NEW YORK (GenomeWeb) – The past year and half has seen increasing attention paid to the legal risks posed by the practice of genomic sequencing as part of a new paradigm of precision medicine.
Spurring much of this is a case, Williams v. Quest/Athena, in which Amy Williams is suing Quest Diagnostics and its subsidiary Athena Diagnostics for what she alleges was the negligent misclassification a genetic variant in her son, Christian Millare, who died in Jan. 2008.
According to Williams, Athena's classification of a gene change they detected in Christian as a variant of unknown significance (VUS) was negligent, causing him to continue to receive treatment that worsened his condition and caused his death, whereas if the variant were classified as pathogenic, he might have been diagnosed with Dravet syndrome and his life could have been saved.
Although the case remains unresolved, attendees and speakers at the annual meeting of the Association for Molecular Pathology earlier this month traded questions, concerns, and convictions that the ongoing debate has brought to light in their community, and tried to broaden the conversation beyond the specificities of this particular suit.
Central to the discussion was the question of how thick the wall is between laboratory testing and the practice of medicine, and whether and how this will change in the future.
A complicating quirk of the Williams case is that as part of its defense strategy, Quest has asked the court to dismiss the case on the grounds that Williams is alleging medical malpractice on its part as a licensed healthcare provider.
This would mean the company avoids going to trial because in the context of medical malpractice, as opposed to negligence, the time has run out for Williams to seek relief.
But in stark contrast to Quest's strategy of positioning itself as a healthcare provider in the context of the Williams case — many speakers at the AMP session expressed anxiety about a legal blurring of the boundaries between the provision of genomic testing and the practice of precision medicine.
Participants at the meeting also grappled with the ongoing challenges of standardizing the classification of genomic variants, how labs report them to the physicians who order tests, and, ultimately, how that data gets shared with the scientific community.
Classification and reclassification
The question of liability in the case of genomic variant interpretation is complicated by the challenge of defining negligence in a field that lacks standardization.
Though more groups are beginning to use guidelines put forth by AMP and the American College of Medical Genetics and Genomics, these guidelines are voluntary, and the field is ever evolving. Moreover, recent studies have illustrated that even when using the same framework, pathologists at different labs may come to not just divergent, but in some cases opposite conclusions about the pathogenicity of a variant.
For example, at the 2016 annual AMP meeting, researchers shared early data from an AMP working group project called VITAL (Variant Interpretation Testing Across Laboratories), which is attempting to uncover the precise reasons that labs might vary in their interpretation of the same genetic variant, even when they employ strategies or use guidelines that have been developed to make this process more uniform.
In one challenge, featuring a variant in the gene PDE11A, responses ranged over the full spectrum from benign to pathogenic with significant proportions on opposite ends of the spectrum.
"Pathogenic, likely pathogenic, VUS, these are not terms handed down on stone tablets from the mountain."
"Pathogenic, likely pathogenic, VUS, these are not terms handed down on stone tablets from the mountain. They're just the best we can do now," said Wayne Grody, a professor of pathology and lab medicine at UCLA, in a session at the AMP meeting last week.
If Quest fails to have the case thrown out on the grounds that it should be considered a healthcare provider and Williams' suit does proceed to trial as a question of simple negligence, a jury will be faced with deciding whether Athena was negligent in classifying Christian's variant as a VUS based on the available literature on the variant in 2007, the lab's internal protocols for classifying variants at the time, and potentially on expert testimony as to the standard of practice in this regard.
"The standards are not going to be the current ACMG standard for interpretation, they are going to be what was sent out with the report at the time, and judge and jury will decide whether that was adequate and accurate." Arizona State professor Robert Cook-Deegan, who is also an unpaid consultant to Williams, said during the AMP session.
Williams' complaint alleges that Athena failed to adhere to its own criteria, which require that a variant that is reported in the literature as disease causing should be deemed pathogenic.
And at the time of the test, Christian's variant had indeed been reported in two separate publications on a single individual, a girl that Cook-Deegan said at the meeting last week is "as far as we know, still alive in Canada, riding her horse."
But Quest has countered that because Christian's variant could not be classified as de novo through requested parental testing (something Williams contends she was never informed of) it is not supported by those two studies, which discuss specifically a confirmed de novo variant.
Elaine Lyon, medical director at Utah's ARUP Laboratories, reiterated what others have said — that based on existing rules or on older guidelines, there are legitimate reasons why a pathologist would have classified Christian's variant as a VUS, which hobbles an argument that Athena was negligent.
"I would say that language is subject to interpretation," Roger Klein, chair of AMP's professional relations committee, said during the session in regard to the question of what Athena's own standards were at the time, and whether it did or did not perform according to them.
"I think it's still going to need an expert witness saying this is the process by which the standard of care was violated," he argued.
"With the papers, one would think that the lab certainly should have known about it since an author was working at Athena," Grody added. "But as Elaine says one case doesn’t carry that much weight and VUS does not mean benign."
