledgement:

I really want to appreciate my thesis advisor, Dr. Donald Barkauskas for his guidance for thesis
and his patience. I also want to thank for my thesis committee Dr. Meredith Franklin and Dr. Ha
Dang for their amending and useful suggestions.

3

TABLE OF CONTENTS
ACKNOWLEDGEMENT ................................................................................................................................................. 2
ABSTRACT ......................................................................................................................................................................... 4
INTRODUCTION .............................................................................................................................................................. 5
DATA AND METHODS ................................................................................................................................................. 6
RESULTS ............................................................................................................................................................................. 8
DISCUSSION .................................................................................................................................................................. 15
APPENDIX ....................................................................................................................................................................... 16
REFERENCES .................................................................................................................................................................. 18

4

Abstract:

We used data from ARST0332, a risk-based treatment for non-rhabdomyosarcoma soft tissue
sarcomas (NRSTS) in pediatric patients from Children’s Oncology Group to study the difference
between patients with body wall primary tumors and patients with non-body wall primary tumors.
There are 529 evaluable patients with NRSTS in the study, 75 having a primary tumor site at body
wall. Results from Fisher’s exact test indicated patients with body wall tumors had statistically
significantly different features, including tumor size (p=0.0383) and tumor depth (p=0.0314), than
patients with non-body wall tumor. The survival probability of event-free survival and overall
survival were not significantly different between body wall and non-body wall patients (EFS p-
value=0.1243, OS p-value=0.9922). Using Cox proportional hazards regression, we found risk
group (low, intermediate, high) to be a strong predictor of overall survival, where high risk had
hazard ratio of 1.981 (95% CI: 0.603, 6.511) vs. intermediate risk and low risk had hazard ratio of
0.180 (95% CI: 0.039, 0.835) vs. intermediate risk. Treatment arm (the therapy patients received)
was also a strong predictor in the event-free survival model, as determined by an interaction term
between risk group and treatment arm.

5

Introduction:

Nearly 4% of childhood malignancies are caused by the non-rhabdomyosarcoma soft tissue
sarcomas (NRSTS) (Bethesda, 1999). A recent study finds that the survival of patients from 15 to
44 years old with sarcoma did not improve much over the past years compared to other age groups
(Bleyer A., 2003). Thus, we should take efforts for children, adolescents and young adults to
improve their survival. Special treatments should be designed for children in pediatric oncology
cases because children had less tolerance in chemotherapy and radiotherapy. The Children’s
Oncology Group (COG) study “ARST0332: Risk-Based Treatment for Non-Rhabdomyosarcoma
Soft Tissue Sarcomas in Patients Under 30 Years of Age” was used to evaluate a novel risk-based
strategy for treating patients, in order to improve the effect of therapy for low-, intermediate- and
high-risk patients (Sheri L.).

We analyzed the data from ARST0332, for the patients with NRSTS’ primary site at body walls,
including anterior chest wall, posterior chest wall, paraspinal, abdominal wall, and breast. Our aim
was to describe the demographics, clinical features, as well as to perform survival analysis and
model selection for overall survival and event-free survival using Cox-proportional hazards
regression.

6

Data and Methods:

Data:
The original data for our study includes demographic data and clinical features. Demographic data
are age, gender, race, ethnicity. Age at enrollment data were categorized into three groups: 0–9,
10–17, 18–30. Race was defined as white, black or African American, other and unknown.
Ethnicity was classified as Hispanic or Latino, Not Hispanic or Latino and unknown.

Clinical features include primary tumor size, tumor depth, tumor grade as determined by the POG
(Pediatric Oncology Group) and FNCLCC (French Federation of Cancer Centers) grading systems,
tumor invasiveness, metastatic disease. Primary tumor size was categorized as less than 5 cm or
not. Tumor depth was categorized as superficial or deep. Tumor grade was categorized as low (1
or 2) or high (3). Tumor invasiveness was shown as “yes” or “no”, as was metastatic disease.

