Outcomes in adolescent and young adult patients (16 to 30 years) compared to younger patients treated for high-risk B-lymphoblastic leukemia: Report from Children’s Oncology Group Study AALL0232

Abstract

Adolescent and young adult (AYA) patients 16–30 years old with high-risk acute lymphoblastic leukemia (HR-ALL) have inferior outcomes compared to younger HR-ALL patients. AALL0232 was a Phase 3 randomized Children’s Oncology Group trial for newly diagnosed HR B-ALL (1–30 years). Between 2004 and 2011, 3,154 patients enrolled with 3,040 eligible and evaluable for Induction. AYA patients comprised 20% of patients (16–21 years, n= 551; 22–30 years, n=46). 5-year event-free survival and overall survival was 65.4±2.2% and 77.4±2.0% for AYA patients compared to 78.1±0.9% and 87.3±0.7% for younger patients (p<0.0001). Five-year cumulative incidence of relapse was 18.5±1.7% for AYA patients and 13.5±0.7% for younger patients (p=0.006), largely due to increased marrow relapses (14.0±1.5% versus 9.1±0.6%; p<0.0001). Additionally, induction failure rate was higher in AYA (7.2±1.1% versus 3.5±0.4%; p<0.001) and post-induction remission deaths were significantly higher in AYA (5.7±1.0% versus 2.4±0.3%; p<0.0001). AALL0232 enrolled the largest number of AYA B-ALL patients to date, demonstrating significantly inferior survival and greater rates of treatment-related toxicities compared to younger patients. Although treatment intensification has improved outcomes in younger patients, they have not been associated with the same degree of improvement for older patients.

Introduction

Although 5-year overall survival (OS) rates for pediatric patients with B-lymphoblastic leukemia (B-ALL) are currently >90%^1–3, survival remains inferior for adolescent and young adult (AYA) patients.^4–7 AYA patients ages 16 to 20 years with B-ALL treated on Children’s Cancer Group (CCG) studies from 1988 to 1995 had 7-year event-free survival (EFS) and OS rates of 63% and 67%.⁴ Comparatively, AYA patients (16 to 20 years) treated on 5 sequential adult Cancer and Leukemia Group B (CALGB) studies from 1988–2001 had 7-year EFS and OS of 34% and 46%.⁴ In the subsequent CCG 1961 study enrolling patients from 1996 to 2002, AYA outcomes were somewhat improved with 5-year EFS and OS of 71.5%±3.6% and 77.5%±3.3%.⁸ Though incompletely understood, reasons for the stark differences in survival between younger and AYA patients treated on pediatric protocols could include the greater incidence of unfavorable biology identified in older patients (i.e., Philadelphia chromosome-positive (Ph⁺) ALL and Philadelphia chromosome-like (Ph-like) ALL)^9–12 and fewer cases of favorable biology (i.e. hyperdiploidy and ETV6-RUNX1 fusion ALL)¹⁰. Other factors that contribute to lower survival in AYA patients could include non-adherence to treatment, higher rates of treatment-related toxicities, lower rates of inclusion into clinical trials, insufficient medical insurance, and/or less access to designated pediatric centers for leukemia care.^4,13–15 As referenced above, when AYA patients with B-ALL have been treated on pediatric-inspired protocols, survival has been significantly higher than in adult studies.^13,16 The CALGB study 10403, modeled after the Children’s Oncology Group (COG) AALL0232 high-risk (HR) B-ALL trial, recently reported results on 295 AYA patients (ages 17–39 years) with B-ALL or T-ALL, median age 24 years, with 3-year EFS 59% and OS 73%; much better than the earlier CALGB trials.¹⁷

COG AALL0232 enrolled 3,154 pediatric and AYA patients with newly diagnosed HR B-ALL between January 2004 and January 2011, reporting 5-year EFS and OS of 75.2%±1.1% and 85.0%±0.9% respectively.¹⁸ Here we report outcome comparisons between AYA (16 to 30 years) and younger patients (<16 years) with HR B-ALL treated on COG AALL0232.

