Association of Mismatch Repair Mutation With Age at Cancer Onset in Lynch Syndrome

1Lynch syndrome (LS) is caused by germline mutations in one of the mismatch repair genes (MLH1, MSH2, MSH6, or PMS2) and is associated with increased cancer risk, including colorectal cancer (CRC), endometrial cancer (EC), and ovarian cancer (OC). Cancer surveillance by colonoscopy has been shown to reduce mortality from CRC. Gynecological surveillance by transvaginal ultrasonography, hysteroscopy, and endometrial biopsy is of unproven benefit. The age at initiation of cancer surveillance is also controversial. The debate is complicated by the rarity of LS and the lack of LS data. Specific gene mutations affect penetrance and influence the age at onset of LS-associated cancers. The aim of this study was to determine whether mutated gene and mutation type affect age at onset of LS-associated cancers.

2A single-center retrospective cohort study was conducted. The Genetic Register Lynch Syndrome Database from Manchester Centre for Genomic Medicine (large quaternary referral center for 5.6 million people) was interrogated for incident CRC, EC, and OC. All patients provided a priori consent for their data to be used in research; data are anonymized and analyzed as part of clinical audit, and so no ethical review was required. Women were censored from EC and OC analysis at the time of hysterectomy or bilateral salpingo-oophorectomy. All those who did not develop cancer were censored at death or the last recorded date of follow-up.

3Gene sequencing of index individuals was performed as previously described by either Sanger techniques or by next generation sequencing (after 2013) and multiple ligation–dependent probe amplification (MLPA). Only those with heterozygous mutations were included. Statistical analysis was conducted using Stata SE (version 13) and Graphpad Prism (version 7) using Kruskal-Wallis and Mann-Whitney tests and Kaplan-Meier analysis. Significance was defined as P ≤.05. Analysis by gene and mutation type was prespecified as the main comparator for cancer incidence.

4The data set included 1063 individuals (495 men, 568 women) with confirmed pathogenic LS germline mutations. Mean follow-up was 68.2 months.

5Overall, 162 ECs (eTable in the Supplement) were diagnosed. Of these, 26 ECs were diagnosed among the 68 women with MSH6 mutations (38%); 83 ECs among the 279 women with MSH2 mutations (30%); and 53 ECs among the 196 women with MLH1 mutations (27%). Women with MSH6 mutations presented with EC at later ages than women with other mutations: median difference from MLH1, 3.9 years (95% CI, 0.9-6.5 years); median difference from MSH2, 5.7 years (95% CI, 2.4-8.5) (analysis of variance P =.002). The median ages of EC onset were 49 (range, 17-71), 47 (range, 32-72) and 53 (range, 42-66) years for women with MLH1, MSH2, and MSH6 mutations, respectively. Cumulative incidence of EC according to mutated gene is shown in the Figure. When stratified by mutation type, women with truncating MLH1 mutations presented with EC at later ages than those with nontruncating mutations (median difference, 6.6 years; 95% CI, 2.7-10.4 years; P =.002). The same was not true for MSH2 carriers and MSH6 mutation carriers.

6Forty-nine OCs were diagnosed, 9 of which were synchronous with, but not metastatic from, EC. The OC rates associated with the evaluated mutated genes were as follows: MSH2, 10% (29 of 279); MLH1, 8% (15 of 196); and MSH6, 7% (5 of 68) (eTable in the Supplement). The median age at OC diagnosis was 47 years (range, 24-70 years); however, women with truncating mutations were older at the time of diagnosis than those with nontruncating mutations (median difference, 6.3 years (95% CI, 0.2-14.2 years; P =.04). This was true across all genes but not when individual genes were analyzed separately.

