The effects of genomic polymorphisms in one-carbon metabolism pathways on survival of gastric cancer patients received fluorouracil-based adjuvant therapy

5-fluorouracil (5-FU) is widely used to treat patients with gastric cancer (GC). However, the response rate is quite heterogeneous. The single nucleotide polymorphisms (SNPs) and their interactions of genes in the one-carbon metabolism (OCM) pathway, including Methylenetetrahydrofolate reductase (MTHFR), Methionine synthase reductase (MTRR), Methionine synthase (MTR), and Thymidylate synthase (TS), significantly affect 5-FU metabolism. In this study, 650 stage II-III patients were recruited from 1998 to 2006. Among them, 251 received 5-FU treatment and other 399 patients were untreated. The Cox regression analysis, log-rank tests and Kaplan–Meier plots were adopted. In the chemotherapy cohort, MTRR 66 GA + GG genotypes decreased death risk, however, the protect effect of MTRR 66 GA + GG disappeared when GC patients simultaneously had MTHFR 677TT + TC or MTR 2756GG + GA genotypes. TS 5′-UTR 2R3R + 3R3R genotypes also prolonged overall survival of patients treated with 5-FU. And this favorable prognosis obviously enhanced when GC patients simultaneously had TS 3′-UTR DD + DI and TS 5′-UTR 2R3R + 3R3R genotypes. Our findings showed that the polymorphisms of MTRR 66 A > G and TS 5′-UTR 3R > 2R may be potential prognostic factors for GC patients receiving 5-FU.

. The median age was 61.0 years (range, 28-83 years) and 328 of them died during the follow-up of 119.0 months. 251 patients underwent 5-FU-based adjuvant chemotherapy and 399 of them did not receive chemotherapy after surgical resections. In the no adjuvant chemotherapy cohort, the survival time was significantly related to varieties in tumor differentiation, lymph node metastasis and TNM stage (all log-rank p < 0.05). In particular, no chemotherapy patients with III level of TNM stage (MST, 36 months) had a 80.9% notably higher risk of death (HR = 1.809, 95% CI = 1.322-2.477, p < 0.001) when compared with those with II level (MST, 71 months). Patients with lymph node metastasis (MST, 38 months) had a 50.0% higher risk of death (HR = 1.500, 95% CI = 1.056-2.129, p = 0.021) when compared to those without lymph node metastasis (MST, 67 months). Meanwhile, patients with well differentiation (MST, 74months) had 43.6% and 67.5% lower risk of death when compared to those with poorly differentiation (MST, 38 months; HR = 1.436, 95% CI = 1.037-1.989) and mucinous or signet-ring cell (MST, 35 months; HR = 1.675, 95% CI = 1.064-2.636; p = 0.038), respectively. In contrast, no remarkable higher death risks were founded among the chemotherapy cohort.
MTRR 66 A > G and TS 5′-UTR 2R > 3R polymorphism prolonged overall survival of GC patients receiving 5-FU-based adjuvant chemotherapy. Cox regression analyses were used to assess relationships of polymorphisms of those genes in the OCM pathway with overall survival in different genetic models among the two cohorts (Table 2). For the patients with adjuvant regimens, significant associations were found between MTRR 66A > G polymorphism and overall survival time in the dominant model. Chemotherapy patients with MTRR 66GA + GG genotypes (MST, 80 months) exhibited a 34.3% obviously lower risk of death (HR = 0.657, 95% CI = 0.446-0.967, p = 0.031, Table 2, Fig. 1) when compared with the patients with AA genotype (MST, 46 months), whereas no associations were founded among the cohort without chemotherapy.
For other polymorphisms including MTHFR 677C > T, 1298A > C MTR 2756A > G and TS3′ -UTR 6 bp ins > del, no significant relationships were founded between polymorphisms in any genetic models and survival Scientific RepoRts | 6:28019 | DOI: 10.1038/srep28019 among patients accepted 5-FU treatment (Table 2). Furthermore, none of the six polymorphisms were correlated with overall survival for patients without 5-FU treatment.