In ARST0332, patients were divided into three risk groups (Figure 1), based on tumor features like
metastatic disease (Yes/No), non-metastatic unresectable (Yes/No), POG grade (low/high),
primary tumor size (≤5 cm/>5 cm), microscopic margins (positive/negative). These characteristics
further divided patients into four treatment arms (see Figure 1).

7

Methods:

We used Fisher’s exact test for categorical variables and two-sample t-test for continuous variables
to compare demographic and clinical features between body wall and non-body wall patients. We
used Fisher’s exact test to get pairwise association between all paired variables within body wall
patients. (Bishop Y , 2007)

We used the log-rank test stratified by risk groups (determined by metastasis mainly) to compare
the event-free survival (EFS) and overall survival (OS) for body wall vs. non-body wall patients.
Kaplan-Meier estimators for survival were plotted stratified by risk groups for EFS and OS. For
EFS, patients with first event date for relapse/recurrence of tumor, development of second
malignant neoplasms (SMN) and death were considered to have had an EFS event; all others were
considered censored. For OS, patients who were alive at last contact were considered censored.
Event times (years) of EFS and OS were calculated by subtracting event from enrollment date for
each patient.

Log-rank tests for each of patient characteristics, stratified by risk group among body wall patients
for all the variables appear in table 1. Variables with log-rank p-values less than 0.2 were included
in a multivariable Cox proportional hazards regression model, and stepwise model selection with
false positive rate α=0.05 was conducted for each EFS and OS. Because we are doing stepwise
model selection here and none of other variables remains in the model, we didn’t consider
collinearity issue for variables like POG grade and FNCLCC grade. Proportional hazards
assumption is assumed for all analyses. Since not all possible combinations of risk group and
treatment group were possible, we created a new variable combining risk group and treatment arm
included this variable in the model for EFS.

All data were organized and analyzed in SAS, version 9.4.

8

Results:

There are 529 evaluable patients with NRSTS in the study and 75 with primary tumor site at body
wall. The median follow-up time for patients alive at last contact for the entire study was 6.50
years (411 patients), for body wall patients was 5.85 years (60 patients), and for non-body wall
patients was 6.58 years (351 patients). Descriptive characteristic including Mean (SD) and N (%)
for patients with or without body wall primary tumor site, along with results from the Fisher’s
Exact Test or two-sample t-test are shown in Table 1.

Fisher’s exact test indicated patients with body wall tumors and patients with non-body wall tumor
had statistically significant difference in tumor size (p=0.0383) and tumor depth (p=0.0314). Two-
sample t-test showed that patients with body wall tumors and patients with non-body wall tumor
had statistically significant difference in primary tumor size (p=0.0080).

9

Table 1. Patient characteristics stratified by primary tumor type (body wall vs. non-body wall) with
results from Fisher’s exact-test comparisons unless noted.

Variables Not Body-wall Body-wall P-value (Fisher’s
exact test) Frequency (%) or Mean (SD)
Gender

Male 214 (47.1%) 31 (41.3%) 0.3829
Female 240 (52.9%) 44 (58.7%)
Race

White 327 (72.0%) 48 (64.0%) 0.2070
Black or African American 69 (15.2%) 11 (14.7%)
Other 18 (4.0%) 6 (8.0%)
Unknown 40 (8.8%) 10 (13.3%)
Ethnicity

Hispanic or Latino 70 (15.4%) 10 (13.3%) 0.0986
Not Hispanic or Latino 371 (81.7%) 59 (78.7%)
Unknown 13 (2.9%) 6 (8.0%)
POG grade

