Pancreatic cancer and a novel MSH2 germline alteration

Abstract

Objectives

To describe a novel MSH2 missense alteration co-segregating with pancreatic cancer.

Methods

Observational study of a kindred in which a novel MSH2 missense alteration was identified.

Results

We report a family in which a MSH2 P349L missense alteration is co-segregating with pancreatic cancers among three nonsmoking first degree relatives. Lynch syndrome-related tumors from individuals carrying this alteration consistently showed loss of immunohistochemical expression of MSH2 and in-silico analyses support interpretation of this DNA alteration as likely pathogenic.

Conclusions

The MSH2 P349L may increase the risk for pancreatic cancer beyond the usual mutations in DNA mismatch repair genes; however studies of additional families with the identical missense alteration are needed to confirm this initial impression.

Introduction

Lynch syndrome (OMIM #s 120435, 609310) is an autosomal dominant cancer predisposition syndrome that underlies about 3–5% of all colorectal cancers.^1-5 It is caused by germline mutations in one of several DNA mismatch repair genes, including MSH2, MLH1, MSH6, and PMS2. The genetic heterogeneity has made diagnostic testing a challenge, such that use of tumor assessment of either DNA mismatch repair deficiency (microsatellite instability) and/or expression of the four DNA mismatch repair gene products has been widely used to screen suspected cases. The Bethesda guidelines⁶ put forth recommendations based upon expert opinion for when tumor testing should be considered. Under that report, the cancers listed as Lynch Syndrome-associated included colorectal, endometrial, stomach, ovarian, pancreas, ureter and renal pelvis, biliary tract and brain (usually glioblastoma as seen in Turcot syndrome) tumors, sebaceous gland adenomas and keratoacanthomas in Muir-Torre syndrome, and carcinoma of the small bowel. Although pancreatic cancer is included in this list, risk usually appears to be only minimally increased, relative to the general population.

Materials and Methods

We have been following a family with a novel MSH2 missense alteration in which pancreatic cancer has been more commonly observed than colon or endometrial cancer (Figure 1). The ancestry is Northern European. None of the affected individuals smoked cigarettes nor had known exposure to unusual environmental agents. There is no family history of melanoma, early onset breast cancer, ovarian cancer, or pancreatitis. Table 1 lists the cancers of all relevant family members, and, where available, the results of tumor immunohistochemical expression of the DNA mismatch repair genes. All testing was done at Mayo Clinic using standard techniques.^7–9 The MSH2 germline change, identified by sequencing, is in exon 6, c.1046C>T, (CCT>CTT), p.Pro349Leu, hereafter called P349L. This variant co-segregates with the development of pancreatic cancer and with the loss of MSH2 expression in tumor tissue in this family. The details of this family have not previously been published; however one aspect of this family’s laboratory results was included in a prior publication that reported on use of BRAF screening as a strategy to simplify HNPCC genetic testing.^{10, 11} No BRAF V600E somatic mutation was found in the MSI-high tumor tested in this family, consistent with this being a Lynch Syndrome family. The kindred is enrolled in an ongoing familial pancreatic cancer registry and an affected individual was studied and found to be negative for CFTR and CDKN2A mutations.

Figure 1.

Pedigree of family with MSH2 P349L missense substitution, showing those diagnosed with pancreatic cancer.

Table 1.

Results of germline and tumor molecular testing in family with three individuals with pancreatic cancer, co-segregating with an MSH2 P349L alteration.

		Tumor immunohistochemistry expression
Relation to proband, cancer, age at diagnosis	MSH2 P349L germline mutation	MSI	MLH1	MSH2	MSH6	PMS2
Paternal grandmother	ND
Breast 73		ND	ND	ND	ND	ND
Father	ND (obligate carrier)
Colon 43			NA	NA	NA	NA
Pancreas 50			Normal	Loss	Loss	normal
Mother	Negative
Chronic lymphocytic leukemia		ND	ND	ND	ND	ND
Proband	Positive
Endometrial 39			Normal	Loss	Loss	ND
Spindle Cell Sarcoma 39			Normal	Normal	Normal	ND
Colon 50		High	Normal	Loss	Loss	normal
Papillary bladder-58			Normal	Normal	Normal	ND
Pancreas 60			Normal	Loss	Loss	normal
Sister	Positive
Pancreas 54			Normal	Loss	Loss	ND

MSI-tumor microsatellite instability; ND-not done.

