|Home | About | Journals | Submit | Contact Us | Français|
Methotrexate (MTX) is an important anticancer drug and the most efficient chemotherapy component in primary CNS lymphoma (PCNSL). A typical side effect of intravenous high-dose MTX is the occurrence of confluent CNS white matter changes (WMC). Because MTX directly interferes with methionine metabolism, we analyzed the impact of genetic variants of methionine metabolism on the occurrence of WMC as a model of MTX toxicity. In a retrospective analysis of 68 PCNSL patients treated with MTX-based polychemotherapy with (n = 42) or without (n = 26) intraventricular treatment, 10 genetic variants influencing methionine metabolism were analyzed. Pearson’s χ2 test and multinominal regression analysis were used to define the relevance of these genetic variants for the occurrence of WMC. In this patient sample, the occurrence of WMC was significantly predicted by the TT genotype of methylenetetrahydrofolate reductase c.677C>T (χ2 = 8.67; p = 0.013; df = 2), the AA genotype of methylenetetrahydrofolate reductase c.1298A>C (χ2 = 13.5; p = 0.001; df = 2), and the GG genotype of transcobalamin 2 c.776C>G (χ2 = 19.73; p < 0.001), in addition to male gender (χ2 = 11.95; p = 0.001). These data strengthen the hypothesis that MTX effects are influenced by methionine metabolism, which may offer new strategies to improve MTX-based therapies.
The anticancer and anti-inflammatory drug methotrexate (MTX) is a competitive inhibitor of several enzymes involved in folate and methionine metabolism, such as dihydrofolate reductase (DHFR), aminoimidazole carboxamide synthase, glycinamide ribonucleotide transformylase, and thymidylate synthethase.1,2 Manifestations of MTX toxicity include nausea, mucositis, myelosuppression, and renal, liver, and pulmonary dysfunction. Neurotoxicity of MTX may present with acute, subacute, and chronic encephalopathy with diffuse reactive astrocytosis and multiple, noninflammatory necrotic foci, as well as with clinically asymptomatic subacute confluent white matter changes (WMC), which are diagnosed on T2-weighted MRI sequences.3 Response to MTX-based chemotherapy and manifestation of toxicity show high interindividual variability. It is therefore important to identify factors influencing MTX effects in order to improve treatment outcome.
Uptake of folates into the cells occurs through organic bidirectional anion carriers such as the reduced folate carrier 1 (RFC1), as well as through the unidirectional folate receptors α and β that mediate folate uptake into cells via endocytosis. These receptors, as well as several transporters that unidirectionally export folates from cells, are also involved in uptake and export of MTX.4 The intracellular DHFR-inhibiting effect of MTX leads to a depletion of 5,10-methylenetetrahydrofolate and thereby to an impairment of both nucleic acid synthesis and homocysteine remethylation via 5,10-methylenetetrahydrofolate reductase (MTHFR) and 5-methyltetrahydrofolate-homocysteine S-methyltransferase (MTR). Because this is the only pathway of homocysteine remethylation in the CNS, application of MTX can lead to a marked increase of neurotoxic homocysteine within the CNS and to a lack of S-adenosylmethionine (SAM), which is necessary for myelination within the CNS, neuronal membrane stability, and neurotransmission.2,5–8
MTX is the most efficient drug in the treatment of primary CNS lymphoma (PCNSL),9 which accounts for 3% of all primary brain tumors in the United States.10 The prognosis of PCNSL is poor, with a median survival of approximately 2 months in untreated patients. Brain irradiation, chemotherapy, and combination of these result in a substantially improved survival.11,12 The “Bonn protocol” to treat PCNSL is based on a high-dose MTX (cycles 1, 2, 4, and 5) systemic polychemotherapy regimen of alkylating agents, vinca alkaloids, and dexamethasone combined with intraventricular MTX, prednisolone, and cytarabine (Ara-C), which has resulted in a 5-year survival fraction of 75% in patients younger than 60 years of age. Details of this protocol have been previously published.13
In a recent study, we observed that the presence of at least one of the genotypes MTHFR c.677TT, MTR c.2756AG/GG, or transcobalamin 2 (Tc2; transport protein of the MTR cofactor vitamin B12) c.776GG was associated with the occurrence of WMC in a sample of 42 MTX-treated PCNSL patients, but we failed to identify a significant independent association of any of these genotypes with WMC.14 In the present study, we extended the sample to 68 patients and analyzed a larger set of eight polymorphisms of methionine metabolism. We identified three functional polymorphisms that independently predict the occurrence of WMC.
The study population consisted of 68 consecutive PCNSL patients (32 male, 36 female) of Caucasian origin who had been treated in two prospective chemotherapy trials. Until December 2002, treatment included intraventricular drug administration (“original Bonn protocol”: MTX, prednisolone, and Ara-C; n = 42).13 Afterward, patients were treated without intraventricular therapy (“modified Bonn protocol”; n = 26; unpublished data of the authors). Median age at diagnosis was 59 years (range, 28–77 years).
