Skip to main content
Premium Trial:

Request an Annual Quote

Incidental Findings Challenges Point to Need for Better Databases, Population Specific References


NEW YORK (GenomeWeb) – Since the American College of Medical Genetics first recommended that clinical laboratories search for and report on pathogenic findings in around 56 genes associated with serious genetic conditions, there has been much debate on the impact of doing so. Critics worried that it would be too much of a burden on laboratories, that patients and physicians alike would not know what to make of the findings, and that it would prove too challenging to figure out the meaning of potentially disease-causing variants in individuals who were so far symptom free.

Now, a study published recently in Genome Research, looked for incidental findings in 112 disease-causing genes in the exomes of 6,503 individuals that were sequenced as part of the National Heart Lung and Blood Institute's Exome Sequencing Project in an attempt to better understand what percentage of individuals would likely have incidental findings and what challenges may arise.

The team, led by Gail Jarvik, who heads the division of medical genetics at the University of Washington, found that although only between 1.1 percent and 2.0 percent of individuals had an actionable highly penetrant pathogenic variant — and only 0.5 to 0.7 percent when just looking at the ACMG genes — there were significant challenges in classifying many of the variants and a lack of data on variants from individuals of African descent.

"There is significant mismatch among genetics professionals in how they classify variants," Jarvik told GenomeWeb. Differences were seen between people in the same laboratory as well as across different labs, she said, in terms of whether a variant was called as possibly pathogenic, unknown significance, or likely benign.

In addition, "research data and clinical data is lacking in people of diverse ancestries," she added.

The UW group is not the only one to tackle the challenges of returning incidental findings from genomic sequencing. Earlier this month, a team from Baylor College of Medicine and Johns Hopkins University published a study in Genetics in Medicine that looked at data from 200 individuals, and the American College of Medical Genetics and Genomics in November updated its recommendations for returning such findings.

The UW team analyzed exome data from 4,300 European- and 2,203 African-ancestry participants in NHLBI's Exome Sequencing Project and looked for variants in 112 medically actionable genes.

First, the researchers compared that data to the Human Gene Mutation Database and found 615 variants. Of those, 116, or 18.9 percent, had a frequency higher than what would be expected for the disorder, and on review none were classified as pathogenic or likely pathogenic.

There were 559 variants in genes associated with dominant disorders, 261 of which were singletons. Most variants were seen more than once. The singleton variants represented 51 of the 64 pathogenic or likely pathogenic variants associated with dominant disorders, a frequency that represented a "significant excess of rare pathogenic variants relative to those variants observed more than once," the authors wrote.

Using stringent criteria to call variants pathogenic or likely pathogenic, the researchers found pathogenic variants in 30 of 4,300 European-ancestry individuals and six of 2,200 African-ancestry individuals. They also found 38 unique variants in 23 genes that were annotated as disease causing in HGMD, and in total found likely pathogenic variants in 65 individuals, 52 of European ancestry and 13 of African ancestry.

Jarvik said that because there is much less data on individuals not from European descent, calling variants as pathogenic or not is more difficult. "One of the most helpful things in determining if a variant is disease causing or not is how commonly you see that variant," she said. It is not that a specific variant will be disease-causing in one population but benign in another, she said, but specific disease-causing variants are likely to be more prevalent in specific populations.

"There are variants that occur in one group, but not another," she said. "Or variants that we rarely see in one population may turn out to be common in another population. That information is helpful, because it means it probably doesn't cause disease," she said.

To see how different individuals and different laboratories classified variants, the researchers selected six variants within groups of varying pathogenicity. Five research and clinical laboratories that are part of the Clinical Sequencing Exploratory Research consortium then blindly classified those. Agreement between all labs was only obtained for the one truncating variant, which was called as pathogenic.

For another variant, the classifications ranged from pathogenic to variant of unknown significance. Different laboratories appeared to have different thresholds for whether or not a variant was called as pathogenic, pointing to the need for standards to make such classifications more consistent.

"Clearly we have a ways to go before we agree on what a variant means," Jarvik said.

Jarvik added that for the most part, variants of unknown significance turn out to be benign, which is something that laboratories should take into account and physicians need to be aware of when getting these results. She worries that a VUS result in, for instance, a breast cancer risk gene, may cause undue worry to physicians and patients and the patient may want to act on it unnecessarily.

There needs to be better education that variants of unknown significance usually turn out to not cause disease, Jarvik said.

In addition, she said, there are not yet agreed upon standards for how to weigh different types of evidence, such as functional data, to determine whether a variant is disease causing or not.

Jarvik said that the results of this study will be used to come up with an estimate of how often clinical laboratories should expect to come across incidental findings, which would be helpful for figuring out how to budget follow-up studies, she said.

The CSER laboratories are also planning to do another round where the labs each blindly analyze the same results to see if there is more of a consensus. "It's a very useful exercise to see where we disagree and why and where guidelines could be clearer," she said.

Jarvik said that she would also like to do a similar study looking at individuals from Asian ancestry, which is the least represented population in genomic databases.

In the future, she said these kinds of studies would help guide clinical implementation of genomic data and precision medicine.