Skip to main content
Premium Trial:

Request an Annual Quote

At ACMG, Researchers Report Data Re-Analysis, Matchmaking Boosts Solved Exome Cases

Tampa Convention Center, site of the 2016 ACMG Meeting

TAMPA (GenomeWeb) – Many clinical exome sequencing studies of hereditary diseases have reported diagnostic yields on the order of 25 percent, meaning that about three quarters of patients are left without a molecular diagnosis. At the American College of Medical Genetics and Genomics annual meeting today, researchers from Baylor College of Medicine and Johns Hopkins University School of Medicine reported on strategies for boosting the diagnostic rate by reanalyzing exome data and using matchmaking tools to find similar cases.

According to Baylor's Pengfei Liu, ACMG recognized the potential utility of data re-analysis several years ago, when the organization recommended that raw genomic data be stored for at least two years, and variant files for as long as possible. At that time, however, there was little evidence to support the idea that reanalyzing data resulted in additional diagnoses.

To test this, Lui and his colleagues looked at about 5,700 patients who underwent clinical exome sequencing at Baylor between 2012 and 2015, and whose data was reanalyzed at a later stage. Initially, the exome test solved 24 percent of cases, but another 5 percent of patients received a diagnosis after their data were reanalyzed. Of the new diagnoses, 62 percent resulted from disease genes that had not been discovered at the initial time of testing; 11 percent derived from the analysis of copy number variants; 10 percent came from studies of parental data, either to confirm that a patient mutation was de novo or that it occurred in trans; and the remainder from clinical correlations, updated variant interpretation, or Sanger sequencing of underrepresented regions.

To assess whether the timeline for the re-analysis made a difference, the researchers also conducted separate analyses for an early cohort of 250 patients – results that were published in 2013 in the New England Journal of Medicine – and a later cohort of 2,000 cases, results from which were published in 2014 in the Journal of the American Medical Association.

Re-analysis increased the diagnostic rate for these 250 cases to 36 percent from 25 percent, but only increase the rate to 30 percent from 25 percent for the 2,000 cases. While most of the new diagnoses in both groups resulted from newly discovered disease genes, those played a greater role in the earlier group.

Liu said that the rate at which new disease genes are discovered has not plateaued yet, and he expects many more diagnoses to emerge from data re-analyses, especially for older cases.

In the later cohort, copy number variants were responsible for a larger percentage of new diagnoses than in the early group, which might reflect the fact that more doctors order clinical exomes as a first test these days, rather than first ruling out pathogenic CNVs by an array-based test. While this is not an ideal scenario, Liu said, testing labs should adapt and attempt to call CNVs from exome data.

As the number of patients undergoing clinical exome testing increases, he said, labs are likely to spend more time on re-analysis. However, this service is currently not reimbursable through health insurance, which only pays for the initial test, so Baylor and most other labs currently provide re-analysis to patients free of charge. Going forward, it will be important to formulate guidelines for performing exome re-analyses in a systematic way, he said, and to establish mechanisms for reimbursement.

Johns Hopkins' Nara Sobreira reported on her team’s strategy for increasing the "solved" rate for exome sequencing through re-analysis of variants with improved bioinformatics tools and by finding patient matches that can help establish a candidate variant as causative.

The Hopkins group — which has been using the PhenoDB database to store and analyze clinical data from exome tests — recently re-analyzed data from 1,063 samples from 477 families. Of these, 97 families had received a diagnosis from the initial analysis. In those samples, they looked for rare functional variants in known imprinted genes, genes in pseudoautosomal regions, genes that escape X inactivation, and genes on chromosome Y. They found that the genes in the pseudoautosomal regions were actually not covered by the baits in their exome test, so they could not assess them. For the others, they detected a number of variants, which they are currently evaluating further for potential causality. They also scoured the exome data for rare homozygous deletions, which they detected in several patients, and for some of these, the deletions could explain their phenotype.

In addition to re-analyzing exome data, the researchers have been working on establishing causality for novel candidate disease genes through patient matches. For this, the team has been using the GeneMatcher website, which allows them to find other clinicians and researchers around the world who have patients, or animal models, with mutations in the same genes as their own patients. Through an API developed by the Matchmaker Exchange project, GeneMatcher submitters can also query the PhenomeCentral and Decipher databases. As of March, more than 4,000 genes had been submitted to GeneMatcher from more than 1,300 submitters in 48 countries, and 1,900 matches had been made, Sobreira reported.

Her team has so far submitted data from 104 families, involving 280 genes, and has had 314 matches so far, involving 113 genes. Several cases have been successes, meaning the researchers could establish that a candidate gene is indeed disease causing, and several others are pending, both from Hopkins and from other groups. The total number of solved cases tracing their success to GeneMatcher is currently unknown, Sobreira said, but the organizers are planning to survey submitters about their success rate in the near future.