Response to: Design of a Core Classification Process for DNA Mismatch Repair Variations of A Priori Unknown Functional Significance


Correspondence to: Lene Juel Rasmussen, Center for Healthy Aging, University of Copenhagen, Blegdamsvej 3B, 2200 Copenhagen, Denmark. E-mail:

We thank Grandval et al. (2013) for their comment on our review “Pathological Assessment of Mismatch Repair Gene Variants in Lynch Syndrome: Past, Present, and Future” [Rasmussen et al., 2012]. In our review, we propose the use of a structured decision tree to evaluate the pathogenicity of variants of uncertain significance (VUS) in DNA mismatch repair genes, as identified in individuals suspected of the common cancer predisposition Lynch syndrome. We propose to initially use “indirect” patient and family-derived data that include segregation, phenotypic analysis, immunohistochemistry (IHC), and microsatellite instability (MSI; step 1). This is followed by the analysis of the VUS using in silico algorithms and an assay that measures functionality of mismatch repair of the VUS, as a “direct” approach to measure functionality of the variant (step 2).

In the comment, Grandval et al. (2013] investigated pathogenicity of a large number of VUS from French patients. This represents a commendable effort by the authors, enabling to define the likelihood of pathogenicity of a significant number of variants. The authors conclude that, although the proposed decision tree is simple and friendly to use, step 1 supplemented with in silico analysis and database review suffices to categorize most variants in a five-stage classification system, ranging from “definitely pathogenic” to “not pathogenic,” as proposed by Plon et al. (2008). Nevertheless, this classical approach for diagnosis, using only “indirect” data, may have a number of potential caveats. For instance, in many countries, first-degree relatives of a carrier of a defined VUS are not analyzed for presence of the VUS, which significantly complicates segregation analysis. Moreover, recent data from the same authors demonstrate a disease penetrance, even of proven pathogenic DNA mismatch repair gene mutations, of less than 50% [Bonadona et al., 2011], which further complicates the use of segregation analysis for diagnostic assessment. Also other “indirect” data used for assessment such as IHC and MSI, or in silico analysis, may yield false positive or false negative results [Lagerstedt Robinson et al., 2007]. It is for these reasons that the approach proposed by Grandval et al. (2013) for diagnosis of VUS requires the availability of a comprehensive set of clinical and pathological data for the assessment of each VUS. Moreover, the “indirect” nature of the data requires extreme scrutiny in their interpretation and integration and, even when all of these conditions are met, false positive or false negative results cannot be excluded.

There is an absolute correlation between biochemical defects in DNA mismatch repair and cancer predisposition in Lynch syndrome. Importantly, the genetics and biochemistry of DNA mismatch repair have been studied in detail for four decades and, consequently, Lynch syndrome is the only major cancer predisposition where the underlying biochemical pathway has been unraveled in great detail [Hsieh and Yamane, 2008]. Functional assays for DNA mismatch repair have been available for quite some time but, unfortunately, have been too complicated to independently validate, let alone to implement in the clinical setting. Furthermore, results published from such assays were frequently discordant, as Grandval et al. (2013), rightfully state. Possibly, the development of simple and cost-effective cell-free activity assays [Drost et al., 2012] will relieve these impediments in the future, and current efforts are aimed to achieve their validation and subsequent implementation in the clinic.

We wish to emphasize that the use of biochemical assays as a standard ingredient in the diagnosis of VUS in DNA mismatch repair genes, even when a complete set of “indirect” data is available, provides a unique opportunity to improve the predictive values of their diagnostic assessment, as also acknowledged by Grandval et al. (2013). Such a procedure should integrate “indirect” data, when available, with information from databases, with in silico analysis and with functional analysis of the VUS, and result in the classification of the VUS as proposed by Plon et al. (2008). Until that time has come, the approach taken by Grandval et al. (2013) certainly is the best alternative.


Disclosure statement: The authors declare no conflict of interest.