University of Southern California Dissertations and Theses

Randomized clinical trial generalizability and outcomes for children and adolescents with high-risk acute lymphoblastic leukemia

(USC Thesis Other)

Randomized clinical trial generalizability and outcomes for children and adolescents with high-risk acute lymphoblastic leukemia

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Copyright 2020 Bo Yuan
Randomized Clinical Trial Generalizability and Outcomes for Children and
Adolescents with High-Risk Acute Lymphoblastic Leukemia
By
Bo Yuan
A Thesis Presented to the
FACULTY OF THE USC KECK SCHOOL OF
MEDICINE UNIVERSITY OF SOUTHERN
CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
Biostatistics
December 2020
ii
Acknowledgments
I would like to express my gratitude to my thesis mentor, Dr. Lingyun Ji, for the support
and engagement throughout my master's thesis learning process. Thank you for sharing your
precious time instructing me on how to analyze data and write a paper.
I would also like to thank Dr. Wendy Jean Mack and Dr. Todd A. Alonzo for being my
committee members and providing valuable comments and remarks to further improve my
thesis.
iii
Table of Contents
Acknowledgments ......................................................................................................................................ii
List of Tables ............................................................................................................................................. iv
List of Figures ............................................................................................................................................. v
Abstract ...................................................................................................................................................... vi
Introduction ................................................................................................................................................ 1
Materials and Methods .............................................................................................................................. 2
RER patients ............................................................................................................................................................ 3
SER patients ............................................................................................................................................................ 4
Statistical analysis ................................................................................................................................................... 4
Results ......................................................................................................................................................... 5
All Patients .............................................................................................................................................................. 5
RER and SER patients ............................................................................................................................................. 6
Randomized and Non-Randomized RER patients .................................................................................................. 7
Randomized and Non-Randomized SER patients ................................................................................................. 10
Toxicity.................................................................................................................................................................. 12
Conclusions ............................................................................................................................................... 15
Discussion .................................................................................................................................................. 16
References ................................................................................................................................................. 18
iv
List of Tables
Table 1. Characteristics of randomized and non-randomized SER vs RER patients .................... 7
Table 2. Comparisons of toxicities (grade 3, 4, and 5) between RER patients who were
randomized vs. not randomized (induction cycle) ........................................................................ 12
Table 3. Comparisons of toxicities (grade 3, 4, and 5) between SER patients who were
randomized vs. not randomized (induction cycle) ........................................................................ 13
Table 4. Comparisons of toxicities (grade 3, 4, and 5) between SER patients who were
randomized vs. not randomized (consolidation cycle) ................................................................. 14
v
List of Figures
Figure 1. CCG-1961 consort diagram ............................................................................................ 7
Figure 2. EFS/OS for RER patients, by randomization groups (combined non-rand groups)....... 9
Figure 3. EFS/OS for RER patients, by randomization groups ................................................... 10
Figure 4. EFS/OS for SER patients, by randomization groups (combined non-rand groups) ......11
Figure 5. EFS/OS for SER patients, by randomization groups .................................................... 12

vi
Abstract
The CCG-1961 study was a Children’s Cancer Group (CCG) Phase III clinical trial,
which investigated 2078 children and adolescents with “higher risk” acute lymphoblastic
leukemia (ALL) who were enrolled between November 1996 and May 2002. After the induction
cycle, patients who met randomization criteria were randomized into treatment arms and
continued treatment. However, a group of patients who met randomization criteria was not
randomized due to physicians'/parents' choices or other reasons. In this thesis, we compared
outcomes of patients who were randomized vs. those who were not randomized to understand if
the two groups of patients had similar Event-free Survival (EFS) or Overall Survival (OS).
Our results showed that patients who met randomization criteria but were not randomized
generally had worse outcomes than patients who were randomized. Both EFS and OS were
statistically significantly worse in non-randomized patients than in randomized patients, which
was true both with Rapid-Early Responders (RER) and Slow-Early Responders (SER) (p<0.0001
for both EFS and OS for RER; EFS: p=0.0001, OS: p=0.0078 for SER). Our findings
demonstrated that patients who were randomized in Phase III clinical trials might not be
representative of the entire patient population eligible for those trials, and as a result, using their
outcomes as historical controls in the design of new clinical trials should be handled with
caution.

Key Points: Randomized and non-randomized patients in a Phase III clinical trial had
significantly different outcomes.

