Discussion

The accelerated development of genetic counselling in cancer during the past few years is due to the feedback and interactive information sharing on genetic studies, clinical management and psychological issues in families with a high risk of cancer. Identification of deleterious variants in such families is essential for accurate assessment of individual risk and, if required, subsequent inclusion into a personalized surveillance programme.

Unfortunately, genetic testing for hereditary cancer frequently fails to identify unambiguous deleterious variants. Erroneous classification of a genetic variant may have a great effect on at-risk familial who undergo genetic testing for risk prediction because it results in incorrect clinical recommendations.

LS is the most common hereditary CRC-predisposing syndrome and accounts for 3% of unselected CRC cases. A significant proportion of DNA variations found in patients suspected of having LS are UVs (32%, 18% and 38% for MLH1, MSH2 and MSH6, respectively) [6]. The pathogenicity of the MLH1 p.Lys618Ala variant remains controversial because of conflicting data [InSiGHT, http://www.insight-group.org] (Figure 5).

Classification of the MLH1 p.Lys618Ala variant according to the InSiGHT database (accessed on 07/2010).

The p.Lys618Ala substitution replaces a charged amino acid with a neutral one, and occurs alongside four charged amino acids that are well conserved in mammals. In silico predictions of the pathogenicity of this variant using the PolyPhen http://genetics.bwh.harvard.edu/pph/ and SIFT http://sift.jcvi.org/ computational program were discordant; the SIFT analysis classified it as a tolerant variant and the PolyPhen analysis classed it as possibly damaging [9]. It has been shown that this variant may reduce the binding ability of MLH1 to PMS2 in HCT116 cells co-transfected with mutated MLH1 and wild-type PMS2 [10]. In contrast, it had no effect on the ability of MLH1 to bind PMS2 in a co-immunoprecipitation assay [9]. Functional analysis using the pCAS ex vivo splicing assay and RNA analysis also demonstrated no effect [11]. Moreover, a significant decrease in MLH1 protein stability has been found for the p.Lys618Ala variant [9].

The results of in silico prediction and functional assays alone are insufficient to determine whether this variant is deleterious or a rare functional polymorphism. For this purpose, it is necessary to integrate indirect evidence with direct genetic evidence involving clinical observations of disease occurrence.

The frequency of variants in unaffected controls is used often to distinguish between neutral and potentially deleterious variants. If the frequency of a variant among a few hundred controls is ≥1%, it is highly unlikely to be a high-risk variant. In such cases, it is still possible that the variant will be associated with a modest risk of the disease [3]. Case-control studies enable quantification of the disease risk associated with the variant. The main disadvantage of such analyses is that a large sample size is required to obtain sufficient power to detect the lower risk level. The sample size required is related inversely to the frequency of the variant in the population. The sample size used in the present study resulted in 80% power to detect an OR of 3.0 (two-sided test; alpha level, 5%). The frequency of the p.Lys618Ala variant in our control series was 2.7% and no significant differences were observed in the sporadic and familial groups, indicating that a high penetrance effect for colorectal carcinogenesis can be excluded. Similar results were reported in case-control studies on Scottish and Danish populations [12,13].

