Pharmacogenetics of methotrexate: toxicity among marrow transplantation patients varies with the methylenetetrahydrofolate reductase C677T polymorphism

Methotrexate (MTX) is an antifolate chemotherapeutic drug,1 and is used in patients undergoing marrow transplantation to prevent graft-versus-host disease (GVHD).2 Toxicities include mucositis and myelosuppression.1 We hypothesized that MTX sensitivity would vary with genetic variability in folate-metabolizing enzymes.

Folate is essential for nucleotide synthesis. The effectiveness of MTX is largely attributable to its role as an inhibitor of dihydrofolate reductase. Its metabolites also inhibit other folate enzymes, including 5,10-methylenetetrahydrofolate reductase (MTHFR),1,3,4 which converts 5,10-methylenetetrahydrofolate to 5,10-methyltetrahydrofolate.

A common MTHFR polymorphism (C677T) results in reduced activity.5 The variant TT genotype, associated with about 30% of wild-type (CC) activity, is present in about 10% to 12% of white and Asian populations. Heterozygotes (CT) (about 60% activity) constitute approximately 40% of the population. Variations are seen in risk of acute lymphocytic leukemia,6colorectal neoplasia,7-10 neural tube defects,11,12 and possibly cardiovascular disease,13 largely in the presence of low folate levels. Because MTX induces a low folate state, we hypothesized that toxicity would be aggravated among patients with the TT and, possibly, the CT genotype.

We undertook a study of patients undergoing marrow transplantation at the Fred Hutchinson Cancer Research Center (FHCRC) from 1992 to 1999. The following criteria applied: (1) first chronic phase of chronic myelogenous leukemia14; (2) first transplantation; (3) conditioning regimen: either busulfan/cyclophosphamide15 or cyclophosphamide/total body irradiation, as described15,16; (4) age at transplantation, at least 18 years; and (5) available DNA. All patients received 4 doses of MTX (intravenously, day 1, at 15 mg/m,2 and days 3, 6, and 11, at 10 mg/m²) and cyclosporine (as previously described15) for prevention of GVHD.2 All patients gave informed consent. The study was approved by the FHCRC Institutional Review Board.

Data were abstracted by a single abstractor (blinded to genotypes) on previous interferon treatment, smoking, MTX administration, and, if applicable, MTX serum levels and leucovorin administration.

Data from the patient database included: (1) conditioning regimen; (2) donor/matching status; (3) demographics; (4) weight, height, and calculated body mass index and surface area; (5) platelet and granulocyte counts; (6) bilirubin and creatinine; and (7) survival status or day of last contact.

Mucositis is a major toxicity associated with folate antagonists. The oral mucositis index (OMI) was developed17 to measure severity of oral mucosal changes; patients were examined every 2 to 3 days. Oral mucositis usually peaks between days 7 and 11 and resolves by days 18 to 21 after transplantation. We assessed mean scores on days 6 to 12 (peak period) and days 1 to 18 (overall mucositis course).

Genotyping for the MTHFR C677T polymorphism was performed,9 blinded to outcome measures; 10% blinded duplicate samples yielded 100% concordance. Genotypes were in Hardy-Weinberg equilibrium.

The primary outcome was oral mucositis, as measured by the OMI (mean, days 6-12, at least 2 assessments; and days 1-18, at least 4 assessments). Secondary outcomes included granulocyte engraftment, assessed as time to the first of 3 consecutive days of counts exceeding 100/μL and 500/μL; platelet engraftment, assessed as time to the first of 7 consecutive days of counts exceeding 10 000/μL and 20 000/μL without transfusion; and bilirubin, days 1 to 18. We were limited in our ability to investigate other measures of MTX toxicity: MTX serum levels, leucovorin administration as a rescue drug, and intestinal mucositis.

The original cohort consisted of 238 patients with chronic myelogenous leukemia (CML). Charts for 227 patients were abstracted. For 7 patients, we could not verify that 4 doses of MTX had been given; 136 and 124 patients, respectively, met inclusion criteria for OMI measurement during days 6-12 and 1-18.