1
Introduction
A clinical trial is an investigation of human subjects that tests the clinical,
pharmacokinetic, or pharmacodynamic effects of an experimental drug or device on those
subjects. The goal of conducting a clinical trial is to ascertain the safety and effectiveness of a
medical product
1
. Before commercialization, a new treatment is usually evaluated in three phases
of clinical trials. Each phase is designed to answer certain questions about the product
2
. Phase I
clinical studies assess the safety in a small group of people. The goal in the phase I trial is to
investigate doses and side-effects of the new treatment. Phase II studies involve a comparatively
larger number of patients (sometimes up to several hundred) with a disease of interest and
gathers preliminary data on the effectiveness of the new treatment. In cancer therapeutics, phase
II studies not only evaluate the effectiveness of treatment but also help to justify the initiation of
a subsequent phase III study
3
. Phase III studies compare a new treatment with the current
standard therapy and further investigate safety as well as efficacy
4
.
When evaluating the efficacy of a new treatment, using outcomes of patients treated on
previously completed trials as historical controls is frequently adopted in oncology and pediatric
clinical trials
5
. This practice provides historical data that are from past studies to current trials,
which may increase the statistical power and reduce enrollment time and expense
6
. Use of
historical controls may occur in both phase II and phase III trials, where investigators use single-
arm studies (all enrolled patients receiving the experimental treatment) and compare the
experimental treatment's therapeutic efficacy to the standard treatment for the study population.
Such trials are often designed to evaluate if the new treatment improves the response rate over a
historical control rate
3
, for example, if the new treatment improves long-term clinical outcomes
such as Event-free Survival (EFS) and Overall Survival (OS). However, if the historical control

2
data are unsuitable or biased due to the lack of randomization, single-arm trials could lead to
biased conclusions on the treatment effects
6
.
Historical control rates are often based on patients' outcomes from completed,
randomized Phase III clinical trials. In Phase III clinical trials, not every trial-eligible patient is
randomized. Using outcomes of randomized patients as historical control rates would be
reasonable if their outcomes are representative of the population.
These issues are particularly relevant to pediatric ALL (acute lymphoblastic leukemia).
Patients often receive an initial induction cycle (IND) and are then randomly assigned to
treatment arms after completion of IND
7
. Among patients who meet the randomization criteria
after IND, there is often a subgroup of patients who are not randomized and are taken off
protocol treatment due to various reasons (e.g., refusal, protocol deviation, worsened disease
status, AEs). In this thesis, our goal was to use a CCG (Children’s Cancer Group) Phase III
clinical trial in ALL, CCG-1961, to investigate if randomized patients had similar outcomes of
interest compared to patients who were not randomized. This research question is of interest
since the comparison between the two groups of patients may enrich our understanding on
whether outcomes of randomized patients are representative of the population, and whether their
outcomes can serve as reliable historical control rates in the design of new Phase II or Phase III
clinical trials.
Materials and Methods
The CCG-1961 Phase III clinical trial was open to enrollment from September 1996 to
May 2002. Patients with ALL aged 1-9 years with initial WBC ≥ 50,000/ 𝜇𝑙 , or aged between 10
and 21 years with any WBC were eligible to be enrolled. Diagnosis was assessed based on
morphologic, histochemical, and immunophenotypic features of leukemia cells
8
.

3
Induction therapy consisted of intravenous VCR 1.5 mg/m
2
per week and daunorubicin
(DNM) 25 mg/m
2
per week for 4 weeks; oral prednisone 10 mg/m
2
per day for 28 days; native e.
coli asparaginase 6000 Units/m
2
intramuscularly thrice weekly for 9 doses; intrathecal (IT)
cytarabine on day 1 and IT methotrexate (MTX) on days 8 and 29. All patients had a bone
marrow aspirate performed on day 8. Per the percentage of bone marrow blasts, patients were
classified as rapid early response (RER) or slow early response (SER)
9
.
Rapid early response (RER) patients with CNS-3 disease or who were Philadelphia
chromosome positive were excluded from randomization. Slow early response (SER) patients
with Philadelphia chromosome positive were excluded from randomization. In addition to the
exclusion criteria for randomization, patients could also be taken off protocol per physician or
parent choice without proceeding to randomization
9
. RER and SER patients who met
randomization criteria and did not refuse randomization or drop from the protocol were
randomized per protocol, details provided below.
RER patients
Patients who had no CNS disease present at diagnosis and were not Philadelphia
Chromosome positive with Day 8 Bone Marrow blasts <25% were categorized as rapid early
response (RER) and were randomly assigned to one of four treatment regimens with a 2 ×2
factorial design
9
:
• Regimen A Standard BFM with one course of Delayed Intensification
• Regimen B Standard BFM with two courses of Delayed Intensification
• Regimen C: Increased intensity (augmented) BFM with one course of Delayed
Intensification
• Regimen D: Increased intensity (augmented) BFM with two courses of Delayed
Intensification