7Colorectal cancer was diagnosed in 546 individuals (241 women and 305 men) at the following mutated gene–associated rates: MLH1, 61% (249 of 409); MSH2, 50% (239 of 479); MSH6, 33% (43 of 129); and PMS2, 32% (15 of 46) eTable in the Supplement). Women presented with their first CRC at later ages than men (median difference, 3.3 years; 95% CI, 1.2-5.4 years; P =.002). Individuals with MSH6 mutations presented at later ages than those with mutations in other genes: MLH1 median difference, 9 years (95% CI, 5-14 years); MSH2 median difference, 8 years (95% CI, 3-12 years); and PMS2, median difference, 6 years (95% CI, −3 to 13 years; (P <.001) (Figure). Furthermore, there was a trend for individuals with truncating MLH1 mutations presenting with CRC at later ages than those with nontruncating mutations, although the difference was not statistically significant. In women, CRC was the sentinel cancer, with CRC being diagnosed before EC (median difference, 2.6 years; 95% CI, 0.2-4.8 years; P =.006).

8The number of biennial colonoscopies and annual gynecology reviews to identify cancer in individuals with LS aged 25 to 39 years is detailed in the Table. If a rate threshold of 0.5% cancers per screen is justifiable, colonoscopies starting at age 25 years for those with MLH1 and MSH2 mutations and at age 30 years for those with MSH6 and PMS2 are appropriate. For that same rate, gynecological surveillance is appropriate from age 30 years for those with MSH2 mutations, from age 35 years for those with nontruncating MLH1 mutations, and from age 40 years for those with MSH6 and truncating MLH1 mutations. Women with heterozygous PMS2 mutations do not warrant gynecological surveillance because their absolute risk of gynecological cancer is very low.

9In this study, MSH6 mutation carriers presented later with CRC and EC than those with MLH1 or MSH2 mutations, consistent with previous studies. Furthermore, women with truncating MLH1 mutations presented with EC on average 6 years later than those with nontruncating mutations, which to our knowledge has not been reported before.

10Among the study patients, CRC had an earlier age at onset than EC. This result is from our research group’s established surveillance program for CRC enabling earlier detection and thus younger ages of diagnosis for screened individuals. When CRC is the presenting cancer in known mutation carriers, the potential benefits of synchronous risk-reducing hysterectomy and bilateral salpingo-oophorectomy should be discussed with women.

11A strength of the present study is the large LS database. Its origin as a clinical database necessitates its accuracy and prospective ongoing maintenance by a dedicated data manager. The center has extensive experience in the clinical application of DNA sequencing. While missing data could have biased the results, only 4% of data sets were incomplete.

12It is interesting to speculate why patients with truncating mutations developed cancer at a later age. In ataxia-telangiectasia, missense ATM mutations lead to a nearly functional protein molecularly similar to the ATM protein that competes with the wild type and creates premature genomic instability and earlier disease onset. Conversely, the truncating ATM mutation protein is nonfunctional, and therefore, genomic stability is preserved while the wild type remains. It is not until the second knockdown mutation occurs that ATM dysfunction ensues. This is called the “dominant negative” phenomenon and could explain the later age at EC onset seen in carriers of the truncating MLH1 mutation.

13Møller et al studied the impact of mutated genes on age at cancer onset and recommended surveillance from age 25 years in carriers of MLH1 or MSH2 mutations and from age 40 years in those with MSH6 or PMS2 mutations. Some national expert guidelines corroborate these recommendations, but others endorse a standardized approach, which is simplistic and easy to implement but crude. Targeted surveillance has a de facto benefit in both reducing health care costs and decreasing patient distress.

14The data in the present study originate from a defined geographical area in the northwest of England, and potential local population factors may limit the generalizability of the conclusions. Rare genetic conditions like LS benefit from collaborative multicenter investigation. The cumulative incidence reported should not be interpreted as the risk for mutation carriers because those diagnosed with cancer are more likely to be tested. We did not make corrections for index testing nor for the proportion of untested relatives who would be carriers.

15To our knowledge, this is the first study that promotes an extra tier of risk stratification according to mutation type, and not just mutated gene. Cancer surveillance could be started later for individuals with MSH6 mutations, and surveillance for EC could be started later in those with truncating MLH1 mutations.