Of the 68 patients, 55 completed therapy (six cycles of polychemotherapy), and all 68 underwent at least two cycles. MRI was carried out before initiation of therapy. The second MRI, which was used to define occurrence of WMC, was performed after two cycles of chemotherapy to assess treatment response as part of the trial protocol.13 None of the patients underwent brain irradiation prior to MRI assessment. We defined WMC as new, confluent, and symmetrical hyperintense lesions of the supratentorial white matter detectable on fluid-attenuated inversion recovery (FLAIR) and T2-weighted images.3
Genotyping was performed by amplification of genomic DNA applying PCR and subsequent restriction enzyme digestion or by allele-specific PCR, followed by agarose gel electrophoresis. For analysis of the 5-aminoimidazole-4-carboxamide ribonucleotide transformylase/inosine monophosphate cyclohydrolase (ATIC) polymorphism c.347C>G,15 we used the PCR primers 5′-AGCTGTAAACCACATGAGTGG and 5′-AGGCAGAGGTTGCAGTCAGC with an annealing temperature of 61°C. The PCR product of 471 bp was digested with the restriction enzyme HpyCH4III (New England Biolabs, Frankfurt, Germany), resulting in fragments of 26 bp, 96 bp, and 349 bp for the G allele and 21 bp, 26 bp, 96 bp, and 328 bp for the C allele. The other PCR and restriction analysis conditions were essentially as described previously.16–23 We used a retrospective approach for genetic analyses and correlations with WMC. Biochemical parameters such as plasma levels of folate, vitamin B6, or vitamin B12 were not available. Diagnosis of WMC was independently made by the chairman of the subdivision of neurooncology (U.S.) and the chairman of the subdivision of neuroradiology at the University Hospital of Bonn (H.U.). For all patients, both made the same diagnosis (presence or absence of newly developed WMC after the second cycle of chemotherapy). Diagnosis of WMC and genetic analyses were performed in a blinded fashion. Misdiagnosis of asymptomatic WMC for severe leukoencephalopathy in MRI-based diagnoses was unlikely, because leukoencephalopathies generally present differently on MR images and develop later than MRI was performed.24
Nominal regression analysis (α = 0.05) was used to test the independent predictive value of the following covariables: polymorphisms, age, gender, KPS score before therapy, and application of intraventricular therapy (yes or no) on the incidence of WMC (definition given above) as end point. For the tested genotypes, the Hardy-Weinberg equilibrium was calculated for all PCNSL patients together (with or without WMC) with a χ2 goodness-of-fit test (α = 0.05). Informed written consent was obtained from all patients or their legal guardians. This study was approved by the local ethics committees.
Of the 68 patients, one patient died after the second cycle, before MRI could be carried out, and suitable MRI studies from two further patients were not available for other reasons. Thus, 65 PCNSL patients were analyzed for the occurrence of WMC (Table 1). The Hardy-Weinberg equilibrium of the observed genotype frequencies was fulfilled for each genetic variant. The C allele of the rare polymorphism cystathionine β-synthase c.833T>C (I278T) was not detected in this sample, and we excluded this polymorphism from further analyses.
Overall, 20 (31%) of the 65 patients showed clinically asymptomatic WMC, and 45 did not. In all 20 cases, WMC was detectable in the MR images after the second cycle of therapy; none of the 55 patients who completed the six cycles of therapy developed WMC later than after the second cycle. Multivariate nominal regression analysis revealed that the occurrence of WMC was predicted by male gender (χ2 = 11.95; p = 0.001; odds ratio [OR] = 2.54; 95% confidence interval [CI] = 0.85–7.58). In addition, three genetic variants independently predicted the occurrence of WMC in our patient sample: MTHFR c.677C>T (χ2 = 8.67; p = 0.013; OR conferred by the TT genotype: 6.0; 95% CI = 1.32–27.2), MTHFR c.1298A>C (χ2 = 13.5; p = 0.001; OR conferred by the AA genotype: 6.00; 95% CI = 1.72–20.9), and Tc2 c.776C>G (χ2 = 19.7; p<0.001; OR conferred by the GG genotype: 2.15, 95% CI = 0.67–6.97). The DHFR c.594+59del19bp variant significantly predicted WMC formation in nominal regression analysis (χ2 = 6.63; p = 0.036), but this result was due to an overrepresentation of heterozygous cases only, which is not plausible. Thus, this result must be disregarded if not confirmed in other studies. The polymorphism ATIC c.346C>G trended toward an association with WMC (χ2 = 4.97; p = 0.083; OR conferred by the GG genotype: 3.5; 95% CI = 0.70–17.4).
