The ACMG Gene Screening Recommendations

57 sauceIn March, the American College of Medical Genetics and Genomics (ACMG) released its much-anticipated Recommendations for Reporting of Incidental Findings in Clinical Exome and Genome Sequencing. The ACMG describes itself as “an organization composed of biochemical, clinical, cytogenetic, medical and molecular geneticists, genetic counselors and other health care professionals committed to the practice of medical genetics.” Its brief mission statement includes a commitment to “Define and promote excellence in the practice of medical genetics and genomics in the integration of translational research into practice.” It publishes the journal Genetics in Medicine, and has previously issued “standards and guidelines” for clinical genetics laboratories and cystic fibrosis carrier screening.

The core recommendation is straightforward: “The ACMG recommends that for any evaluation of clinical sequencing results, all of the genes and types of variants in the Table should be examined and the results reported to the ordering physician.” Reading this in light of the definitions section and the rest of the report, it seems to mean this: Whenever a lab is requested to do any “clinical sequencing” (more below on what this means), it should examine the 57 genes listed on the Table and report any significant mutations it finds. It is the responsibility of the clinician who ordered the initial sequencing “to provide comprehensive pre- and post-test counseling to the patient.” In what has become the most controversial aspect of the Recommendations, the ACMG recommends the test findings “be reported without seeking preferences from the patient and family and without limitation due to the patient’s age.” In other words, patients should be given the 57-gene screening whether they want it or not and told the results even if they say they don’t want them—in effect, if you consent to any clinical sequencing, you automatically consent to this further screening and to hearing the results. The same holds true for the parents of minor patients.

In assembling the list of 57 genes, the ACMG focused on conditions “most likely to be verifiable by other diagnostic methods and amenable to medical intervention.” Moreover, not all mutations should be reported, but “only variants with a higher likelihood of causing disease.” The list includes the BRCA1&2 breast and ovarian cancer genes (the first two entries), as well as genes associated with Lynch Syndrome (a much-increased likelihood of developing colon cancer), Marfan Syndrome (heart and other problems, often in tall, thin people, and suspected in the sudden deaths of several high-level basketball players), hypertrophic cardiomyopathy, Long QT (enhanced likelihood of sudden death), and familial hypercholesterolemia.

Controversy erupted as soon as the Recommendations were issued. In fact, two themes have dominated. Some critics argue that the Recommendations are an affront to the principle of patient autonomy (which includes the right not to know), which has always been a core principle of medical and research ethics, and is enshrined as such in the 1979 Belmont Report that still guides bioethics. A second line of criticism emphasizes how much is not known about the pathway from mutations to disease (except in a few instances), and also challenges some of the ACMG’s conclusions about what is “amenable to medical intervention.” Taken together, the two branches of the critique assert that medical genetics is not sufficiently developed to justify the recommended assault on patient autonomy.

The ACMG report itself deals with some aspects of the controversy preemptively, and the authors of the Recommendations have already published a “Clarification” in Genetics in Medicine that deals with the major issues raised by the critics. The Clarification stresses that the clinician’s return of results to the patient should be part of a “shared decision-making process” tailored “to the clinical circumstances,” thus respecting patient autonomy. It also points out that “the list [of 57 genes] will evolve as more information is acquired about these and other diseases/variant,” and will thus be subject to “further curation.” (And I can’t help remarking on what a truly odd word that is in a document intended, according to its title, to clarify.)

The debate over how big an affront this is to the autonomy principle and whether these are the right genes is not likely to be resolved anytime soon. With this as background, I want to focus on what the Recommendations might mean from a legal and practical perspective.

When Are The Recommendations Triggered?

I found myself in a totally unexpected controversy when I described the Recommendations as follows in my post on the Myriad decision: “In a nutshell, the ACMG recommends that doctors and labs doing any genetic testing should always sequence 57 listed genes and report the results to the patient, regardless of whether the patient wants those results.” A geneticist friend chastised me for my naïve glossing of “any evaluation of clinical sequencing results” (the precise language of the Recommendations) as “any genetic testing.” My friend—a recognized expert—pointed out that anybody would understand (well, anybody else, I guess) that “clinical sequencing” could refer only to whole genome (or whole exome) sequencing. Thus, he said, what the Recommendations actually mean is the 57 genes should be reported only in instances where a lab was doing a WGS or WES. In other words, they apply only where the lab had already sequenced everything, and need only instruct its computer to spit out these results in addition to anything else that had been requested. He also said that this would be a relatively small subset of all genetic testing.

Well, this guy would know, so if that’s what he said the authors meant, then that’s what they meant. But authors don’t control the lives or their texts or the meanings of their words—readers do. And for “normal” readers—defined here as all non-geneticists, a category that potentially includes lawyers, judges, and regulators—this is a very strained meaning. (I say this having written on linguistic topics for 30 years.)

First, the Recommendations define clinical sequencing (in the Introduction, not, curiously, in the Definitions) as “exome and genome sequencing”—not a W to be found. And they do so in the first sentence of the report, right after the Abstract. Second, the Abstract itself says this: “We recommend that laboratories performing clinical sequencing seek and report mutations of the specified classes or types in the genes listed here. That sounds consistent with my original reading—“doctors and labs doing any genetic testing should always sequence 57 listed genes and report the results to the patient.” Third, if clinical sequencing = genome sequencing = whole genome sequencing, then that ubiquitous W in “WGS” or “WES” is and has long been wholly redundant. That would be an odd linguistic development. Usually, speakers and writers “mark” (or put an extra label on) a version of a word or phrase to distinguish it from the generic “unmarked” version, as in “split-fingered fastball” versus “fastball.” So the whole analysis begins with a major ambiguity. As I argue at the very end, it is an ambiguity with potential legal significance.

