Clinical and demographic factors contributing to asparaginase-associated toxicities in children with acute lymphoblastic leukemia

Abstract

A total of 548 patients (age range: 1–22 years, 60.4% Hispanic, 55.8% male) diagnosed with acute lymphoblastic leukemia were reviewed for pegaspargase-associated hypersensitivity (14.8%), hyperbilirubinemia (9.7%), venous thromboembolism (VTE, 9.7%), and pancreatitis (5.3%). Odds ratios (OR) and 95% confidence intervals (CI) evaluated associations between clinical factors and each toxicity, cumulative number of toxicities, and toxicity clusters identified using k-mode analysis. Most (68.9%) did not experience any toxicity, 24.6% experienced one toxicity, and 6.3% two or more. Age >10 years was associated with hyperbilirubinemia (OR = 3.83; 95% CI: 1.64–8.95), pancreatitis (OR = 3.72; 95% CI: 1.29–10.68), VTE (OR = 4.65; 95% CI: 1.96–11.02), and cumulative toxicity burden (OR = 3.28, 95% CI: 1.97–5.47); high-risk therapy with hypersensitivity (OR 2.25; 95% CI 1.25–4.05); and overweight with cumulative toxicity burden (OR = 1.76, 95% CI: 1.20–2.57). Eight unique toxicity profiles were identified. Older age, overweight, and treatment intensity contribute to pegaspargase-associated toxicities.

Introduction

Each year, approximately 3,500 children are diagnosed with acute lymphoblastic leukemia (ALL), accounting for 26% of all childhood cancers in the United States (US) (1). Asparaginase is a critical component of contemporary protocols for pediatric ALL. Pegaspargase, a pegylated form of asparaginase, is the long-acting asparaginase preferentially used in ALL therapy and is administered either intramuscular (IM) or intravenous (IV). Pegylation of asparaginase decreases the immunogenicity and increases the therapeutic half-life of the enzyme (2, 3). Although pegaspargase therapy significantly improves long-term survival in pediatric patients with ALL (4), its use is commonly associated with various dose-limiting toxicities, including hypersensitivity, hyperbilirubinemia, pancreatitis, and venous thromboembolism (VTE) (4–6).

The reported incidence of pegaspargase-associated toxicities varies considerably in children and young adults. For example, rates of hypersensitivity range between 3% and 24% while VTE occurs in 5% to 12% of patients (7–10). Furthermore, because most studies report the prevalence of individual toxicities, the cumulative burden of pegaspargase-associated toxicities remains largely unknown. This is particularly true among understudied ethnic minority populations, such as Hispanic and non-Hispanic Black populations. Because treatment-associated toxicities may necessitate treatment interruption, potentially jeopardizing relapse-free survival, additional information on the cumulative incidence and clinical predictors of toxicity are needed to identify patients at highest risk of adverse treatment responses (11). To address this need, this study aimed to describe the cumulative burden and patterns of occurrence of four toxicities commonly attributed to pegaspargase (hypersensitivity, hyperbilirubinemia, pancreatitis, and VTE), evaluating clinical and demographic factors associated with the incidence of these toxicities in a multi-ethnic cohort of pediatric patients with ALL.