Stratified analyses of MTRR 66A > G polymorphism on reducing the death risk of GC patients accepted 5-FU-based chemotherapy. To further explore the relationships between survival time and MTRR 66A > G polymorphism among GC patients with 5-FU-based chemotherapy, stratified analysis were conducted according to the histological types, sizes, sites, TNM stages, depth of invasion, lymph node metastasis, distant metastasis and chemotherapy (  Fig. 3).
The effects of combined polymorphisms on 5-FU-based chemotherapy. Considering the complicated interactions of the genes outlined above, we investigated the effect of polymorphism interactions on the prognosis of GC patients who accepted 5-FU treatment (Supplementary Tables S1-S3). In univariate analysis, MTRR 66 GA + GG genotypes remarkably reduced death risk, but the elongated survival time disappeared  Table 5). In these cases, the MSTs were reduced from 80 months to 60 and 64 months, respectively. In addition, it seemed that patients simultaneous with TS 3′ -UTR DD + DI and TS 5′ -UTR 2R3R + 3R3R genotypes showed higher reduction of death risk (HR = 0.332, 95% CI = 0.134-0.822, log-rank p = 0.046, Table 5) than those with TS 5′ -UTR 2R3R + 3R3R (HR = 0.498, 95% CI = 0.259-0.960) genotypes alone.

Discussion
We systematically investigated the effects of SNPs in the OCM pathway and their interactions on the survival of stage II-III GC patients, who received 5-FU base adjuvant chemotherapy. Our results suggested that the polymorphisms of MTRR 66A > G and TS 5′ -UTR 2R > 3R remarkably prolonged the survival of Chinese stage II-III GC patients treated with 5-FU base regimen. The MST of GC patients was also significantly influenced by interactions between MTRR 66A > G, MTR 2756A > G, TS 5′ -UTR 2R > 3R and MTHFR 677C > T.
As an inhibitor of DNA synthesis, 5-FU displays potent anticancer activity by inhibiting TS activity through forming a ternary complex with 5, 10-MTHF 21 . The SNPs of MTRR 66 A > G and MTR 2756 A > G has been shown to decrease MTR enzyme activity. MTRR is used to maintain MTR in its active form, but the activity of MTR could not be preserved at a high level because the substitution of Ile22 to Met22 in MTRR caused by the 66A > G mutation decreased MTR activity 22 . The reduced MTR activity resulted in an accumulation of 5-MTHF. Subsequently, it increases the level of 5,10-MTHF that ultimately enhances the effect of 5-FU treatment 20 . Our result indicated that MTRR 66GA + GG genotypes significantly improved the survival of GC patients in the chemotherapy cohort (HR = 0.657, 95% CI = 0.446-0.967, Table 2, Fig. 1). This is consistent with other studies 23,24 . Further stratified analysis on pathological features showed that many factors, including histological type, tumor differentiation, depth of invasion, lymph node metastasis and TNM stages of patients, strongly influenced     25 . However, we found that this polymorphism had no obvious association with prognosis of patients receiving 5-FU-based chemotherapy. In the combination analysis, although the favorable response was maintained when patients simultaneously carry MTRR 66GA + GG and MTR 2756AA genotypes; interestingly, MTR 2756GG + GA genotypes neutralized the protective effect of MTRR 66GA + GG (HR = 0.871, 95% CI = 0.443-1.713, Table 5). The similar neutralization was also existed when patients simultaneous with MTRR 66GG + GA and MTHFR 677TT + TC genotypes ( Table 5). The neutralization maybe caused by the interactions of those key enzymes involved in the OCM pathway. Asp868 of MTR and Ile22 of MTRR may play vital role in the interactions between MTRR and MTR. Met22 conversion of MTRR may breakdown MTRR-MTR interaction, but Gly868 