All analyses were performed to investigate differences in outcomes by MTHFR genotype: individuals with the CT and TT genotypes were each compared to those with the CC genotype. Linear regression was used for continuous variables. Adjusted OMI means were calculated for all genotypes with covariates set at mean levels. Comparisons across genotypes were made using t tests. Measures of engraftment were analyzed using proportional hazards. Patients were censored on day 29 because surveillance was incomplete after this time. Patients who died before day 29 were censored on day of death (n = 2).

All analyses were adjusted for age, sex, conditioning regimen (corresponding to donor type), total delivered MTX dose, weight, and height. Engraftment analyses were also adjusted for creatinine. Other covariates that did not affect the relationship between genotype and outcome were race, smoking status, birthplace, and previous interferon treatment.

Characteristics of the patients are shown in Table1. Patients with lower MTHFR activity had a higher OMI during days 1 to 18 (Figure1). The mean (95% confidence interval) for patients with the CC (wild type) genotype was 15.9 (13.2-18.6); for the CT genotype, 17.5 (14.8-20.3); and for the TT genotype, 21.6 (16.8-26.4) (P < .05 compared to CC). The increase across CT (+1.6) and TT genotypes (+5.7) corresponds to the approximate residual MTHFR enzyme activity among those who carry the variant allele (CT, 60% activity; TT, 30% activity). Similar trends were observed for days 6 to 12: OMI differences [compared to the mean for CC genotype = 18.4 (15.2-21.7)]: CT genotype: +2.7,P = .25; TT genotype: +3.8, P = .27). Higher levels of mucositis among patients with a TT genotype correspond to a 36% increase in the mean OMI during days 1 to 18.

Recovery of platelet counts was slower among patients with variant genotypes than among those with the CC genotype (Table2). Recovery to 10 000/μL was 24% slower for the TT genotype (P = .23), and recovery to 20 000/μL, 34% slower (P = .08). Recovery of granulocytes was slightly slower among individuals with variant genotypes, although neither was statistically significant nor showed a trend.

Mean bilirubin levels and increases in bilirubin levels between days 1 and 18 were not related to genotype (data not shown).

To our knowledge, this is the first study illustrating a relationship between a folate-related variant and MTX toxicity. As hypothesized, patients with the MTHFR TT genotype experienced higher toxicity, evidenced by increased oral mucositis and a trend toward delayed platelet recovery. Imbalances in folate pools in those with the TT genotype are predictable18 and have been demonstrated.19 Low folate (dietary or MTX induced) may further decrease availability of folate for nucleotide synthesis.9 

A model of oral mucositis has recently been described.20In patients undergoing marrow transplantation, mucosal toxicity induced by the conditioning regimens predominates. We showed that patients with the MTHFR TT genotype experience higher OMI levels during days 1 to 18, with an effect size similar to that of cyclophosphamide/total body irradiation conditioning over busulfan/cyclophosphamide conditioning (> 30%). Among these patients, reduced DNA repair capacity, attributable to a decreased synthesis of nucleotides, probably results in delayed healing.

Engraftment of platelets was similarly affected. One might predict that the donor genotype would be more relevant for this outcome. However, the observed delay in recovery may indicate that the presence of the much larger number of recipient somatic cells affects folate pools. The product of MTHFR is 5′-methyl-tetrahydrofolate reductase, the transport form of folate. Thus, patients with a TT genotype may show a decreased availability of folate.19,21-23 For consent reasons we were not able to genotype the donors. Future studies should obtain both patient and donor genotypes to enhance understanding of this possible interaction and its impact.

The high prevalence of the MTHFR variant allele and the ability to perform genotyping for this single nucleotide polymorphism, rapidly and inexpensively, may render it a candidate for expanding tailored therapy24; genotyping may reduce MTX toxicity among a subgroup of patients. Although alternatives to MTX for GVHD prevention are limited, dose adjustment or newly evaluated drugs25may be appropriate.

We are indebted to the staff members who were involved in the conduct of this study: Lori Hubbard who performed the chart abstraction, Gary Schoch who provided support with the master patient database, and Sheryl Marks, Eileen van Hollebecke, and members of the FHCRC Collaborative Data Services who performed the data entry of the chart abstraction data. We thank Anajane Smith, Eric Mickelson, and Dr Effie Petersdorf for support in DNA retrieval, Michele Lloid for performing the OMI exams, Juanita Leija for technical assistance with the genotyping, and Mari Nakayoshi for graphical and word processing assistance.