4
SER patients
Patients with Day 8 Bone Marrow (BM) blasts >25% (M3, slow early response, SER)
who had < 50% CD19 antigen positive or CD19 negative blasts in their marrow at diagnosis
continued to be treated with standard BFM Induction. Patients with M1(< 5% BM blasts) or M2
(5 - 25% BM blasts) marrows at the end of induction (Day 29) received augmented BFM
consolidated Interim Maintenance (IM) I and were then randomly assigned to two courses of
Delayed Intensification with either doxorubicin or idarubicin.
• Augmented BFM with two courses and Delayed Intensification plus Doxorubicin
• Augmented BFM with two courses and Delayed Intensification plus
Idarubicin/Cyclophosphamide
SER patients with marrow blasts ≥25% (M3) at the end of IM#1 were excluded from
randomization
9
.
Any adverse events (AEs) during the trial were evaluated per Common Terminology
Criteria for Adverse Events and reported through CCG Toxicity and Complications data
collection forms. The site, measure, and grade for all toxicities (grades 3 and 4) were required to
be reported
10
.
Statistical analysis
For the comparison of the patients who were randomized and those who were not
randomized, two primary clinical endpoints were evaluated: Event-Free Survival (EFS) and
Overall Survival (OS). EFS and OS were calculated from day 29 of induction for RER patients
and day 57 of the first IM phase for SER patients; for patients who were taken off protocol
earlier, the date off protocol was used as day 0. EFS was defined as the length of time to first
event (relapse, second malignancy, or death) or last follow-up for patients who were event-free.

5
Overall Survival (OS) was defined as the time to death or last follow-up for patients who were
alive
7
. Cumulative survival estimates were calculated by the Kaplan-Meier procedure; the
standard error of the cumulative survival estimates used Greenwood’s method. The log-rank test
was used to compare the EFS and OS outcomes between randomized vs. non-randomized
patients; these comparisons were conducted separately for RER and SER patients. Hazard ratios
(HR) between the randomized and two non-randomized patient groups were compared using the
Cox proportional hazards regression. Patient characteristics were examined between randomized
and non-randomized groups using Chi-squared tests for homogeneity of proportions. Pearson's
Chi-squared test and Fisher's Exact tests were employed to test for homogeneity of proportions
of grade 3+ toxicities between randomized and non-randomized patients; these comparisons
were also conducted separately for RER and SER patients. All p-values were reported as two-
sided. Statistical analysis was performed using Stata 15 software
11
.
Results
All Patients
A total of 2,078 people were enrolled in CCG-1961 (Figure 1). Twenty-one participants
were ineligible for the study (six patients due to improper consents, two patients were
administered chemotherapy before signing informed consent, eight patients were found to have
malignancies other than ALL, two patients had steroid exposure longer than 48 h before
diagnosis, two standard-risk patients were erroneously enrolled on 1961, and one patient did not
have an evaluable bone marrow result). Among the 2,057 eligible patients, 27 patients died in
induction, 24 patients had Day 29 Bone marrow blasts greater than 25%, and 36 patients had
invaluable Day 8 or Day 29 marrow aspirates. The remaining 1970 patients were classified as
either RER or SER patients.

6
RER and SER patients
For the 1970 subjects, 71.4% were RER patients (n=1406) and 28.6% were SER patients
(n=564). The 5-year and 10-year EFS rates (SE) for RER patients were 77.5% (1.2%) and 75.2%
(1.2%). The 5-year and 10-year OS rates (SE) were 84.9% (1.0%) and 84.1% (1.1%). The 5-year
and 10-year EFS rates (SE) for SER patients were 74.3% (2.1%) and 70.1% (2.3%). The 5-year
and 10-year OS rates (SE) were 79.0% (2.0%) and 76.6% (2.1%). As described in the “Materials
and Methods” section, RER patients who met the randomization criteria were randomly assigned
to one of four treatment arms (standard or longer and stronger post-induction intensification
therapy). SER patients who met the randomization criteria were randomly assigned to one of two
treatment arms (two courses of Delayed Intensification with either doxorubicin or idarubicin).
We evaluated all RER patients and SER patients and identified the subgroup of patients
who were not randomized, with the intention to compare their outcomes to patients who were
randomized. A total of 1406 patients were defined as RER patients. Among those patients, 77
patients were excluded from randomization due to CNS-3 or Philadelphia chromosome positive
at diagnosis. The remaining 1329 RER patients were potentially eligible for randomization.
There were 564 SER patients in total, and 33 patients were not randomized because they were
Philadelphia chromosome positive at diagnosis. The remaining 531 SER patients were
potentially eligible for randomization. Trial outcome comparisons between randomized and non-
randomized patients in the RER and SER groups are summarized below.