High 328 (72.3%) 55 (73.3%) 0.8901
Low 126 (27.8%) 20 (26.7%)
FNCLCC grade

Indeterminate 1 (0.2%) 0 (0.0%) 0.7503
High 201 (44.3%) 35 (46.7%)
Low 252 (55.5%) 40 (53.3%)
Tumor size
(categorical)
≤ 5.0 cm 159 (35.0%) 36 (48.0%) 0.0383
> 5.0 cm 295 (65.0%) 39 (52.0%)
Tumor depth

Deep 382 (84.1%) 55 (73.3%) 0.0314
Superficial 72 (15.9%) 20 (26.7%)
Invasiveness

Invasive 265 (58.4%) 35 (46.7%) 0.0605
Non-invasive 189 (41.6%) 40 (53.3%)
Age at
enrollment
(categorical)
<10 125 (27.5%) 21 (28.0%) 0.8962
10-17.999 257 (56.6%) 44 (58.7%)
≥18 72 (15.9%) 10 (13.3%)
Risk group

High 72 (15.9%) 8 (10.7%) 0.1662
Intermediate 199 (43.8%) 28 (37.3%)
Low 183 (40.3%) 39 (52.0%)
Treatment Arm

Arm A 169 (37.2%) 36 (48.0%) 0.1101
Arm B 14 (3.1%) 3 (4.0%)
Arm C 94 (20.7%) 17 (22.7%)
Arm D 177 (39.0%) 19 (25.3%)
Age at enrollment date (continuous) 13.02 (5.55) 12.82 (5.65) 0.7744*
Primary tumor size (continuous) 8.87 (5.38) 7.28 (4.62) 0.0080*
*Two-sample t-test

10

Table 2. Univariate associations of patient characteristics among body wall patients (p-value by
Fisher’s Exact test).
Statistical significance is marked by red (p<0.05) and blue (marginally significant: p<0.1)

Race Ethnicity
POG
grade
FNCLCC
grade
Tumor
size
Tumor
depth
Invasive-
ness
Age
group
Risk
group
Treatment
Arm
Gender 0.1550 0.2790 1.0000 1.0000 0.2412 0.7828 0.2509 0.1550 0.3223 0.5201
Race 0.0006 0.2071 0.0496 0.7084 0.0496 0.7066 0.7399 0.0436 0.0856
Ethnicity 0.2209 0.4445 0.9147 0.0864 0.8381 0.5711 0.7629 0.2044
POG grade <0.0001 0.0081 <0.0001 0.0354 0.0033 <0.0001 <0.0001
FNCLCC grade 0.0107 0.0354 0.1080 0.3411 0.0014 0.0071
Tumor size 0.0002 <0.0001 0.0441 <0.0001 <0.0001
Tumor depth <0.0001 0.0078 0.0002 <0.0001
Invasiveness 0.0279 <0.0001 <0.0001
Age group 0.0146 0.0403
Risk group <0.0001

The Fisher’s Exact tests suggest statistically significant (p<0.05) relationships between race and
ethnicity and as well as race and risk group. POG grade is statistically significantly associated with
FNCLCC grade, tumor size, tumor depth, invasiveness, age group, risk group, and treatment arm.
FNCLCC grade has significant relationships with tumor size, tumor depth, risk group, and
treatment arm. Tumor size has significant relationships with tumor depth, invasiveness, age group,
risk group and treatment arm. Tumor depth has significant relationships with invasiveness, age
group, risk group, treatment arm. Invasiveness has significant relationships with age group, risk
group, treatment arm. Age group has significant relationships with risk group, treatment arm. Risk
group has significant relationship with treatment arm. Treatment arm has significant relationship
with every variable except race and ethnicity. (Table 2)

11

Kaplan-Meier curves of event-free and overall survival for all patients stratified by risk group and
grouped by body wall vs. non-body wall:

Figure 3.1: EFS by risk group and body wall primary tumor

12

Figure 3.2: OS by risk group and body wall primary tumor

The results of EFS and OS for all patients stratified by risk group are in Table 3.

Table 3. Log-rank test to compare survival distribution between body wall and non-body wall
stratified by risk group.