Subjects and Methods

Subject Characteristics

AALL0232 opened to subject entry in January 2004 and closed to enrollment in January 2011 [NCT00075725]. Subjects with newly diagnosed NCI HR B-ALL (1–9 years old with an initial white blood cell count (WBC) ≥50,000/microliter or 10–30 years old with any WBC) were eligible. Patients with Down syndrome were initially included on AALL0232 but were later excluded from enrollment due to increased toxicity. The few patients with Down syndrome that did enroll (n=41) were excluded from this AYA analysis. The diagnosis was determined by morphologic and immunologic features.^19,20 Central nervous system (CNS) status was defined based on cerebrospinal fluid (CSF) obtained prior to therapy as CNS1 (no blasts on cytocentrifugation), CNS2 (CSF WBC <5/microliter with blasts on cytocentrifugation), or CNS3 (CSF WBC ≥ 5/microliter with blasts on cytocentrifugation and/or clinical signs of CNS leukemia). AALL0232 was approved by the NCI and institutional review boards of participating institutions. Informed consent was obtained from subjects or a parent/guardian prior to starting protocol therapy in accordance with Department of Health and Human Services guidelines. For purposes of this study, AYA was defined as ages 16 to 30 years as a comparator to the prior CALGB and CCG 1961 studies reporting AYA outcomes.⁴ Additionally, a receiver-operating characteristic (ROC) analysis incorporating both EFS and age (limited to those ≥10 years) was performed to define the age cutoff associated with inferior outcome and identified the optimal age cutoff to be 16 years (HR=1.51, [1.274, 1.784]; p=0.0000018), consistent with the a priori lower age limit used to define AYA in this report.

Characterization of cytogenetic and genetic features

As described previously, all patients had cytogenetic, fluorescence in situ hybridization (FISH), and DNA index (to identify hypodiploidy) studies performed locally or centrally and reviewed centrally_. FISH studies included assessment for simultaneous trisomies of chromosomes 4, 10 and 17 (triple trisomy, TT), ETV6-RUNX1 (TEL-AML1) fusion, KMT2A (MLL) gene rearrangements (MLL-R), and Ph+ ALL. A subset of 861 patients had testing, performed retrospectively, for the Ph-like gene expression profile as described previously. ^{21 9,10,22}

Minimal residual disease (MRD)

MRD at day 29 of induction was measured by flow cytometry at one of two central reference laboratories as previously described ¹⁸

Treatment

AALL0232 utilized a 2 × 2 factorial design with a COG modified augmented intensity Berlin-Frankfurt-Münster (BFM) backbone.²⁰ Eligible subjects were randomized at study entry to receive dexamethasone (10 mg/m²/day) days 1–14 versus prednisone (60 mg/m²/day) days 1–28 during Induction and high dose methotrexate with leucovorin rescue (HD-MTX) versus Capizzi escalating dose MTX (without leucovorin rescue) plus pegaspargase (C-MTX) during Interim Maintenance 1.¹⁸ Early response was used to stratify treatment. Patients were classified as Rapid Early Responders (RER) if they had an M1 marrow (<5% blasts) by Induction day 15 and <0.1% MRD in the bone marrow (BM) at day 29. Patients who achieved morphologic remission (M1) by day 29 but had either an M2 (5–25% blasts) or M3 (>25% blasts) BM on Induction day 15 or day 29 BM MRD ≥0.1% were considered Slow Early Responders (SER). SER subjects were non-randomly assigned to receive a second Delayed Intensification, a second Interim Maintenance with C-MTX, and 12 Gy of prophylactic cranial irradiation. Additionally, subjects with an M2 BM or ≥1% MRD at day 29 received two additional weeks of Induction therapy and were considered SER if the day 43 marrow was M1 with <1% MRD, otherwise they were considered Induction failures and removed from protocol therapy, as were those with an M3 marrow at day 29. CNS3 subjects were non-randomly assigned to receive HD-MTX and 18 Gy cranial irradiation. Those with testicular leukemia at diagnosis and those who received >48 hours of corticosteroid therapy in the week prior to diagnosis participated in the Induction steroid randomization but were non-randomly assigned to HD-MTX with two Interim Maintenance and Delayed Intensification phases. If testicular involvement was not resolved at end Induction, 24 Gy testicular irradiation was given during Consolidation. Subjects with Very High Risk (VHR) B-ALL defined as Philadelphia chromosome-positive, severe hypodiploidy (<44 chromosomes and/or DNA index < 0.81), primary Induction failure, or SER with MLL-R, were removed from protocol therapy following Induction; however, they were included in the EFS/OS analyses. Therapy was continued for two years for females and three years for males from the beginning of Interim Maintenance 1. Details of therapy are shown in Supplemental Table 1.