7

Figure 1. CCG-1961 consort diagram

Randomized and Non-Randomized RER patients
A total of 1329 RER patients were potentially eligible for randomization, among whom
1299 were randomized in the 2 × 2 design to the four regimens described above, and 30 (2.3%
of total RER patients) patients were not randomized despite meeting randomization criteria
(Figure 1). Among the non-randomized patients, 22 patients refused to participate in the
randomization, and 8 patients were not randomized for various reasons (e.g., protocol deviation,
worsened disease status, AEs, or reasons not documented well). There were no significant
differences found in presenting features (Table 1).

Table 1. Characteristics of randomized and non-randomized SER vs RER patients
Variables RER Patients SER Patients
Rand
(N=1299)
Non-rand
(N=30)
p-value
b
Rand
(N=447)
Non-rand
(N=84)
p-value
b

Age (yrs) 0.49 0.57
1-9 477 (36.7%) 10 (33.3%) 178 (39.8%) 35 (41.7%)
10-15 658 (50.7%) 18 (60.0%) 215 (48.1%) 36 (42.9%)
16+ 164 (12.6%) 2 (6.7%) 54 (12.1%) 13 (15.5%)

8
Table 1. Characteristics of randomized and non-randomized SER vs RER patients
Variables RER Patients SER Patients
Rand
(N=1299)
Non-rand
(N=30)
p-value
b
Rand
(N=447)
Non-rand
(N=84)
p-value
b

Sex 0.58 0.79
Male 758 (58.4%) 19 (63.3%) 273 (61.1%) 50 (59.5%)
Female 541 (41.6%) 11 (36.7%) 174 (38.9%) 34 (40.5%)
Race 0.68 0.008
Caucasian 939 (87.8%) 23 (85.2%) 317 (87.3%) 56 (81.2%)
Black 88 (8.2%) 2 (7.4%) 37 (10.2%) 6 (8.7%)
Other
a
43 (4.0%) 2 (7.4%) 9 (2.5%) 7 (10.1%)
Unknown 229 3 84 15
WBC x 10
3
/mm
3
0.45 0.46
<50 662 (51.0%) 15 (50.0%) 188 (42.2%) 32 (38.1%)
50–100 315 (24.3%) 5 (16.7%) 124 (27.8%) 21 (25.0%)
100+ 320 (24.7%) 10 (33.3%) 134 (30.0%) 31 (36.9%)
Missing 2 0 1 0
Immunophenoty
pe
0.22 0.34
B lineage 723 (67.8%) 15 (53.6%) 213 (61.6%) 37 (58.7%)
T lineage 237 (22.2%) 10 (35.7%) 108 (31.2%) 24 (38.1%)
Mixed 106 (10.0%) 3 (10.7%) 25 (7.2%) 2 (3.2%)
Unknown 233 2 101 21
a. Other includes Hispanic, Oriental, Hawaiian, Native American, Indian Subcontinent, Filipino, and other
b. P-values: Pearson's chi-square, excluding missing/unknown

We found that both EFS and OS were statistically significantly better for the randomized
group than the non-randomized group (Log rank p<0.0001 for both EFS and OS, Figure 2).
Among those randomized patients, the 5-year and 10-year EFS rates (SE) were 76.9% (1.2%)
and 74.3% (1.2%), and the 5-year and 10-year OS rates (SE) were 84.3% (1.0%) and 83.2%
(1.1%). For those RER patients who met the inclusion criteria but were not randomized, the 5-
year and 10-year EFS rates (SE) were 49.8% (9.2%) and 35.6% (9.8%), and the 5-year and 10-
year OS rates (SE) were 56.3% (9.1%) and 47.5% (9.6%).

9
A. EFS of RER patients, rand vs. non-rand B. OS of RER patients, rand vs. non-rand

Figure 2. EFS/OS for RER patients, by randomization groups (combined non-rand groups)

We also compared EFS and OS of RER patients who met randomization criteria but
refused randomization to those who were not randomized due to various other reasons (e.g.,
protocol deviation, worsened disease status, AEs, or reasons not documented well). The non-
randomized group, due to other various reasons, had worse outcomes compared with patients
who refused randomization, a subgroup of them having early EFS events (Figure 3). The EFS
HR at the same time for non-randomized RER patients with other various reason was 13.93
times than that of the randomized patients (p<0.001, 95% CI: 6.58, 29.50). The OS HR of non-
randomized RER patients with other various reason was 19.25 when compared with the
randomized group (p<0.001, 95% CI: 8.52, 43.48).
Patients who refused randomization showed better EFS/OS than patients who were not
randomized due to other reasons, but outcomes of the former group were still worse than patients
who were randomized. The EFS HR for non-randomized RER patients who refused to enroll
randomization was 2.45 times than that of the randomized patients (p= 0.003, 95% CI: 1.34,
4.48), while The OS HR was 3.8 (p=0.001, 95%CI=1.63, 6.20).

