Toxicity from asparaginase during acute lymphoblastic leukemia induction: a report from the Children’s Oncology Group

Key Points

• •Obesity and age, but not BSA, increase risk for AAT in ALL.
• •AAT during induction did not adversely influence disease response.

Visual Abstract

Abstract

Asparaginase-associated toxicities (AAT) often compromise therapy for acute lymphoblastic leukemia (ALL), affecting relapse risk and survival. There are conflicting data on the contributions of older age, obesity by body mass index (BMI), and/or large body surface area (BSA) to AAT. We examined the association of these risk factors with AAT and the impact of AAT on disease response. Induction data were examined from 4925 patients aged between 1 and 30 years enrolled in the Children’s Oncology Group ALL trials AALL0232 and AALL0434, which included a single dose of pegaspargase (2500 IU/m²) without a maximum dose. The associations of age, BMI, and BSA with hyperbilirubinemia, elevated alanine aminotransferase (ALT), thrombosis, and pancreatitis were evaluated. The impact of AAT on minimal residual disease (MRD) positivity (≥0.01%) at the end of induction (EOI) was assessed. Increased risk of developing at least 1 AAT was observed in patients aged ≥10 years (P = .002) and in those with obesity and high BSA (P < .0001) but not with high BSA alone. Risks for hyperbilirubinemia, ALT elevations, and thrombosis were all increased in patients with obesity and high BSA (odds ratio [OR], 3.5; 95% confidence interval [CI], 2.2-5.7; OR, 3.3; 95% CI, 1.7-6.6; and OR, 3.1; 95% CI, 1.5-6.5, respectively). AAT were not associated with EOI MRD positivity. To our knowledge, we report the largest data set of AAT in children, adolescents, and young adults. Preventive strategies are indicated for older patients and for those with obesity and high BSA, but not with high BSA alone. These trials were registered at www.clinicaltrials.gov as NCT00075725 and NCT00408005.

Introduction

Asparaginase was approved by the US Food and Drug Administration in 1978 and is integral to the treatment of acute lymphoblastic leukemia (ALL).¹ Since its initial incorporation into ALL regimens, asparaginase has improved remission rates and survival.2, 3, 4 Pediatric treatment regimens integrating intensive asparaginase have also enhanced survival among adolescent and young adult (AYA) patients with ALL.^5,6 Unfortunately, off-target toxicities from asparaginase complicate its integration into multiagent regimens for both children and AYAs. Hepatotoxicity, venous thromboembolism (VTE), acute pancreatitis, and severe coagulopathy are among the most common asparaginase-associated toxicities (AAT) that limit further asparaginase use and often compromise the delivery of concomitant chemotherapy.7, 8, 9 Most AAT occur early in chemotherapy, with the induction phase most commonly affected.¹⁰ Interruption of planned chemotherapy from AAT, including truncation of asparaginase dosing, is associated with increased risk of relapse and poorer survival.11, 12, 13, 14, 15, 16, 17, 18

Multiple studies have explored risk factors for severe AAT in children and AYA patients. A greater risk of developing AAT has been associated with older age, higher body surface area (BSA), and/or obesity (by body mass index [BMI]), but reports are conflicting.^10,19, 20, 21, 22, 23, 24, 25, 26, 27 These patient risk factors are often correlated, and it is challenging to separate their individual contributions to the risk of AAT. In an attempt to mitigate pegylated asparaginase toxicity, varying recommendations have nonetheless been proposed for patients considered to be at risk for AAT. In 2019, the increased risk of AAT in adult patients aged ≥22 years prompted a label change for lower pegaspargase (PEG) dosing (2000 IU/m²) than that indicated for pediatric populations (2500 IU/m²). Several studies have similarly attempted to limit AAT by using even lower dosing (500-2000 IU/m²) and/or capping dosing at a maximum single dose of 3750 IU (a non–evidence-based pragmatic threshold equivalent to a single dosing vial).^28,29 Recent and ongoing randomized trials have also investigated integrating protective agents to limit AAT, including for thromboprophylaxis³⁰ and hepatoprotection.¹⁶ Although long-term event-free survival is often because of a combination of disease, patient, and treatment factors, minimal residual disease (MRD) at the end of induction (EOI) is one of the strongest prognostic factors for outcome in pediatric and adolescent ALL.^31,32 Toxicity-induced adjustments to induction therapy may adversely affect EOI MRD, signaling an impact on future outcome.31, 32, 33, 34 To better understand AAT, it is critical to address 2 key knowledge gaps: (1) determine the risk factors for AAT to delineate at-risk populations and (2) understand the impact of AAT on early disease response.