Initial classification is one aspect of the Williams case, and one major source of potential risk for labs, but reclassification is also at question in both the lawsuit and among pathologists.
Williams has contended that Athena’s failure to notify her of the later reclassification of her son's VUS earlier than it did was a recurring failure to comply with CLIA regulations, and thus a recurring harm.
But participants at the AMP meeting said that guidelines from laboratory regulators don't actually mandate any particular schedule for re-querying the scientific literature and reclassifying variants, only that this be done at some point.
"Some aspects of this case are not so new," said Grody during the AMP session. "There have always been VUS even with single-gene sequencing."
Before the US Supreme Court ruled that Myriad Genetics could no longer patent isolated DNA sequences, Grody said that he sent samples to the company for testing, which he said despite being unpopular in the lab community "did a great job following up on VUS."
"Eight years later I would get a letter that said "'Remember that woman from eight years ago?' — which I didn’t — 'Well we have just reclassified it."
"But they only had two genes to deal with. Can a lab really be responsible for the same in regard to thousands of genes?" he asked.
"Unfortunately this is not probably going to be the last case of its type … and even though this case started before the NGS era [we have to put it in that] context because this is where the future lawsuits are going to come up," he added.
In relation to the Williams case, Lyon added that while it's not clear what additional information Quest/Athena used to later reclassify Christian's VUS as disease-causing, its possible that it was his own case.
Whether or not this is acceptable practice is an open question, she said. "It really comes down to how many cases do you need to see. Can you use the [actual] proband you have included in that number? I don’t have any answers for you … I just have more questions."
"We discuss in our lab, do we use this data to push it to the next class or do we wait for the next case … because of bias of ascertainment," Lyon added.
Sharing data
Intimately related to the question of interpretation standards is the issue of data sharing. If information about the pathogenicity of variants is not entered into the public sphere, it can't be used — no matter how standardized guidelines for interpretation become or how consistently labs execute them.
While there are no current requirements that labs publicly share data, leaders like Partners' Healthcare lab director Heidi Rehm have been working with payors, regulators, and lab-accrediting bodies to encourage test providers to submit to ClinVar, an archive launched by the NIH more than three years ago, with some success.
The FDA has also said it wants to recognize public genetic variant databases that meet certain standards, though such moves risk being lumped in with the agency's possible move to exert oversight of laboratory-developed tests, something AMP and others have been strongly opposed to.
Rehm and others involved in ClinGen have been in discussions with the College of American Pathologists about the possibility of requiring that labs share variant information in a public database as a condition for accreditation, though CAP said last year that so far nothing had been brought back for formal consideration.
Cook-Deegan bemoaned the fact that Quest/Athena has as yet still not entered information on Millare's variant into databases.
"This would have been the second case that pushes it across the line to being probably pathogenic," he said. "So the tragedy here is that … someone else who gets the same mutation will be in the same situation right now. There is no further case reporting that has gone on — even though that child has died," he added.
What about cancer?
Most of the discussion of legal risks at AMP was in the context of germline analysis, as was performed in the case of Williams' son Christian.
But Washington University School of Medicine pathologist John Pfeiffer pushed the question further by bringing in cancer sequencing tests and the additional challenges they pose to interpretation and reporting of variants, not just in terms of VUS.
"These questions and issues are also intrinsic in somatic testing … in ways that are somewhat unsettling," Pfeiffer said.
Somatic variants can occur at a much wider range of variant allele frequencies (VAF), and so the clinical significance can be unclear even if the significance is straightforward, he added. "You may have a mutation that is absolutely correlated with response to a particular therapy, but if it is present at three or five percent… in that patient, at that VAF does it have any meaning?"
In the context of a discussion of legal risk, he said, "I'm over here sweating, thinking, is someone is going to say to me 'You reported a VAF correctly but there was no indication of whether someone should be treated'?"
Cancer also further complicates the question of standardization. Some labs include VAF in their reports and some don't, Pfeiffer added.
Moreover, he said, the cancer field is increasingly finding itself dealing with the same questions of inherited variant reporting that plague the clinical genetics field, because of the fact that somatic sequencing can pick up germline mutations associated with hereditary cancer syndromes.
Allele frequency also complicates this process, because variants can appear at frequencies that don’t define them clearly as either germline or somatic. In that case, Pfeiffer asked, how are pathologists expected to report them?
As an example, he presented the case of a young woman diagnosed with lung cancer, who was retested by his lab after failing to respond to an EGFR inhibitor.
"We confirmed the initial EGFR mutation," Pfeiffer said. "And we found an explanation for her [lack of response] in a downstream PIK3CA alteration, but we also found a variant in PTEN at .67 AF.
"What is that?" he asked the audience. "So I looked at her chart and she also had a history of osteosarcoma and thyroid cancer … This is a young woman in her 30s."