Materials and Methods

Study design and population

This retrospective chart review evaluated electronic medical records for children, adolescents and young adults (ages 1–22) newly diagnosed with ALL at Texas Children’s Hospital (Houston, Texas) between September 1, 2011 and December 31, 2017. Eligible participants were either enrolled to or treated according to Children’s Oncology Group (COG) pediatric B- or T-ALL treatment protocols. At Texas Children’s Hospital, pegaspargase was administered at a dose of 2,500 mg/m² IM with the total number of doses varying by protocol, randomization, and risk stratification as follows: AALL0031 (ClinicalTrials.gov identifier: NCT00022737, beginning in Consolidation, × 2 doses) (12–14), AALL0232 (NCT00075725, beginning in Induction, × 11 doses for slow responders and × 7 for rapid responders in Arm DC/PC; × 9 doses for slow responders and × 5 doses for rapid responders in Arm DH/PH) (15), AALL0434 (NCT00408005, beginning in Induction, × 7 doses for Arms A and B, × 5 doses for Arms C and D) (16, 17), AALL0932 (NCT01190930, beginning in Induction, × 2 doses) (18), AALL1122 (NCT01460160, beginning in Consolidation, × 5 doses), AALL1131 (NCT02883049, beginning in Induction, × 5 doses for high risk and × 7 doses for very high risk) (19), AALL1231 (NCT02112916, beginning in Induction, × 9 doses for standard and intermediate risk, × 12 doses for very high risk) (20). One protocol (AALL0031) also included two doses of L-asparaginase (unpegylated) at a dose of 6,000 IU/m² IM. The study was approved by institutional review boards at Baylor College of Medicine and The University of Texas Health Science Center at Houston (UTHealth).

Clinical and demographic variables

Information on clinical and demographic variables, including age at diagnosis, height, weight race/ethnicity, gender, and ALL risk group (standard-risk or high-risk) was abstracted from electronic health records. Race/ethnicity of individuals was categorized as non-Hispanic White, Hispanic, non-Hispanic Black, and ‘other.’ Height (in meters) and weight (in kilograms) measured at diagnosis were also recorded to calculate body mass index (BMI) categories for each patient. For patients less than two years of age at diagnosis, sex-specific weight-for-recumbent length percentiles were calculated per Centers for Disease Control and Prevention (CDC) guidelines while age- and sex-specific BMI percentiles were calculated from CDC growth curves for individuals age 2–19 years (21, 22). BMI for patients ≥20 years old was calculated by dividing weight (kg) by the square of height (m). Individuals were categorized as underweight or normal weight if their BMI percentile <85 (if < 20 years of age) or BMI < 25 kg/m² (if ≥20 years of age) and as overweight or obese if their BMI percentile was ≥ 85 (if < 20 years of age) or BMI ≥ 25 kg/m² (if ≥ 20 years of age) (22).

Definition of toxicities

The primary outcomes evaluated in this study were incident cases of four toxicities occurring after exposure to pegaspargase therapy: hypersensitivity, hyperbilirubinemia, pancreatitis, and VTE. Individuals were followed from diagnosis of ALL through induction and post-induction phase of the treatment (i.e., consolidation, interim maintenance, delayed intensification) up to the start of maintenance therapy. We defined clinically significant toxicities as those requiring treatment modification and/or those meeting the Common Terminology Criteria for Adverse Events (CTCAE v5.0) criteria for grade 3+ adverse events. Specifically, individuals with reactions to pegaspargase that required a switch to Erwinase, an analogue of asparaginase derived from Erwinia chrysanthemi with lower immunogenicity than pegaspargase (23), met the definition for hypersensitivity. Similarly, clinically significant VTE was based on physician diagnosis and introduction of an anticoagulant (e.g., low molecular weight heparin). A diagnosis of CTCAE v5.0 grade 3+ hyperbilirubinemia was based on bilirubin values exceeding five-times the institutional upper limit of normal (ULN), while CTCAE v5.0 grade 3+ pancreatitis was based on lipase values exceeding three-times the institutional ULN.

