Classification of Ambiguous Mutations in DNA Mismatch Repair Genes Identified in a Population- Based Study of Colorectal Cancer
Total Page:16
File Type:pdf, Size:1020Kb
HUMAN MUTATION 29(3), 367^374, 2008 RESEARCH ARTICLE Classification of Ambiguous Mutations in DNA Mismatch Repair Genes Identified in a Population- Based Study of Colorectal Cancer Rebecca A. Barnetson,1Ã Nicola Cartwright,1 Annelot van Vliet,1 Naila Haq,1 Kate Drew,1 Susan Farrington,1 Nicola Williams,2 Jon Warner,2 Harry Campbell,3 Mary E. Porteous,2 and Malcolm G. Dunlop1 1Colon Cancer Genetics Group, University of Edinburgh Cancer Research Centre, School of Molecular and Clinical Medicine and Medical Research Council (MRC) Human Genetics Unit, Western General Hospital, Edinburgh, United Kingdom; 2South East (SE) Scotland Clinical Genetics Service, Western General Hospital, Edinburgh, United Kingdom; 3Public Health Sciences, University of Edinburgh, United Kingdom Communicated by Albert de la Chapelle Identification of germline mutations in DNA mismatch repair genes in colorectal cancer probands without an extensive family history can be problematic when ascribing relevance to cancer causation. We undertook a structured assessment of the disease-causing potential of sequence variants identified in a prospective, population- based study of 932 colorectal cancer patients, diagnosed at o55 years of age. Patient samples were screened for germline mutations in MLH1, MSH2,andMSH6. Of 110 carriers, 74 (67%) had one of 33 rare variants of uncertain pathogenicity (12 MLH1,11MSH2,and10MSH6). Pathogenicity was assessed by determining segregation in families, allele frequency in large numbers of unaffected controls, effect on mRNA for putative splice- site mutations, effect on protein function by bioinformatic analysis and tumor microsatellite instability (MSI) status and DNA mismatch repair protein expression by immunohistochemistry. Because of the ambiguous nature of these variants and lack of concordance between functional assays and control allele frequency, we devised a scoring system torankthedegreeofsupportforapathogenicrole.MLH1 c.200G4Ap.G67E,MLH1 c.2041G4A p.A681T, and MSH2 c.263415G4C were categorized as pathogenic through assimilation of all available data, while 14 variants were categorized as benign (seven MLH1,threeMSH2, and four MSH6). Interestingly, there is tentative evidence suggesting a possible protective effect of three variants (MLH1 c.2066A4G pQ689R, c.2146G4A p.V716M, and MSH2 c.965G4A p.G322D). These findings support a causal link with colorectal cancer for several DNA mismatch repair gene variants. However, the majority of missense changes are likely to be inconsequential polymorphisms. Hum Mutat 29(3), 367–374, 2008. r 2007 Wiley-Liss, Inc. KEY WORDS: DNA mismatch repair; Lynch syndrome; HNPCC; colorectal cancer; MLH1; MSH2; MSH6 INTRODUCTION et al., 1997; Quehenberger et al., 2005], it is often not possible to obtain an informative pedigree. Therefore, other criteria have Lynch syndrome or hereditary nonpolyposis colorectal cancer been proposed for the testing of probands for Lynch syndrome, (HNPCC) (MIM 120435) is an autosomal dominant syndrome that take into account early age at onset, gender, tumor site, and that accounts for about 1 to 3% of all colorectal cancer cases [Aaltonen et al., 1998; Samowitz et al., 2001]. The majority of Lynch syndrome cases have a germline mutation in one of the DNA mismatch repair genes MLH1, MSH2,orMSH6 [Akiyama The Supplementary Material referred to in this article can be et al., 1997; Bronner et al., 1994; Fishel et al., 1993; Leach et al., accessed at http://www.interscience.wiley.com/jpages/1059-7794/ suppmat. 1993; Papadopoulos et al., 1994]. In most cases, this results in Received 21February 2007; accepted revised manuscript 7 August tumor microsatellite instability (MSI) because of defective DNA 2007. mismatch repair. Mutation carriers are at a greatly increased risk of ÃCorrespondence to: Rebecca A. Barnetson, Colon Cancer Genet- developing colorectal cancer at an early age and are also ics Group, MRC Human Genetics Unit, Western General Hospital, predisposed to carcinomas of the endometrium, ovary, gastro- Edinburgh, EH42XU,United Kingdom. intestinal tract, and urinary tract [Vasen et al., 1996]. Recently, we E-mail: [email protected] have shown that 6% of males and 3% of females diagnosed with Grant sponsor: Cancer Research UK; Grant number: C348/A3758; colorectal cancer aged less than 55 years carry such mutations Grant sponsor: Scottish ExecutiveChief Scientist’sO⁄ce;Grant num- bers: CZB/4/11; and K/OPR/2/2/D333; Grant sponsor: Medical Re- [Barnetson et al., 2006]. The Amsterdam criteria have been search Council; Grant number: G0000657-53203. developed for the diagnosis of Lynch syndrome, based on the DOI 10.1002/humu.20635 patient’s family history of colon