NEW YORK – Investigators from Australia, leading an international research team, have developed a new set of guidelines for interpreting variants in BRCA1 and BRCA2 genes, which were approved by the ClinGen consortium last year.
Published this week in the American Journal of Human Genetics, the team described its efforts to develop these improved criteria under an international consortium called ENIGMA.
ENIGMA was formed in 2015 at the request of ClinGen as an external panel focused on harmonizing qualitative and quantitative classification criteria for BRCA1 and BRCA2 variants from clinical and research settings.
In 2020, the panel sought to become an internal ClinGen Variant Curation Expert Panel to align with the US Food and Drug Administration's recognized processes for contributions to ClinVar. The FDA recognized ClinGen as its first designated public genetic variants repository in 2018, meaning that genetic test developers can access data from the repository through the ClinVar archive and use it to support claims for their diagnostic tests without the need for additional FDA review.
ENIGMA's first step to adapting its multifactorial variant classification methodology for entry into ClinGen was to align its own curations with the increasingly used classification criteria developed by the American College of Medical Genetics and Genomics and the Association of Molecular Pathology.
The crux of this work was converting the ENIGMA group's classification tiers, which grouped multiple sources of information — such as variant type, frequency data, and tumor pathology — to match ACMG/AMP evidence code structure. The team also shared pilot study results demonstrating the value of their more detailed specifications to resolve discordances and uncertainty in variant classification.
In a statement, the study's first author Michael Parsons, a researcher at Australia's QIMR Berghofer Medical Research Institute, said that although gene variant interpretation has become more standardized in recent years, it can still vary between diagnostic laboratories. This means two individuals with identical results might receive completely different treatment from their doctors.
According to the ENIGMA investigators, its refined recommendations for variant classification are also intended, in part, to help clinicians decide whether to test close relatives to aid prevention and early detection.
Amanda Spurdle, senior author of the study and head of the ClinGen Variant Curation Expert Panel for BRCA1 and BRCA2, said in an interview that the results of the group's realignment work is a set of clearer guidelines that the team believes will lead to more definitive variant classifications, fewer variants of unknown significance, and, ultimately, improvements in the clinical management of patients.
Among other findings, the Enigma investigators determined in their analyses that 13 of the existing ACMG/AMP code descriptions were non-applicable to BRCA1 and BRCA2 or overlapped with other criteria.
Spurdle said that the team also found that some methods developed for predicting the effect of gene variants in general did not perform well for BRCA1 and BRCA2 specifically.
Using statistical methods including logistic regression analysis, heterogeneity analysis, and likelihood ratio estimation, the team created rules to calibrate the strength of evidence for different data types used to inform variant classification.
The group derived what they called likelihood ratio estimates toward or against pathogenicity for a given evidence type using defined reference sets of benign and pathogenic variants. The LR estimates were then used to assign weights for or against pathogenicity following recommendations arising from Bayesian modeling of the ACMG/AMP guidelines.
Once they had solidified an initial harmonized classification methodology that allowed the ACMG/AMP evidence codes to be applied to more complex variant pathogenicity evidence, investigators performed a pilot analysis of 40 variants. The preexisting ClinVar classification for the pilot variants was 13 conflicting or uncertain, 11 pathogenic, three likely pathogenic, one likely benign, and 12 benign.
A group of volunteer curators then used their harmonized evidence coding strategy to classify the 40 variants. Based on initial results, the team then refined their specifications. And in a final curation effort, volunteers were able to resolve six of eight uncertain calls to a single classification. Five of six likely pathogenic variants could also be resolved to pathogenic, and all three likely benign were converted to benign.
Compared to the original ClinVar assessments, classification was resolved for 11 of 13 uncertain variants, the team reported.
According to the authors, the results have relevance beyond just BRCA. The ENIGMA team's findings have already led to other variant curation panels adopting aspects of its approach to refining classification evidence codes. Spurdle highlighted recent work on TP53, and the study authors added that some adaptations have been incorporated into a draft iteration of the next version of the ClinGen-promoted classification guidelines for application to any Mendelian disease.
Spurdle said that her team is now working on trying to reclassify all the conflicting or uncertain BRCA variants that they can. "We can see we are going to resolve a whole bunch of them, but we're now calling for more data for the ones that are remaining."
In other non-cancer settings like rare disease or pediatric genetics, some of the more notable issues have arisen when a variant of unknown significance is upgraded without patients being informed. For example, the 2016 lawsuit by Amy Williams against Quest Diagnostics arose from a variant reclassification affecting her epileptic son.
Spurdle said that her observation in cancer has been that clinicians struggle more with downgrades than upgrades. Her lab has been conducting its own research on the impact of variant reclassification and spoke with a clinician who had seen patients make clinical decisions they shouldn't have based on VUS findings that were later downgraded to likely benign.
As a part of this research, Spurdle's lab has been asking the clinicians and labs what they do when faced with a reclassification. In Australia, as in the US, most small labs don't have the bandwidth to proactively recontact clinicians, and clinicians largely don't have the bandwidth to chase down reclassifications from labs, she said.
Group members are also working on ways to semi-automate their evidence curation and coding process, which Spurdle said will likely put them ahead of what gets an official stamp from the FDA through the ClinGen system, which requires that each variant classification be reviewed by three independent individuals. One immediate approach would be algorithmic code assignment based on frequency and computational information in the BRCA Exchange portal.
The team plans to also make their aligned codes publicly available through the BRCA Exchange portal while the longer-term FDA-aligned process of ClinGen curation continues. With tens of thousands of variants in the database, the task is a considerable one, Spurdle said.