Clinical and germline risk factors for multiple treatmentrelated toxicities in pediatric acute lymphoblastic leukemia

Treatment-related toxicity (TRT) causes morbidity in acute lymphoblastic leukemia (ALL). A minority of patients suffer multiple TRT (mTRT). We characterized the incidence and risk factors for ALL mTRT in 1,240 patients between 1998-2013. The mTRT incidence was 10.7% with the most common mTRT combination being bone and neurotoxicity in 40%. There was no difference in leukemia-free survival (LFS), event-free (EFS), or overall survival (OS) following mTRT. Five clinical/laboratory factors (older age (≥10 years), female sex, high-risk leukemia, low albumin and elevated Y glutamyl transferase (GGT) during induction therapy) and one germline MUC16 single nucleotide polymorphism (SNP) (rs78342591; P=2.24x10-⁸) were associated with mTRT risk. The burden of TRT can be devastating for patients and clinicians. The occurrence of mTRT has not been well studied but can impair chemotherapy delivery and may be associated with an increased relapse risk. It is unknown what predisposes individuals to mTRT. Possible susceptibilities include organ dysfunction, delayed drug excretion, drug-drug interactions, genetic predisposition, constitutional syndromes and physiological factors such as age or sex. mTRT is likely exacerbated by intensive ALL therapy. Genome-wide association studies (GWAS) have identified germline risk factors associated with TRT but have focused on individual TRT. In ERASE (Evaluation of Risk of ALL Treatment-related Side-Effects), we undertook a retrospective study of Australian pediatric ALL patients diagnosed between 1998-2013, including annotation of treatment, survival, TRT and a germline GWAS (Online Supplementary Tables S1, S2). This analysis focused on mTRT, their impact on survival, and identifying clinical and germline factors associated with mTRT risk. The ERASE study including the GWAS have been published.^1,2 mTRT was defined as experiencing ≥2 TRT and controls as 0 or 1 documented TRT and who were followed for ≥18 months from diagnosis. The mTRT phenotype included bone (osteonecrosis or fractures), central or peripheral neurotoxicity, symptomatic venous thromboembolism (VTE) and insulin requirement. The mTRT GWAS cohort included 707 individuals, with five excluded due to lack of mTRT information, leaving 64 mTRT cases and 638 controls. The number of directly genotyped and imputed SNP with a minor allele frequency (MAF) >0.05% was 10999498 and with a MAF>2% was 7780980.

The median age was 59 months (range, 9-218 months) with a median follow-up of 78 months (range, 3-186 months). The 5-year OS, EFS and LFS of the ERASE cohort was 92±0.8%, 83.8±1.1% and 85.6 ±1.1%. mTRT occurred in 133 of 1,240 (10.7%) with the majority being CTCAE grade ≥2 severity (123/1240, 9.9%). The incidence of individual TRT included neurotoxicity in 7.6% (94/1240), insulin requirement in 6.9% (85/1,240), bone toxicity in 6.0% (75/1,240) and VTE in 5.5% (68/1,240) (Table 1). Bone and neurotoxicity was the most frequent combined mTRT. There was no difference in LFS, EFS or OS in mTRT patients (N=133), compared to controls (N=1,107). The 5-year LFS was 88,8±2,9% (mTRT), versus 85,9±1.1% (control, P=0.276), 5-year EFS was 84.9±3.3% (mTRT) versus 84.2±1.2% (control, P=0,595) and 5-year OS was 89.1±2.9% (mTRT) versus 92.5±0.9% (no control, P=0.138) (Online Supplementary Figure S1).

Risk factors for mTRT were assessed using univariable and multivariable logistic regression analyses. Twenty-six of 38 factors were significant in univariable analysis. Univariable associations with the mTRT phenotype included factors present at diagnosis and treatment during the early dose-intensive phases of chemotherapy (Table 2). Eighteen variables were carried into multivariable regression and five were independently associated with mTRT: age ≥10 years, female sex, high-risk ALL treatment, low serum albumin (<20 g/L during induction/consolidation), elevated GGT (>5x upper limit of normal during induction/ consolidation) (Table 2).

Table 1.Incidence of individual toxicities and combinations of multiple toxicities observed in the ERASE cohort (N=133).

The GWAS identified 28 candidate SNP (P<5x10^-6), mapping to eight genes including MUC16, SMYD3, FAM155A, UQCRFS1, FMO1, PIGF, LOC105371611, LOC105372352 (Table 3). Most candidate SNP (20/28) were associated with a reduced odds ratio of mTRT. Three SNP, associated with increased mTRT risk fell within MUC16 introns (rs78342591, rs62118276 and rs2341321). One reaching genome-wide significance (rs78342591; P=2.24x10^-8) (Table 3). Four SNP in SMYD3 were associated with increased mTRT risk (Table 3). The MUC16 rs78342591 risk allele (C) was examined with 640 individuals with informative data. Individuals with at least one rs78342591 risk allele C accounted for 17 of 64 (26.6%) of the GWAS cohort of children affected by mTRT. Four individuals were homozygous for the risk allele, with 50% (2/4) experiencing mTRT. Seventy-three individuals were heterozygous for the risk allele (CT), with 20.5% (15/73) experiencing mTRT. In contrast, 563 patients were homozygous for the non-risk allele (TT) with 8.3% (47/563) experiencing mTRT. Splicing analysis using Introme predicted the introduction of a polypyrimidine tract-binding protein (PTB) binding site from the rs62118276 SNP.

Table 2.Univariable and multivariable analysis of risk factors associated with multiple treatment-related toxicities.

Table 3.Top single nucleotide polymorphisms associated with multiple treatment-related toxicities phenotype in the ERASE cohort.