Methods

Patient population

This study included children and AYAs with newly diagnosed ALL who were enrolled in frontline Children’s Oncology Group (COG) trials for B-cell ALL (B-ALL) or T-cell ALL (T-ALL). Patients with National Cancer Institute/Rome high-risk B-ALL were enrolled in COG AALL0232 (2004-2011) and those with T-ALL in COG AALL0434 (2007-2014). All patients with ALL in these trials were included in this analysis; patients with T-cell lymphoblastic lymphoma enrolled into AALL0434 were excluded. In summary, patients aged 1 to 30 years (inclusive) received COG-modified augmented Berlin-Frankfurt-Münster therapy, as previously described.^35,36 A common 4-drug induction was used for both trials, consisting of a glucocorticoid (prednisone for 28 days or dexamethasone for 14 days), weekly vincristine and daunorubicin for 4 doses, and a single dose of PEG. Chemotherapy dosing was calculated by BSA. All patients, including those aged ≥22 years, were prescribed PEG 2500 IU/m², as per approved dosing at the time, without recommended dose capping and administered either IV or intramuscularly once on days 4, 5, or 6.

Toxicity capture

Toxicity data were collected from routine clinical trial adverse event reporting. Adverse events were graded using Common Terminology Criteria for Adverse Events, version 4.0. AAT of interest were moderate or severe toxicities that were dose limiting and/or treatment impacting, specifically grade ≥3 hyperbilirubinemia, grade ≥4 elevation in alanine aminotransferase (ALT), grade ≥2 thromboembolic event, and grade ≥3 acute pancreatitis. As contemporary chemotherapy dose modification thresholds for hyperbilirubinemia fell between Common Terminology Criteria for Adverse Events grades 3 and 4, grade ≥4 hyperbilirubinemia was captured as a secondary liver end point. Toxicity data were captured for induction, except for pancreatitis, where consolidation toxicity was included because of the potential delayed onset from induction PEG exposure.

Statistical considerations

The primary objective of this study was to determine the contributions of BMI, BSA, and/or age to the risk of developing at least 1 AAT during early therapy. Secondary objectives were risk factors for each AAT and the association of AAT during induction with the presence of MRD at EOI. MRD positivity was defined for this analysis with the contemporary threshold of ≥0.01% in the marrow using multiparameter flow cytometry. Obesity was defined per the Centers for Disease Control and Prevention guidelines using BMI percentile ≥95% for those aged <20 years and BMI ≥30.0 for those aged ≥20 years.³⁷ Obesity was further classified as class 1 (BMI, 30-34.9 or ≥95% to <120% of 95th percentile), class 2 (BMI, 35.0-39.9 or ≥120% to <140% of 95th percentile), or class 3 (BMI, ≥40 or ≥140% of 95th percentile).³⁸ High BSA was defined for this analysis as >1.5 m², the BSA threshold equivalent to dose capping for the PEG dose (ie, 2500 IU × 1.5 m² = 3750 IU, 1 vial). AYA was defined as the age of ≥15 years per the most recent National Comprehensive Cancer Network (NCCN) guidelines.³⁹ Univariable and multivariable analyses were performed for the end point of developing at least 1 AAT and for each individual ATT. Multivariable analyses retained variables significant in the univariable analyses and included a combination variable categorizing BSA and obesity together (vs without obesity; high vs low BSA). For analysis of EOI MRD, postinduction pancreatitis was excluded. All analyses were 2 sided, with significance set at P <.05. Data analyses were performed using SAS version 9.4 software.

Institutional review board approval was obtained and maintained for each of the trials from which data were obtained, and this study was conducted in accordance with the Declaration of Helsinki.

