Incidentalome: Accidental Gene Findings You May Not Want To Know

What is it? Dr. Robert Green , a medical geneticist at Brigham and Women’s Hospital and Harvard Medical School, offers this helpful analogy: Say you get an X-ray because you think you might have cracked a rib and the radiologist notices a shadow on your lung that appears to be a tumor. That finding is “incidental,” because it turned up even though you weren’t looking for it.

Translate that to the dawning era of genomics in medicine. As DNA analysis gets exponentially cheaper and eventually becomes routine in our care, the labs that sequence our full genomes will be turning up all sorts of extraneous information. A scan ordered for, say, a heart problem could turn up a genetic variant that means you’re predisposed to cancer. Or that you’re at heightened risk for diabetes or Alzheimer’s disease.

What to do? The gene finding only suggests an increased or decreased risk, mind you, not a certainty. Should the lab that finds it tell the doctor who’s treating you? And should your doctor tell you?

'What is the value of telling people something you’re pretty damned sure they’re going to jump to the wrong conclusion about?'

These incidentalome questions arose first in the context of large-scale genome research, Dr. Green notes, among scientists who wondered what their obligation was to tell subjects about genetic findings that they had stumbled upon. (What if, for example, a subject who has grown up male turns out to have the genes of a female?)

But it is poised to become a question for the clinic: The Laboratory for Molecular Medicine at Partners Center for Personalized Genetics Medicine expects to begin offering whole-genome sequencing as a clinical service as early as this July, he said, and nationally, a few other institutions — Baylor, Wisconsin — are already offering clinical sequencing.

So now might be a time for us all to start thinking: What would you want to know? You may not have been curious enough to pay for one of the direct-to-consumer services that analyze people’s genes, but what if your doctor in the not-too-distant future has the information anyway?

And more urgently, now is the time for the medical community to start figuring out its own standards for what to share, Dr. Green argues. To help launch that discussion, he has just published a paper in the journal “Genetics in Medicine” that he describes as a “quick and dirty poll” of how medical specialists feel about incidental findings. And he has been leading a committee of the American College of Medical Genetics on creating a more formal consensus statement on the topic. He is scheduled to present it publicly on March 28 at the college's annual meeting, and anticipates controversy.

“Among experts," he says, "opinions range from ‘You absolutely should not return anything except the directed question that’s being looked at — If you’re looking at a heart problem you should only return genes related to the heart problem — while others say, ‘You must return everything. You should return all the known mutations, all the unknown mutations, all the risk variants, all the pharmacogenomic variants — you should return everything because it would be unethical to do otherwise.’”

I must say, I personally side with the “I want to know everything” camp. (I hear there’s a special term for the genomes of people like me: The Narcissome.)

But Dr. Green points out that what’s at issue just now differs from the earlier direct-to-consumer genomics debate over whether people have the right to pay to slake their curiosity about their own genes.

Rather, this is the moment to figure out standards of care for laboratories and clinicians as genomics enters the clinic. What should be the minimum set of gene findings a doctor should return to you? The Genetics in Medicine paper, he said, is the first to grapple with actual variants on the path to determining that standard.

He explains the context: “There are many different levels of certainty when you find a mutation. In some mutations, we’re fairly certain that if you have the mutation you will get the disease, but in many other mutations we truly have no idea whether the mutation will be predictive of a disease or not.”

The trouble is, he says, that both in the medical and the general population, genetics is so associated with determinism, “with the notion that if you have a certain variant you’re going to get the disease. It’s very hard to communicate to either doctors or patients that in some situations we truly don’t know.”

“So then you find yourself entangled in a philosophical question: If you truly don’t know what a finding means, is it reasonable to inform somebody about it? When it’s almost impossible to hear the part of the statement that says, ‘OK, there’s a cancer gene that has a mutation but I have no idea if that puts you as increased risk for cancer or not.’? You almost can’t hold those two ideas in your head at the same time. Our brain is such that we leap to the conclusion that that risk indeed exists. So the philosophical question becomes: What is the value of telling people not only something you don’t really know about but something you’re pretty damned sure they’re going to jump to the wrong conclusion about?”

We could argue about that all night, but it’s actually not the question at hand. The immediate question, he says, is something closer to: What information should doctors give patients because they would be remiss not to?

