Limits of Gene Variants as Biomarkers Aired in Cohort Study

Pam Harrison

January 06, 2016

NASHVILLE, TN — The prevalence of gene variants previously associated with long-QT syndrome and Brugada syndrome, two conditions known to elevate the risk of sudden death, turned out not to correlate significantly with development of ventricular arrhythmia or even with evidence of a conduction defect, in a cohort study that highlights some limits of gene-based disease prediction[1].

"Until now, with some exceptions, the way in which we understand the genetic architecture of human traits has been to do targeted genotyping. And many people have a vision that we're going to have that kind of sequence data available to clinicians on a routine basis," according to Dr Dan Roden (Vanderbilt University Medical Center, Nashville, TN).

That's unlikely, he said, going by the current prospective study of 2022 persons analyzed for 84 sequenced genes involved in drug responses, including SCN5A and KCNH2, which are encoded for ion channels involved in drug interactions; the genes have variants associated with long-QT syndrome and Brugada syndrome.

"I don't want to come across as paternalistic," he told heartwire from Medscape, "but it's misguided to believe that we can find poorly understood variants on sequencing selected genes and then tell apparently healthy persons that 'you have this genetic variant that might make you drop dead.' " Because, he added, "what do people do with that information?"

He is senior author on the report published in the January 5, 2016 issue of the Journal of the American Medical Association, with lead author Dr Sara L Van Driest (Vanderbilt University).

A Call for Uniform Standards

Out of the entire cohort, 223 participants, or about 11%, harbored one or more of a total of 122 rare, nonsynonymous, or splice-site variants in the two ion-channel genes. Three expert laboratories designated 42 of the variants in 63 participants as being potentially pathogenic, but, as the authors confirmed, the three laboratories interpreted the data very differently. Indeed, only four variants were designed as potentially pathogenic by all three laboratories.

"One of the most important lessons from this paper is that we really do need relatively uniform standards in terms of understanding how to assign pathogenicity," Roden said.

Importantly, on review of phenotype data in electronic medical records (EMR) and electrocardiographic data, only 35% of variant carriers showed ECG or clinical evidence of an arrhythmia phenotype. And there was no difference in prevalence of arrhythmia diagnoses or ECG-defined arrhythmia phenotypes among variant carriers compared with noncarriers.

"If I do this kind of genetic testing in patients who have fainting spells or an abnormal ECG, my suspicion that they may be harboring a variant that is going to have serious health consequences for them is high, and a positive test makes it much higher," Roden explained.

As such, he said, "I think the study is all about Bayes's theorem." But that has other implications, too. "If physicians do the same test in a 75-year-old patient with a normal ECG, my suspicion that they are going to harbor a variant that is going to have serious health consequences for them is very low. And even if the test comes back positive, the probability that they are harboring a variant is still very low," according to Roden.

Still, he said, "How can you test patients and then not give them the results? So there's a real awkwardness in all of this, and there's a level of education needed for both healthcare providers and patients, because there's real uncertainty about interpreting results such as ours."

"A Duty to Report Incidental Detection"

"The lack of association among pathogenic variants and meaningful clinical outcomes in the study population calls into question how existing knowledge of genetic variants translates to predicting outcomes in unselected and ostensibly healthy individuals," according to Dr William Gregory Feero (Maine Dartmouth Family Medicine Residency, Fairfield) in an accompanying editorial[2].

Both SCN5A and KCNH2 are listed by the American College of Medical Genetics and Genomics as "actionable genes," he writes, "implying a duty for molecular laboratories and clinicians to report incidental detection of pathogenic findings in these genes."

That agreement between the expert laboratories involved in the study was "disconcertingly low," combined with the lack of evident disease in the majority of patients who carried a pathogenic variant, "provides de facto evidence of a high false-positive rate for variant interpretation."

The current study, Feero writes, "exposes some shortcomings in the existing ability to meaningfully predict the consequences of at least some genetic variations currently thought to be causally related to serious disorders." For now, he continues, "caution should be exercised when considering clinical interventions informed by the presence of 'pathogenic' variations in healthy individuals, families, and populations."

Van Driest and Roden had no relevant financial relationships; disclosures for the coauthors are listed in the article. Feero reported no relevant financial relationships.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.