Due to an increased incidence of osteonecrosis observed in children ≥10 years randomized to dexamethasone during Induction, therapy amendments were made during the conduct of AALL0232. Consequently, AALL0232 was amended in 2008 to halt steroid randomization in these older subjects but the steroid randomization was continued in subjects <10 years. In addition, all patients subsequently received discontinuous dexamethasone during Delayed Intensification, and prednisone during Maintenance.

Toxicity Assessment

Adverse events were reported using NCI Common Terminology Criteria (CTC) version 3.0 through December 2010 and then version 4.0 thereafter with data from CTCv3.0 mapped to CTCv4.0 using NCI-established algorithms. Adverse event reporting was supplemented with NCI’s Adverse Event Expedited Reporting System (AdEERS) reports and MedWatch reports (for reporting adverse events with commercial agents to the US Food and Drug Administration and NCI).

Statistical Analysis

The study was originally designed as a 2 × 2 randomized factorial design, testing two steroid formulations (prednisone versus dexamethasone) in Induction and HD-MTX versus C-MTX in the Interim Maintenance phases. Study design was described earlier.¹⁸

EFS was defined as time from study entry to first event (Induction failure, Induction death, relapse, second malignancy, remission death), or date of last follow-up for event-free subjects. Subjects who crossed over to the HD-MTX arm after it was found to be superior to C-MTX, were censored at the time of cross-over. OS was defined as time from study entry to death or date of last follow-up. All outcome analyses are based on intent-to-treat; hence irrespective of therapy received all patients are followed for an EFS/OS event and included as such in the outcome analyses. Survival rates were estimated using the Kaplan-Meier method with standard errors of Peto et al.^{23 24}, and survival curves were compared using the log-rank test. Multivariable Cox regression analyses of outcomes including various risk factors (Age, Sex, Race, WBC, body mass index (BMI), Cytogenetics, MRD) were also performed. Cumulative incidence rates were calculated for relapse (overall and by type of relapse), secondary malignant neoplasm (SMN), and remission deaths separately. Competing risks included Induction failures, Induction deaths, relapses, SMN, and remission deaths. Cumulative incidence rates between groups were computed using the cumulative incidence function for competing risks, and comparisons were made using the K-sample test.²⁵ A p-value less than 0.05 was considered as significant for all comparisons. All analyses were performed using SAS^® software version 9.4 (SAS Institute, Cary, NC). All graphics were generated using R (http://www.R-project.org, version 2.13.1).

Results

Subjects

AALL0232 enrolled 3,154 subjects; of the 3,040 non-Down syndrome (DS) patients who were eligible and evaluable for Induction therapy, 2,443 were <16 years of age and 597 were AYA 16 to 30 years. The median age for the AYA patients was 17 years. Of these, 2,662 were eligible and evaluable post induction (Figure 1; CONSORT diagram) to the four treatment regimens; Prednisone/C-MTX (PC, n=805), Prednisone/HD-MTX (PH, n=822), Dexamethasone/C-MTX (DC, n=474), Dexamethasone/HD-MTX (DH, n=561). Induction failures or VHR ALL features [BCR-ABL1 (157), hypodiploidy < 44 chromosomes (85), and MLL-R with SER (25)], led to the removal of 378 subjects from protocol therapy after Induction.

Figure 1:

CONSORT DIAGRAM

Age distribution of the 3,040 eligible and evaluable patients for Induction therapy ranged from 12 months to 30 years as follows: 1–9 years (33%), 10–15 years (47%), and 16–30 years (20%) including 2% of patients who were 21 years or older. Fifty-six percent were boys.

Compared to younger patients with NCI HR B-ALL, AYAs were more likely to have the Ph-like ALL gene expression profile (17.7% versus 11.5%, p=0.015) and less likely to have ETV6-RUNX1 fusion (3.8% versus 16.4%, p<0.0001). There were no significant differences between AYA and younger HR B-ALL patients regarding incidence of double or triple trisomy (favorable biology), BCR-ABL1 fusion, hypodiploidy or MLL-R (unfavorable biology) (Table 1).

Table 1.