10
A. EFS of RER patients, randomized vs. non-
randomized
B. OS of RER patients, randomized vs. non-
randomized

Figure 3. EFS/OS for RER patients, by randomization groups

Randomized and Non-Randomized SER patients
A total of 531 SER patients potentially met the randomization criteria. Among those
patients, 447 patients were randomized and received one of two treatments described in the
Methods section. A total of 84 SER patients (15.8% of total SER patients) were not randomized.
Within the non-randomized group, 37 patients refused randomization and 47 patients were not
randomized due to various other reasons. Randomized and non-randomized SER patients’ race
was found significantly different in presenting features (Table 1). Non-randomized SER patients
were more likely to be other race, and randomized patients were more likely to be Caucasian or
Black. Outcomes for SER randomized patients were significantly better than SER non-
randomized patients (EFS: Log rank p=0.0003, OS: Log rank p=0.0042). The 5-year and 10-year
EFS rates (SE) for SER randomized patients were 71.2% (2.0%) and 67.9% (2.1%), and the 5-
year and 10-year OS rates (SE) were 76.9% (1.9%) and 74.9% (2.0%), respectively. For SER
patients who were not randomized, the 5-year and 10-year EFS rates (SE) were 54.7% (5.4%)

11
and 54.7% (5.4%), and the 5-year and 10-year OS rates (SE) were 65.6% (5.3%) and 65.6%
(5.3%), respectively (Figure 4).
A. EFS of SER patients, randomized vs. non-
randomized
B. OS of SER patients, randomized vs. non-
randomized

Figure 4. EFS/OS for SER patients, by randomization groups (combined non-rand groups)

Further investigation of subgroups of SER non-randomized patients showed that the
patients who were not randomized due to other reasons had worse outcomes compared with the
non-randomized group who refused randomization (Log rank p=0.0001 for EFS, Log rank
p=0.0078 for OS, Figure 5). The EFS HR for non-randomized SER patients with other reason
was 2.53 times than that of the randomized patients (p<0.001, 95% CI: 1.65, 3.88). The OS HR
of non-randomized patients with other reason was 2.15 when compared with the randomized
group (p=0.004, 95% CI: 1.28, 3.59).
The non-randomized patients with other reasons had early event (relapse, second
malignancy, or death) compared to patients who refused to enroll randomization. Both EFS and
OS HR of refusal group were not significant (EFS: HR=1.34, p=0.319, 95% CI: 0.756, 2.36; OS:
HR=1.49, p=0.21, 95% CI: 0.80, 2.78).

12
A. EFS of SER patients, randomized vs. non-
randomized
B. OS of SER patients, randomized vs. non-
randomized

Figure 5. EFS/OS for SER patients, by randomization groups

Toxicity
RER patients
For RER patients, 601 (46.27%) of randomized patients and 18 (60.00%) of non-
randomized patients developed grade 3+ adverse events (AEs) during the induction phase (Table
2), which were not significantly different (p=0.14). Examining AEs of different toxicity sites, the
grade 3+ toxicity observed among RER randomized and non-randomized patients were generally
similar (p>0.05), except that the proportion of patients with infections significantly differed
between the two groups (p=0.004). The proportion of patients with grade 3+ infections among
the RER randomized group was 6.62%, compared to 16.67% for the RER non-randomized
group. The most frequent grade 3+ toxicities among RER randomized patients was pancreas
toxicities (with 18.63%, n=242), but were infections for RER non-randomized patients (26.67%,
n=8).
Table 2. Comparisons of toxicities (grade 3, 4, and 5) between RER patients who were
randomized vs. not randomized (induction cycle)
Toxicity Site Randomized
(n=1299)
Not Randomized
(n=30)
p-value

Any Site 601 (46.27%) 18 (60.00%) 0.14*
1. Blood 0 (0.00%) 0 (0.00%)