Statistical approach

Descriptive statistics were calculated for demographic and clinical variables, including counts and percentage of the total for categorical variables. Age at diagnosis was categorized as: 1–5 years, 6–10 years, and >10 years. Cumulative incidence was calculated separately for each of the four individual toxicities. Multivariable logistic regression models were constructed to examine the association between clinical (BMI, ALL risk group, immunophenotype) and demographic factors (age, sex, and race/ethnicity) and each individual toxicity. Three different approaches were used to characterize toxicities profiles: 1) separate multivariable logistic regression models for no toxicities vs any one toxicity, no toxicities vs any two toxicities, and no toxicities vs any three toxicities; 2) ordinal logistic regression comparing no toxicities vs any one toxicity vs any two toxicities, vs any three or more toxicities; and 3) K-modes cluster analysis. Briefly, K-modes clustering is similar to K-means or hierarchal clustering approaches, but is intended for the analysis of categorical data (24, 25).To determine the optimal number of clusters, two criteria were used – purity of clusters and number of clusters with fewer than five individuals. The purity of clusters is defined as the ratio between the dominant class in the cluster and the size of the cluster. Therefore, the optimal K-modes solution maximized the purity score while not identifying any clusters with fewer than 5 individuals. All clinical and demographic variables were evaluated and odds ratios (ORs) and 95% confidence intervals (CI) were calculated using multivariable regression models, with a two-sided p-value <0.05 indicating statistical significance. All analyses were performed in R (version 3.6.1).

Results

We retrospectively reviewed the electronic health records for 548 eligible patients with ALL treated with pegaspargase. The demographics and clinical characteristics of the study population are presented in Table 1. The mean age at ALL diagnosis was 7.6 years (range: 1.1 to 22.2 years). Most patients in the study population were males (55.8%) of Hispanic ethnicity (60.4%) and treated with standard-risk treatment protocols (55.7%).

Table 1:

Clinical and demographic characteristics of patients treated on acute lymphoblastic leukemia protocols

	Study Population (N = 548)
	No toxicity (n=378)	≥1 toxicity (n=170)
Age at diagnosis, n(%)
1–5 years old	179 (47.4)	45 (26.5)
6–10 years old	113 (29.9)	35 (20.6)
> 10 years old	86 (22.7)	90 (52.9)
BMI at diagnosis, n(%)
Underweight and Normal	213 (56.3)	71 (41.8)
Overweight and Obese	165 (43.7)	99 (58.2)
Race/Ethnicity, n(%)
Non-Hispanic white	100 (26.5)	42 (24.7)
Hispanic white	224 (59.3)	107 (62.9)
Non-Hispanic black	26 (6.8)	12 (7.1)
Others	28 (7.4)	9 (5.3)
Gender, n(%)
Male	205 (54.2)	100 (58.8)
Female	173 (45.8)	70 (41.2)
ALL risk group, n(%)
Low/Standard Risk	239 (63.2)	66 (38.8)
High/Very High Risk	139 (36.8)	104 (61.2)

The percentages computed across categories.

Hypersensitivity was observed in 14.8%, hyperbilirubinemia in 9.7%, VTE in 9.7%, and pancreatitis in 5.3% of patients (Table 2). In multivariable logistic regression models accounting for age at diagnosis, race/ ethnicity, gender, BMI category at diagnosis, and treatment risk group, age >10 years at diagnosis was associated with a higher frequency of hyperbilirubinemia (OR = 3.83, 95% CI: 1.64 – 8.95), VTE (OR = 4.65, 95% CI: 1.96–11.02), and pancreatitis (OR = 3.72, 95% CI: 1.29 – 10.68), while a higher frequency of hypersensitivity was observed in patients treated on high-risk therapy (OR = 2.25, 95% CI: 1.25–4.05). Overall, the majority of patients (68.9%) did not experience clinically significant hyperbilirubinemia, pancreatitis, hypersensitivity, or VTE (Table 3). Of the remainder, 24.6% of patients experienced just one toxicity, and 6.3% experienced two or more of the toxicities evaluated. The results of the ordinal logistic regression model (Table 3) indicate that patients 10 years of age or older (OR: 3.28 [95%bCI: 1.97–5.47]) and patients who are overweight or obese at diagnosis (OR: 1.76 [95% CI: 1.20–2.57]) experience a greater overall burden of toxicities commonly associated with pegaspargase therapy. Less than quarter of individuals (22.8%) with no toxicity were >10 years of age at diagnosis, compared to 50.4%, 57.5%, and 77.8% of the individuals with one, two, and three or more toxicities, respectively. Similarly, 88.9% of the individuals with three or more toxicities were overweight or obese at diagnosis compared to 43.7% of those without toxicity.

