Background and aims: Hereditary non-polyposis colorectal cancer (HNPCC) may be caused by mutations in the mismatch repair (MMR) genes MLH1, MSH2 or MSH6. Family history (Amsterdam criteria) has traditionally been used to select patients for mutation testing. It has been demonstrated that germline mutations in the MMR genes are associated with lack of the corresponding gene product as assessed with immunohistochemistry (IHC) in tumour specimens. The aim of the study was to assess the value of the Amsterdam criteria II and IHC in predicting germline mutations.
Methods: Fifty-six families that were previously tested for MLH1, MSH2 and MSH6 mutations were selected for this study. All pedigrees were extended and verified and the families were scored according to the original (I) and the revised Amsterdam criteria (II). The probabilities for MLH1 and MSH2 mutations were calculated by logistic regression. In addition, all available tumour material from indexed family members was examined by IHC for the presence of the three gene products.
Results: Three out of seven (39%) families where the mutation could be identified complied with the Amsterdam criteria I, while all seven (100%) met the Amsterdam criteria II. All families carrying a MLH1 or MSH2 mutation had > 15% calculated probability of finding a mutation. Tumours from all seven mutation carriers lacked the immunohistochemical expression of the corresponding MMR gene.
Conclusion: The results indicate that the Amsterdam criteria II in combination with immunohistochemistry of the mismatch repair proteins in tumours may be a cost-effective approach to select families for mutation analysis.