13
Table 2. Comparisons of toxicities (grade 3, 4, and 5) between RER patients who were
randomized vs. not randomized (induction cycle)
2. Marrow 2 (0.15%) 0 (0.00%) 1**
3. Liver 199 (15.32%) 6 (20.00%) 0.45**
4. Pancreas 242 (18.63%) 4 (13.33%) 0.46*
5. Renal and
Genitourinary
55 (4.23%) 2 (4.23%) 0.37**
6. Gastrointestinal 83 (6.39%) 4 (13.33%) 0.13**
7. Pulmonary 64 (4.93%) 2 (6.67%) 0.66**
8. Cardiac 49 (3.77%) 1 (3.33%) 1**
9. Nervous System 41 (3.16%) 3 (10.00%) 0.07**
10. Skin 15 (1.15%) 0 (0.00%) 1**
11. Allergy 4 (0.31%) 0 (0.00%) 1**
12. Coagulation 142 (10.93%) 3 (10.00%) 1**
13. Hearing 0 (0.00%) 0 (0.00%)
14. Electrolytes 86 (6.62%) 5 (16.67%) 0.19**
15. Infection 112 (8.62%) 8 (26.67%) 0.004**
16. Fever 9 (0.69%) 0 (0.00%) 1**
17. Local 3 (0.23%) 0 (0.00%) 1**
18. Mood 10 (0.77%) 0 (0.00%) 1**
19. Vision 1 (0.08%) 0 (0.00%) 1**
20. Weight Change 13 (1.00%) 0 (0.00%) 1**
21. Performance
(Karnofsky %)
14 (1.08%) 1 (3.33%) 0.29**
*P-values: Pearson's Chi-squared test
** P-values: Fisher’s Exact tests

SER patients
For SER patients, 216 randomized patients (48.32%) and 42 (50.00%) non-randomized
patients had grade 3+ toxicity reactions during the induction cycle (Table 3). The proportion of
patients with any toxicities was not statistically significantly different between the two groups
(p=0.78). Three toxicity sites that showed significant group differences were pulmonary
(p=0.014), electrolytes (p=0.005), and Karnofsky score (p=0.003). The most frequent grade 3+
toxicities for the randomized SER group was pancreas toxicity with 19.02% (n=85), while for
non-randomized patients was liver toxicity (n=17, 20.24%).

Table 3. Comparisons of toxicities (grade 3, 4, and 5) between SER patients who
were randomized vs. not randomized (induction cycle)
Toxicity Site Randomized
(n=447)
Not Randomized
(n=84)
p-value
Any Sites 216 (48.32%) 42 (50.00%) 0.78*
1. Blood 0 (0.00%) 0 (0.00%)

14
Table 3. Comparisons of toxicities (grade 3, 4, and 5) between SER patients who
were randomized vs. not randomized (induction cycle)
2. Marrow 2 (0.15%) 1 (1.19%) 0.40**
3. Liver 80 (17.90%) 17 (20.24%) 0.61*
4. Pancreas 85 (19.02%) 14 (16.67%) 0.61*
5. Renal and Genitourinary 14 (3.13%) 4 (8.00%) 0.51**
6. Gastrointestinal 27 (6.04%) 8 (9.52%) 0.24*
7. Pulmonary 14 (3.13%) 8 (9.52%) 0.01**
8. Cardiac 17 (3.80%) 6 (7.14%) 0.24**
9. Nervous System 18 (4.03%) 5 (5.95%) 0.39**
10. Skin 0 (0.00%) 0 (0.00%)
11. Allergy 0 (0.00%) 0 (0.00%)
12. Coagulation 50 (11.19%) 10 (11.90%) 0.85*
13. Hearing 0 (0.00%) 0 (0.00%)
14. Electrolytes 20 (4.47%) 11 (13.10%) 0.005**
15. Infection 37 (8.28%) 10 (11.90%) 0.28*
16. Fever 1 (0.22%) 2 (2.38%) 0.07**
17. Local 5 (1.12%) 0 (0.00%) 1**
18. Mood 4 (0.89%) 3 (3.57%) 0.08**
19. Vision 0 (0.00%) 0 (0.00%)
20. Weight Change 1 (0.22%) 2 (2.38%) 0.07**
21. Performance (Karnofsky %) 1 (0.22%) 4 (4.76%) 0. 003**
*P-values: Pearson's Chi-squared test
** P-values: Fisher’s Exact tests

After completing the induction cycle, the SER patients were also enrolled in the
consolidation cycle (IM I) before randomization. AE data were available in 447 randomized and
62 non-randomized patients (Table 4). Twenty-two SER patients dropped from the consolidation
cycle after the completion of the induction cycle. Including any grade 3+ toxicity reaction, we
did not observe a significant difference between SER randomized and SER non-randomized
groups (p=0. 57). Electrolytes and pancreas toxicities were statistically significantly different
between randomized and non-randomized groups (p=0.005, p=0.018, respectively). The liver
toxicity reaction had the highest frequency among both randomized and non-randomized groups,
with 19.69% (n=88) and 22.58% (n=14) of the patients, respectively.