Patient Characteristics

Patient Characteristic		<16 Years N=2443 n (%)	≥16 Years N=597 n (%)	P-value
Gender	Male	1303 (53.3)	385 (64.5)	<0.0001
	Female	1140 (46.7)	212 (35.5)
Race	White	1832 (84.1)	475 (89.3)
	Black or African American	184 (8.4)	30 (5.6)	0.010
	Other	163 (7.5)	27 (5.1)
Ethnicity	Hispanic or Latino	560 (23.8)	151 (26.4)	0.192
	Not Hispanic or Latino	1794 (76.2)	421 (73.6)
WBC (×1000 per μL)	<50	1234 (50.5)	480 (80.4)
	≥ 50	1209 (49.5)	117 (19.6)	<0.0001
CNS	CNS1	2034 (83.4)	504 (84.7)
	CNS2	349 (14.3)	78 (13.1)	0.728
	CNS3	57 (2.3)	13 (2.2)
RER/SER	RER	1852 (80.2)	373 (66.5)
	SER	458 (19.8)	188 (33.5)	<0.0001
BM Induction day29	M1	2284 (97.2)	545 (95.6)
	M2	42 (1.8)	13 (2.3)	0.079
	M3	24 (1.0)	12 (2.1)
MRD Induction day29	MRD < 0.01%	1723 (73.9)	316 (55.9)
	0.01% ≤ MRD < 0.1%	233 (10.0)	83 (14.7)
	0.1% ≤ MRD < 1.0%	202 (8.7)	82 (14.5)	<0.0001
	1.0% ≤ MRD < 10.0%	118 (5.1)	58 (10.3)
	MRD ≥ 10%	54 (2.3)	26 (4.6)
BMI	<30	2304 (95.2)	475 (80.5)
	≥30	117 (4.8)	115 (19.5)	<0.0001
Ph-like Status	Yes	58 (11.5)	53 (17.7)
	No	445 (88.5)	247 (82.3)	0.015
ETV6-RUNX1	Yes	350 (16.4)	19 (3.8)
	No	1782 (83.6)	487 (96.2)	<0.0001
Triple Trisomy (+4, +10, +17)	Yes	258 (12.4)	60 (12.0)
	No	1826 (87.6)	439 (88.0)	0.828
Double Trisomy (+4, +10)	Yes	337 (16.1)	81 (16.2)
	No	1751 (83.9)	418 (83.8)	0.960
KMT2A (MLL-R) rearrangement	Yes	81 (3.9)	20 (4.0)
	No	1991 (96.1)	476 (96.0)	0.899
Hypodiploidy	Yes	67 (2.7)	18 (3.0)
	No	2365 (97.3)	575 (97.0)	0.711
BCR-ABL1 positive	Yes	120 (4.9)	37 (6.3)	0.189
	No	2316 (95.1)	554 (93.7)
Number of patients who completed protocol therapy		1608 (65.8%)	300 (50.3%)	<0.0001

Toxicity of Therapy

The patterns of toxicity differed between AYA and younger patients throughout protocol therapy (Table 2). This study did not collect data on treatment adherence or compliance; however, protocol therapy completion rate was significantly lower for AYA patients compared to younger patients (50.3% (n=300) versus 65.8% (n=1608); p<0.0001) (Table 1). A total of 22.6% of AYA and 15.5% of younger patients did not complete protocol therapy due to an on-therapy event and 6.4% of AYA and 4.9% of younger patients were removed at end induction per protocol due to VHR ALL features (e.g. severe hypodiploidy, Induction failure, KMT2A-rearranged and treatment slow early responder, Philadelphia chromosome-positive). The reasons for non-completion among other patients (20.7% of AYA and 13.8% of younger patients) included toxicity, patient/family refusal, and physician’s choice.

Table 2.