Results

From both trials, 4925 patients were included in the analysis, of which most were treated for B-ALL on trial AALL0232 (63%) and the remainder for T-ALL on AALL0434 (37%). As described in Table 1, 25% of the cohort were AYAs, but only 2.3% were aged ≥21 years. By BMI, 904 patients (18%) presented with obesity; of these, the distribution of obesity class was 69% class 1, 21% class 2, and 10% class 3. Of those with obesity, most patients presented concurrently with a high BSA (62%). Conversely, 39% of the cohort were classified as having a high BSA, but less than one-third (29%) of these patients presented with concurrent obesity.

Table 1.

Patient characteristics

Patient characteristics	Any AAT		No AAT		P value
	n∗	%∗	n∗	%∗
Total	290	6	4635	94
Age, y					<.0001
Median (IQR)	14	10-6	11	5-14
<10	54	2.7	1928	97.3
10-14.9	123	7.1	1598	92.9
15-20.9	98	8.9	1009	91.1
≥21	15	13.0	100	87.0
Race					.051
White	219	6.1	3398	93.9
Black or African American	15	3.2	450	96.8
Other	20	7.7	239	92.3
Unknown	36	6.2	548	93.8
Ethnicity					.121
Hispanic or Latino	69	7.0	910	93.0
Not Hispanic or Latino	207	5.5	3552	94.5
Unknown	14	7.5	173	92.5
BMI, kg/m2					<.0001
Underweight	8	3.3	238	96.7
Normal weight	134	4.5	2828	95.5
Overweight	47	6.0	741	94.0
With obesity	101	11.2	803	88.8
BSA, m2					<.0001
Median (IQR)	1.6	1.3-1.9	1.3	0.8,1.7
BSA ≤1.5	105	3.5	2865	96.5
BSA >1.5	185	9.6	1745	90.4

^∗Unless otherwise indicated.

Risk factors for AAT

The overall prevalence of at least 1 AAT during induction was 6% (290/4925). The prevalence of AAT increased with age and BSA and in those with obesity (Table 1). The prevalence of AAT increased from class 1 (8.6% [54/628]) to class 2 (16.9% [32/189]) and class 3 (17.2% [15/87]) obesity (P = .001). There were no significant associations of race or ethnicity with the presence of induction AAT.

After accounting for regimen and age, multivariable analyses demonstrated increased risk for developing at least 1 AAT in those with obesity vs those without obesity (odds ratio [OR], 2.5; 95% confidence interval [CI], 1.88-3.24) and with high vs low BSA (OR, 2.0; 95% CI, 1.39-2.81; Table 2). However, in analyses combining BMI and BSA (Table 3), only obesity and higher BSA were associated with increased risk for developing an AAT (OR, 3.3; 95% CI, 2.22- 4.77); patients with high BSA without obesity were not at an increased risk (OR, 1.4; 95% CI, 0.94-2.04). The risk of developing at least 1 AAT also increased with age; multivariable analyses accounting for BMI and BSA demonstrated a greater than twofold increase in odds for developing an AAT for all age groups ≥10 years old vs younger children (Tables 2 and 3).

Table 2.

Univariable and multivariable analyses for the presence of any AAT during induction

Patient characteristics	Univariable analysis		Multivariable model 1∗		Multivariable model 2∗
	OR (95% CI)	P value	OR (95% CI)	P value	OR (95% CI)	P value
Age, y		<.0001		<.0001		<.0001
<10	Reference		Reference		Reference
10-14.9	2.7 (1.98-3.81)		2.6 (1.88-3.65)		1.9 (1.27-2.78)
15.0-20.9	3.5 (2.47-4.88)		3.2 (2.28-4.55)		1.8 (1.10-2.88)
≥21	5.4 (2.92-9.82)		4.9 (2.63-8.96)		2.8 (1.38-5.57)
Race		.057
White	Reference
Black or African American	0.5 (0.3-0.88)
Other	1.3 (0.81-2.09)
Ethnicity		.122
Hispanic or Latino	Reference
Not Hispanic or Latino	0.8 (0.58-1.02)
Regimen		.039		.553		.342
AALL0434	Reference		Reference		Reference
AALL0232	1.3 (1.01-1.69)		1.1 (0.83-1.41)		1.1 (0.87-1.47)
BMI category		<.0001		<.0001
Normal	Reference		Reference
Underweight	0.7 (0.34-1.47)		0.7 (0.36-1.53)
Overweight	1.3 (0.95-1.88)		1.3 (0.90-1.79)
Obese	2.7 (2.03-3.48)		2.5 (1.88-3.24)
High BSA, m2		<.0001				.0001
≤1.5	Reference				Reference
>1.5	2.9 (2.26-3.70)				2.0 (1.39-2.81)

