The Genotype of MLH1 Is An Independent Predictor of Outcome In Diffuse Large B-Cell Lymphoma Treated with R-CHOP: a Training-Validation Study

Several drugs utilized in diffuse large B cell lymphoma (DLBCL) rely on DNA damage for tumor killing. This study aimed at verifying whether single nucleotide polymorphisms (SNPs) of genes involved in DNA repair may contribute to prognostication of DLBCL treated with R-CHOP. The study utilized a training-validation design. The training cohort (n=163) was a mono-institutional, prospectively collected, consecutive series of DLBCL homogeneously treated with the same chemotherapeutic regimen both at diagnosis (R-CHOP21) and at relapse/progression (R-DHAP ± BEAM conditioned autologous stem cell transplant, ASCT). The validation cohort (n=156) was a multi-institutional retrospective series of DLBCL treated with R-CHOP at diagnosis. Candidate SNPs (n=35) were selected by an educated guess approach, and included SNPs belonging to genes involved in: i) mismatch repair (MLH1); ii) base excision repair (XRCC1, OGG1); iii) nucleotide excision repair (ERCC1, ERCC2, ERCC4, ERCC5, ERCC6, XPA, XPC); iv) double strand break repair (BRCA1, BRCA2, LIG4, XRCC2, XRCC3, XRCC4, XRCC6); and v) direct reversal (MGMT). Clinical endpoints were progression free survival (PFS) after R-CHOP, overall survival (OS) from diagnosis, and OS from salvage treatment. Univariate analysis controlled for multiple comparisons identified MLH1 rs1799977 as the sole SNP predicting DLBCL OS in the training series (AG/GG genotype: HR: 3.23; 4-year OS: 55.5% vs AA genotype: 4-year OS: 80.9%; p<.001; q=.009) (Fig. 1A). Multivariate analysis identified the MLH1 rs1799977 AG/GG genotype (HR: 3.14; p<.001) as an independent predictor of OS, along with IPI score (HR: 1.38; p=.037) and bulky disease (HR: 2.56; p=.004). The prognostic relevance of MLH1 rs1799977 in the DLBCL training series was due to the impact on risk of failing both R-CHOP21 (AG/GG genotype; HR: 2.02; 4-year PFS: 47.5%; AA genotype: 4-year PFS: 65.6%; p=.007) (Fig. 1B) and second line treatment (AG/GG genotype: HR: 3.04; 2-year OS from salvage: 16.0%; AA genotype: 2-year OS from salvage: 57.3%; p=.007) (Fig. 1C). Multivariate analysis identified the MLH1 rs1799977 AG/GG genotype (HR: 1.96; p=.010) as an independent predictor of PFS after R-CHOP21, along with IPI score (HR: 1.41; p=.002) and bulky disease (HR: 1.96; p=.012). By bivariate analysis, MLH1 predicted OS from salvage treatment independent of having (p=.002) or having not (p=.049) consolidated with ASCT. The DLBCL validation series did not differ from the training series in terms of clinical features at presentation, median follow-up (p=.429), OS (p=.331), PFS (p=.416), OS from salvage (p=.987), and prevalence of MLH1 rs1799977 genotypes (p=.378). The MLH1 rs1799977 AG/GG genotype was confirmed as a predictor of poor outcome in the DLBCL validation series when considering all clinical endpoints, including: i) OS (unadjusted HR: 3.22, p=.001; adjusted HR: 3.15; p=.001); ii) PFS (unadjusted HR: 1.98, p=.017; adjusted HR: 1.86, p=.018); and iii) OS from salvage treatment (unadjusted HR: 2.95, p=.027). Pooling of the training and validation series (n=319) revealed that MLH1 AG/GG predicts DLBCL OS within subgroups defined by IPI. The biologic plausibility of the association between MLH1rs1799977 genotype and DLBCL outcome is supported by four lines evidence: i) MLH1rs1799977 is a nonsynonymous SNP causing the I219V amino acidic substitution in MLH1, a gene of the mismatch repair pathway; ii) in silico, MLH1rs1799977 is predicted to have deleterious consequences; iii) in vitro, the G variant allele of MLH1rs1799977 associates with reduced MLH1 protein expression; iv) loss of MLH1 expression in tumor cells is known to induce refractoriness to doxorubicin and platinum compounds.