Chips used to detect single-nucleotide polymorphisms—the type of DNA microarrays used by direct-to-consumer genetic testing companies to detect variants in a person’s genome—have a false discovery rate of more than 85 percent when screening for very rare variants, according to a preprint published on bioRxiv on June 9.
Charleston Chiang, a human geneticist at the University of South California Keck School of Medicine who was not involved in the study, says he thinks the paper raises awareness of a potentially important problem in the direct-to-consumer (DTC) genetic testing industry. Unreliable results may find their way to customers, who could “interpret the results at these rare variants literally, without accounting for any possible laboratory errors,” Chiang says. “That would be a legitimate concern.”
The chips tested in the study were supplied by the manufacturer Affymetrix, owned by Thermo Fisher Scientific. These specific chips are used by some DTC genetics testing companies such as LivingDNA, and raw data from these companies can be uploaded by customers to third-party services that claim to be able to further extrapolate information about one’s health based on the person’s genetic code. Such services are plentiful, inexpensive or free, and largely unregulated. While the companies warn customers not to regard their results as a medical diagnosis, the level of uncertainty may not be clear to all consumers, Chiang says.
Even for DTC genetics testing companies such as MyHeritage and 23andMe, which claim to use chips with higher accuracy, very rare variants will be susceptible to such false positives, says study coauthor Caroline Wright, a genomicist at the University of Exeter College of Medicine and Health.
To investigate the accuracy of Affymetrix’s chips, Wright and her colleagues evaluated data from 50,000 participants of the open-access UK Biobank. They found that, while genetic variants prevalent in more than 0.1 percent of individuals were detected accurately more than 99 percent of the time, variants classified as very rare—those found in less than 0.001 percent of people in the dataset—had a false positive rate of more than 84 percent, when validated by sequencing.
Wright says she hopes that the results of the study will reach consumers of these mail-in genetic tests and clinicians alike. She wants it to be clear that “the rarer the variant, the less likely [the result] is to be accurate.” Thermo Fisher Scientific did not respond to multiple requests for comment before deadline.
An example of variants of such rarity are the BRCA genetic variants that are linked with breast and ovarian cancers. There are thousands of known variants of BRCA1 and BRCA2, and each is quite uncommon. The online genealogy platform MyHeritage, which has more than 105 million users, announced in May of this year that it would be expanding its DNA testing services to offer health reports that included genetic risk for hereditary breast cancer. In March 2018, 23andMe, which has more than 10 million customers worldwide, had also announced that they had received authorization from the US Food and Drug Administration to begin providing a DTC genetic test for three genetic variants found on the BRCA1 and BRCA2 genes.
MyHeritage tells The Scientist that it is aware of the issues the study highlights. Company researchers have “therefore developed multiple layers that substantially increase the [positive predictive value] of MyHeritage’s BRCA tests,” the company writes in an emailed statement. Employees double-check and validate every potentially pathogenic variant using sequencing. In addition, they have on average 2.6 DNA probes for each BRCA variant, whereas the paper looked at only one probe; multiple probes increase accuracy.
23andMe also sequences the BRCA genes before giving a report of the variants identified, and further notes that consumers ought not to trust the results from third-party services. In an emailed statement, the company writes that raw data “should not be used for anything beyond educational and research purposes” and not be considered medical or diagnostic proof.
Philippe Froguel, a genomicist at Imperial College London who was not involved in the study, agrees with the authors of the study about the dangers of relying solely on DNA arrays to identify genetic variants. And in fact, if doctors need to sequence a patient’s DNA to get an accurate read of the genome, they might as well start there. “If you want to identify pathogenic mutations causing diseases . . . the only thing to do is to develop a [sequencing-based] test of high quality,” he says in an email.
Frederick Roth, a genetics researcher at the University of Toronto who was not involved in the research, says in an email that the new study is in line with a growing understanding of the best genetic approaches for identifying genetic markers of interest. “[T]hese results take us in the same direction that the field is already going,” he writes. “If you care about extremely rare variants, then sequencing is the way to go!”
Wright agrees, emphasizing that these DTC genetic tests are “absolutely not a replacement test” for a diagnosis from a healthcare professional. She worries that patients who do not understand the uncertainty surrounding the results given by third-party services may act on results before seeking to confirm the diagnosis in a clinical setting. “These tests could be causing a whole lot of worry and potentially actual physical harm, for something that isn’t there,” Wright says.
Grace Browne is a freelance science writer living in Dublin, Ireland. Follow her @gracefbrowne.