Somatic Nonepigenetic Mismatch Repair Gene Aberrations Underly Most Mismatch Repair–Deficient Lynch-Like Tumors

Abstract

Individuals with Lynch syndrome have a pathogenic germline variant in one of the mismatch repair (MMR) genes (MLH1, MSH2/EPCAM, MSH6, or PMS2) and are often recognized based on MMR-deficient (dMMR) colorectal cancers (CRCs) or endometrial cancers (ECs).1Lynch H.T. et al.Clin Genet. 2009; 76: 1-18Crossref PubMed Scopus (555) Google Scholar,2Kunitomi H. et al.Oncol Lett. 2017; 14: 3297-3301Crossref PubMed Scopus (17) Google Scholar Approximately 15% of CRCs and 20%–30% of ECs are dMMR, mostly due to MLH1-promoter hypermethylation. dMMR tumors without a pathogenic germline variant or MLH1-promoter hypermethylation (Lynch-like) may have 2 somatic nonepigenetic MMR aberrations (somatic dMMRs).3Herman J.G. et al.Proc Natl Acad Sci U S A. 1998; 95: 6870-6875Crossref PubMed Scopus (1628) Google Scholar, 4Mensenkamp A.R. Vogelaar I.P. et al.Gastroenterology. 2014; 146: 643-646Abstract Full Text Full Text PDF PubMed Scopus (211) Google Scholar, 5Haraldsdottir S. et al.Gastroenterology. 2014; 147: 1308-1316Abstract Full Text Full Text PDF PubMed Scopus (233) Google Scholar, 6Geurts-Giele W.R. Leenen C.H. et al.J Pathol. 2014; 234: 548-559Crossref PubMed Scopus (96) Google Scholar, 7Pearlman R. Haraldsdottir S. et al.J Med Genet. 2019; 56: 462-470Crossref PubMed Scopus (19) Google Scholar Knowledge on the contribution of Lynch syndrome–associated germline variants and somatic dMMR as a cause of dMMR for different genes and age groups is important for tumor surveillance strategies. However, the percentage of germline variants in MMR genes that are missed by current methods is unclear. Here, we show an unprecedented prevalence of somatic dMMR in dMMR tumors of patients who were referred for diagnostic testing for Lynch syndrome after exclusion of MLH1-promoter hypermethylation. Study cohort 1 included individuals (N = 304) with a dMMR CRC or EC without somatic MLH1-promoter hypermethylation who were counseled by a clinical geneticist at the Radboudumc (Nijmegen, the Netherlands) between 2009 and 2019. The cohort included individuals (n = 151) with a pathogenic MMR germline variant and individuals with a dMMR CRC (n = 130) or EC (n = 23) without a pathogenic MMR germline variant. Study cohort 2 consisted of individuals (N = 125) with a dMMR CRC (n = 101) or EC (n = 28) in which Lynch syndrome and MLH1-promoter methylation were excluded elsewhere and only underwent somatic MMR mutation analysis at the Radboudumc (Supplementary Figure 1A). Thirty tumors from study cohort 1 and 8 tumors from study cohort 2 with insufficient tumor material were excluded. This study (CMO-2018-4922) was approved by the local ethical committee of the Radboudumc. Genomic DNA extracted from formalin-fixed, paraffin-embedded normal and dMMR tumor tissues (252 tumors of 248 patients) was analyzed with targeted sequencing for the MMR genes. Tumors with no or 1 identified somatic event were subjected to multiplex ligation-dependent probe amplification (MLPA) to detect somatic MMR gene deletions and duplications and methylation-specific MLPA to evaluate MSH2-, MSH6-, and PMS2-promoter methylation (for details, see the Supplementary Methods). Pathogenic MMR gene variants of somatic dMMR CRCs identified in this study and the study by Pearlman et al7Pearlman R. Haraldsdottir S. et al.J Med Genet. 2019; 56: 462-470Crossref PubMed Scopus (19) Google Scholar were included in a mutational signature analysis. In total, we performed somatic MMR mutation analyses for 205 dMMR CRCs and 47 ECs. Somatic dMMR was identified in 88.8% and 80.9% of analyzed dMMR CRCs (182/205) and ECs (38/47), respectively (total, 87.3%; 220/252 tumors) (Figure 1A and Supplementary Table 1). The majority of somatic inactivating events were identified by targeted sequencing (211/220 tumors; 95.9%), and exon deletions were detected by MLPA analysis in 3.6% (8/220 tumors) and PMS2-promoter methylation in 0.5% of dMMR tumors (1/220 tumors). Loss of heterozygosity occurred more frequently for MLH1 (78.3%) than for the other MMR genes (<43.4%) (Supplementary Figure 1B). To investigate the distribution of patients with Lynch syndrome or somatic dMMR throughout different age groups, study cohort 1 was divided into 6 age cohorts (Figure 1B). The mean age at diagnosis of patients with somatic dMMR tumors is significantly higher compared to that of patients with Lynch syndrome (52 vs 48 years; P < .01). Furthermore, the fraction of patients with somatic dMMR was associated with an older age at diagnosis (Figure 1B and Supplementary Figure 1C–F). The proportion of somatic dMMR tumors differs per inactivated gene (P < .01) (Figure 1C). For MLH1, the proportion of tumors with somatic dMMR is larger than with a germline MLH1 pathogenic variant (50.5% and 34.7%, respectively), whereas for MSH2, the proportion is similar (42.0% and 44.4%, respectively), and most MSH6 and PMS2 tumors are the result of a pathogenic germline variant (76.2% and 73.9%, respectively) (Figure 1C). The proportion of patients with Lynch syndrome (51.5% vs 53.8%) and somatic dMMR (34.6% vs 30.8%) was independent of whether tumors were or were not screened for dMMR before referral of the patients to clinical genetic counseling (P > .89). Next, we investigated, by means of mutational signature analysis, whether a general mutational mechanism underlies the collective somatic mutation pattern of somatic dMMR CRCs, but we did not identify a single responsible mutational process (Supplementary Figure 1G). The majority of dMMR tumors without a pathogenic MMR germline variant or somatic MLH1-promoter hypermethylation are the result of 2 somatic MMR gene aberrations (87.3% of tumors). Whereas the majority of somatic MMR inactivation can be detected by using targeted sequencing, a fraction of somatic dMMR tumors harbor somatic exon deletions that are missed by sequencing. Unexpectedly, 1 tumor presented with PMS2-promoter methylation, which, to the best of our knowledge, has not been described before. In a minority of the dMMR tumors, only 1 or no somatic inactivating event was detected that might be due to missing somatic or germline variants that stay undetected by the analysis techniques used. In general, somatic dMMR tumors are diagnosed at an older age compared to Lynch syndrome–associated dMMR tumors. Their mean age at diagnosis is likely underestimated, because in the Netherlands, patients with CRCs and ECs diagnosed after 69 years of age are not referred for MMR gene testing irrespective of dMMR status. In CRCs diagnosed before 70 years, MLH1-promoter hypermethylation was shown to occur 4.5 times more frequently than somatic dMMR.8Vos J.R. et al.International Journal of Cancer. 2020; 147: 2150-2158Crossref PubMed Scopus (8) Google Scholar Especially in young patients with somatic dMMR, it remains of interest to determine the cause of these somatic mutations. However, we did not detect a common mutational mechanism in the CRCs investigated. To conclude, the majority of dMMR CRCs and ECs in the absence of pathogenic MMR germline variants or somatic MLH1-promoter hypermethylation are caused by somatic genetic MMR aberrations. These findings indicate that the diagnostic yield of somatic MMR gene testing is high and significantly reduces the number of patients who remain uncertain about their genetic susceptibility to Lynch syndrome after germline testing. Therefore, a combined analysis of germline and somatic MMR gene testing, potentially including somatic MMR promoter-methylation and -exon deletions, should be considered for patients with dMMR tumors without MLH1-promoter hypermethylation who are eligible for genetic testing. The authors thank all study participants. Furthermore, the authors would like to thank Evelien Hoenselaar, Neeltje Arts, and Hanneke Volleberg-Gorissen, Monique Goossens. Sandra Hendriks-Cornelissen, Mandy Hermsen, for their contributions to this project. Dutch LS-like study group (alphabetical order): Fonnet E. Bleeker, Antoni van Leeuwenhoek Hospital; Charlotte J. Dommering, Amsterdam University Medical Centers; Mirjam M. de Jong, University Medical Center Groningen; Nicoline Hoogerbrugge, Rachel S. van der Post, Brigit Wapstra, Radboud university medical center; Edward M. Leter, Maastricht University Medical Center+; Tom G.W. Letteboer, University Medical Center Utrecht; Maartje Nielsen, Leiden University Medical Center. Lisa Elze, MSc (Data curation: Equal; Formal analysis: Equal; Investigation: Equal; Validation: Equal; Writing – original draft: Equal; Writing – review & editing: Equal); Arjen R. Mensenkamp, PhD (Data curation: Equal; Formal analysis: Equal; Investigation: Equal; Validation: Equal; Writing – review & editing: Equal); Iris D. Nagtegaal, MD, PhD (Data curation: Equal; Formal analysis: Equal; Validation: Equal; Writing – review & editing: Equal); Wendy van Zelst-Stams, MD, PhD (Data curation: Equal; Resources: Equal; Writing – review & editing: Equal); Richarda M, de Voer, PhD (Conceptualization: Equal; Formal analysis: Equal; Supervision: Equal; Writing – original draft: Equal; Writing – review & editing: Equal); Marjolijn J.L. Ligtenberg, PhD (Conceptualization: Equal; Formal analysis: Equal; Investigation: Equal; Supervision: Equal; Validation: Equal; Writing – review & editing: Equal). Immunohistochemistry for the detection of MLH1, MSH2, MSH6, and PMS2 and somatic MLH1-promoter hypermethylation of MLH1-deficient tumors using methylation-specific MLPA (MS-MLPA) (MRC-Holland) were performed by using standard procedures during routine diagnostic procedures. Tumors were categorized based on the immunohistochemical pattern as MLH1 deficient (loss of MLH1 and PMS2 staining), PMS2 deficient (loss of PMS2 staining), MSH2 deficient (loss of MSH2 and MSH6 staining), and MSH6 deficient (loss of MSH6 staining). Genomic DNA was extracted from white blood cells or formalin-fixed, paraffin-embedded normal and dMMR tumor tissue. MLH1 (NM_000249.3), MSH2 (NM_000251.2), MSH6 (NM_000179.2), and PMS2 (NM_000535.5) were sequenced by Sanger- and AmpliSeq-based sequencing (Life Technologies, Bleiswijk, The Netherlands) according to the manufacturer’s instructions and/or single-molecule molecular inversion probe–based next-generation sequencing, as described previously.1Eijkelenboom A. et al.J Mol Diagn. 