Will the Recommendations Have Any Legal Significance?

What does it mean to the law when a learned medical body like the ACMG offers Recommendations (or guidelines, or practice standards, some other near-synonym)? In particular, do those Recommendations become by definition the standard of care for the specialty, immediately or in the near future?

The Recommendations begin with an italicized “Disclaimer,” which seems intended to prevent the Recommendations from being interpreted as a new standard of care. It begins with these statements:

This guideline is designed primarily as an educational resource for medical geneticists and other health care providers to help them provide quality medical genetic services. Adherence to this guideline does not necessarily assure a successful medical outcome. This guideline should not be considered inclusive of all proper procedures and tests or exclusive of other procedures and tests that are reasonably directed to obtaining the same results.

But the legal concept of standard of care, which is central to malpractice litigation, is ultimately in the hands of judges and juries. Professional malpractice is usually defined as failing to meet the applicable standard of care in a way that proximately causes an injury to the patient or client. Standards of care used to be highly localized—a doctor in rural Mississippi wasn’t held to the same standards as one practicing at a fancy hospital in Boston. But they are becoming nationalized as a result of the telecommunications revolution. That rural doctor might not have access to the same facilities as her big-city counterpart, but she should have access to the same information.

Imagine that today a doctor orders limited clinical genome sequencing for a patient. The lab does a WGS (here’s where that ambiguity comes in—would our analysis be different if the lab did less than WGS?) and reports back only on the genes the doctor mentioned. The doctor’s order, and the labs report, do not include the genes linked with Lynch syndrome. Several years later, but before the relevant statute of limitations has expired, the patient is diagnosed with advanced and untreatable colon cancer. He also gets further genetic testing and is found to have Lynch syndrome. He sues the original doctor for malpractice, contending that (1) that doctor should have gotten the Lynch syndrome genes sequenced and (2) had that been done, he would’ve been given rigorous colon screening, which would probably have detected the cancer at a treatable stage.

What role will the Recommendations have in this lawsuit? In principle, practice guidelines can be both a sword and a shield. A plaintiff can argue that a doctor who fails to follow them has thereby failed to meet the standard of care. But a doctor who does follow them can use that fact defensively—“by definition, I did meet the standard.”

What does the law say? One of the best judicial analyses I’ve seen was done by U.S. District Judge Carlton Reeves of Mississippi in a case called Conn v. U.S., 880 F. Supp. 2d 741 (S.D. Miss. 2012). The plaintiff sued the Veterans Administration for failing to diagnose and treat his 90% arterial blockage, after which he suffered a massive heart attack. The plaintiff’s expert relied on clinical care guidelines promulgated by the American College of Cardiology Foundation and the American Heart Association in determining the standard of care that he claimed the VA hospital had violated. Judge Reeves granted summary judgment to the defendant VA hospital, dismissing the plaintiff’s claim before trial.

The judge wrote that “reliance on the Guidelines implicates an issue arising more and more often in the nation’s courts: whether the ACCF/AHA Guidelines, or any other set of published clinical practice guidelines, can establish a national standard of care in a medical malpractice lawsuit. To put it mildly, the question does not enjoy a consensus answer.” After reviewing cases that came out both ways, and acknowledging the value of “consensus standards of conduct that are both clearer and more rational than those currently used to identify professional negligence, he concluded that Mississippi law “permits expert witnesses to rely on clinical practice guidelines if the conduct prescribed by those guidelines does indeed describe the specific actions that would be taken by a minimally competent physician.”

Judge Reeves’ conclusion embodies the following points: (1) Clinical practice guidelines in a given specialty do not automatically (or per se, as lawyers would say) become the legal standard of care. (2) The applicable standard of care must be established by expert testimony. (3) But experts can rely on guidelines in reaching an opinion on what the standard of care is, (4) if and only if the guidelines are specifically relevant the situation faced by the doctor in the case.

How might this reasoning apply to my ACMG Recommendations hypothetical case? We would first conclude that the Recommendations would not automatically comprise the standard of care, whether used as a sword by the plaintiff patient or a shield by the defendant doctor. But a court would allow an expert (same rules for plaintiff’s and defendant’s experts) to factor them into his or her determination of the standard of care, and to testify to the jury about doing so—but only if they were specifically relevant.

At this point, focus on how the Recommendations are written. They couldn’t be more specific: here are 57 genes that you should always sequence and report. In a case that hinged on one of the 57 not being reported, I think most judges would conclude that they do “indeed describe the specific actions that would be taken by a minimally competent physician.” The expert could rely on them in setting the standard of care, and could tell the jury about the reasons for that reliance. In fact—at least through this back door—the Recommendations would, in a practical sense, become the standard of care.

It’s not clear whether the people who wrote and approved the Recommendations worked through scenarios like this. But, as I said earlier, authors don’t control the lives of their texts. It seems abundantly clear that doctors ordering any kind of genetic testing need to be aware of them and follow them, unless there are compelling, case-specific reasons to do otherwise. And I’d also advise them to follow my original interpretation of when the Recommendations apply—doctors and labs doing any genetic testing should always sequence 57 listed genes and report the results to the patient, regardless of whether the patient wants those results—rather than the more limited one that the ACMG apparently intended.