When he called the patient's physician to raise the possibility of an inherited cancer disposition syndrome, "the answer was a one word expletive,"
When he called the patient's physician to raise the possibility of an inherited cancer disposition syndrome, "the answer was a one word expletive," Pfeiffer said.
"And that's what it was, an inherited cancer syndrome that only came to attention based on somatic testing for targeted therapy with an allele fraction that doesn’t make any sense."
So as pathologists in the hereditary genetics field are grappling with what their legal risk is when they report uncertain variants, it's clear, Pfeiffer said, that some of the same issues are present in cancer sequencing.
"Most of these variants that are at an allele frequency that looks like it might be inherited, most of the times it turns out that they are not," he said. "But often when they are, it’s the first time anyone has locked onto the fact that there is a cancer syndrome in this family."
How and when to report
Beyond the question of classifying a variant, Williams v Quest/Athena has also brought up questions for the lab community about the fundamental relationship between pathologists and physicians.
Many at the AMP meeting expressed dismay that the neurologist and other clinicians treating Christian did not (according to Williams) tell her about the VUS result, or use the finding to inform other investigations into the possibility that her son had Dravet syndrome.
Williams can't claim damages against the doctor in question because of the statute of limitations concerning medical malpractice, and Quest is hoping the same will be true for itself. But if blame is to be applied, the opinion of pathologists at the meeting was largely that it rests on the child's physicians.
Unfortunately, the boundaries are somewhat indistinct. For example, University of Utah genetic counselor Jamie McDonald said during the session that in her opinion, Athena's initial lab report didn't include enough information about what VUS means for a non-geneticist to understand.
But others questioned whether making sure that clinicians fully understand genetic testing findings is, or can be, the responsibility of labs, considering the ever-increasing volume of these tests being performed.
"It seems to me that in 2005 anyway, the fact this was classified as uncertain was more defensible that the fact that the report contains so little explanatory information," McDonald argued.
However, she added, "Even as reported … a thoughtful, astute doctor should have taken note of possible medical complications at the time instead of putting this aside as one more inconclusive test."
According to Grody, he and his colleagues make a practice of contacting physicians directly when they report a VUS, which they do only if their analysis deems it to be potentially actionable.
However, he said "I think it’s a two-way street between lab and clinician. Anyone ordering this should know what a VUS is or they shouldn’t be ordering it."
"It’s the responsibility of the clinicians to integrate lab findings into their treatment and it’s the clinician who decides what and how to tell the patient and whether and how to act on it," he added.
"I feel like this VUS was actionable," Grody said, considering Christian's documented failure to respond to standard anticonvulsive treatments at increasing doses. "Whether you call it VUS or not …I think the physician had a responsibility to give the Dravet drugs a try."
But Lyon stressed that there are risks both to over- and under-analyzing the potential clinical impact of variants, and labs are under pressure in both directions.
McDonald echoed this as well. "My main thought is that avoiding legal risk and doing the right thing are intimately related," McDonald said. "However," she added, "my biggest concern in terms of doing right by patients is that increasing concern about liability might make it harder for labs to do the right thing."
Lyon raised two hypothetical examples that reflect the dichotomy of over- versus under-calling variants.
In one, two individuals receive carrier screening and one has a pathogenic variant returned in a gene. Sequencing of the second "reproductive partner" reveals a VUS, which, based on current guidelines for variant interpretation, would not be reported.
These two then go on to have a child with a disease.
Lyon's second example was the case of two people who both carry a mild variant, something known to cause disease, but only in trans with a severely pathogenic variant on the opposite chromosome — never in a homozygous state.
The risk to a child, which is not 1 in 4 as might be expected, may be hard for clinicians to understand. "Is there a risk there of a claim for wrongful termination of a pregnancy?" Lyon asked at the meeting.
Are are we going to be driven to defensive medicine?" Lyon asked, "…to overcall because there is less of a risk of a wrongful termination than there is legal risk for wrongful life or wrongful birth?"
Lyon was challenged by an audience member who argued that if guidelines hold that one shouldn’t return a VUS to a healthy subject, maybe those guidelines are wrong.
"I think it's appropriate to report," the speaker said. "It means you found something and you don’t know the significance, but it puts into the physician's mind that down the line, some more significance could be attached."
But considering the Williams case, does responsibility rest on labs to make sure that physicians understand the nuance of a VUS? Or is the responsibility of a doctor who orders these tests to think critically about the result, especially, as some argued at the meeting, in cases like Christian's?
Cook-Deegan and others on the stage largely concluded that there do need to be better rules for making sure results are communicated, updated, and transmitted appropriately.
In regard to Williams, "the bottom line is, somebody should have told [her]." he said. "The standards for variants … that's evolving, but the standards of who communicates to who, that's a lot more fuzzy."
"I'm not sure what they should be," he added. "But until we figure that out, boy we've got a problem."