The antifolate MTX is an important anticancer drug. Understanding mechanisms involved in MTX toxicity is important to improve MTX therapies. Mild MTX-induced neurotoxicity causes WMC, which in its extreme variant may present as necrotizing encephalopathy after intraventricular MTX application25 or as a dementing disease after combination of systemic MTX with whole-brain radiotherapy.26 Using this model, we analyzed whether genetic variants of methionine metabolism modify susceptibility to MTX toxicity. It is unclear to what extent these results may be applied to different treatment protocols.
Male gender was significantly associated with WMC. Methionine metabolism is influenced by endogenous steroids,27 which may explain gender-specific differences of WMC rates. In contrast to our data, Hoekstra et al. reported that MTX withdrawal as putative sign of side effects was more frequent in female than in male patients treated with low-dose oral MTX (≥15 mg/week).28 Speculatively, the impact of gender on MTX toxicity may differ between therapy regimens (long-term low dose vs. high dose) and organs, but gender-specific MTX effects and side effects have not been clearly delineated.
We have previously reported preliminary data suggesting that a combined genotype of polymorphisms of methionine and folate metabolism may be associated with the occurrence of WMC in MTX-treated patients who carry at least one of the following: genotypes MTHFR c.677TT, MTR c.2756AG/GG, or Tc2 c.776GG. This previous study did not identify any significant independent association of one of these genotypes with WMC.14 In the present study, we attempted to identify clinical and genetic risk factors for the formation of WMC in a larger series of PCNSL patients treated with high-dose MTX. In multivariate analysis, the Tc2 c.776C>G polymorphism and the two MTHFR missense polymorphisms c.677C>T and c.1298A>C were significantly associated with the occurrence of WMC, and the missense polymorphism ATIC c.346C>G trended toward an association.
The MTHFR polymorphism c.677C>T (p.A222V) leads to a reduced MTHFR activity, resulting in a lower remethylation rate of homocysteine to methionine and thus lower SAM concentrations and increased homocysteine levels (Fig. 1).17 The missense polymorphism Tc2 c.776C>G (P259R polymorphism) alters the affinity of Tc2 to cobalamin, also leading to reduced remethylation of homocysteine to methionine.16,29,30 Thus, the T allele of MTHFR c.677C>T and the G allele of Tc2 c.776C>G may both plausibly produce the homocysteine-increasing/SAM-decreasing effect of MTX, leading to higher risk of MTX-induced WMC, as observed in our study. Anecdotally, we have previously reported serial data on cerebrospinal fluid (CSF) analyses of four PCNSL patients treated with high-dose MTX showing that the two patients with the GG genotype of Tc2 c.776C>G had the lowest CSF levels of SAM. One of these Tc2 c.776GG carriers was the only one of those four patients who developed WMC.31 In that study, we also reported that during MTX treatment, homocysteine CSF levels increased up to 17-fold in patients treated with intra-ventricular MTX. Because we have found that even slight elevations of homocysteine levels led to neuronal dysfunction in cultured cells as well as in primary neurons,14 the increase of neurotoxic homocysteine during MTX treatment is a possible mechanism of MTX neurotoxicity in addition to SAM depletion. Polymorphisms of methionine metabolism may be relevant modifiers.
Concerning the second genetic variant with significant influence on the occurrence of WMC in multivariate analysis, MTHFR c.1298A>C (E429A polymorphism), we have shown that this polymorphism influences MTHFR activity in human fibroblasts (~80% activity in CC fibroblasts as opposed to AA fibroblasts).32 However, the A allele, which was associated with higher MTHFR activity in our as well as in other studies,32,33 was observed in association with WMC. This is in contrast to our hypothesis on elevated homocysteine levels and reduced SAM levels as reasons for MTX neurotoxicity. However, the MTHFR c.1298A>C and c.677C>T variants are in linkage disequilibrium; that is, c.677T allele is linked to c.1298A.34 Thus, an overrepresentation of c.677T alleles accounts for an overrepresentation of c.1298A alleles. The limited number of patients with different haplotypes in this series precludes the identification of a MTHFR c.677C>T; c.1298A>C haplotype predisposing to the development of WMC. Thus, we cannot determine whether the linkage disequilibrium with MTHFR c.677C>T or a different mechanism may explain the observed association of the wild-type A allele of MTHFR c.1298A>C with WMC.
The enzyme ATIC is involved in folate metabolism catalyzing the two final steps of purine synthesis. Previous studies showed no biological effect of the ATIC polymorphism c.346C>G (p.T116S),15 which in the present study showed a trend toward association with WMC. Thus, the effect of the ATIC polymorphism on the occurrence of WMC needs further study in an independent patient sample.
In summary, within the limitations of any association study, the present study supports the hypothesis that methionine metabolism modifies MTX (neuro)toxicity. The manipulation of methionine metabolism, for example, by supplementation with SAM (Fig. 1), may be a promising strategy to improve MTX tolerance.
This study was supported by the “Deutsche Krebshilfe,” grant 106262 (M.L. and U.S.).