Optimising clinical care through CDH1-specific germline variant curation: improvement of clinical assertions and updated curation guidelines

Abstract

Background: Germline pathogenic variants in CDH1 are associated with increased risk of diffuse gastric cancer and lobular breast cancer. Risk reduction strategies include consideration of prophylactic surgery, thereby making accurate interpretation of germline CDH1 variants critical for physicians deciding on these procedures. The Clinical Genome Resource (ClinGen) CDH1 Variant Curation Expert Panel (VCEP) developed specifications for CDH1 variant curation with a goal to resolve variants of uncertain significance (VUS) and with ClinVar conflicting interpretations and continues to update these specifications.

Methods: CDH1 variant classification specifications were modified based on updated genetic testing clinical criteria, new recommendations from ClinGen and expert knowledge from ongoing CDH1 variant curations. The CDH1 VCEP reviewed 273 variants using updated CDH1 specifications and incorporated published and unpublished data provided by diagnostic laboratories.

Results: Updated CDH1-specific interpretation guidelines include 11 major modifications since the initial specifications from 2018. Using the refined guidelines, 97% (36 of 37) of variants with ClinVar conflicting interpretations were resolved to benign, likely benign, likely pathogenic or pathogenic, and 35% (15 of 43) of VUS were resolved to benign or likely benign. Overall, 88% (239 of 273) of curated variants had non-VUS classifications. To date, variants classified as pathogenic are either nonsense, frameshift, splicing, or affecting the translation initiation codon, and the only missense variants classified as pathogenic or likely pathogenic have been shown to affect splicing.

Conclusions: The development and evolution of CDH1-specific criteria by the expert panel resulted in decreased uncertain and conflicting interpretations of variants in this clinically actionable gene, which can ultimately lead to more effective clinical management recommendations.

Keywords: Gastrointestinal Diseases; Genetic Predisposition to Disease; Genetic Variation; Genetics, Medical.

Figure 1:. PVS1 decision tree for CDH1.

A gene-specific decision tree was developed to clarify rules for application of PVS1 (adapted from Tayoun et al 2018). Initiation codon variants and nonsense or frameshift variants predicted to undergo NMD should justify the PVS1 criterion at very strong strength. For truncations in the NMD-resistant zone, PVS1 should be applied at strong or moderate strength depending on whether the location of the variant is upstream or downstream of the most 3’ well-characterized pathogenic variant [NM_004360.5:c.2506G>T (p.Glu836Ter)]. Refer to Supplementary Table S5 for key splice site information including the recommended strength of the PVS1 criterion for each donor and acceptor site.

Figure 2:. Comparison of CDH1 VCEP variant classifications to previous ClinVar assertions.

Using the CDH1 specifications to the ACMG/AMP guidelines, 33% (90/273) variant classifications were altered and the change depicted here. Our data are limited in that ClinVar assertions for each variant were not extracted at the time of the curation and final expert interpretation of each CDH1 variant. Therefore, the number of improved CDH1 variant curations may be higher than reported therein. Despite this limitation in our analysis, CDH1 expert curations still include resolution of 97% (36/37) variants with conflicting interpretations to a non-VUS classification. Eighteen variants with conflicting interpretations were resolved to benign, 16 to likely benign, one to VUS, one to likely pathogenic, and one to pathogenic. Fourteen VUS were resolved to likely benign and 1 to benign. Ten variants with a combination of benign/likely benign assertions and four variants like a likely benign assertion were resolved to benign, and 14 variants with a combination of pathogenic/likely pathogenic assertions and ten variants with a likely pathogenic assertion were resolved to pathogenic. Overall, 52 improvements resulted in resolution of a clinically significant conflict and 38 improvements resulted in resolution of a confidence conflict.

Figure 3:. Summary of 273 CDH1 VCEP variant interpretations and frequency of application of each evidence code.

(a) The CDH1 VCEP has curated and deposited 273 variant interpretations to the ClinVar database: 51 benign, 38 likely benign, 34 variants of uncertain significance, 28 likely pathogenic, and 122 pathogenic. (b) Frequency of application of each evidence code and its accompanying strength. The most frequently applied non-variant specific evidence codes applied were PS4, PM2, PP1, BS2, and BP2, demonstrating the importance of collating published and unpublished lab data provided by diagnostic laboratories for variant classification.