Use of “Coldspot” Regions in Variant Classification

Determining the clinical significance of genetic variants is a process that involves gathering evidence from multiple sources, assessing the relative strength of each piece of evidence, and then formally assigning a level of certainty expressing the likelihood of being disease-causing based on the combination of all available data. This classification of genetic variants may be utilized to identify or confirm the cause of disease or to help guide treatment decisions. Given the importance of variant classification in medical management, in 2015 the American College of Medical Genetics and Genomics (ACMG) and the Association for Molecular Pathology (AMP) published a landmark guideline for variant classification that provides an evidence-based framework for classifying variants (1). This guideline defined 28 criteria or evidence types, each with an assigned direction (favoring benign or favoring pathogenic) and relative weight. A final classification is derived by combining these criteria according to a rubric that places variants for Mendelian disorders on a significance gradient ranging from “pathogenic” (P) to “benign” (B), with intermediate categories being “likely pathogenic” (LP), “variant of uncertain significance” (VUS), and “likely benign” (LB). Because these guidelines were developed to be broadly applicable across genes, inheritance patterns, and diseases, instructions regarding the applicability of each criterion for any given gene and disease were not provided. Instead, the authors noted “that those working in specific disease areas should continue to develop more focused guidance regarding the classification of variants in specific genes given that the applicability and weight assigned to certain criteria may vary by gene and disease” (1).

In line with that recommendation, in their recent publication, Dines et al. proposed that variant classification rules for these genes should be amended to incorporate data supporting regions without essential function that tolerate variation and assign weight toward a benign classification (2,). By utilizing missense variant classifications in BRCA1 and BRCA2 from ClinVar, the authors calculated the odds of pathogenicity for missense variants in regions throughout each gene and then compared with Bayesian modeling of the ACMG/AMP guidelines to determine the relative weight of these data (3). Although the occurrence of a missense variant in a critical or well-established functional domain that does not tolerate variation is considered a moderate piece of evidence toward pathogenicity (ACMG/AMP code PM1), the inverse—the concept of a missense variant in a region without essential function—was not included in the ACMG/AMP guideline.

The study by Dines et al. adequately highlights this gap and rectifies the issue by suggesting that “coldspot” evidence could be used to support a benign classification under the ACMG/AMP BP4 criterion, currently defined as “multiple lines of computational evidence suggest no impact on gene or gene product.” Much effort on refining the pathogenic criteria in the ACMG/AMP guidelines is directed toward better differentiating P and LP variants from VUSs. Although equally critical in reducing the number of VUSs, refinement of the benign criteria is traditionally overlooked, making the work by Dines et al. highly impactful.

However, in our opinion, further thought needs to be given to the strength assigned to this evidence to align the data well with the existing framework. Dines et al. show that the relative strength of evidence supporting a coldspot varies (ranging from strong for variants in exon 11 in BRCA1 and exons 10 and 11 in BRCA2 and moderate for missense variants in the coiled-coil domain in BRCA1) and propose that missense variants in the regions that reach benign strong odds be classified as LB.

However, the ACMG/AMP framework requires at least 2 pieces of benign evidence to reach an LB classification: (a) 1 benign strong piece of evidence and 1 benign supporting piece or (b) ≥2 benign supporting pieces of evidence. A laboratory assigning strong weight to a missense variant in the coldspot regions, as defined by the authors, would therefore reach a classification of only LB when additional benign evidence, aside from occurrence in a coldspot, is available. Alternatively, the ACMG/AMP rubric for combining rules could be modified either globally or for this specific set of genes.

Refinement of the ACMG/AMP criteria regarding BRCA2 and BRCA1 is especially important because these genes have the first (32 777 submissions) and third (21 928 submissions) highest numbers of submissions in ClinVar, respectively (ClinVar data accessed March 7, 2020), testimony to a high volume of testing across laboratories. Interestingly, BRCA2 and BRCA1 also have the second and third highest numbers of VUS vs LB/B conflicts in ClinVar, respectively (ClinVar data accessed through ClinVar Miner March 7, 2020; https://clinvarminer.genetics.utah.edu/). In addition, of the 573 ClinVar variants with VUS vs LB/B conflicts in BRCA1, 217 are missense variants located in the defined BRCA1 coldspots. For BRCA2, of the 912 ClinVar variants with VUS vs LB/B conflicts, 346 are missense variants located in the defined BRCA2 coldspot. This type of conflict is of particular interest regarding coldspot data because these conflicts suggest that some evidence already shows a benign interpretation and that incorporation of coldspot data could lead to reclassification of the VUS interpretations.

