Sometimes Wrong, Often in Doubt

In personalized medicine, uncertainty is the rule, not the exception.

My first job out of college was working the night shift at a residential academic program for adolescents with “severe developmental disabilities” (almost all were lumped into this vague category). I can’t say that it was glamorous or even intellectually stimulating; mostly I mopped the floors and cleaned the toilets.

But my time there changed me forever. It was a stark introduction to the ways in which our society devalues people with severe behavioral and intellectual challenges; to the limits of psychiatry, neurology, special education, and behavior modification; and to the enormous emotional and financial toll borne by families with children with special needs (and how they so often manage to bear that toll with patience and grace).

Three decades later the world often looks much different. Not far from my office — and in dozens of clinics around the world — kids with similar developmental problems are having their exomes (and sometimes genomes) sequenced. In many cases this information leads directly to a diagnosis and gives parents concrete information about their child’s condition and recurrence risks; it can end diagnostic odysseys, lead to early interventions, and occasionally even alter a child’s care. In the last few years we have seen the growth of this amazing diagnostic technology and the simultaneous emergence of remarkable therapies for certain genetic and infectious diseases as well as many types of cancer.

We want our stories to convey the most current, most accurate understanding of the science (and the policy that governs it) as it exists today.

At the same time, much of today’s world looks all too familiar. The U.S. healthcare system remains fragmented, reactive, unevenly distributed, bureaucratic, inefficient, and obscenely expensive. Innovation is as hard as ever. Effective drugs exist for only a handful of the thousands of hereditary conditions; those that do are already challenging third-party payers with their six-figure price tags. And while the menu of behavioral interventions for the kinds of kids I looked after in the 1980s has grown dramatically, we still lack rigorous studies of most of them. Meanwhile, it will probably not come as a shock to hear that the personalized medicine space has its fair share of snake oil salesmen, people who would peddle hope to the vulnerable in the absence of data.

It is an exciting but precarious moment. Patients and research participants are steeped in uncertainty: What does this genetic variant mean? Does this drug really work? If it does, then can I endure the side effects? Should I participate in a clinical trial? This one or that one? If I do, then will my insurer cover it?

In a recent issue of The New England Journal of Medicine, David Hunter, a professor at Harvard’s T.H. Chan School of Public Health, wrote about the raft of uncertainties that accompany so-called precision medicine and our woeful lack of preparedness to deal with them. As the new editor-in-chief of Genome, a big part of my job is to remain cognizant of such uncertainties. We have no other choice as we reckon with the present and make plans for the future. Meanwhile, with respect to this magazine, here’s what I am certain about:

Good science is absolutely necessary.

This is both obvious and non-negotiable. We want our stories to convey the most current, most accurate understanding of the science (and the policies that govern it) as it exists today. Very often that will leave us in a state of uncertainty, whether we’re talking about epi-genomes, microbiomes, drug prices, or insurance coverage.

Good science is not sufficient.

Smoking tobacco is bad for you — don’t do it. The data are irrefutable and, save a few lonely “smoking truthers,” no one would argue the point. But what about the terminally ill patient who, in his waning days, craves a cigarette after dinner? Would we wag our fingers and deny him that pleasure? My point is, irrefutable data don’t exist in a vacuum; there are real people who make choices about how they will live — or die — in the face of those data. Such choices are rarely scientific; often they will be uncertain — and irrational. There’s a name for this: the human condition.

Health disparities are real.

A few months ago, Harvard Medical School’s Zak Kohane and colleagues in Boston and Copenhagen published an analysis of more than 10 years’ worth of data on genetic variants that were thought to cause hypertrophic cardiomyopathy (HCM), a thickening of the heart muscle that can lead to sudden death. They found that patients of African ancestry were more likely to receive reports that their DNA variants caused HCM based on what was known at the time. Yet all of those variants have since been reclassified as benign. How many of those “certain” diagnoses led to unnecessary procedures?

Doubt makes us uncomfortable; like an unresolved chord at the end of a song, we have a hard time just letting it sit there.

And why the misclassification? Because African-Americans are terribly underrepresented in genetic and genomic research — without a large dataset we can’t draw robust conclusions. Of course, the ugly history of how minorities have been treated in American research makes such underrepresentation completely understandable. That said, if promising genomic technologies do not include diverse populations, then they will only exacerbate existing health disparities and probably create new ones.

Genetics is too important to be left to geneticists.

While the landscape has changed for the better, now and again I still hear geneticists, regulators, and ethicists sounding the alarm about “toxic knowledge” and “lack of actionability” and how only a board-certified medical geneticist has the requisite abilities to help a patient or a research participant navigate his or her DNA. Last year an FDA official publicly compared people who wanted their raw sequencing data to anti-vaxxers. This type of condescension offends me for three reasons: 1) it indulges the fantasy that there are enough clinical geneticists in white coats to act as gatekeepers of the Tree of Knowledge and that there’s nothing wrong with telling a family they have to wait six or 12 or 22 months for an appointment (I recently saw the 22-month figure on a genetics clinic’s website); 2) it discounts the skills of thousands of eminently capable genetic counselors; and 3) it presumes that patients and families are both too frail and too stupid to understand the uncertainty of most of their own genomic data — that they have nothing to contribute to the proceedings other than their time, tissue, and passive compliance. If we insist that heredity is only comprehensible to those with advanced degrees, then we are creating a self-fulfilling prophecy.

Could there be exceptions to these tenets? Could I be wrong? Sure. I suffer from doubt just like most everyone else. Doubt makes us uncomfortable; like an unresolved chord at the end of a song, we have a hard time just letting it sit there.

But this is where we are. “Medicine’s ground state is uncertainty,” Atul Gawande has written. “And wisdom — for both the patients and doctors — is defined by how one copes with it.” Of this we can be sure.