conversion of MTR may compensate and recover those interactions to a certain extent by changing the three dimensional spatial   structure 26 . Similar interactions probably also exist between MTRR and MTHFR. Because our results were from a single cohort study with limited number of patients, the results would be validated in other cohorts with enlarged sample sizes. The stratified analysis of different combined chemotherapy regimens showed that patients with MTRR 66GA + GG genotypes treated with L-OHP and 5-FU had remarkably favorable prognosis than those with AA genotypes (HR = 0.538, 95% CI = 0.299-0.970, p = 0.035, Table 4, Fig. 3). Whereas, no obviously improved survival was founded among patients receiving DDP and 5-FU chemotherapy (HR = 0.839, 95% CI = 0.506-1.393, p = 0.494, Table 4, Fig. 3). This is not surprising because L-OHP displays much more potent apoptotic activity than DDP 27 . Although L-OHP forms fewer intrastrand cross-links with adjacent guanines (GG) and occasionally adenines (AG) than DDP at equimolar concentrations, L-OHP-Pt-DNA adducts are more lethal than CDDP-Pt-DNA adducts 28 . The other reason might be differential DNA repair of L-OHP and DDP lesions. Intact mismatch repair is the main way for DDP induced cytotoxicity; whereas, L-OHP adducts are poorly recognized by mismatch repair proteins [29][30][31] .
A number of studies showed that the polymorphisms of TS affect the sensitivity to 5-FU, but the results were controversial. It was discovered that the 3R3R genotype was related with poor outcomes for colorectal cancer patients treated with 5-FU 32,33 . For GC patients receiving 5-FU regimens, Ott et al. also found that 3R3R was a risk factor for the survival 34 . Whereas, some other studies reported that no significant differences were observed for the survival of patients with different genotypes 35,36 . Jakobsen et al. reported that patients with 3R3R genotype responded better to 5-FU and had longer survival time 11 . Obviously, the effects of TS 5′ -UTR polymorphism on 5-FU treatment are very complicated. The explanation for those controversies may involve in cancer type, genetic background and gene-gene interactions. Our finding was similar to Jakobsen's result (HR = 0.498, 95% CI = 0.259-0.960, p = 0.032, Table 2). The 3R allele could enhance TS expression when compared with the 2R allele, and the higher TS expression level caused by 2R3R and 3R3R genotypes would increase the formation of the ternary complex, leading to the improvement of sensitivity to 5-FU. The inhibition of TS may give rise to insufficient dTMP supply in DNA synthesis, dUMP misincorporation into DNA and double-strand breaks followed by cell apoptosis 37,38 . Controversial data on TS 3′ -UTR 6 bp ins > del polymorphism were also reported. In colorectal cancer patients receiving chemotherapy, some studies showed a better outcome of 6bp ins > del polymorphism on the survival 10,39 , while another report exhibited a higher response rate for patients with TS 6bp ins/ins genotype than those with TS 6bp del/del genotype 40 . Several other studies, similar to our findings, demonstrated that no obvious different survival times were identified among the patients with different genotypes 41,42 . Interestingly, although no obvious relativity was found between TS 3′ -UTR 6bp ins > del polymorphism with patients′ survival, the protective effect of TS 5′ -UTR 2R > 3R polymorphism was enhanced when chemotherapy patients simultaneous with TS 3′ -UTR DI + DD and TS 5′ -UTR 2R3R + 3R3R genotypes (HR = 0.332, 95% CI = 0.134-0.822, log-rank p = 0.046, Table 5). The interaction between TS 5′ -UTR and 3′ -UTR may be the explanation for this synergistic effect.