Table 2:

Odds ratios (OR) and 95% CI from multivariable logistic regression of individual pegaspargase - associated toxicities in pediatric patients with ALL (n=548)

	Hypersensitivity	Hyperbilirubinemia	VTE	Pancreatitis
Number of Events	81	53	53	29
Race/ethnicity
Non-Hispanic White	Ref	Ref	Ref	Ref
Hispanic White	1.08 (0.60 –1.94)	1.16 (0.55 –2.41)	1.06 (0.52 –2.16)	0.5 (0.20 –1.20)
Non-Hispanic black	1.20 (0.42 –3.37)	1.45 (0.45 –4.70)	0.72 (0.18 –2.83)	0.92 (0.22 –3.81)
Others	1.42 (0.51 –3.95)	0.71 (0.14 –3.47)	0.62 (0.12–2.99)	1.41 (0.34 – 5.7)
Age at diagnosis
1–5 years	Ref	Ref	Ref	Ref
6–10 years	1.15 (0.59 – 2.22)	1.43 (0.59 –3.50)	1.44 (0.60 – 3.52)	0.92 (0.26 –3.24)
> 10 years	1.37 (0.72 –2.59)	3.83 (1.64 –8.95)*	4.65 (1.96 –11.02)*	3.72 (1.29 – 10.68)*
BMI categories
Underweight or normal	Ref	Ref	Ref	Ref
Overweight or Obese	1.47 (0.91 – 2.41)	1.74 (0.96 –3.18)	1.72 (0.94 –3.15)	2.06 (0.92 – 4.61)
ALL Risk Group
Low/Standard Risk	Ref	Ref	Ref	Ref
High/Very High Risk	2.25 (1.25 –4.05)*	1.09 (0.52 – 2.31)	0.77 (0.37 –1.65)	1.08 (0.41 –2.82)
Gender
Male	Ref	Ref	Ref	Ref
Female	0.81 (0.50 –1.34)	0.84 (0.45 –1.55)	0.71 (0.38 –1.33)	1.17 (0.52 – 2.60)

^*Denote significant results at the nominal significance level of 0.05

Table 3:

Association of risk factors with co-occurrence of pegaspargase associated toxicities

	No Toxicity	1 toxicity	2 toxicities	≥3 toxicities	Ordinal Logistic Model
	(n=378)	(n=135)	(n=26)	(n=9)	OR (95% CI)
Age at diagnosis
1 –5 years	179 (47.4%)	37 (27.4%)	7 (26.9%)	1 (11.1%)	Ref
6–10 years	113 (29.9%)	30 (22.2%)	4 (15.4%)	1 (11.1%)	1.28 (0.77 –2.13)
> 10 years	86 (22.8%)	68 (50.4%)	15 (57.7%)	7 (77.8%)	3.28 (1.97 –5.47)*
BMI categories
Underweight or Normal	213 (56.3%)	58(43.0%)	12 (46.2%)	1 (11.1%)	Ref
Overweight or Obese	165 (43.7%)	77 (57.0%)	14 (57.7%)	8 (88.9%)	1.76 (1.20 –2.57)*
Race/ethnicity
Non-Hispanic white	100 (26.5%)	35 (25.9%)	5 (19.2%)	2 (22.2%)	Ref
Hispanic White	224 (59.3%)	84 (62.2%)	19 (73.1%)	4 (44.4%)	0.97 (0.62 –1.51)
Non-Hispanic black	26 (6.9%)	9 (6.7%)	2 (7.7%)	1 (11.1%)	0.94 (0.42 –2.12)
Others	28 (7.4%)	7 (5.2%)	0 (0.0%)	2 (22.2%)	0.86 (0.36 –2.04)
Gender
Male	205 (54.2%)	79 (58.5%)	16 (61.5%)	5 (55.6%)	Ref
Female	173 (45.8%)	56 (41.5%)	10(38.5%)	4 (44.4%)	0.83 (0.56 –1.21)
ALL risk group
Low/Standard Risk	239 (63.2%)	54 (40.0%)	9 (34.6%)	3 (33.3%)	Ref
High/Very High Risk	139 (36.8%)	81 (60.0%)	17 (65.4%)	6 (66.7%)	1.50 (0.95 – 1.36)

