Senior doctors have called for a crackdown on consumer genetic tests, following an influx of patients who have been wrongly told they are carrying dangerous mutations linked to cancer or other devastating conditions.
Women have been incorrectly informed by companies that they have faulty BRCA genes, which convey a high risk of breast and ovarian cancers. One patient was scheduled for preventive breast-removal surgery after a consumer genetic test suggested she had a BRCA mutation. The surgery was called off at the last moment when an NHS laboratory revealed the result to be a false positive.
Doctors warn that the issue is placing an increasing burden on GP surgeries and NHS genetics clinics, which are left to deal with the fallout when people receive alarming or confusing results.
“The NHS is incredibly irritated by these results because [in the case of BRCA] they’re more often wrong than right,” said Prof Anneke Lucassen, a clinical geneticist at Southampton University and chair of the British Society for Genetic Medicine. Lucassen said it was “terrifying” that one of her patients – and possibly others – had planned irreversible breast surgery on the basis of flawed results.
“I’m not saying ban these tests, but they need more regulating,” she said.
A recent analysis of a genetic testing chip that has been used by direct-to-consumer companies, found that for typical disease-causing BRCA mutations, when the test gave a positive result it was far more likely than not to be wrong.
Lucassen said she is personally aware of about 10 cases of patients being referred to genetics services after having tested positive for BRCA or bowel cancer mutations, which were later shown to be glitches when NHS labs conducted more rigorous tests. “I feel this is the tip of the iceberg,” she said.
Consumer genetics tests, offering insights on health and ancestry, have had a surge in popularity in the past few years, with hundreds of thousands of people in the UK thought to have undergone testing. The major consumer genetics companies, such as 23andMe or AncestryDNA, typically offer just ancestry testing or health readouts restricted to a handful of conditions. However, consumers can download their raw genetic data and send it to third-party companies, which for as little as £20 provide results and analysis on thousands of genes.
The problem is that the chip technology used by 23andMe and AncestryDNA is not designed to give accurate results for rare mutations, which secondary companies, such as Promethease and LiveWello, provide information about.
Although these companies normally state that any findings should not be treated as a medical diagnosis, consumers – and even doctors – may be unaware of just how unreliable such results can be.
Recent analysis of a commercial DNA chip found that it was correct just 17% of the time when it detected a rare pathogenic BRCA mutation of the kind that’s substantially increases risk of breast and ovarian cancer. The test was also found to miss more than half of those with actual BRCA mutations, in a dataset is nearly 50,000 individuals, raising concerns that people with family histories of breast cancer could be falsely reassured by negative results.
Caroline Wright, a genomics researcher at Exeter University who led the analysis, said: “The findings are shockingly bad. At the point where they’re this bad, you have to think about not sharing the data. It can be so damaging to the individual if they find out this completely erroneous information.”
Wright said that the genotyping chips used by some companies might perform slightly better, but that the technology is fundamentally not designed to detect rare mutations.
Lucassen described how one of her female patients had been scheduled for a double mastectomy on the basis of findings that were shown to be flawed when the genetic analysis was redone in NHS laboratories, which use more powerful genotyping techniques. When the repeat tests came back negative, Lucassen said the patient asked: “Can I have the surgery anyway just to be on the safe side?” Although she was ultimately dissuaded, “she wasn’t at all reassured by finding this was a false positive,” Lucassen said.
Jan Cobben, a clinical genetics consultant at Northwick Park hospital in north-west London, said he had encountered two separate cases of parents who tested positive for mutations linked to deafness and thought the genes could be affecting their children. In one case, the family had a deaf child and thought they had found an explanation. A different set of parents had been dealing with the fairly common frustration that their child often seemed to ignore what they were saying, but the genetic result led them to fear that there could be a serious underlying medical condition.
“They think: ‘Oh my God, it’s related,’” said Cobben. “There’s a gene for hearing where abnormalities are not very rare. We repeated the test and the supposed abnormality was just a normal variant.”
Cobben said he would advise anyone considering private genetic screening: “Stay away, don’t do it.” “You shouldn’t do any medical testing without a reason,” he said. “It’s a basic medical principle.”
In response to questions, Kwame Iwegbue, CEO of LiveWello, which carries out secondary analysis of DNA data, said the company does not alert customers to the risk of false positive or false negative results because these are “qualities of diagnostic tests”. “LiveWello does not provide any diagnostic testing,” he said.
Iwegbue said it was “absolutely” reasonable for NHS clinics to cover the cost of interpreting and validating private genetic tests. “A direct consequence of access to third-party online tools like LiveWello and access to ubiquitous sources of online medical knowledge is that patients nowadays are empowered and informed,” he said. “At the very least the information they acquire from these sources provide talking points for a more rewarding doctor-patient encounter … If the information presented by the patient is in fact wrong, the doctor can use this opportunity to explain the reasons why, and then present the patient with what they consider to be the correct information.”
Greg Lennon, a co-founder of Promethease, said: “When we can, we specifically warn our users of likely miscalls, and in addition, we remind people that reported associations depend on the accuracy of the data.”
Lennon sent screenshots of example reports where a false positive is thought to be likely, which prominently alert the user to this possibility.
A spokeswoman for 23andMe said: “There is a distinction between the validated variants we report on in our … reports and the raw, unvalidated data customers may choose to download on their own. We do not advise customers to utilise third-party interpretation services and note the risks of doing so very clearly on our website and within customers’ 23andMe accounts.”