The 677 C > T polymorphism induces an Ala-to-Val switch at codon 222 in MTHFR, resulting in thermolabile variant of the enzyme with reduced activity 43 . The substitution of Glu-to-Ala at codon 429 in MTHFR caused by 1298 A > C mutation also reduced enzyme activity, but at a lower degree compared to 677 C > T 43,44 . Although decreased MTHFR activity improve the chemotherapy sensitivity of 5-FU by increasing the level of 5,10-MTHF, the reported data were also inconsistent. For the chemotherapy cohort, some groups reported that MTHFR 1298 A > G and 2756 C > T polymorphism responded negatively to 5-FU for colorectal and gastric cancers patients [45][46][47] . Whereas, Omar et al. showed that the polymorphism of the MTHFR 677C > T was associated with longer TTP and OS for chemotherapy patients 48 . Some reported that no remarkable relationship was found between MTHFR 1298 A > G and 2756 C > T polymorphisms and patient's survival 12,49,50 , which was consistent  with our results. The inconsistent results may be caused by diverse schedules, the varied doses of 5-FU used, various cancer stages or types, and various ethnic populations among those studies.
In conclusion, our study systematically investigated the effects of polymorphisms and their interactions of genes in OCM pathway on clinical prognosis of stage II-III GC patients treated with 5-FU-based regimens. Our findings showed that the polymorphisms of MTRR 66A > G, TS 5′ -UTR 2R > 3R may serve as potential prognostic markers for GC patients receiving 5-FU-based regimens. Results from this study indicated that gene-gene interactions may enhance or neutralize a drug's anticancer effects, emphasizing that further investigations should be conducted in different cohorts with larger number of patients.
Patients and methods. Patient samples. From January 1999 to December 2006, 650 stage II-III patients histopathologically diagnosed with GC were recruited in our study. They underwent surgery at the Yixing People's Hospital (Yixing, Jiangsu Province, China). None of them had received chemotherapy or radiotherapy before operation. 251 patients had 5-FU-based adjuvant chemotherapy, and the other 399 patients had not due to various reasons. The chemotherapy regimens were 5-FU in combination with DDP or L-OHP. Chemotherapy was not conducted until the patients met the following characteristics: platelet counts ≥ 100 × 10 9 /L, neutrophil counts ≥ 1.5 × 10 9 /L, hemoglobin level ≥ 8 g/dl and no sign of organ toxicity. The general clinical characteristics of patients were summarized in Table 1. The median follow-up time is 68.5 months and the maximum follow-up time is 119.0 months. According to Lauren's Standard, the histopathology of GC was classified into intestinal or diffuse types 51 . The TNM stages were evaluated according to the TNM classification of the American Joint Committee on Cancer (AJCC cancer staging manual, seventh edition). This study was approved by the Institutional Review Board of Nanjing Medical University (Nanjing, China). And the methods were carried out in accordance with the approved guidelines. All patients enrolled in this study signed an informed consent on using clinical specimen in medical research.
Genotyping. Genomic DNA was extracted from tumor specimens as described previously 23 . Polymorphisms of the above genes were determined by SNaPshot technology with either ABI florescence-based assay allelic discrimination assay (Applied Biosystems, Forster City, CA) or automated sequencing 52 . The primers used for sequencing in this study are listed in Supplementary Table S4. Genotypes were determined by Genemapper 4.0 software (Applied Biosystems) and the polymorphisms were analyzed in ABI3130 genetic analyzer. About 10% samples were randomly selected to validate genotypes by two people independently. The results were 100% concordant.
Statistical analysis. The overall survival time was calculated from the date of surgery to the last follow up time or death (March, 2009) and mean survival time was chosen if the median survival time (MST) could not be calculated. The relationships of each genotype and combinations of genotypes with clinicopathologic characteristics were elucidated using Student t-test and the Pearson chi-square test according to the types of variables. Log-rank tests and Kaplan-Meier plots were performed by SPSS version 20.0 with two-sided tests and p < 0.05 was considered statistically significant (SPSS Inc, Chicago, IL, USA). Univariate or multivariate Cox regression analysis was adopted to estimate adjusted hazard ratios (HRs) and 95% confidence intervals (CIs).