Table 4. Comparisons of toxicities (grade 3, 4, and 5) between SER patients who
were randomized vs. not randomized (consolidation cycle)
Toxicity Site Randomized
(n=447)
Not Randomized
(n=62)
p-value
Any Sites 192 (42.95%) 29 (46.77%) 0.57*

15
Table 4. Comparisons of toxicities (grade 3, 4, and 5) between SER patients who
were randomized vs. not randomized (consolidation cycle)
1. Blood 0 (0.00%) 0 (0.00%)
2. Marrow 0 (0.00%) 0 (0.00%)
3. Liver 88 (19.69%) 14 (22.58%) 0.59*
4. Pancreas 17 (3.80%) 7 (11.29%) 0.02**
5. Renal and Genitourinary 4 (0.89%) 0 (0.00%) 1**
6. Gastrointestinal 52 (11.63%) 6 (9.68%) 0.65 *
7. Pulmonary 11 (2.46%) 1 (1.61%) 1**
8. Cardiac 9 (2.01%) 3 (4.84%) 0.17**
9. Nervous System 25 (5.59%) 5 (8.06%) 0.39**
10. Skin 12 (2.68%) 1 (1.61%) 1**
11. Allergy 9 (2.01%) 1 (1.61%) 1**
12. Coagulation 16 (3.58%) 1(1.61%) 0.71**
13. Hearing 0 (0.00%) 0 (0.00%)
14. Electrolytes 6 (1.34%) 5 (8.06%) 0.005**
15. Infection 54 (12.08%) 7 (11.29%) 0.86*
16. Fever 1(0.22%) 0 (0.00%) 1**
17. Local 2 (0.45%) 0 (0.00%) 1**
18. Mood 4 (0.89%) 1 (1.61%) 0.48**
19. Vision 0 (0.00%) 1 (1.61%) 0.12**
20. Weight Change 9 (2.01%) 0 (0.00%) 0.61**
21. Performance
(Karnofsky %)
3 (0.67%) 1 (1.61%) 0.41**
*P-values: Pearson's Chi-squared test
** P-values: Fisher’s Exact tests

Conclusions
In this thesis, we evaluated patients enrolled in CCG-1961 and identified those who were
randomized vs. those who were eligible but not randomized. We compared EFS and OS between
the randomized groups and the non-randomized groups. In general, non-randomized patients had
significantly worse outcomes than those who were randomized. Further evaluation of the non-
randomized groups showed that patients who were documented to have clearly refused
randomization generally had better outcomes than those who were not randomized for various
other reasons.
Considering the differences in outcomes between patients who were randomized vs. those
who were not randomized, we concluded that in Phase III clinical trials, outcomes of randomized
patients might not represent the outcomes of the entire population. As a result, when designing

16
new clinical trials, use of outcomes of randomized patients from completed Phase III clinical
trials as historical control rates could potentially lead to some bias. The determination of
historical control rates needs to be conducted with caution.
We further examined the proportion of patients with grade 3+ AEs among the randomized
and non-randomized groups in order to understand whether the non-randomized patients were
not randomized due to more severe AEs occurring in the induction cycle (RER) or the induction
or consolidation cycles (SER). Although there was no consistent pattern for higher AE rates in
the non-randomized groups when compared to the randomized groups, the non-randomized
groups showed higher AE rates in specific toxicity sites. Besides AEs, the factors that caused
patients to be taken off protocol without randomization needs further investigation.
Among RER patients, we observed that the proportion of patients having infections in the
non-randomized group was significantly higher than in the randomized group. For SER patients,
electrolyte abnormalities in the induction cycle or the consolidation cycle were significantly
more frequent in the non-randomized group than in the randomized group. Pulmonary toxicities
and Karnofsky conditions were also significantly worse in the non-randomized SER patients
compared to the randomized SER patients during the induction cycle.
Discussion
In this thesis, our analysis of CCG-1961 data showed that the randomized groups had
better outcomes than non-randomized groups. These results indicated that for ALL Phase III
clinical trials, outcomes of randomized patients may not be representative of the entire
population. Hence, when designing a new clinical trial, if one determines the historical control
rates by solely utilizing outcomes of randomized patients, that may lead to biased estimate of the
historical rates.

17
The magnitude of this potential bias could depend on the differences in outcomes of
randomized vs. non-randomized patients, as well as the proportion of patients who did not
proceed to randomization, among other factors. For CCG-1961, the percentage of SER patients
who did not proceed to randomization was 15.8%, representing a non-ignorable group of
patients.
Further research will be conducted to evaluate the proportion of non-randomized patients
in a broader choice of Phase III clinical trials in ALL. Associations between this proportion vs.
intensity of treatment or stage of cancer will be assessed. Reasons for patients not to proceed to
randomization will be further investigated. After determining the extent and magnitude of
differences in outcomes of randomized patients vs. non-randomized patients, recommendations
will be made in the determination of historical control rates based on such Phase III clinical
trials.

