Verification of the three-step model in assessing the pathogenicity of mismatch repair gene variants

Abstract

In order to assess whether variations affecting DNA mismatch repair (MMR) genes are pathogenic and hence predisposing to Lynch syndrome (LS), a three-step assessment model has been proposed. Where LS is suspected based on family history, STEP1 is dedicated to the identification of the causative MMR gene and the variation within it. Thereafter, in STEP2 of the assessment model, the effect of the variation on the function of the protein is assessed in an in vitro MMR and in silico assays. Where LS cannot be confirmed or ruled out in STEP2, the more specific biochemical laboratory assays such as analyzing the effect of the variation on expression, localization, and interaction of the protein are required in STEP3. Here, we verified the proposed three-step assessment model and its ability to distinguish pathogenic MMR variations from variants of uncertain significance (VUS) by utilizing the clinical as well as the laboratory and in silico data of 37 MLH1, 26 MSH2, and 11 MSH6 variations. The proposed model was shown to be appropriate and proceed logically in assessing the pathogenicity of MMR variations. In fact, for MMR deficient MSH2 and MLH1 variations the first two steps seem to be sufficient as STEP3 provides no imperative information concerning the variant pathogenicity. However, the importance of STEP3 is seen in the assessment of MMR proficient variations showing discrepant in silico results as their pathogenicity cannot be confirmed or ruled out after STEP2. MSH6 variations may be applicable to the model if appropriate selection in terms of ruling out MLH1 and MSH2 variations and MLH1 promoter hypermethylation is ensured prior to the completion of STEP2. In conclusion, taking into consideration the susceptibility gene the three-step model can be utilized in an appropriate and efficient manner to determine the pathogenicity of MMR gene variations.

Figure 1

A three-step decision tree proposed to facilitate the functional assessment of VUS. (Modified from Couch et al. [2008].)

Figure 2

Schematic illustration of (A) MLH1 (B) MSH2, and (C) MSH6 showing the known functional domains, locations of the studied variations, and the amount of steps required for their assessment of pathogenicity. Each required step of the three-step assessment model is represented with a circle. STEP1 is divided to indicate the accordance of the family history with the Amsterdam criteria I/II (lower half: black, AC fulfilled; white, AC not fulfilled; gray, AC fulfilled by some families; diagonal line, data not available) and the MSI and IHC results of the tumor (upper half: black, MSI-H and/or reduced protein expression [IHC]; white, no MSI-H and no problems in protein expression (IHC); gray, contradicting data between several families; diagonal line, data not available). STEP2 is divided to indicate variant protein in vitro MMR activity (lower half: black, deficient; white, proficient) and in silico results (upper half: black, pathogenic effect predicted by SIFT and MAPP-MMR; white, neutral effect predicted by SIFT and MAPP-MMR; gray, discordant or only one result, either pathogenic or nonpathogenic available; diagonal line, data not available). STEP3 assay results are combined to assess VUS pathogenicity (black, results indicating pathogenicity; white, results indicating nonpathogenicity; gray, STEP inconclusive; diagonal line, data not available).