Induction and Post-Induction Toxicities

	Age <16 Years	Age ≥16 Years	P-value
Induction
Hyperglycemia	319 (15.4%)	118 (23.6%)	<0.0001
Hyperbilirubinemia	91 (3.7%)	41 (6.9%)	0.0007
Thrombosis	28 (1.2%)	9 (1.5%)	0.470
Pancreatitis	12 (0.5%)	3 (0.5%)	0.972
Febrile neutropenia	338 (13.8%)	44 (7.4%)	<0.0001
Post-Induction
Mucositis during IM #1	229 (11.7%)	80 (18.2%)	0.0002
Peripheral motor neuropathy (overall)	190 (7.8%)	72 (12.1%)	0.001
Febrile neutropenia	1161 (56.8%)	207 (45.2%)	<0.0001
Hyperbilirubinemia	194 (9.5%)	79 (17.3%)	<0.0001
Hepatic failure	7 (0.3%)	6 (1.3%)	0.009
Toxicity comparison restricted to BMI ≥30
	Age < 16 Years N=117	Age ≥ 16 Years N=115	P-value
Induction
Hyperglycemia	35 (37.6%)	38 (40.9%)	0.652
Hyperbilirubinemia	15 (12.8%)	23 (20.0%)	0.140
Thrombosis	3 (2.6%)	4 (3.5%)	0.721
Pancreatitis	1 (0.9%)	2 (1.7%)	0.620
Febrile neutropenia	8 (6.8%)	6 (5.2%)	0.604
Infection (grade 4 and 5)	5 (4.3%)	11 (9.6%)	0.112
Post-Induction
Mucositis during IM #1	14 (16.7%)	20 (26.7%)	0.125
Peripheral motor neuropathy (overall)	10 (8.6%)	16 (13.9%)	0.195
Febrile neutropenia	41 (46.6%)	37 (45.1%)	0.848
Hyperbilirubinemia	24 (27.3%)	20 (24.4%)	0.668
Hepatic failure	1 (1.1%)	2 (2.4%)	0.610
Infection (grade 4 and 5)	9 (10.2%)	9 (11.0%)	0.874

Induction Toxicity

Comparing patients randomized to receive dexamethasone vs. prednisone during Induction, AYA patients experienced higher rates of Grade ≥3 hyperglycemia (23.6% versus 15.4%, p<0.0001) and Grade ≥3 hyperbilirubinemia (6.9% versus 3.7%, p=0.0007) compared to younger patients (Table 2). There were no differences in the rates of Grade ≥3 thrombosis and pancreatitis between the two patient groups. AYA patients had lower rates of Grade ≥3 febrile neutropenia during Induction (7.4% versus 13.8%, p<0.0001) and there was no significant difference between the rates of Induction mortality in comparison to younger patients (2.2% versus 1.6%, p=0.366). Induction toxicities further divided based on age: <16 years, 16 to 21 years and ≥22 years are reported in Supplemental Table 2.

Post-Induction Toxicity

Following Induction, AYA patients experienced higher rates of Grade ≥3 mucositis during Interim Maintenance (18.2% versus 11.7%, p=0.0002) and Grade ≥3 peripheral neuropathy throughout treatment (12.1% versus 7.8%, p=0.001) compared to younger patients (Table 2). Post-induction rates of hyperbilirubinemia (17.3% versus 9.5%, p<0.0001) and hepatic failure (1.3% versus 0.3%, p=0.009) were higher in the AYA cohort compared to younger patients. Similar to the Induction phase, AYA patients had lower rates of Grade ≥3 febrile neutropenia during the post-induction period (45.2% versus 56.8%, p<0.0001) but there were significantly more deaths in remission in comparison to younger patients (5.7% versus 2.4%, p<0.0001). Grade 5 infections accounted for most of the remission deaths in both age groups (Supplemental Table 3). Post-induction toxicities further divided by age: <16 years, 16 to 21 years and ≥22 years are reported in Supplemental Table 2.

Toxicity comparisons in obese patients

We also examined toxicity in obese patients, defined as those with a BMI sssss30. Among patients 1–15 years, 4.8% were obese as compared with 19.5% (p<0.001) of AYA patients (Table 1). While the rates of many toxicities were higher in obese versus non-obese patients, the rates of toxicities were similar between obese AYA and younger patients.

Treatment Response and Outcome

AYA patients had a slower response to Induction therapy measured by both early BM morphologic response at days 8 and 15 as well as end Induction MRD levels (Table 1). RER status was achieved in 66.5% of AYA patients compared to 80.2% of younger patients (p<0.0001). Patients of any age identified as SER (n=646) were less likely to complete planned protocol therapy compared to RER (n=2225), (42.9% (n=277) versus 72.8% (n=1619); p<0.0001). Day 29 BM MRD <0.01% was achieved in 55.9% of AYA patients compared to 73.9% of younger patients (p<0.001). Additionally, day 29 BM MRD ≥1.0% was reported in 14.9% of AYA compared to 7.4% of younger patients (p<0.001). Using the protocol definition of Induction failure (see above), AYA patients had a significantly higher rate than younger patients, s7.2% vs. 3.5% (p<0.001; Table 3) even when using the more stringent definition of M3 Induction failure (2.0% versus 1.0%, p=0.038).

