Colon Cancer Germline Genetics: The Unbelievable Year 1993 and Thereafter

First in a series of seminal discoveries was the linkage between a dinucleotide repeat marker (D2S123) and colorectal cancer in two large families (3). The experimental findings were robust and definitive. Confirming Henry Lynch's work, this finding proved for the first time that a dominantly inherited gene mutation exists that confers high risk of colorectal cancer. Furthermore, the physical assignment of the locus to the short arm of chromosome 2 provided a starting point for efforts to physically map and clone the culpable gene.

The second discovery was directly related to the mapping by linkage. Assuming that the culpable gene was of tumor suppressor type likely being susceptible to loss of heterozygosity (LOH), a search for LOH in the region was initiated. It did not disclose LOH, but instead widespread alterations in short repeat sequences (microsatellites) were found (4). The alterations were termed RER for “replication error” but later became known as microsatellite instability (MSI). A remarkable feature was that MSI occurred in the majority of tumors derived from individuals with hereditary colorectal cancer, whereas it occurred in only 13% of sporadic cancers. The main conclusions were 2-fold. First, the alterations (deletions and/or additions) hinted that the culpable gene had to do with DNA synthesis or repair. Second, the absence of LOH and presence of MSI suggested a previously undescribed molecular pathway. Significantly, in a third article in the same issue of Science, MSI was described in 28% of colorectal cancer tumors. MSI correlated with right-sided tumor location and with increased patient survival (5). These data strongly supported the previous findings. Finally, a month after the above three articles had appeared, an article was published showing essentially the same patterns of MSI as the previous articles, supporting the idea of a novel pathway (6).

The obvious thing to do next was to find and characterize the gene that by linkage had to be located next to D2S123 in chromosome 2p. Previous descriptions of the mutH, mutL, and mutS mismatch repair pathways in bacteria and yeast helped guide the cloning and characterization of a human mismatch repair gene, which was named hMSH2 (7). As it mapped to precisely the same locus in chromosome 2 that had been implicated by linkage in hereditary cancer families (3), it was a highly likely candidate gene for hereditary colorectal cancer. It just remained to be shown to harbor germline mutations. The authors found an intronic T to C variant close to a splice site in several colorectal cancer patients and concluded that this proved that MSH2 was the culprit gene. It later turned out that the proposed mutation was not pathogenic, but the conclusion was nevertheless correct (7). This was published in the December 3 issue of Cell. In the next issue, on December 17, another group of researchers reported having identified and cloned MSH2. In this case, several pathogenic mutations, both germline and somatic, were detected in patients with hereditary colorectal cancer, sealing the evidence implicating MSH2 (8).

The year 1993 was still to see further developments. In the December 17 issue of Cell, another article on mismatch repair and MSI was published (9). MSI-positive and MSI-negative cell lines plus tumor and germline tissue from colorectal cancer patients were utilized to show that the mutation rates of CA repeats were at least 100 times higher when MSI was present. Relying on extensive previous research, a profound defect in strand-specific mismatch repair was demonstrated in association with MSI and occurred not only as microsatellite but also as single base–base mismatches. These experiments provided a biochemical basis for what was called the “mutator phenotype.” We note here that the last author (P. Modrich) received the 2015 Nobel Prize in chemistry for his work on the fundamentals of DNA mismatch repair.

Finally, what about the article (1) that caused this commentary to be solicited? It is a forerunner of the voluminous stream of translational research that followed (and continues to follow) as a consequence of the 1993 discoveries. In particular, researchers working on hereditary colon cancer were beginning to confirm a message that had already been repeatedly pronounced by Dr. Lynch: in families with colorectal or endometrial cancer, there are (many) other cancers as well. The article (1) was a first attempt at determining the true spectrum of cancers in Lynch syndrome or, more precisely, to determine what cancers comprised subsets of tumors with MSI. Six organs were studied, and the results were as follows: MSI was found in 10% of colorectal, 18% stomach, and 22% endometrial cancers, whereas no MSI was seen in breast, testis, or lung cancers. The result suggested that the MSI phenotype occurs in subsets of some but not all tumors known to occur in multicancer families. With the increasing use of DNA sequencing of the mismatch repair genes and widespread research into families suggestive of having Lynch syndrome, the “true” spectrum of Lynch syndrome soon became better defined. Toward 2005, the tumor spectrum often was listed as follows: colorectal, endometrium, ovary, stomach, small bowel, hepatobiliary tract, pancreas, upper uroepithelial tract, and brain. Some evidence suggested that the list might include the adrenal, sarcomas, prostate, and neurofibromatosis.

