On this basis, we could consider two (different clinico-pathological) subsets of early onset CRC: the greatest percentage represented by left sided CRC without important family history (no Amsterdam Criteria fulfilled) and the lowest percentage represented by LS related CRC, with Amsterdam II criteria fulfilled and

typical features of the syndrome. Our major concern was whether we should have performed a molecular screening in both subsets of early onset CRC. In order to address this issue and considering that all Lynch syndrome associated CRC display MSI-H [4], we performed a logistic regression model to identify features predictive of MSI-H. The regression tree revealed, indeed, that using the combination of the two features “No Amsterdam Criteria” and “left sided BAY 80-6946 mw CRC” to BAY 11-7082 concentration exclude MSI-H, has an accuracy of 89.7% (Figure 2). Interestingly, in the group with no family history, we identified OTX015 molecular weight 3 MSI-H cases. The germline mutation analysis did not confirm LS diagnosis in any of the patients as MMR deleterious mutations were not found. Despite this, we observed

an acquired MLH1 promoter hypermethylation in one case, with loss of PMS2 expression at IHC. Lack of MLH1 expression affects PMS2 protein stability and explains its loss at IHC, thus we classified this case as “sporadic colorectal cancer” [41]. Moreover, we identified a single nucleotide polymorphism (c.116G > A; p.Gly39Glu; rs1042821) in the MSH6 gene, in two cases in which IHC detected a normal expression of the corresponding protein. This polymorphism (MSH6 G39E) encodes a non-conservative amino acid change where it is unknown whether the variant affects protein function. MSH6 G39E is reported, in one study to confer Farnesyltransferase a slight risk of CRC in males (OR 1.27; 95% CI 1.04 to 1.54), higher in MSI-H than MSS (OR 1.30; CI 95%) [38]. Other authors reported in

MSH6 G39E homozygous patients an increased risk of rectal cancer only [42]. The observed association should be interpreted with caution, since no association was found between the MSH6 variant and the overall CRC, probably due to the small number of rectal cases included in the study. The secondary aim of the present study was to compare the diagnostic accuracy of IHC and MSI analysis in early onset CRC to select the best technique to start with in the suspected LS. We observed that MSI analysis had a higher diagnostic accuracy (95.7% vs 83.8%) sensitivity (100% vs 75%), specificity (94.8% vs 85.6%) and AUC (0.97 vs 0.80) than IHC (Figure 1). In fact, had we not used MSI analysis, we could have missed four LS cases not detected by IHC in the group with Amsterdam II Criteria. Even in the early-onset group, IHC was misleading as it showed a lack of expression of MMR genes in three MSS patients in which the germline mutation analysis did not reveal any deleterious mutation.