Response: Re: MLH1 93G>A Promoter Polymorphism and the Risk of Microsatellite-Unstable Colorectal Cancer

We thank Drs Hubner and Houlston for their interest in our study. Linkage disequilibrium with a mutation or another single-nucleotide polymorphism is a possibility that we considered and discussed in our article. In Ontario and Newfoundland, all case patients with a high-frequency microsatellite instability status (MSI-H) and/or mismatch repair immunohistochemical deficiency undergo germline mutation screening for the MLH1, MSH2, and MSH6 genes. This screening includes exon-by-exon genomic DNA sequencing and the detection of large genomic insertions–deletions with a multiplex ligation-dependent probe amplification assay. Among the 117 patients with colorectal cancers who had MSI-H tumors in the Ontario Familial Cancer Registry, 27 met Amsterdam I criteria—eight (30%) of whom carried germline MLH1 gene mutations. Of these eight case patients, five were homozygous wild type for the MLH1 polymorphism -93G>A, two were heterozygous for it, and one was homozygous for the variant. In the remaining 90 non-Amsterdam MSI-H colorectal cancer patients, we identified three (3.3%) with MLH1 germline mutations (one for each of the MLH1 -93 respective genotypes). For the Newfoundland MSI-H colorectal cancer patients, no germline MLH1 mutations were detected. Thus, the sample of MSI-H MLH1 germline mutation carriers in our populations (whether meeting Amsterdam criteria or not) was small and did not demonstrate evidence for linkage of MLH1 -93G>A with germline MLH1 mutations. Moreover, although enriched with familial colorectal cancers, a large proportion of case patients in our study did not in fact have familial colorectal cancers and, therefore, our results should not be compared with those from the study by Lagerstedt Robinson et al. ( 1 ), which clearly focuses on high-risk colorectal cancer families. Hubner and Houlston make reference to the MLH1 mutation -42C>T, which was reported in one Newfoundland family. This variant is not a common variant ( 2 ) and was not present in any Newfoundland colorectal cancers in our study. We have in fact sequenced the promoter region of MLH1 from germline DNA of selected case patients from both populations meeting Amsterdam I criteria, from those meeting Amsterdam I criteria and having a MSI-H colorectal cancer, and from those with microsatellite stable tumors. We did not find the MLH1 -42C>T or the other reported MLH1 promoter variants-28A>G ( 3 , 4 ) or -28A>T ( 5 ) in any of the case patients that we examined. Although to date we have not found evidence of linkage of -93G>A with other MLH1 genetic sequence changes, we continue to explore the possibility of such linkage as well as other plausible mechanisms. Only a very small proportion of the 117 Ontario MSI-H colorectal cancer patients carried germline MLH1 (11 tumors), MSH2 (10 tumors), and/or MSH6 (one tumor) mutations. In Newfoundland, only one of the 33 MSI-H colorectal cancer patients carried a germline MSH2 mutation. The remaining MSI-H colorectal cancers are likely attributed to methylation of the MLH1 promoter. Indeed, recently, the MLH1 -93G>A polymorphism was associated with MLH1 promoter methylation in endometrial and colorectal cancers ( 6 ). Interestingly, a more recent study found that the MLH1 promoter tends to be methylated on a chromosome bearing a variant -93 “A” allele ( 7 ). Such studies, including our own, provide evidence for the involvement of MLH1 -93G>A in cancer, although its exact role remains unclear. We disagree with the suggestion of selection bias with respect to MSI testing in the case populations in our study. All colorectal cancers within the two registries underwent MSI analysis, and the unavailability of MSI results for our case populations was not a matter of bias, as shown by similarity in the general clinical and pathologic characteristics between our populations and those of previous studies cited in our article. We addressed the issue of selection bias with respect to the willingness to participate in the registries; however, the proposed selection bias would play a role only if the lack of MSI data was itself influenced by the MLH1 -93G>A genotype, which cannot be the case. It is important that the findings in our study were replicated in two independent populations. After examining Hardy–Weinberg equilibrium and the minor allele frequencies of the MLH1 -93G>A in control subjects and case patients between each population, we found no differences between these two populations for this common single-nucleotide polymorphism. As such, we feel justified in using the Newfoundland population to validate our findings. Furthermore, in a large case–control study of colon cancer, there is evidence of a similar association between MLH1 -93G>A and MSI-H colon cancers (Ulrich CM and Potter JD: personal communication). Because our findings have been replicated in our own study and in a separate independent population and because the MLH1 -93G>A polymorphism has been reported in a variety of cancers, until there is evidence of linkage to mutation, this polymorphism should still be considered a low-penetrance susceptibility allele.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.