^*Denote significant results at the nominal significance level of 0.05

After considering the purity of the clusters and number of individuals in each identified cluster, a solution of eight clusters was selected for analysis (Supplemental Figure 1, Supplemental Table 1). The largest cluster (cluster 1) consisted of the 378 patients (69%) without any of the four toxicities evaluated (Supplemental Table 2). Other clusters contained individuals with only hypersensitivity (cluster 2, n=61), only hyperbilirubinemia (cluster 3, n=31), both hypersensitivity and hyperbilirubinemia (cluster 4, n=5), and pancreatitis alone or in combination with hypersensitivity or hyperbilirubinemia (cluster 5, n=20). The remaining three toxicity profiles consisted of individuals with VTE: VTE and hyperbilirubinemia with or without hypersensitivity (cluster 6, n=11), VTE alone or in combination with hypersensitivity (cluster 7, n=32), and VTE and pancreatitis with or without other toxicities (cluster 8, n=9). Supplemental Table 3 presents the distribution of clinical and demographic factors across the eight identified clusters. Age >10 years at diagnosis was associated with a 2.8- to 8.3-times higher likelihood of belonging to a cluster comprised of individuals with toxicity (Supplemental Table 4). Although not statistically significant for every cluster, being overweight or obese at diagnosis and being treated on high- or very high-risk protocols was consistently associated with an increased likelihood of belonging to a cluster with toxicities (e.g., clusters 2–8). For example, cluster 1 (no toxicities) had the highest percentage of patients between the ages of 1 to 5 years at diagnosis (47.4% vs <33% for all other clusters, p<0.001), treated on low/standard risk therapy (63.2% vs <47% for all other clusters, p<0.001), and underweight/healthy weight at diagnosis (56.3% vs <46% for all other clusters, p=0.03). We did not observe statistically significant differences in toxicity profiles by gender or race/ethnicity.

Discussion

In this multi-ethnic cohort of more than 500 children and adolescents undergoing treatment for ALL, the majority of patients tolerate contemporary ALL chemotherapy relatively well. However, nearly a third of patients experienced clinically significant toxicities commonly associated with pegaspargase chemotherapy. Hypersensitivity was the most frequently reported toxicity (14.8% of patients) in this study population, which was largely consistent with previous studies (3–24%) (7, 10). Similarly, the incidence of pegaspargase - associated pancreatitis (4.7%) and VTE (9.7%) in our population was comparable to previously reported rates (26, 27). However, the incidence of hyperbilirubinemia in the current study (9.7%) was somewhat higher than rates in the published literature (4–5%) (4, 5, 26, 28).This increase can be explained by higher proportion of obese/overweight patients in our study population. Previous studies have shown that obesity/overweight is associated with hyperbilirubinemia (29). Although individual pegaspargase - associated toxicities have been studied extensively, the cumulative burden of these toxicities has received relatively limited attention.

Our study investigates the cumulative frequency of four common pegaspargase - associated toxicities. Of the 548 patients, 378 (68.9% of patients) did not have any of the four toxicities evaluated, 135 (24.6% of patients) had one toxicity, 26 (4.7% of patients) had two toxicities, and nine (1.6% of patients) had three or more toxicities. This study also used a novel K-modes clustering approach to identify reoccurring patterns in the incidence of these toxicities. The results of this analysis suggest that the majority of toxicities occur without evidence of the other toxicities evaluated. For example, among patients with evidence of hypersensitivity, 75.3% did not experience any of the other toxicities. Similarly, 59.6% of patients with hyperbilirubinemia, 55.2% of patients with pancreatitis, and 65.9% of patients with VTE experienced only a single clinically relevant toxicity. Among the 35 individuals with multiple toxicities during the evaluation period, most (n=19; 54.3%) experienced hypersensitivity requiring a switch from pegaspargase to Erwinia chrysanthemi.