The ERASE study collected mTRT data across two major ALL treatment platforms, creating an opportunity to undertake the first study of clinical and genetic risk factors for mTRT in pediatric ALL. At least 10% of ALL patients experienced mTRT, but mTRT did not impact on ALL survival, a finding, whilst counterintuitive, aligns with the observation of Yeoh and co-workers who did not observe an increase in relapse risk in ALL patients experiencing treatment delay during the intensive phase of ALL therapy.³

The strongest independent risk factor for mTRT was older age (≥10 years), which is a risk factor for VTE, osteonecrosis, fractures, methotrexate neurotoxicity, vincristine-induced neuropathy and insulin requirement.^2,4-8 The association between high-risk ALL and mTRT is likely correlated with dose intensity and/or cumulative chemotherapy dosing. Female sex was an independent significant risk factor for mTRT, but female sex has not consistently been identified as a TRT risk factor across different studies.^4,5,9 There was an association between hypoalbuminemia and mTRT, independent of risk group and age, pointing to a link between therapy intensity and serum albumin, as low serum albumin often occurs during severe illness. Hypoalbuminemia is a likely consequence of treatment with asparaginase, malnourishment and/or underlying disease severity. A tentative association between albumin and osteonecrosis has been reported.⁵ Hypoalbuminemia has been associated with delayed methotrexate clearance.¹⁰ Treatment-related GGT elevation was associated with mTRT. Elevated GGT has been identified as a risk factor for symptomatic VTE¹¹ as well as decreased survival in multiple cancers including breast, ovarian, endometrial and melanoma treated with checkpoint inhibitors.

The mTRT GWAS identified 28 SNP mapping to eight genes with P values <5x10^-6. Six loci were associated with a reduced mTRT risk and two with increased mTRT risk. One SNP, rs78342591, reached genome-wide significance. MUC16 encodes a large transmembrane, mucinous, glycoprotein normally found on bronchial, endometrial, ovarian and corneal epithelia.¹² Multiple MUC16 functions have been identified including as an anti-microbial barrier, providing immune-protection from the innate immune system, enhancing metastasis and cancer cell proliferation and when knocked down promoting apoptosis and cell cycle arrest.¹² All MUC16 SNP identified in the GWAS are intronic, raising the hypothesis that the SNP might influence MUC16 splicing. Although Introme analysis did not link these SNP with a high probability splicing change,¹³ the rs62118276 SNP is predicted to introduce a polypyrimidine tract-binding protein (PTB) binding site. PTB regulates alternative splicing by exon inclusion/exclusion. Within the Expression Atlas, MUC16 is expressed within the liver and kidney but not within bone, postnatal brain, nerve cells, vascular endothelium, or hematopoietic cells. The role of MUC16 in mTRT remains to be clarified. We hypothesize that dysregulated hepatic and renal MUC16 expression following cytotoxic chemotherapy exposure results in dysregulated local cytokine production and inflammation increasing the risk of treatment-related toxicity.

This study has several limitations arising from the retrospective design and findings will require validation. Following the ERASE study, we are collecting data on two additional Australian ALL cohorts, one retrospective and one prospective, to replicate these findings. Data were collected by retrospective chart review, a resource- and time-intensive methodology which limits data collection to information easily and reproducibly documented in the medical record. With electronic medical records, automatically extracting adverse events in ALL patients has been demonstrated by the Children’s Oncology Group,¹⁴ suggesting that automated collection of relevant TRT in the future is feasible. There are substantial differences between the risk stratification and treatment algorithms used in different ALL treatment platforms. The data analysis was as recorded by the treating clinician and center based on the local risk stratification and treatment allocation without adjusting for differences between protocols (Online Supplementary Table S2). Toxicities analyzed in ERASE reflect those occurring at a reasonable frequency (≈5%) and consistently captured. Replication followed by functional validation of the MUC16 SNP on chemo-toxicity may provide clearer evidence regarding the mechanism of ALL mTRT Although blinatumomab is effective and well tolerated in patients with B-ALL,¹⁵ most ALL treatment platforms are adding blinatumomab to existing chemotherapy rather than substituting chemotherapy with blinatumomab suggesting that TRT from conventional chemotherapy will continue to be a problem for the foreseeable future. Functional validation and biomarker studies may provide tools for early diagnosis and intervention in children who are at high-risk of mTRT This study provides clinically relevant information that can be used for counseling ALL patients and their families, regarding factors that may lead to increased risk of mTRT Adolescent girls diagnosed with high-risk ALL have the highest chance of mTRT Through improved understanding of clinical and germline factors associated with mTRT risk, it may be possible to devise strategies to reduce mTRT

Footnotes

• Received August 25, 2025
• Accepted October 23, 2025

Correspondence

^#TNT and GMM contributed equally as senior authors

T. TRAHAIR - Toby.Trahair@health.nsw.gov.au

Disclosures

No conflicts of interest to disclose.

Contributions

Funding

This work was supported by the Kids Cancer Alliance (a Translational Cancer Research Center of Cancer Institute NSW), Cancer Institute NSW (grant ECF181430) (to MKM), the Anthony Rothe Memorial Trust (to TNT and MKM), Royal Australasian College of Physicians - Kids Cancer Project Research Entry Scholarship (to MKM) and a Cancer Therapeutics CRC (CTx) PhD Clinician Researcher Top-Up Scholarship (to MKM).

Acknowledgments

We would like to acknowledge the contribution of Dr Françoise Mechinaud, Director of the Children’s Cancer Center, The Royal Children’s Hospital, Melbourne, Australia, prior to her retirement. The authors thank the Sydney Children’s Tumour Bank Network for providing samples for this study, with support from the Cancer Council NSW, NHMRC Australia and Tour de Cure.