Results

In silico analyses. 17% of all mutations in MSH2 are missense mutations.¹² The P349L variant is not listed in the Mismatch Repair Genes Variant Database from the Memorial University of Newfoundland (http://www.med.mun.ca/MMRvariants/search_results.aspx) nor is it included in the paper or supplemental materials in the MAPP-MMR database.¹³ It is also not reported in the MMR Gene Unclassified Variants Database (www.mmruv.info), although an MSH2 P349R mutation at the same site is reported by three in silico models, suggesting pathogenicity.¹³ The P349L variant has not been included in functional studies of pathogenicity of MSH2 missense variants.^{14, 15} However, the P349L variant is located in the lever domain of the MSH2 gene, a large domain that connects the ATP binding subunits to the clamp domains to mediate signals between the ATP- and the DNA-binding portions of the protein. Two of three missense substitutions studied functionally in the lever domain manifest lower stability and defective DNA mismatch repair and loss of expression in tumors, which is consistent with studies of homologous positions in yeast MSH2, in which half of missense alterations lead to inefficient expression of the gene.^{14, 15} The Uniprot database, referring to the Domingo report of this family,¹⁰ cites the Pro349Leu variant as possibly pathogenic (http://www.expasy.org/cgi-bin/variant_pages/get-sprot-variant.pl?VAR_043763). A BLOSUM score of -3 is reported in Uniprot.¹⁶ This is supported by in silico analyses using Align-GVGD, with Grantham Variation 0 and the Grantham Deviation 97.78 resulting in a C65 score for MSH2 P349L.^{17, 18} These findings together indicate that the residue is evolutionarily constrained, and predicts that this missense alteration is very likely to have functional consequences.

In order to derive a quantitative classification of pathogenicity, we performed Bayes factor analysis of variant segregation data using methods described previously.¹⁹ Calculations assumed age-specific relative risks for colorectal cancer, endometrial cancer, and other Lynch Syndrome-related cancers (including pancreatic cancer) as estimated in Quehenberger et al. (2005).²⁰ This analysis also provided odds in favor of pathogenicity of 35.7:1, translating to a probability of pathogenicity of 0.97 for this variant. MSH2 P349L would thus be considered class 4 (likely pathogenic), based on the IARC 5 class classification system that is linked to posterior probability estimates.²¹

Discussion

Available data suggest that, in general, penetrance for pancreatic cancer in Lynch Syndrome is low. Prior to discovery of the genetic basis of the Lynch Syndrome, Watson and Lynch (1993) studied 1,424 at-risk persons from 23 large families (with 287 colorectal cancers) who were suspected of having this disorder.²² Six pancreatic cancers were recorded, compared with 4.1 expected, which was not statistically significantly different. In l999, Aarnio et al. had studied 360 gene carriers of 50 families with gene mutations (94% in MLH1 and 6% in MSH2), and found 3 pancreatic cancers, giving a Standardized Incidence Ratio of 4.5, but with a 95% CI of 1.0–14.²³ In 2008, in a study that assessed extracolonic cancer risk among 6,041 members of 261 families with documented mutations in MLH1 (60%) or MSH2 (40%), cancers of the biliary tract, liver and pancreas combined accounted for 1.09% of the cancers in this study, giving a hazard ratio of 1.869 and a cumulative incidence of 4.1% to age 70 years.²⁴ Most recently, risk of pancreatic cancer alone was addressed in 6,342 individuals from 147 families with MMR mutations (37.4% in MLH1, 55.1% in MSH2, and 7.55% in MSH6). A cumulative risk of pancreatic cancer was calculated as 1.31% (95% CI=0.31–2.32%) to age 50, and 3.68% (95% CI=1.45–5.88%) to age 70, which is an 8.6-fold (95% CI=4.7–15.7%) increase compared with the general population.²⁵ In summary, we present a family in which a novel P349L missense substitution in MSH2 that co-segregates with disease in a Lynch Syndrome family appears particularly to be associated with a high risk for pancreatic cancer. All three cancer affected individuals carry the MSH2 P349L missense substitution, and there is loss of expression of MSH2/MSH6 in each of their pancreatic tumors. Together with in silico predictions, these data provide support that this alteration is the causative mutation. It would be of interest to learn of other families with the same missense alteration to determine if predisposition to pancreatic cancer is consistently associated with this MSH2 change.