^∗Model 1 is inclusive of BMI; model 2 is inclusive of BSA.

Table 3.

Combined multivariable model for the presence of any AAT during induction

Variable	OR	95% CI	P value
Age, y			.002
<10	Reference	Reference
10-14.9	2.0	1.36-3.03
15-20.9	2.2	1.33-3.56
≥21	3.3	1.65-6.8
Regimen			.547
AALL0434	Reference
AALL0232	1.1	0.83-1.41
Combined BSA and BMI category∗			<.0001
Without obesity, low BSA	Reference
Without obesity, high BSA	1.4	0.94-2.04
With obesity, low BSA	1.6	0.92-2.9
With obesity, high BSA	3.3	2.22-4.77

^∗Low BSA, ≤1.5 m²; high BSA, >1.5 m².

The prevalence of individual AAT varied among patients with older age, obesity, or higher BSA (supplemental Table 1). The presence of hyperbilirubinemia was further increased with the class of obesity (P = .001) from class 1 (5.1% [32/628]) to class 2 (13% [24/189]) to class 3 (16% [14/87]). Multivariable analyses for the end points of individual AAT showed that only patients with both obesity and a high BSA were associated with elevated ALT or thromboembolism, whereas obesity, irrespective of BSA, was associated with developing hyperbilirubinemia (Figure 1; supplemental Table 2). Obesity also remained associated with more severe grade ≥4 hyperbilirubinemia (P = .0007), both in those with high BSA (OR, 2.6; 95% CI, 1.22-5.59) and low BSA (OR, 4.7; 95% CI, 1.22-17.94). Obesity and BSA were not significantly associated with acute pancreatitis. In multivariable analyses, older age was significantly associated only with hyperbilirubinemia and not with elevated ALT, thromboembolism, or acute pancreatitis. The treatment regimen was not associated with a difference in overall AAT (Table 3) but was associated with risk for conjugated hyperbilirubinemia and thromboembolism (supplemental Table 2).

Figure 1.

Association of combined BMI and BSA with individual AAT. Results from multivariable analyses for each AAT examining the combination of patients with high vs low BSA (>1.5 vs ≤1.5 m²) and with high BSA with obesity vs without obesity by BMI (≥95%; >30, by age). OR with associated 95% CIs presented; all OR adjusted for variables significant in univariable analyses. See “Methods.” AP, acute pancreatitis; TE, thromboembolism.

MRD

In those with EOI MRD available (90.3% [4447/4925]), the prevalence of MRD ≥0.01% did not differ in those who developed at least 1 AAT (36.6% [90/246]) vs no AAT (33.5% [1407/4201]) during induction (P = .318). The prevalence of EOI MRD ≥0.01% in those with individual AAT vs those without individual AAT also did not differ for hyperbilirubinemia grade ≥3 (37.6% vs 37.7%; P = .297) or grade ≥4 (38.3% vs 33.6%; P = .499), elevated ALT (36.4% vs 33.6%; P = .703), thromboembolic event (35.4% vs 33.6%; P = .767), or acute pancreatitis (52.4% vs 33.6%; P = .069). AAT during induction was not associated with risk of EOI MRD ≥0.01% (Figure 2).

Figure 2.

Association of AAT during induction with MRD. Association of AAT with EOI MRD positivity ≥0.01%. Episodes of pancreatitis that occurred after EOI MRD were excluded from this analysis.