Incorporation of coldspot data may result in more likely benign classifications; however, laboratories should be cautious about using this line of evidence blindly for all missense variants in a defined coldspot region without carefully ruling out the existence of evidence supporting pathogenicity. First, a missense variant could be disease-causing given the underlying nucleotide substitution affecting splicing as opposed to the amino acid substitution at the protein level. The coldspot publication notes 6 P/LP missense variants in BRCA1 that affect splicing, all but 1 of which are nucleotide substitutions of the last nucleotide in an exon. These data suggest that laboratories should consult splicing prediction data before applying coldspot data to missense variants and cautiously not apply coldspot evidence to missense variants affecting the last nucleotide of an exon. If coldspot evidence is included in future iterations of ACMG/AMP guidelines, this line of evidence should include the caveat “Beware of changes that impact splicing rather than at the amino acid/protein level,” similar to the caveat provided for pathogenic criteria implicating a codon due to other pathogenic missense variants (ACMG/AMP PS1 and PM5). Second, although a large region may appear as a coldspot (the suggested coldspot regions in BRCA1 and BRCA2 are 1142 and 2015 residues in length, respectively), there may be selected residues within a coldspot that could be critical. Without taking these considerations into account, a P/LP missense variant in a coldspot could initially be misclassified as LB.

Overall, such a coldspot analysis requires large cohort studies and/or the presence of many observed and classified variants in public databases. Because this volume of data will not be available for all genes, alternative approaches for identifying such coldspot regions could be considered. Approaches may include comparison of expected to observed variation in unselected or healthy populations. Regional constraint scores are available for the Exome Aggregation Consortium (ExAC) data set for genes that have ≥2 regions with significantly different levels of missense constraint—that is, one region showing significant depletion of expected missense variation and another region showing significant excess of expected missense variation (4,). However, using this population data set has important caveats because missense constraint is expected only for disease genes that affect reproductive fitness (4). In addition, this approach considers only the number of observed and expected variations in the gene and does not account for the location and frequency of disease-causing variants within the gene. Thus, a combined approach that incorporates knowledge gained from large population data sets regarding tolerance of variation and variant classifications from data sets such as ClinVar may be the best option for identifying both candidate coldspots and hotspots in disease genes.

Approaches similar to that of Dines et al. for other genes would have a clinically significant impact on variant classification because the majority of variants identified in Mendelian genes are rare and of uncertain significance (5,); however, for most disease areas, a consortium approach will be needed because of the overall lower volume of clinically classified variants. In 2013, the NIH-funded Clinical Genome Resource (ClinGen) consortium was formed to develop standards and processes for evaluating genes and genomic variation to enhance clinical validity and utility (5). A core goal is expert interpretation of variants, and a growing number of variant curation expert panels have been formed to focus on specific genes or groups of genes. These expert panels are tasked with providing specifications to the ACMG/AMP guidelines for genes–diseases dyads and interpreting variants according to these rules (https://clinicalgenome.org/working-groups/clinical-domain/). The approach for defining coldspot regions piloted by Dines et al. is highly synergistic with the goals of ClinGen’s variant curation expert panels and is providing a valuable basis for further refinement and implementation across other genes. Identification of coldspots and incorporation of these data into variant classification could significantly reduce the number of VUSs returned to patients.

Nonstandard Abbreviations

ACMG, American College of Medical Genetics and Genomics; AMP, Association for Molecular Pathology; P, pathogenic; B, benign; LP, likely pathogenic; VUS, variant of uncertain significance; LB, likely benign; ExAC, Exome Aggregation Consortium; ClinGen, Clinical Genome Resource.

Human Genes: BRCA1, BRCA1 DNA repair associated; BRCA2, BRCA2 DNA repair associated.

Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved.

Authors’ Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:

Employment or Leadership: B. Funke, Sema 4, Clinical Genome Resource.

Consultant or Advisory Role: B. Funke, Genomics PLC.

Stock Ownership: B. Funke, Sema 4.

Honoraria: None declared.

Research Funding: None declared.

Expert Testimony: None declared.

Patents: None declared.

References