18
References
1. E6(R2) Good Clinical Practice: Integrated Addendum to ICH ...
https://www.fda.gov/media/93884/download. Published March 2018. Accessed September 14,
2020.
2. Types and Phases of Clinical Trials: What Are Clinical Trial Phases? American Cancer
Society. https://www.cancer.org/treatment/treatments-and-side-effects/clinical-trials/what-you-
need-to-know/phases-of-clinical-trials.html. Published August 18, 2020. Accessed September 14,
2020.
3. Thall PF, Simon R. Incorporating historical control data in planning phase II clinical
trials. Statistics in medicine. 1990;9(3):215-228. doi:10.1002/sim.4780090304
4. National Comprehensive Cancer Network. Phases of clinical trials.
https://www.nccn.org/patients/resources/clinical_trials/phases.aspx. Accessed September 15,
2020.
5. Lewis CJ, Sarkar S, Zhu J, Carlin BP. Borrowing From Historical Control Data in Cancer
Drug Development: A Cautionary Tale and Practical Guidelines. Statistics in Biopharmaceutical
Research. 2019;11(1):67-78. doi:10.1080/19466315.2018.1497533

6. Viele K, Berry S, Neuenschwander B, et al. Use of historical control data for assessing
treatment effects in clinical trials. Pharm Stat. 2014;13(1):41-54. doi:10.1002/pst.1589

7. Thall PF, Simon R. Incorporating historical control data in planning phase II clinical
trials. Statistics in medicine. 1990;9(3):215-228. doi:10.1002/sim.4780090304

8. Seibel NL, Steinherz PG, Sather HN, et al. Early postinduction intensification therapy
improves survival for children and adolescents with high-risk acute lymphoblastic leukemia: a
report from the Children's Oncology Group. Blood. 2008;111(5):2548-2555. doi:10.1182/blood-
2007-02-070342

9. Teinherz PG, Seibel NL, Sather H, et al. Treatment of higher risk acute lymphoblastic
leukemia in young people (CCG-1961), long-term follow-up: a report from the Children's
Oncology Group. Leukemia. 2019;33(9):2144-2154. doi:10.1038/s41375-019-0422-z

10. Common Terminology Criteria for Adverse Events (CTCAE) Version 5.0.
https://ctep.cancer.gov/protocoldevelopment/electronic_applications/docs/CTCAE_v5_Quick_R
eference_8.5x11.pdf. Published November 17, 2017. Accessed October 11, 2020.

11. StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Clinical outcomes of allogeneic hematopoietic stem cell transplant in acute lymphoblastic leukemia patients: a quality improvement project and systematic review meta-analysis

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

Induction therapy in relapse adult acute lymphoblastic leukemia

PDF

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

PDF

The effect of cytomegalovirus on gene expression of pediatric acute lymphoblastic leukemia

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

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

Identifying prognostic gene mutations in colorectal cancer with random forest survival analysis

PDF

Biomarker-driven designs in oncology

PDF

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

PDF

The effects of late events reporting on futility monitoring of Phase III randomized clinical trials

PDF

The interplay between tobacco exposure and polygenic risk score for growth on birthweight and childhood acute lymphoblastic leukemia

PDF

Effects of post-menopausal hormone therapy on arterial stiffness in the ELITE trial

PDF

An assessment of the dependence of routine assessments of the scheduling the estimation of event rates in follow‐up trials

PDF

Racial/ethnic differences in patient experiences with health care in association with the Gleason’s score level at prostate cancer diagnosis: findings from the SEER-CAHPS data

PDF

Elevated fasting free fatty acids in overweight Latino children and adolescents

PDF

Need for tissue plasminogen activator for central venous catheter malfunction and its association with occurrence of vVenous thromboembolism

PDF

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

PDF

High intensity interval training for breast cancer patients receiving anthracycline-based chemotherapy

Asset Metadata

Creator Yuan, Bo (author)

Core Title Randomized clinical trial generalizability and outcomes for children and adolescents with high-risk acute lymphoblastic leukemia

School Keck School of Medicine

Degree Master of Science

Degree Program Biostatistics

Publication Date 12/07/2020

Defense Date 12/02/2020

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

Tag acute lymphoblastic leukemia (ALL),Children’s Cancer Group (CCG) Phase III clinical trial,non-randomized patients,OAI-PMH Harvest,outcomes,randomized patients

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Ji, Lingyun (committee chair), Alonzo, Todd A. (committee member), Mack, Wendy Jean (committee member)

Creator Email bonnieyuan21@gmail.com,byuan616@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-408710

Unique identifier UC11668807

Identifier etd-YuanBo-9186.pdf (filename),usctheses-c89-408710 (legacy record id)

Legacy Identifier etd-YuanBo-9186.pdf

Dmrecord 408710

Document Type Thesis

Rights Yuan, Bo

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