Discussion

Although AAT result in significant morbidity for patients, treatment-impacting AAT, as defined for this study, were relatively uncommon in the overall pediatric population, affecting <10% of the cohort. Nonetheless, identifying populations at highest risk for severe AAT is critical for optimizing safe delivery of asparaginase in the context of multiagent chemotherapy. Distinguishing the contribution from age, BSA, and obesity to the risk of developing AAT has been previously complicated by their correlation within patient populations. Leveraging data from 2 frontline ALL trials conducted by the COG allowed us to study a sufficiently large number of patients to identify the contribution from each of these risk factors. By doing so, we found that the risk for developing at least 1 severe AAT was associated with obesity but not with high BSA. Closer inspection demonstrated that the risk for AAT from obesity was predominantly driven by the contribution of asparaginase to hepatic injury and thromboembolism, without affecting the development of acute pancreatitis. Additionally, older age at diagnosis was found to be an independent predictor of developing AAT, specifically hyperbilirubinemia, whereas treatment with the T-ALL regimen was associated with increased risk for thromboembolism. This latter finding is intriguing, possibly reflecting presenting features more commonly found in patients with T-ALL, such as hyperleukocytosis or vascular compression from a mediastinal mass. Importantly, developing AAT during induction was not associated with poorer disease response as measured by EOI MRD.

Among the potential AAT, hyperbilirubinemia has a profound impact on concomitant induction chemotherapy,⁴⁰ and early severe hyperbilirubinemia is associated with poorer survival.^16,18 In our cohort, patients with obesity as well as children aged ≥10 years and AYAs constituted at-risk groups for hyperbilirubinemia. These findings are consistent with reports from other consortia. In the pediatric-inspired National Clinical Trial Network interconsortium CALGB C10403 trial, obesity was significantly associated with a greater risk for developing hyperbilirubinemia during induction, with the prevalence correlating with increasing class of obesity.¹⁰ Obesity was similarly reported as a risk for hyperbilirubinemia in AYA and adult patients treated within the Dana-Farber Cancer Institute, United Kingdom National Cancer Research Institute (UKALL14 trial), and German Multicenter Study Group in Adult ALL (GMALL 07/2003 trial) consortia.^17,41,42 Conversely, in 911 pediatric patients treated in Dana-Farber Cancer Institute trials 05-001 and 11-001, BSA and age, but not obesity, were associated with the risk of developing at least 1 AAT.²¹ However, in this secondary analysis, hyperbilirubinemia and other individual AAT were not analyzed separately, and although not described, reported rates of obesity from the trial⁴³ suggest that relatively few patients with obesity would have been included in the analysis. This likely limits the ability to tease out these correlated effects. In our study, multivariable models inclusive of these covariables demonstrated a clear trend for AAT risk from obesity but not BSA, likely because of the larger combined cohort available. Pragmatically, our data support that pediatric and adolescent patients who are “tall and skinny” (ie, high BSA alone) are not at greater risk for AAT than their peers and do not constitute an at-risk group warranting additional scrutiny or preemptive dose modifications or interventions.

Although patients who are at risk of AAT likely benefit from closer monitoring,⁴⁴ there is a lack of evidence-based guidelines on toxicity prevention. Integrated preventive approaches are being trialed for individual AAT. In the international PREVAPIX-ALL (ACCL1331) trial, the direct oral anticoagulant apixaban was successful in reducing VTE in those with obesity,⁴⁵ although not in the overall population.³⁰ A key lesson from this trial was the potential benefit of preventing AAT through targeted use of preventive strategies in high-risk subsets of patients. Levocarnitine¹⁶ is currently being tested in the United States and Canada, within a COG-led National Cancer Institute (NCI) National Clinical Trials Network (NCTN) intergroup randomized trial for AYA patients, to protect against induction hyperbilirubinemia, with a priori end points focused on the efficacy overall and also in higher-risk subsets (ClinicalTrials.gov identifier: NCT05602194). Alternate preventive strategies include recommendations for dose capping, dose reductions, and/or therapeutic drug monitoring to potentially reduce the risk of AAT,^7,29,40 but data supporting the efficacy of these approaches are inconsistent.^25,41,46, 47, 48 However, the aforementioned prospective trial of levocarnitine includes a range of asparaginase dosing and asparaginase therapeutic drug monitoring that may inform the potential connection between asparaginase activity and AAT. Similarly, although integrating dietary and physical activity modifications during induction has demonstrated improved disease response,⁴⁹ the impact of these interventions on reducing obesity, which in turn may reduce AAT, has not been studied. In our cohort, using BSA as a surrogate for total asparaginase dose, data for patients with obesity showed a dose effect for some AAT (ie, BSA correlated with increased risk for VTE or elevated ALT) but not for others (ie, no association of BSA with risk for hyperbilirubinemia). Future prospective studies are needed to study the impact of empiric dose reductions and/or dose capping on AAT, including individual AAT, so as not to obscure differences between toxicities.

