Pharmacogenetic Predictors of Chemotherapy Treatment-Related Toxicities in Paediatric and Adolescent Acute Lymphoblastic Leukemia: A Systematic Review, Meta-Analysis and Literature-Based Candidate Prioritization
TL;DR
The review supports routine TPMT and NUDT15 genotyping and highlights the need for harmonised definitions, multi-ethnic studies, standardised data representation, and prospective validation of identified variants, which remains to be evaluated prospectively.
Abstract
Background: Childhood and adolescent acute lymphoblastic leukemia (ALL) survival outcomes have improved significantly, but treatment-related toxicities (TRTs) remain a major concern affecting dose intensity and quality of life. Germline pharmacogenetic variations contribute to inter-individual differences in chemotherapy response, yet consistent replication of associations has been limited by differences in treatment protocols, ethnic backgrounds, and toxicity definitions. Methods: This systematic review and meta-analysis (PROSPERO CRD42021229748) included 68 studies in the qualitative synthesis, with 42 high-quality studies undergoing structured synthesis and, where appropriate, quantitative meta-analysis. Studies focused on the toxicities of thiopurines, methotrexate, glucocorticoids, vincristine, and asparaginase. Results: Meta-analyses showed strong evidence linking the NUDT15 rs116855232 variant to thiopurine-induced myelosuppression (OR 19.0, 95% CI 1.6–224; I2 = 86.5%) and a significant association between the TYMS enhancer repeat polymorphism and osteonecrosis (OR 6.66, 95% CI 3.67–12.11; I2 = 0%). In contrast, MTHFR variants and VDR polymorphisms showed no significant associations with methotrexate-induced myelosuppression or osteonecrosis, respectively. Narrative synthesis highlighted clinically actionable associations: TPMT and NUDT15 with thiopurine toxicity; CEP72 with vincristine neuropathy; and HLA haplotypes with asparaginase hypersensitivity. Population allele frequency comparisons (IndiGenomes, gnomAD, UK Biobank) showed strong concordance (r = 0.919–0.997), supporting the broad generalisability of identified variants, which remains to be evaluated prospectively. The main limitations included heterogeneous toxicity definitions, non-uniform genetic models, evolving treatment protocols, population heterogeneity, and a lack of harmonised reporting. Conclusions: The review supports routine TPMT and NUDT15 genotyping and highlights the need for harmonised definitions, multi-ethnic studies, standardised data representation, and prospective validation. A literature-based candidate gene list for future PGx association studies was generated.