2016; 18: 851-863Abstract Full Text Full Text PDF PubMed Scopus (64) Google Scholar, 2Mensenkamp A.R. Vogelaar I.P. et al.Gastroenterology. 2014; 146: 643-646Abstract Full Text Full Text PDF PubMed Scopus (218) Google Scholar, 3de Voer R. Diets I. Post R. et al.Clin Gastroenterol Hepatol.2021Google Scholar Tumors without 2 somatic events by Sanger and ion semiconductor sequencing were resequenced by using single-molecule molecular inversion probe–based next-generation sequencing. Sequence primers and probes are available upon request. For all germline analyses and for MMR-deficient tumors without 2 somatic events after targeted sequencing and with available matched normal tissue were subjected to MLPA analysis to detect exon deletions and duplications in EPCAM, MLH1, MSH2, MSH6, and PMS2 and MS-MLPA to reveal promoter methylation of MSH2, MSH6, or PMS2 by using the MRC-Holland SALSA probe mix assays (MRC-Holland) according to the manufacturer’s instructions. Tumors were considered to be somatic dMMR when 1 of the following mutation type combinations was identified in the tumor tissue: (1) 2 (likely) pathogenic somatic mutations in 1 MMR gene, (2) a (likely) pathogenic mutation with loss of heterozygosity (LOH) in the same MMR gene, (3) tumors with a variant of unknown significance and either a (likely) pathogenic somatic variant or LOH in the same MMR gene, or (4) tumors with promoter methylation (Supplementary Table 1). Tumors needed to fulfill 1 of the following criteria defined to have 1 somatic event: (1) 1 (likely) pathogenic event, (2) only LOH, or (3) 2 (likely) pathogenic events of which at least 1 is presumed a subclonal event. Tumors needed to fulfill 1 of the following criteria defined to have no somatic event: (1) 1 (likely) pathogenic-variant presumed subclonal event, (2) only variants of unknown significance or (likely) benign variants, or (3) no variants. The distributions of tumors with dMMR in the 4 MMR genes were compared between study cohorts 1 and 2. Only the first detected dMMR tumor of each patient without a germline mutation or MLH1-promoter hypermethylation was included. One dMMR tumor with 2 MMR genes that presented with somatic dMMR was considered MLH1 deficient in this comparison. Next, the nature of the second hit in somatic dMMR tumors was analyzed. Sequencing changes and exon deletions were considered as a second variant, whereas LOH and complete gene deletions were considered as LOH in the analysis. One tumor with a PMS2 methylation was excluded for this analysis. The age and gene distribution of individuals with Lynch syndrome or somatic dMMR tumors of study cohort 1 were compared. Six age groups were built to compare the age at diagnosis. Tumors with germline analysis before 2009 were excluded. In the gene distribution analysis, 1 microsatellite instable tumor that was not subjected to somatic mutation analysis was excluded because the specific immunohistochemical status was unknown. The proportion of patients who were ascertained by the Clinical Genetic Center of the Radboudumc were compared to the proportion of patients ascertained by a routine screening. Tumors with germline analysis before 2009 were excluded. Over the years, the criteria used to refer a patient to a Clinical Genetic Center were changed to less stringent criteria. Somatic variants of CRCs with somatic dMMR found in study cohorts 1 and 2 were included. Additionally, somatic variants in dMMR CRCs identified by Pearlman et al4Pearlman R. Haraldsdottir S. et al.J Med Genet. 2019; 56: 462-470Crossref PubMed Scopus (22) Google Scholar were included as a third cohort. Mutational signature analysis was performed with the R package DeconstructSigs tool.5Rosenthal R. et al.Genome Biol. 2016; 17: 31Crossref PubMed Scopus (440) Google Scholar The characteristics of patients with somatic dMMR mutated CRC and EC were summarized with a mean. Moreover, characteristics of study cohort 1 were compared to patients with Lynch syndrome and unsolved patients not subjected to somatic sequencing using the Student t test and chi-square test. Download .xlsx (.05 MB) Help with xlsx files Supplementary Table

Topics & Concepts