Although almost 5000 patients were evaluated, there are several limitations to note. Foremost, the included COG trials predominantly enrolled younger children and adolescents; only 2% of enrolled patients were aged ≥21 years. Although these patients experienced the highest incidence of AAT (>10%), the ability to generalize this finding of BMI vs BSA in older AYA patients is limited. Similarly, data describing additional potential clinical contributors to liver injury, such as infection, critical illness, and preexisting metabolic dysfunction–associated fatty liver disease, were not routinely collected; future studies focused on AAT should include detailed information on host, treatment sequelae, chemotherapy delivery, and asparaginase administration with dose response. To that end, COG is now collecting serum asparaginase activity levels in current and future frontline ALL trials. All patients in these trials received PEG as the formulation of pegylated l-asparaginase. PEG continues to be the predominant asparaginase formulation used worldwide and in adult ALL; these data provide insights into AAT for these regimens. Recently, calaspargase pegol, which has a longer half-life, has replaced PEG in frontline regimens for pediatric ALL in the United States. Although no difference in AAT has been described between pegylated Escherichia coli asparaginase formulations to date,50, 51, 52 the exclusive use of calaspargase pegol in the United States for pediatric ALL warrants ongoing dedicated monitoring for AAT in patients with obesity. It is important to note that this study focuses only on the induction phase, as this is the period at highest risk for developing AAT on COG ALL protocols. Potential impact on disease response was also limited to analysis of EOI MRD as a proximal and prognostic marker of outcome from preceding induction toxicity; however, additional studies are necessary to understand the impact of this early toxicity on longer-term outcomes.

This report describes the largest data set of AAT in children and younger AYAs receiving PEG without dose cap recommendations during induction therapy for the treatment of ALL. Obesity, but not high BSA alone, conferred significantly increased risk for treatment-modifying AAT, apart from acute pancreatitis. In pediatric patients, dose capping of pegylated E coli asparaginase is a reasonable, albeit untested, approach to limit toxicity in patients with obesity and is likely not necessary for those with high BSA alone. Moreover, older patients were independently at risk for developing AAT, highlighting the unique risks facing AYA patients with ALL. Prospective studies to reduce risk for AAT are a critical next step to support safer dosing of asparaginase in frontline treatment settings.

Conflict-of-interest disclosure: E.O. reports consulting fees from Jazz Pharmaceuticals, Syndax Pharmaceuticals, and Mercor AI. L.D.M. reports consulting fees from Jazz Pharmaceuticals, honoraria from Servier Pharmaceuticals and Syndax Pharmaceuticals, and speaker’s bureau participation for Jazz Pharmaceuticals. R.E.R. reports consulting fees from Jazz Pharmaceuticals and Servier Pharmaceuticals and honoraria and travel support from Jazz Pharmaceuticals and Amgen. E.A.R. serves on a data and safety monitoring board for Bristol Myers Squibb. S.P.H. reports honoraria from Jazz Pharmaceuticals and Servier Pharmaceuticals and stock ownership in Amgen. J.L.M. reports consulting fees from Jazz Pharmaceuticals. R.J.S. reports consulting fees from Jazz Pharmaceuticals and Syndax Pharmaceuticals. M.L.L. reports service on an advisory board for Jazz Pharmaceuticals, participation in a speaker’s bureau for Amgen, and service on an advisory board for Illumina. The remaining authors declare no competing financial interests.

The current affiliation for A.L.A. is Servier Pharmaceuticals LLC, Boston, MA.