NEW YORK (GenomeWeb) – Using the recent example of a wrongful death case against Quest Diagnostics and Athena Diagnostics, a high-profile, independent committee of the National Academies of Sciences, Engineering, and Medicine last week considered the liabilities labs are facing as genetic testing becomes more ubiquitous in healthcare.
Earlier this year, Amy Williams filed a lawsuit accusing Quest and Athena of misclassifying an SCN1A gene mutation for her son, Christian, who was suffering from a severe epileptic condition. His doctors ran a lot of tests to try to diagnose the cause, including an analysis from Athena of the SCN1A gene. Mutations in this gene are well known to cause an intractable form of epilepsy, called Dravet, although some patients don't have such mutations.
Athena classified Christian's specific variant (1237T>A, Y413N) as having unknown links to the disease. Williams says Athena got it wrong, and points in her complaint to two papers published before Athena issued Christian's June 2007 genetic test report that reference the case of a little girl with the same SCN1A variant and same epileptic condition as Christian. She also cites a patent, which was part of the IP that Athena exclusively licensed to provide SCN1A testing in the US and includes a description of Christian's specific mutation as disrupting biological processes in a way that would cause epilepsy. Moreover, she points to Athena's own classification criteria outlined in the 2007 test report, which explains that for a variant to be classified as a "known disease-associated mutation" it needed to be documented in the literature.
Based on this, Williams believes Athena should have classified the variant as causing Dravet. But because Athena said the variant's relationship to Dravet was uncertain, Williams claims that Christian's doctors continued to treat him as if he had a mitochondrial disorder and gave him increasing doses of drugs known to worsen seizures in Dravet patients.
Christian had a fatal seizure in January 2008 at the age of two. There is a risk of sudden death among Dravet patients, studies show. However, Williams believes that if Athena had interpreted the SCN1A mutation as pathogenic, doctors would have diagnosed and treated her son appropriately, and Christian would still be alive.
Quest and Athena, which did not present before the National Academies' Committee on Science, Technology, and Law (CSTL) last week, have argued the case should be dismissed on statute of limitations grounds. The lawsuit is currently pending in US District Court in the District of South Carolina, and Judge Margaret Seymour has yet to decide whether Williams' case should be dismissed or decided by a jury.
Using this lawsuit as an example, lab director Heidi Rehm, policy expert Robert Cook-Deegan, legal scholar John Conley, and Laurel Coons, a pharmacology PhD candidate at Duke University who is helping Williams with the technical details of her case, discussed how labs and healthcare providers can be held increasingly liable when patient care relies on the accurate classification of a genetic variant.
At the meeting, Conley, a law professor at the University of North Carolina, noted the questions raised by Williams v Quest/Athena, for which there is little or no legal precedence. For example, can genetic testing labs be considered healthcare providers covered under malpractice law — which is what Quest has argued? To what extent is the lab or the doctor responsible for staying up on the latest evidence on genetic variants? Can violations of federal lab regulations under CLIA serve as proof of negligence in tort cases? Is the lab protected if it just follows standard industry practices or can the law force adoption of a new standard?
These experts urged the committee to study these difficult questions, advance recommendations that will help make data sharing on genetic variants the norm in the life sciences space, and encourage genetic test providers to work together to resolve discrepancies in variant classification (see related story).
One can disagree professionally with the interpretation because it is based upon judgment, but I'm not sure there was true error there.
Room for professional judgment
Currently, when it comes to classifying variants, labs have their own processes, but many use guidelines put forth by the American College of Medical Genetics & Genomics, the most recent version of which were developed jointly with the Association for Molecular Pathology. But even using these guidelines, genetic testing professionals may come to divergent conclusions about the pathogenicity of a variant, much like any area of medicine requiring professional judgement, Rehm, director of the Laboratory for Molecular Medicine at Partners HealthCare Personalized Medicine, told CSTL members.