Outcome Chi-Square(log-rank) DF p-value
EFS 2.3622 1 0.1243
OS 0.0001 1 0.9922

The log-rank test stratified by risk group showed that there was not statistically significant
difference between body wall and non-body wall patients in EFS (p=0.1243) and OS survival
(p=0.9922) (Table 3).

In order to investigate what factor is important for body wall patients, the rest of analysis will be
on body wall only.

13

The results for univariate stratified log-rank tests are in Table 4.

Table 4. Log-rank test stratified by risk group among body wall patients for each variable,
associated with EFS and OS.
Covariates considered in the Cox proportional hazard regression are marked in blue (p-value<0.2)

Covariates EFS OS
Stratified log-rank test χ
2
DF p-value Stratified log-rank test χ
2
DF p-value
Gender 0.2765 1 0.5990 0.0036 1 0.9523
Race 0.4564 3 0.9284 1.8236 3 0.6098
Ethnicity 3.9540 2 0.1385 2.1306 2 0.3446
POG grade 3.2781 1 0.0702 1.8546 1 0.1732
FNCLCC grade 1.8839 1 0.1699 0.3632 1 0.5468
Tumor size 0.9992 1 0.3175 0.1983 1 0.6561
Tumor depth 3.8381 1 0.0501 2.4445 1 0.1179
Invasiveness 7.2214 1 0.0072 1.3651 1 0.2427
Age group 3.1325 2 0.2088 1.3247 2 0.5156
Treatment arm 14.9962 2 0.0006 6.6004 2 0.0369

The variables in Table 4 above with p-value less than 0.2 were marked with blue and were used in
the stepwise model selection using Cox proportional hazard regression. We kept risk group in the
final model as a known marker because risk group is a known prognostic factor in NRSTS. The
final model for EFS contains risk group and treatment arm interaction (Table 5.1). The final model
for OS survival model contains only risk group (Table 6.1).

The final model of EFS contains risk group-treatment arm. Compared to low risk-treatment arm B
as the baseline, intermediate risk-treatment arm C had hazard ratio of 0.269 (95% CI: 0.049, 1.484);
intermediate risk-treatment arm D had hazard ratio of 1.153 (95% CI: 0.250, 5.312); high risk-
treatment arm C had hazard ratio of 1.088 (95% CI: 0.098, 12.113); high risk-treatment arm D had
hazard ratio of 2.921 (95% CI: 0.569, 15.006); and low risk-treatment arm A had hazard ratio of
0.132 (95% CI: 0.024, 0.727).

For the overall survival model, only risk group remains in the final model. Using intermediate risk
as the baseline, high risk had hazard ratio of 1.981 (95% CI: 0.603, 6.511); and low risk had hazard
ratio of 0.180 (95% CI: 0.039, 0.835).

14

Table 5.1. Cox-Regression model for EFS survival among body wall patients

Parameter DF Parameter
estimate
SD Χ
2
p-
value
Hazard
Ratio
95% hazard
ratio CI
Risk
group/
treatment
arm
Baseline
(Low/B)
— — — — — — —
Intermediate/C 1 -1.31313 0.87147 2.2704 0.1319 0.269 0.049 1.484
Intermediate/D 1 0.14225 0.77950 0.0333 0.8552 1.153 0.250 5.312
High/C 1 0.08438 1.22956 0.0047 0.9453 1.088 0.098 12.113
High/D 1 1.07194 0.83497 1.6481 0.1992 2.921 0.569 15.006
Low/A 1 -2.02449 0.86997 5.4153 0.0200 0.132 0.024 0.727

Table 5.2. Type III Wald tests examine the significance of partial effects in EFS