Table 3.

Induction Events

	Age <16 yr (n= 2443)	Age ≥16 yr (n= 597)	P-value
Induction death	40 (1.6%)	13 (2.2%)	0.366
Induction failure (per protocol definition)	86 (3.5%)	43 (7.2%)	<0.001
Induction death or failure (per protocol definition)	126 (5.2%)	56 (9.4%)	<0.001
M3 Induction failure	24 (1.0%)	12 (2.0%)	0.038

M3: >25% leukemia blasts present in the bone marrow

The 5-year EFS/OS rates were 65.4±2.2% / 77.4±2.0% for AYA patients compared to 78.1±0.9% / 87.3±0.7% for younger patients (<0.0001) (Figure 2A and 2B). When the patients with VHR cytogenetic features that mandated removal from protocol therapy at end induction are excluded from the analyses, the 5-year EFS (67.3±2.3% for AYA vs. 80.5±0.9% for younger patients, p<0.0001) and OS (79.7±2.0% for AYA vs. 89.3±0.7% for younger patients, p<0.0001) were similar and significantly better in younger patients. AYA patients had a higher overall 5-year cumulative incidence rate of relapse (18.5±1.7% versus 13.5±0.7%, p=0.0006) compared to younger patients (Supplemental Figure 1). This was largely due to higher cumulative incidence rates of BM relapse ± extramedullary disease (14.0±1.5% versus 9.1±0.6%, p<0.0001). There was no significant difference between the cumulative incidence rates of isolated CNS relapse (3.9±0.8% versus 3.6±0.4%, p=0.83) in the AYA patients compared with younger patients (Table 4).

Figure 2:

(A) EFS for <16 vs. ≥16 years old; 5-year EFS: 78.1±0.9% vs. 65.4±2.2%; (B) OS for <16 vs. ≥16 years old; 5-year OS: 87.3±0.7% vs. 77.4±2.0%

Table 4.

Cumulative Incidence Rates by Event type

Event	Age <16 yr 5-year rate ±standard error	Age ≥16yr 5-year rate ±standard error	P-value
Relapse	13.5±0.7%	18.5±1.7%	0.0006
Marrow ± EMD	9.1±0.6%	14.0±1.5%	<0.0001
Isolated CNS Relapse	3.6±0.4%	3.9±0.8%	0.830
Relapse, other	0.9±0.2%	0.6±0.4%	0.567
SMN	0.8±0.2%	1.0±0.4%	0.818
Remission Death	2.4±0.3%	5.7±1.0%	<0.0001

EMD: extramedullary disease; CNS: central nervous system; SMN: secondary malignant neoplasm

Obesity was a significant risk factor predicting inferior event-free survival regardless of age, however obese AYA patients had significantly lower 5-year EFS (50.8±5.4% (n=115) as compared to obese younger patients 66.9±4.8% (n=117); p=0.006) (Supplemental Figure 2). The cumulative incidence of remission deaths was significantly higher in the AYA patients compared to younger patients (5.7±1.0% vs. 2.4±0.3%, p<0.0001) (Supplemental Figure 3). Overall, AYA patients had statistically significant inferior 5-year EFS and OS on all treatment regimens DC, DH, PC, and PH (Supplemental Table 4). Breaking the AYA group further based on age (≥22 years vs. 16 to 21 years) showed no difference in rates of relapse (15.0±5.7% versus 18.8±1.8%, p=0.88) or deaths in remission (5.5±1.0% versus 7.5±4.2%, p=0.69) (Supplemental Table 5).