Evaluating Christian's SCN1A variant according to the 2015 ACMG/AMP guidelines, of which Rehm was the senior author, she demonstrated that they could be used to support Athena's 2007 classification as a variant of uncertain significance, since Christian's parents weren't tested to establish whether the variant was inherited or de novo like more than 90 percent of pathogenic mutations in Dravet.
While the guidelines also state that multiple cases seen with the variant add weight to a classification, the two published reports cited by Williams reference the same girl. Although parental testing was done to establish the de novo status of the variant in that patient, Rehm noted that the researchers did not confirm paternity and maternity. This is a check that the 2015 guidelines recommend in order to make full use of the strength of de novo evidence, although this analysis step is not standard among labs.
Williams claims that neither Christian's doctor nor the lab informed her that an SCN1A test had been done, so she couldn't have gotten the follow-up testing strongly recommended in Athena's original test report. Once she learned about the test in 2014, she requested and received a copy of the 2007 report. Several months later, in 2015, Athena reclassified the variant as pathogenic, but the report doesn't point to new evidence that led to the change, and parental testing still hadn't been performed.
When Christian got tested, the ACMG guidelines that would have been relevant were published in 2000, which are far less detailed than the latest recommendations. The ACMG updated the recommendations again in 2007 and published them in 2008. The 2015 recommendations provide more specific advice on how to weigh different kinds of evidence and suggeste analysis steps such as paternity and maternity testing.
All of this is a reflection of how much the genetics field has changed in the last decade and illustrates "the complexities and subjective nature of variant classification even when using a set of professional guidelines on variant interpretation," wrote Jeanette McCarthy, a professor at Duke University Medical Center, and Bryce Mendelsohn, resident physician in the pediatrics department at University of California, San Francisco, in a July Medscape article discussing the Williams case.
They also evaluated Christian's mutation according to the 2015 ACMG/AMP guidelines, and noted the caveats Rehm did. But they also pointed out that the variant did not meet any of the dozen criteria for a benign classification. Moreover, the authors did not find the variant in three major databases containing genetic sequences from healthy adults, though these repositories represent "a thin slice of humanity" and were not available in 2007.
This is a false-negative call. It's clear that it was wrong. Whether it was a laboratory error is the subject of litigation.
Rehm acknowledged that Athena may have erred in other ways, such as not documenting the existence of the published evidence in the test report and not indicating the reasons for reclassifying the variant as pathogenic in 2015. But she told CSTL members that in the practice of medicine there must be room for professional judgment, and this is no different when classifying variants. "One can disagree professionally with the interpretation because it is based upon judgment, but I'm not sure there was true error there," she said.
Coons, who is helping Williams with her case, disagreed. After the CSTL meeting, where she presented the details of the plaintiff's complaint, Coons told GenomeWeb her view that if a variant does fulfill some pathogenic criteria but does not meet any benign criteria, then "this excludes an interpretation of uncertain significance according to the 2015 guidelines." Moreover, strict adherence to parental testing as a standard for classification would render children with one parent or adopted children ineligible for genetic testing through Athena, she said.
"This is a false-negative call," Cook-Deegan, a professor at Arizona State University's School for the Future of Innovation and an unpaid expert for the plaintiff in Williams v Quest/Athena, said during the meeting. "It's clear that it was wrong. Whether it was a laboratory error is the subject of litigation."
Liability at every step
Although the CSTL meeting was about liabilities for labs, Williams v Quest/Athena also raises questions about the responsibilities doctors have when using genetic test results to make diagnosis and treatment decisions for patients.
"There is liability and requirements at every step," Rehm said, noting that a lab has the responsibility to interpret variants using professional standards and expert judgment; the doctor needs to ensure follow-up testing is done and communicate results to patients; and the healthcare system should be set up in a way so medical professionals can easily access test results and the latest information on genetic variants and apply that to patient care.
"The best current tort law analysis would be that liability is shared between the labs and the physicians that order the tests. How that sorts out in a given case will be facts specific," Conley said. "The harder question is liability for what? What is the standard? What constitutes a breach of care by the lab or the physician?" This is where he thinks an organization that can look at the bigger picture could help, rather than a court considering a specific case.