Effect DF Wald Chi-square p-value
Risk group-treatment arm 5 26.8797 <0.0001

Table 6.1. Cox-Regression model for OS survival among body wall patients

Parameter DF Parameter
estimate
SD Χ
2
p-value Hazard
Ratio
95% hazard
ratio CI
Risk
group
Baseline
(Intermediate)
— — — — — — —
High 1 0.68349 0.60717 1.2672 0.2603 1.981 0.603 6.511
Low 1 -1.71299 0.78227 4.7951 0.0285 0.180 0.039 0.835

Table 6.2. Type III Wald tests examine the significance of partial effects in OS

Effect DF Wald Chi-square p-value
Risk group 2 7.6724 0.0216

15

Discussion:

In this study of ARST0332, a risk-based treatment for non-rhabdomyosarcoma soft tissue
sarcomas (NRSTS) in pediatric patients from Children’s Oncology Group, we find that body wall
patients had significantly different features from non-body wall patients in terms of tumor size and
tumor depth (smaller size and more superficial tumors), and that tumor invasiveness is marginally
significant different ( i .e . body wall tumors are invasive less often). For Kaplan-Meier survival
curves for EFS and OS, the difference between body wall patients and non-body wall patients is
not statistically significant, but from the survival curves, for EFS, we can see that at each risk
group body wall patients have higher survival probability than non-body wall patients. For OS, the
survival curves show that body wall patients have similar survival probability to non-body wall
within high and intermediate risk groups and higher survival probability in low risk group.

Among body wall patients, although ethnicity, POG grade, FNCLCC grade, tumor depth,
invasiveness and treatment arms were used in the multivariable Cox proportional hazards model
selection for EFS, only the risk groups and treatment arm remained in the final model. Due to the
definition of risk group and treatment arm, some combinations of risk group and treatment arm
were not possible. So, we added risk-treatment (a new term that includes information of risk group
and treatment arm) and redo the model selection process, only the interaction term between risk
group and treatment arm remained in the final model. This result is expected given that the risk
group showed a significant relationship with treatment arms and treatment arms had relatively
lowest p-value in log-rank test. Although, POG grade, tumor depth and treatment arms were used
in the model selection for OS at first, there were only risk group included in the final model.

For the final model of EFS among body wall patients, not all groups are significantly different
from the baseline. High risk with treatment arm D is significantly different from baseline low risk
with treatment arm B. For the overall survival model, the high risk is not significantly different
from the intermediate risk, the low risk is significantly different from the intermediate risk.

16

Appendix:
Figure 1.

17

Figure 2.

18

References:
1. Bishop, Y ., Fienberg, S., & Holland, P. (2007). Discrete Multivariate Analysis Theory and
Practice. New York, NY: Springer New York.
2. Sheri L. Spunt, et al. Protocol of Risk-Based Treatment for Non-Rhabdomyosarcoma Soft
Tissue Sarcomas (NRSTS) in Patients under 30 Years of Age. Children’s Oncology Group
(ARST0332).
3. Bleyer, A., Budd, T., & Montello, M. (n.d.). Lack of Participation of Older Adolescents and
Young Adults with Cancer in Clinical Trials: Impact in the USA. In Cancer and the Adolescent,
Second Edition (pp. 32–45). Blackwell Science Ltd.
4. Cancer incidence and survival among children and adolescents: United States SEER Program,
1975-1995. (1999). Bethesda, Md.]: National Cancer Institute.

Abstract (if available)

Abstract We used data from ARST0332, a risk-based treatment for non-rhabdomyosarcoma soft tissue sarcomas (NRSTS) in pediatric patients from Children’s Oncology Group to study the difference between patients with body wall primary tumors and patients with non-body wall primary tumors. There are 529 evaluable patients with NRSTS in the study, 75 having a primary tumor site at body wall. Results from Fisher’s exact test indicated patients with body wall tumors had statistically significantly different features, including tumor size (p=0.0383) and tumor depth (p=0.0314), than patients with non-body wall tumor. The survival probability of event-free survival and overall survival were not significantly different between body wall and non-body wall patients (EFS p-value=0.1243, OS p-value=0.9922). Using Cox proportional hazards regression, we found risk group (low, intermediate, high) to be a strong predictor of overall survival, where high risk had hazard ratio of 1.981 (95% CI: 0.603, 6.511) vs. intermediate risk and low risk had hazard ratio of 0.180 (95% CI: 0.039, 0.835) vs. intermediate risk. Treatment arm (the therapy patients received) was also a strong predictor in the event-free survival model, as determined by an interaction term between risk group and treatment arm.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Infants in non-rhabdomyosarcoma soft tissue sarcoma