cancer [Vasen et al., 1991]. As Published online 21 November 2007 in Wiley InterScience (www. many families are small and penetrance is incomplete [Dunlop interscience.wiley.com). r 2007 WILEY-LISS, INC. 368 HUMAN MUTATION 29(3), 367^374, 2008 extracolonic carcinomas [Barnetson et al., 2006; Rodriguez-Bigas ambiguous variants identified in a series of colorectal cancer cases, et al., 1997]. using a number of complementary approaches. These included Many DNA mismatch repair mutations are short insertions, comparison of variant frequency in cases and large numbers of deletions, or nucleotide changes that create premature stop population-based controls, tumor MSI and DNA mismatch repair codons, or nucleotide changes at the intron–exon boundaries that protein expression analysis, cosegregation with cancer in families, affect mRNA splicing. Such mutations are readily categorized as mRNA analysis, and bioinformatic assessment of the functional pathogenic, because they encode a truncated protein, resulting in consequence of the amino acid change. The findings have defective DNA mismatch repair. However, the majority of putative considerable relevance to the clinical management of colorectal mutations identified in this study were missense changes that alter cancer probands and their relatives, as well as to the investigation a single amino acid or are close to splice junctions and were of DNA mismatch repair function. therefore of ambiguous relevance to pathogenicity. Where tumor material is available, microsatellite analysis can be MATERIALS AND METHODS a useful indicator of DNA mismatch repair proficiency. However, in studies that select cases on the basis of family history or very A population-based series of patients from throughout Scotland, early age at onset, the likelihood of MSI is considerable because who were diagnosed with colorectal cancer when they were less such criteria enrich for mutation carriers. This does not necessarily than 55 years of age, were recruited to the study between February mean that every variant identified will be disease-causing, as it 1999 and June 2004. During the same period, unaffected controls may be another DNA mismatch repair mutation that is involved. were ascertained from a population-based register (community This is particularly an issue for variants identified in cases that health index) and were invited to participate. Blood samples for have been ‘‘pre-screened’’ by MSI analysis of tumor samples. In DNA analysis were obtained and a family history was taken from addition, MSI is not a universal indicator of pathogenicity as the all cases and controls recruited to the study. majority of tumors from MSH6 mutation carriers are microsatellite Sequence analysis of MLH1, MSH2, and MSH6 was undertaken stable (MSS) or have low-level microsatellite instability (MSI-L), using DNA samples from all patients recruited to the study, and with only one out of five markers affected [Barnetson et al., 2006; primer sequences and mutation nomenclature were based on Lagerstedt Robinson et al., 2007; Wu et al., 1999]. Furthermore, NCBI reference sequences U40960.1 (MLH1), U04045.1 84% of colorectal carcinomas with high-level microsatellite (MSH1), and U73732.1 (MSH6), respectively, with 11 as the instability (MSI-H) lack MLH1 expression as a result of promoter A of the ATG initiation codon. Mutation analysis comprised PCR hypermethylation [Herman et al., 1998]. Immunohistochemistry amplification of all exons of MLH1, MSH2, and MSH6 and can be useful but the presence of a DNA mismatch repair protein analysis of products by dHPLC on the WAVE (Transgenomics, does not unequivocally establish that the expressed protein is Omaha, NE, USA) or by direct sequencing as described previously functional. [Barnetson et al., 2006]. Variant alleles were confirmed by Segregation of the variant with the disease within a family can sequencing (ABI 3730 DNA Analyser; Applied Biosystems, provide evidence of pathogenicity. However, DNA is not always Warrington, UK). MLH1 and MSH2 were also screened for large available from all family members, many carriers are from small genomic deletions by multiplex ligation-dependent probe ampli- nuclear families, and Lynch syndrome mutations are not fication (MLPA; MRC-Holland, Amsterdam, The Netherlands) completely penetrant [Dunlop et al., 1997; Quehenberger et al., and samples were run on an AB3100 Genetic Analyser and 2005]. Even if a variant does segregate with the disease, it may be analyzed with Genotyper (Applied Biosystems) and custom linked to another mutation that is pathogenic. Determination of software (www.ngrl.org.uk/Manchester/Publications.htm]MLPA). the variant frequency in unaffected controls to assess pathogeni- Allelic discrimination assays were designed to determine whether city requires large sample sizes, as most are found at low