In 1994, several research groups contributed to the cloning of the remaining major mismatch repair genes MLH1, PMS2, and (later) MSH6. It began to be shown that the proportion of all colorectal cancer patients that have Lynch syndrome varies between populations, being typically of the order of 1% to 3%. When patients affected by a Lynch syndrome cancer are molecularly studied and mutations identified, cascade testing of family members typically reveals other mutation-positive, usually unaffected, individuals. The value of diagnosing Lynch syndrome followed by clinical surveillance is beginning to be realized. Adequate counseling is all important. Colonoscopy with removal of polyps as well as surgical removal of the uterus and ovaries are highly successful in preventing cancer and/or death from cancer. The risk of cancer in organs other than the colorectum, endometrium, ovary, small bowel, and stomach is only moderately increased; however, surveillance of organs, such as the hepatobiliary tract, uterus, and pancreas, should be done whenever possible. Here is where the research initiated in the lead article (1) is still relevant. As mentioned above, the study found no evidence of MSI in breast, testis, or lung cancers. Subsequently, numerous studies have addressed these questions. Specifically, we ask whether MSI and defective mismatch repair occur in subsets of common cancers (breast, prostate, testis, and lung). A survey of the literature gives conflicting answers. Virtually, dozens of articles on each of these cancers have been published, and the conclusions are markedly variable. In each cancer, some studies clearly suggest that subsets of patients have MSI tumors, whereas in other studies, the result is negative. Two conclusions are apparent: (i) the evidence for an involvement is strong enough to prompt further studies. Once next-generation sequencing has been successfully performed in large series of the tumors, we shall have the answers; (ii) at present, it appears prudent to assume that MSI-positive tumors occur in these cancers, perhaps with the exception of lung cancer.

Not all patients with MSI have Lynch syndrome; on the contrary, some 60% to 70% do not. In the majority of these, the MSI and mismatch repair deficiency is caused by a somatically acquired hypermethylation of the MLH1 promoter leading to knockdown of MLH. It is important to find out whether MSI is caused by a germline mutation or by promoter methylation, because in the former case family members are at risk; in the latter case, they are not. Methylation analysis provides the answer. Interestingly, it has only recently begun to look as if most tumors that have MSI, but neither a germline mismatch repair gene mutation nor promoter hypermethylation, are caused by biallelic acquired mutations (“two somatic”). To identify these tumors is important in the counseling context.

Strong evidence has been presented supporting previous data that MSI tumors have better prognosis than non-MSI tumors. Moreover, among patients receiving adjuvant chemotherapy, those with MSI apparently show no benefit. These findings will no doubt be expanded, for example, in relation to the recent surge of interest in immunotherapy (see below).

The invention of the various next-generation sequencing systems is having a major impact on the field of colorectal cancer. Gene sequencing panels specifically designed for colorectal cancer are already available and accurate. With falling prices, they are increasingly useful. It seems likely that we are moving toward first-line use of next-generation sequencing panels.

Finally, there is significant recent news regarding treatment options. As described already in 1993, mismatch repair–deficient cells displayed greater than 100-fold numbers of mutations in an in vitro system (9). This led to the hypothesis that neoantigens produced by mutated genes might be targeted by blocking immune checkpoints with antibodies against genes such as programmed death 1 (PD1; ref. 10). In a small phase II trial, end-stage patients with MSI-positive tumors (n = 20) and MSI-negative tumors (n = 21) were treated with pembrolizumab, a PD-1 antibody. The result was strikingly positive. MSI-positive cancers showed a progression-free survival of 78% as compared with 11% in mismatch repair–proficient cases. In the cohort with mismatch repair deficiency, there were 11 colorectal tumors and 9 tumors in other sites (bile duct, endometrial, small bowel, and gastric carcinoma), so the effect was not confined to colorectal cancer (10).

A correlation exists between the success of treatment and the prevalence of mutations (and neoantigens). In the study referred to here (10), mismatch repair–deficient tumors had an average of 1,782 coding mutations compared with 73 in non-MSI tumors. The efficacy of PD-1 blockade in MSI tumors other than colorectal (10) is no doubt inspiring researchers to look for MSI and/or high mutation load in tumors of various organs with the intent to try therapies, including anti-PD-1. Evidence so far is restricted to late-stage metastatic disease. It is safe to predict that studies will soon begin to appear in the literature and in meeting abstracts. We note with some satisfaction that the circle closes on the original 2½ page article that inspired this commentary (1). There is ample reason to be optimistic that novel cohorts of patients with cancer will benefit from this research.

No potential conflicts of interest were disclosed.