PDF

Extremity primary tumors in non-rhabdomyosarcoma soft tissue sarcoma: survival analysis

PDF

Lymph node metastases in non-rhabdomyosarcoma soft tissue sarcoma

PDF

A novel risk-based treatment strategy evaluated in pediatric head and neck non-rhabdomyosarcoma soft tissue sarcomas (NRSTS) patients: a survival analysis from the Children's Oncology Group study...

PDF

Comparative study of the POG and FNCLCC grading systems in non-rhabdomyosarcoma soft tissue sarcomas

PDF

Distant metastatic nonrhabdomyosarcoma soft tissue sarcomas in children and adolescents: clinical features and survival outcome among patients in Children's Oncology Group Phase 3 Study ARST0332

PDF

Survival of children and adolescents with low-risk non-rhabdomyosarcoma soft tissue sarcomas (NRSTS) treated with surgery only: an analysis of 234 patients from the Children’s Oncology Group stud...

PDF

Risk factors and survival outcome in childhood alveolar soft part sarcoma among patients in the Children’s Oncology Group (COG) Phase 3 study ARST0332

PDF

Effects of demographic and tumor characteristics on outcomes in children with malignant peripheral nerve sheath tumors

PDF

Comparison for efficacy and toxicity in 2 vs 3 cycles of cisplatin consolidation therapy in children nasopharyngeal carcinoma

PDF

An assessment of impact of early local progression on subsequent risk for the treatment failure in adolescent and young adult patients with non-metastatic osteosarcoma

PDF

An analysis of disease-free survival and overall survival in inflammatory breast cancer

PDF

Predicting ototoxicity evaluated by SIOP in children receiving cisplatin

PDF

Long-term effects on vertebral body height growth of dose sculpting intensity modulated radiation therapy for children with neuroblastoma

PDF

Body size and the risk of prostate cancer in the multiethnic cohort

PDF

An assessment of necrosis grading in childhood osteosarcoma: the effect of initial treatment on prognostic significance

PDF

Predictive factors of breast cancer survival: a population-based study

PDF

Pathogenic variants in cancer predisposition genes and risk of non-breast multiple primary cancers in breast cancer patients

PDF

Comparison of secondary and primary thyroid cancer in adolescents and young adults (AYA)

PDF

Predicting hospital length of stay (LOS) using the National Inpatient Sample

Asset Metadata

Creator Qian, Qian (author)

Core Title Body wall primary tumors in non-rhabdomyosarcoma soft tissue sarcoma

Contributor Electronically uploaded by the author (provenance)

School Keck School of Medicine

Degree Master of Science

Degree Program Biostatistics

Publication Date 11/16/2018

Defense Date 11/15/2018

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

Tag body wall primary tumors,Cox-proportional hazards regression,non-rhabdomyosarcoma soft tissue sarcoma,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Advisor Barkauskas, Donald (committee chair), Dang, Ha (committee member), Franklin, Meredith (committee member)

Creator Email qianq@usc.edu,qianqian9410@yeah.net

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-106944

Unique identifier UC11675858

Identifier etd-QianQian-6972.pdf (filename),usctheses-c89-106944 (legacy record id)

Legacy Identifier etd-QianQian-6972-1.pdf

Dmrecord 106944

Document Type Thesis

Format application/pdf (imt)

Rights Qian, Qian

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 a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA