Skip to main content
Premium Trial:

Request an Annual Quote

Population Genetic Screening Would Reveal Patients With Underdiagnosed Disorders, Study Suggests

Population Genomics People DNA.jpeg

NEW YORK – A population-based genomic screening study suggests that many genetic disorders may be underdiagnosed and that exome sequencing could be used to identify carriers of disease-associated variants.

In the study, published on Friday in Cell Reports Medicine, researchers from the Icahn School of Medicine at Mount Sinai assessed the diagnostic utility of exome sequencing in ostensibly healthy individuals.

Adam Buchanan, chair of the genomic health department at Geisinger College of Health Sciences who was not part of the study, said in an email that understanding the risks associated with pathogenic or likely pathogenic (P/LP) variants in an unselected population is one of the key programmatic and clinical challenges in population genetic screening. This, he explained, is largely because uncertainty about a P/LP variant's penetrance makes it difficult to determine which genes to screen for and how to counsel patients with a variant in those genes about risk and risk management.

"Existing lists of actionable genes," he said, such as those from the US Centers for Disease Control and Prevention and the American College of Medical Genetics and Genomics, "are not intended to determine which genes to screen for in this context, so a study like this can provide empiric evidence to support a list of genes for population screening."

For this study, the investigators sought to identify gene variants associated with a higher risk of disease in an ostensibly healthy cohort, mimicking how screening would occur in a general population. In particular, they used UK Biobank (UKB) sequence and health records data to select genes for which having a pathogenic variant was also associated with having a diagnostic code in the EHR. Then they identified people in Mount Sinai's BioMe biobank with a pathogenic variant from that UKB-derived gene list, some of whom exhibited symptoms of the related disease.

"One of the overarching questions in population genomic screening is which genes to screen for," Buchanan said. "Right now, there's not a great method for answering that question. Groups mostly rely on expert opinion that wasn't really intended for that purpose. Here, the authors are trying to make this question of gene selection a data-driven one."

Buchanan said that this strategy demonstrates the novelty of the study, which otherwise showed findings consistent with similar programs that also matched genomic data with EHR data to identify previously undiagnosed individuals with genetic diseases. "As such, it's a useful replication of this earlier work," he said.

The investigators focused their study on a set of nine disorders, including breast, colon, prostate, and uterine cancer as well as amyotrophic lateral sclerosis, cardiomyopathy, hypercholesterolemia, and the rare eye disease retinitis pigmentosa.

They screened more than 29,000 individuals from the BioMe cohort for P/LP or loss-of-function (LoF) variants in genes related to these conditions and identified 303 P/LP or LoF variants in 54 genes within the exomes of 614 individuals.

Most of these individuals lacked a corresponding clinical diagnosis, however, leading the investigators to conduct a more focused evaluation of those patients' EHRs. This revealed evidence of untreated symptomatic disease in 75 cases.

Two undiagnosed individuals, for instance, had variants in the PKP2 gene. The finding triggered the researchers to conduct a targeted evaluation of cardiac phenotypes in their EHRs, which, in turn, supported a diagnosis of cardiomyopathy. Similarly, the identification of LDLR variants in three other undiagnosed individuals led to an assessment of their cardiovascular and lipid EHR profile that indicated a diagnosis of familial hypercholesterolemia.

"I think it's effective as a proof of concept and pushing the field to find empiric ways to select genes for population screening, but has some limitations," Buchanan said.

For example, he said, relying on EHR data at the time of the study without bringing patients in for additional diagnostic evaluation likely underestimates the rate of undiagnosed, yet symptomatic patients. Future analyses that disclose the genetic result and then bring patients in for targeted phenotyping that includes procedures not already in the EHR, he said, could refine penetrance estimates.

He also noted that the UKB is a relatively young, healthy, and not racially diverse population, which raises questions about how to design an equitable population genomic screening program that provides medically actionable genetic information across diverse groups.

According to Buchanan, the method used in the study seems most useful as a design feature for population genomic screening programs and informing professional groups' assessments of which genes to include in such efforts, with the caveat that the concerns about cohort diversity and equity in variant interpretation would need to be addressed.