Univariate and Multivariable Analyses of Risk Factors

Univariate and Cox multivariate analyses were performed to identify risk factors predictive of EFS (Table 5). Age <16 years vs. ≥16 years, age as a continuous variable, sex, white vs. black race, WBC <50,000/microliter vs. ≥50,000/microliter, end induction MRD <0.01% vs. ≥0.01%, favorable cytogenetics vs. neutral or unfavorable cytogenetics, and BMI (<30 vs. ≥30) were all significant risk factors for inferior EFS in univariate analysis. These factors all retained significance in multivariable analyses. When analyzing favorable cytogenetics (triple trisomy (+4, +10, +17) or ETV6/RUNX1) based on age (Age <16 vs. ≥16 years), the age effect was found not to be significant [HR: 0.725 (0.383, 1.369); p=0.321], and not significant in ETV6/RUNX1 cases [HR: 0.983 (0.238, 4.069); p=0.982]. For patients harboring favorable cytogenetics in triple trisomy (+4, +10, +17), there was no significant difference in 5-year EFS between AYA (87.1% ±5.0%) and younger patients (89.4% ±2.1%; p=0.38) (Supplemental Figure 4).

Table 5.

Univariate and Multivariable Cox Regression Analyses for EFS

Univariate analyses
Parameter	Hazard Ratio	95% CI for HR	P-value
Age (<16 vs ≥16 yrs)	0.558	(0.469,0.663)	<0.0001
Age as a continuous variable	1.060	(1.045,1.075)	<0.0001
Sex (Female vs Male)	0.846	(0.724,0.988)	0.035
Race (Black vs White)	1.458	(1.108,1.919)	0.007
(Other vs White)	0.841	(0.588,1.203)	0.343
WBC (<50k vs ≥50k)	0.868	(0.744,1.011)	0.069
EOI MRD (<0.01% vs ≥0.01%)	0.239	(0.203,0.280)	<0.0001
Cytogenetics (Unfavorable vs Favorable)	2.373	(1.454,3.873)	0.0005
(Neutral vs Favorable)	2.511	(1.986,3.174)	<0.0001
BMI (30–40 vs <30)	2.104	(1.636,2.707)	<0.0001
BMI (>40 vs <30)	3.340	(2.060,5.413)	<0.0001
Multivariable analyses
Parameter	Hazard Ratio	95% CI for HR	P-value
Age (<16 vs ≥16 yrs)	0.773	(0.625,0.956)	0.018
Age as a continuous variable	1.042	(1.024,1.060)	<0.0001
Sex (Female vs Male)	0.859	(0.719,1.025)	0.092
Race (Black vs White)	1.458	(1.090,1.950)	0.011
(Other vs White)	0.797	(0.553,1.149)	0.225
WBC (<50k vs ≥50k)	0.719	(0.602,0.859)	0.0003
EOI MRD (<0.01% vs ≥0.01%)	0.282	(0.237,0.336)	<0.0001
Cytogenetics (Unfavorable vs Favorable)	1.633	(0.871,3.061)	0.127
(Neutral vs Favorable)	2.391	(1.818,3.146)	<0.0001
BMI (30–40 vs <30)	1.671	(1.235,2.262)	0.0009
BMI (>40 vs <30)	1.780	(1.009,3.137)	0.046

Subjects with VHR features who were removed from the AALL0232 study post-Induction are not included in the above analyses.

Unfavorable: BCR-ABL1, hypodiploid, KMT2A (MLL) rearranged; Favorable: Triple Trisomy (+4, +10, +17) or ETV6/RUNX1 (TEL-AML1)

Discussion

AYA patients with HR B-ALL have inferior EFS and OS compared to younger patients due to higher rates of both relapse and treatment-related mortality. Historically many young adults with B-ALL have been treated on protocols designed for adult patients which tend to be less intensive than pediatric protocols.^4,7,26–32 Recently, Stock et al demonstrated in CALGB 10403 that using the PC arm of the AALL0232 pediatric regimen for AYA patients was tolerable and resulted in significantly improved 3-year EFS (59%) and OS (73%) compared to results seen in prior CALGB trials.¹⁷ AALL0232 represents the largest single cohort of AYA patients with HR B-ALL treated on a randomized study that is reported to date. However, the outcomes reported appear to be very similar to AYA patients treated on the earlier CCG 1961 study (ages 16 – 21 years) that enrolled patients 20 years ago (1996 to 2002).⁸ Although the CCG 1961 study included a much smaller cohort of AYA patients (n=262), 5-year EFS and OS were 71.5%±3.6% and 77.5%±3.3% respectively. In that trial, 12.7% of the total number of patients enrolled were AYA compared to 19% of those aged 16–21 years who enrolled on AALL0232. Five-year outcomes on AALL0232 for these patients were similar to 1961, with EFS and OS 65.1%±2.3% and 77.4%±2.0% respectively. Importantly in both trials, outcomes for AYA patients were significantly inferior to that of younger NCI HR B-ALL patients.