The latest ACMG/AMP guidelines may also be interpreted differently about physicians' responsibilities when dealing with variants of uncertain significance. On the one hand, the guidelines advise against using such a result to make clinical decisions but on the other hand recommend that "additional monitoring" of the patient "may be prudent."
In Williams v Quest/Athena, the plaintiff is not suing Christian's doctors, however. This, Conley said, was a tactical judgment on the part of the plaintiff, but the labs might challenge the doctors' role if the case goes forward.
Williams' goal in bringing this case was to make sure that a case like this doesn't happen again.
Amy Williams' goals
More importantly for Cook-Deegan, the lawsuit shines a light on the fact that genomic knowledge is falling through cracks in the healthcare system, and it can have devastating consequences for patients. "The biggest thing that's missing is the flow of data from research labs and clinical labs into the public databases like ClinVar," he said. "I think what this case exposes is that it isn't happening on a regular basis, and it matters."
After searching all the available variant databases, including the Human Gene Mutation Database, Leiden Open Variation Database, GenBank, 1,000 Genomes, ClinVar and several others, the girl with epilepsy in the two studies is still the only person recorded as having the mutation Christian had. Eight years after his death, Christian's case in not in any database.
"I think a broader issue is that Athena was not and still isn't completely sharing all of their data in a public database," Rehm told the committee. "That practice allows better information around variant knowledge."
Rehm pointed to the critical need for clinical laboratories, researchers, and publications to deposit interpreted variants and supporting evidence into ClinVar — a database of genotype-phenotype relationships launched by the NIH — to ensure that patient reports benefit from access to all available evidence and professional opinions on a variant. She has been working with payors, regulators, and lab-accrediting bodies with some success to encourage test providers to submit to ClinVar.
Athena has submitted 281 variants as of October 2015 (but not on Christian's variant), while Quest has contributed nearly 500 variants as of July, and has launched a data-sharing effort for BRCA1 and BRCA2 genes. Quest does not comment on pending litigation, and the company did not make public comments at the CSTL meeting.
Cook-Deegan further added that Williams has resorted to the courts precisely because the healthcare system isn't set up to avoid these types of problems. "Wouldn't it be nice if we designed systems [where] ... in most cases liability doesn't even arise?" he said. "The one virtue of the tort system is that it's there even if nothing else is, and that's why we've resorted to it this time."
He told the committee that Williams' goal in bringing this case was to make sure that a case like this doesn't happen again. If the lawsuit advances to settlement or gets decided by a trial — a stage that less than 2 percent of federal civil cases reach — then Williams will ask Quest and Athena to commit to sharing genetic variant information in a public database. She would also like funds for a scholarship and lecture series to train genetic counselors; backing for a National Academies' study on data-sharing standards for genetic testing labs; and resources to form an independent, advisory committee that will monitor that these activities are being carried out.
"We were quite surprised when we learned that [CSTL was] going to have a meeting on this, because we have been proposing that if this [case] ever gets to the merits … one of the things we were going to ask for is to have an expert body ... do a study on this," said Cook-Deegan.
Although the committee considered the liability issues around variant interpretation in the context of a pending lawsuit, Conley suggested that matters related to industry standards are better dealt with outside of courts. The tort litigation system is slow, often derails complex cases on technical grounds, like statute of limitations, and more geared toward paying out or denying damages than impacting broader industry issues. "If the Williams case were to go through to a final decision and that decision involved some setting of standards, [those standards] would be based on something people did in 2007, and the decision would probably be reached in 2020," he said.
"There is some urgency," Cook-Deegan said, while it's still early days in terms of applying genomics to patient care, and national projects like the Precision Medicine Initiative and the Cancer Moonshot are just starting up. "Right now, there are business models antithetical to sharing of data," he said. "If we don't get the basic rules of the road set, so the business models develop with data sharing as a foundational principle, we're going to be in trouble in 10 years."
This is one of two articles on a recent National Academies of Sciences, Engineering, and Medicine's meeting that explored liabilities for diagnostic labs. See related story.