Let’s return with Dr. Green to the X-ray analogy: If you get a rib X-ray and the radiologist sees a possible tumor shadow in your lungs, it’s clearly the radiologist’s duty to report that. That’s the kind of standard that genomics needs too, he says. You don’t get to decide whether the radiologist reports the tumor, nor should you.

“There is absolutely no standard that yet exists for what is the fiduciary duty of a lab performing one of these first whole-genome analyses," he says, "and prior to this paper, no one had ever actually asked a group of specialists what they would recommend."

When he asked, Dr. Green reports in his paper, he found a high degree of agreement on returning results on highly predictive genes ranging from predispositions for cancer to heart disease.

From the paper’s press release:

"This is the first study to ask specialists in genetics and laboratory medicine about the conditions they would like to see returned to clinicians who order genome sequencing," said Dr. Green. "While there was not perfect agreement between them, it was heartening that the majority of specialists agreed that many incidental genetic findings should be returned."

The investigators asked 16 specialists to independently evaluate 99 commonly ordered genetic conditions and select which ones they would recommend reporting to the ordering physician if discovered incidentally through whole exome or whole genome sequencing. Respondents provided separate recommendations for disclosing each condition based on whether the patient was an adult or a child and on the strength of evidence that the particular genetic variant was pathogenic. The central question was this: Would well-meaning specialists agree on which findings to disclose?

Specialists did agree on many counts, unanimously favoring disclosure of 21 conditions to adult patients. Concordance was generally high for adult patients when the genetic variant was known to be pathogenic, with at least 80% of the specialists recommending return of results for 64 different conditions, and particularly for conditions with potential for medical intervention (such as cancer predisposition syndromes).

The specialists differed more when the patient was a child, when the variant was only “presumed” or “predicted pathogenic,” and for conditions without potential for medical intervention (for example, neurodegenerative disorders such as Alzheimer’s disease).

“Whole-exome and whole-genome sequencing are valuable pieces of information, but they are just a piece of the clinical picture,” said [Medical College of Wisconson's Dr. Howard J.] Jacob.  “As a lab value, this data is important, but we also need to be cautious and judicious when we return information to patients. This study provides us valuable insight; the next logical step is learning what patients would want to know about their genome.”

Readers, can we give them a jumpstart on this next step? What would you want your doctor to tell you? 

P.S. A footnote for future lexicographers: Coinage credit for the term "incidentalome" goes to Harvard Medical School's Dr. Isaac Kohane. He has a long history of creative coinages and catchy acronyms; I asked him how he came to this one and he replied:

In 2001, I started noticing in gene sequences of supposedly health individuals mutations that had textbook associations with disease. I then investigated the reasons why this might be happening and published them in a JAMA paper in 2006 (Kohane, I. S., Masys, D. R. & Altman, R. B. The incidentalome: a threat to genomic medicine. Jama 296, 212-215, doi:296/2/212 [pii] 10.1001/jama.296.2.212 (2006).)

The full rationale was given in a paper I wrote recently (Kohane, I. S., Hsing, M. & Kong, S. W. Taxonomizing, sizing, and overcoming the incidentalome. Genetics in medicine : official journal of the American College of Medical Genetics, doi:10.1038/gim.2011.68 (2012)) with the key paragraph:

The incidentalome can be taxonomized into four components. In order of increasing challenge, there is first, the substantial proportion of “textbook cases” of mutations documented to cause human disease in a highly penetrant Mendelian fashion, but they are incorrectly annotated in the databases. The second is the technical or measurement error rate in genome-scale sequencing. Third is the incorrect assignment of prior probabilities for much of our genetic and genomic knowledge.
The fourth derives from testing multiple hypotheses across millions of variants. We will describe here the nature of these components, provide rough estimates for the magnitude of the problem, and point out existing approaches that will serve to control the growth of these aspects of the incidentalome. First, however, it is helpful to understand the magnitude of the interpretive challenge and the risks of false-positives by performing an example analysis of a whole-genome sequence (WGS) using the genomes of nine individuals of European descent sequenced by Complete Genomics....

The name itself was a turn on the term incidentaloma which is the way radiographic incidental findings are glibly described by docs.