One fundamental reason for the poor outcome for AYA patients with B-ALL is likely the underlying leukemia biology. Previous reports have documented that B-ALL in AYA patients is more likely to have unfavorable biology and less likely to have favorable characteristics compared to younger patients.⁹ AALL0232 included comprehensive testing for specific sentinel cytogenetic alterations in all patients and a subset was tested for the Ph-like ALL gene expression profile (discovered during the course of this trial)^33,34, allowing a comparison of the incidence of specific lesions in AYA patients versus younger children with HR B-ALL. The AYA patients were more likely to have Ph-like ALL (17.7% versus 11.5%, p=0.015) associated with inferior survival and much less likely to have the favorable ETV6-RUNX1 fusion (3.8% versus 16.4%, p<0.0001) compared to younger patients. However, there were no differences between younger HR B-ALL patients and AYA patients in the rates of the other major favorable sentinel genetic lesions, triple trisomy of chromosomes 4,10 and 17 or double trisomy of chromosomes 4 and 10 (Table 1), or unfavorable lesions (BCR-ABL1 fusion, MLL-R, and hypodiploidy).

AYA patients have traditionally experienced more and higher degree toxicities with chemotherapy compared to younger patients. This phenomenon was also observed in AALL0232. During post-induction phases of protocol therapy, AYA patients experienced higher rates of Grade ≥3 mucositis during IM phases and higher rates of peripheral neuropathy throughout treatment compared to younger patients. Importantly, there was a significant difference between the rates of death during remission for AYA patients compared to younger patients. While the rates of various toxicities were similar in obese AYA and younger patients, the rate of obesity was much higher in AYA patients, and EFS was significantly lower in obese AYA as compared to younger patients. In multivariable analyses, older age, male sex, black race, neutral or unfavorable cytogenetics, end Induction MRD, and BMI ≥30 were all predictive of inferior EFS.

In the recently reported CALGB 10403 trial, 295 AYA patients, median age 24 years, with B-ALL (n=223) and T-ALL (n=71) were treated with a pediatric inspired regimen derived from the control arm of AALL0232 (PC Arm).¹⁷ The 3-year OS and EFS for this cohort were 73% and 59% with no significant difference for B-ALL versus T-ALL or by age subgroups. There was a reported trend toward better outcomes for the younger (ages 16 to 24 years) patients, however survival was nearly identical to the 20 to 29 and 30 to 39-year-old patients. The incidence of Ph-like ALL retrospectively assessed on CALGB 10403 was 31.1% compared to 17.5% on AALL0232, likely the result of older patients enrolling on the CALGB trial (25%; ages 30 – 39 years). Additionally, obesity was identified as a risk factor for poor outcomes with close to 30% of patients being classified as obese on 10403. As the 10403 three-year survival results are similar to 5-year outcomes of AYA patients on AALL0232, it is likely that the same underlying reasons of unfavorable biology and increased toxicity contribute to these poor outcomes. Although AALL0232 was not designed with sufficient power to make firm conclusions about the relative efficacy of HD-MTX and C-MTX in the AYA population, we note that the EFS and OS on the prednisone induction backbone (PC and PH arms, Supplementary Table 4) are very similar among AYA patients, which may have important implications if there are significant concerns about the toxicities of HD-MTX among older patients.

Conclusions

COG AALL0232 enrolled the largest number of AYA patients to date on a pediatric B-ALL study and demonstrated significantly inferior EFS and OS and greater rates of treatment related toxicity for AYA compared to younger patients. Although AYA patients had inferior early response, a higher frequency of Ph-like ALL, and a much lower frequency of ETV6-RUNX1-positive ALL, older age retained prognostic significance in multivariable analyses. Importantly, our results show no improvement in AYA outcomes compared to studies reported >15 years ago. Although treatment intensification strategies have improved outcomes in younger patients, these have not translated into better survival in those older. Thus, future trials must identify novel strategies to both improve outcomes and further reduce toxicity in the AYA cohort. Current clinical trial strategies being tested by COG include incorporating tyrosine kinase inhibitors for recently described Ph-like ALL subsets as well as novel therapies such as blinatumomab, inotuzomab ozagamicin and/or chimeric antigen receptor T-cell therapies.