We also investigated clinical and demographic predictors of individual pegaspargase- associated toxicities and cumulative toxicity burden. Older age at diagnosis was associated with a higher frequency of most of the toxicities evaluated. Although this finding may be partly explained by the more aggressive therapy children and adolescents older than 10 years of age typically receive these associations remained after adjusting for treatment risk group (30–32). In fact, ALL chemotherapy risk group was only associated with a higher likelihood of hypersensitivity after accounting for age at diagnosis (33). Considered collectively, these findings suggest that treatment intensity likely contributes to some toxicity, such as hypersensitivity, but biological difference across ages also appear to impact susceptibility.

The relationship between overweight/obesity and adverse events related to treatment is not fully elucidated. Several studies have reported that overweight/obese patients have a higher risk of anti-leukemia treatment related toxicity, including asparaginase-associated toxicity (29, 34, 35). Consistent with these findings, our study observed that overweight and obese patients had a higher likelihood of multiple toxicities (OR 1.76; 95% CI 1.20 −2.57). In our k-modes analysis, overweight and obesity were consistently, albeit not significantly, associated with an increased odds of belonging to a cluster characterized by clinically relevant toxicity, with a particularly elevated odds of belonging to the cluster with the highest toxicity burden (Cluster 8). A cohort study with 2,008 children had results consistent with our study – increased odds of hepatic toxicity (OR 1.32, 95% CI 1.32–1.51) and pancreatic toxicity (OR 1.53; 95% CI 1.22–1.92) in obese patients compared to normal /overweight patients (36). Obesity may have direct or indirect impact of on liver and pancreatic function (37, 38).

These findings should be interpreted in light of some limitations. By virtue of the retrospective study design, potentially important risk factors and confounders that were not consistently documented in the medical record could not be evaluated. Although this study included a relatively large multiethnic cohort of patients with ALL, the small numbers of individuals observed in certain toxicity profiles limited our statistical power to detect clinically meaningful differences and resulted in imprecise effect estimates. Additionally, the treatment of pediatric ALL includes multiagent chemotherapy, so that we cannot exclude the potential contribution of other agents to these outcomes. Although hypersensitivity, pancreatitis, hyperbilirubinemia, and VTE are common pegaspargase-associated toxicities, pegaspargase may contribute to the development of other clinically significant toxicities which were not included in this analysis (26). Lastly, while we used CTCAE definitions to define hepatotoxicity and pancreatitis, cases of hypersensitivity and VTE were identified by documentation in the medical record. As a result, variation in toxicity detection or reporting between providers may have impacted the observed frequency of these outcomes.

Conclusion

This research adds to our current understanding of individual pegaspargase -associated toxicities by providing information on the cumulative burden of hepatotoxicity, pancreatitis, VTE, and hypersensitivity. The majority of patients tolerate ALL therapy; however, we found that nearly one in every 15 patients experiences multiple pegaspargase associated toxicities during contemporary ALL chemotherapy. Notably, multiple toxicities appear to occur most frequently in older patients and those who are overweight or obese. Although some factors, such as age and treatment intensity, may not be candidates for intervention, potentially modifiable risk factors, such as BMI, may be targets for future research seeking to reduce toxicity burden and improve quality of life in patients with ALL. Efforts toward toxicity prevention should consider interventions that support weight management, physical activity, and nutritional monitoring to mitigate pegaspargase toxicities in at-risk individuals.

Data Availability:

The data that support the findings of this study are available from the corresponding author upon reasonable request.