Treatment-related toxicities can be severe, life-threatening events that impact on delivery of ALL therapy and impair quality of life. The occurrence of multiple (≥ 2) toxicities in any one individual is of particular clinical concern, due to cumulative short and long-term impacts on health and functioning. Few studies have examined the co-occurrence of and risk factors for multiple non-infectious, non hematological toxicities in children with ALL.

We evaluated clinical factors and germline single nucleotide polymorphisms (SNPs), to determine the incidence of multiple toxicities and associated risk factors; and to provide insights into their pathogenesis and prevention. We hypothesized that susceptibility factors associated with a two-fold increased risk of multiple toxicity could be identified, and that these may serve as predictive biomarkers.

A systematic retrospective review of 1251 Australian children (1-18 years) treated on consecutive ALL protocols, between 1998-2013, was performed. Selected toxicities were venous thromboembolism (VTE), insulin-requirement, bone toxicity (osteonecrosis and fractures); neurotoxicity (central and peripheral). Toxicities chosen were those with a reported incidence >5%, robustness of retrospective data capture and clinical significance. Multiple toxicities were defined as ≥2 selected toxicities occurring in the one patient during first-line ALL therapy. Multivariate logistic regression, Cox regression and Kaplan-Meier survival analyses were performed. Two-tailed P values <0.05 were considered significant. A genome-wide association study (GWAS) was conducted on 702 Caucasian individuals within the cohort, which included 64 cases (≥2 toxicities) and 638 controls (≤ 1 toxicity). Samples were genotyped on the Illumina Oncoarray platform and imputed to the 1000G reference set. Genome-wide significance levels for the GWAS were P < 5E-08.

There were 1033 children treated on BFM-based protocols and 218 children on COG-based protocols (Table 1, "Baseline characteristics"). The median age of the cohort was 59 months (range 9-218 months) and median follow-up time was 78 months (range 3-186 months).

Five-year overall survival (OS), event-free survival (EFS) and leukemia-free survival (LFS) were 92% ±0.8%, 83.8% ±1.1% and 85.6% ±1.1% respectively. Multiple (≥ 2) toxicities occurred in 133/1240 evaluable children (10.73%). There was no difference in survival for children who experienced multiple toxicities (n=133) compared to the rest of the cohort (n=1107) ( P =0.138 for OS, P= 0.595 for EFS, P = 0.276 for LFS).

Independent clinical risk factors in multivariate analysis were age ≥ 10 years, female sex, and high-risk ALL. Additional laboratory test risk factors in induction/consolidation were an elevated gamma glutamyl-transferase (GGT) > 5 x upper limit normal (ULN) and serum albumin <20g/L (Table 2). The analysis was adjusted for treatment platform (BFM vs COG), gender, age and immunophenotype (T vs B).

An intronic SNP rs78342591 in the MUC16 locus reached genome-wide significance for multiple toxicity ( P =2.42E-08, odds ratio 5.89, 95% confidence interval (CI) 3.23-10.74). This was an imputed SNP.

Clinical and germline factors could identify children who may benefit from targeted supportive care and surveillance for acute toxicity following ALL therapy. Future research directions include replication, direct sequencing and subsequent functional validation, focusing on MUC16 expression and modulation of drug toxicity. Translation of this research may involve analysis of both risk and protective mechanisms associated with development of multiple toxicities in childhood ALL.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.