Discussion
It has long been known that individual treatment responses and tolerance in pediatric ALL are influenced by genetics. Certain SNPs in the TPMT gene were previously identified as a predictor of 6-MP and 6-TG toxicity in ALL treatment and assessment of TPMT genotype is now standard of care in the upfront setting. The role of MTHFR SNPs in ALL pathogenesis and treatment has been extensively studied yet remains controversial. Many studies are small, contain a multitude of patients on different treatment protocols, contain genetically and ethnically homogenous populations, or have been limited by their retrospective nature. Here, our study examines a large number of patients sourced from a genetically heterogenous population treated with similar COG-based protocols and suggests two MTHFR genotypes, homozygous 677TT and compound heterozygous 677CT/1298AC, are associated with increased MTX toxicity and require decreased oral MTX dosing during ALL maintenance therapy. Based upon this data, one could surmise that appropriate dosing of oral MTX for patients with either homozygous 677TT or compound heterozygous 677CT/1298AC would benefit from initial dose reductions as final average dosing was around 50% of the recommended starting dose for both groups. With both homozygous 677TT and compound heterozygous 677CT/1298AC genotypes having known reduced MTHFR enzymatic activity (30% and 50-60% respectively) relative to heterozygous 677CT and heterozygous 1298AC genotypes (60-70% and >80% respectively), our findings correlate well with known enzymatic function. Surprisingly, our results did not completely correlate with toxicity to IV methotrexate formulations, although could be a limitation of our study given the limited sample size of patients receiving either C-MTX or HDMTX. As far as predicting MTHFR genotype, myelosuppression was the only clinical factor significantly associated with detection of a clinically significant variant. No other toxicities demonstrated statistical significance and were not as common. We also assessed the association of specific MTHFR genotypes and other MTX toxicities, namely methotrexate leukoencephalopathy and thrombosis. Here we found that no MTHFR genotype was shown to increase risk of either toxicity. It should be noted that thrombosis risk in relation to MTHFR polymorphisms is attributed to high homocysteine levels and previously literature suggests that these levels may be mildly elevated at most. Transaminitis and mucositis are also major toxicities associated with methotrexate as well as other chemotherapeutic agents received by ALL patients. We found a large proportion of patients regardless of MTHFR status developed mucositis at some point during treatment although were not able to elaborate on the severity or causal relationship with MTX versus other therapies given limitations of information in the EHR.
There is also evidence that MTHFR status can influence outcomes, namely the likelihood of relapse or eventual death from ALL disease. Within our cohort of patients tested for MTHFR SNP, only one had refractory disease (1 patient with heterozygous 677CT SNP) and ultimately achieved remission after CAR T cell therapy and bone marrow transplant. Compared to most recent statistics showing greater 90% survival among ALL patients, this certainly does not demonstrate worse outcome for patients with clinically meaningful MTHFR genotypes, although would not speculate with regards to a clinical benefit given the small sample size, large variation of time since diagnosis and treatment, as well as significant population of patients within cohort that never received oral or IV methotrexate or had MTHFR genetic testing evaluated given early relapsed/refractory disease.
Being retrospective in nature, there are significant limitations to our study, mainly that information extracted is reliant on accurate medical documentation which is sometimes incomplete. We also combined patients whose MTHFR genotyping was not tested with those who tested wildtype, which was an assumption that may not have been true. In addition, it is likely certain biases played a role in the decisions to adjust oral MTX dosing. By knowing MTHFR status, providers may have preferentially decreased oral MTX dosing in response to toxicity whether it was a clinically significant genotype or not. Given the frequency of MTHFR SNPs in the population combined with targeted testing and the fact this was a single institutional cohort, our sample sizes were limited, especially among different genotype groups. It is very likely additional patients with MTHFR variants, both clinically significant and insignificant, exist within the control population and were never tested. Although this would be expected to further reinforce the significance of our results, it does limit our ability to extrapolate information such as targeted oral MTX dosing and it also may limit some of our other comparisons that did not achieve statistical significance.
Nonetheless, our study was able to demonstrate significantly decreased tolerance to oral MTX among patients with homozygous 677TT and compound heterozygous 677CT/1298AC genotypes as demonstrated by level of myelosuppression. These results do warrant further investigations. Ideally, this would be done through a prospective study. However, a more practical approach could involve a multi-institutional effort where genetic data can be harvested through existing biobanks and MTX dose and toxicity could retrospectively be reviewed for each sample. As was demonstrated with TPMT related to 6-MP toxicity, MTHFR may also influence MTX toxicity and therefore, identification of MTHFR genotypes upfront may lead to reduced ALL treatment complications in the future.
Conflicts of Interest: The authors declare no conflict of interest.
Acknowledgments: CH is supported by the National Institutes of Health (T32 CA060441) and Hartwell Foundation.