Determining actionability of genetic findings

How to handle inadvertent findings is a major subject of discussion in the genomics community.

It has happened to all of us. A diagnostic or screening test is ordered for a specific indication and yields an entirely unexpected result. Sometimes the result is related to the initial reason for the test, but commonly, and often without welcome, it isn't.

The classic example is the chest X-ray ordered for fever, dyspnea and cough that radiology describes as “a vague 1-cm opacity in a single view, cannot rule out malignancy. Clinical correlation suggested. Additional views or chest CT may be of value in characterizing the finding further.” Occasionally, incidental findings are lifesaving, although they often lead to medical misadventures that incur cost, morbidity and mortality.

How to handle inadvertent findings is a major subject of discussion in the genomics community. Right now, the “return of results” issue is a critical point of debate for investigators conducting clinical research studies involving genome scale sequencing. The medical, ethical and legal challenges of incidental findings in routine clinical care, especially with the advent of very low-cost exome or whole-genome sequencing, are daunting and just beginning to be grappled with.

We are approaching a phase of exponential growth in use of sequencing technologies in clinical care. Yet, targeted panels of clinically useful, patent-protected genetic tests will soon be more expensive than sequencing the patient's full genome.

Already there are examples in the research community in which patients have been sequenced for one clinical purpose, only to find other potentially important abnormalities. In fact, in each of our genomes there are variations that could be considered clinically significant at some point in our lives. Some of the variations might predict response to medications and others the risk of passing on a serious condition to offspring, while still others may provide insights into future health risks.

Additionally, there is a lot of variation that amounts to “noise” in our genomes, including variations that confer low or uncertain risk for a wide variety of traits and diseases. And even with the best sequencing and interpretive technologies, the size of the human genome ensures that errors in the sequence data will be admixed with the significant findings.

The National Human Genome Research Institute has recently funded a number of Mendelian Disorders Sequencing Centers and a Clinical Sequencing Exploratory Research Program to better understand the issues surrounding the clinical application of inexpensive sequencing technologies.

Investigators from these groups have formulated a “Return of Results Consortium” to explore the particular issue of incidental findings. The first meeting was held in April 2012. Diverse opinions on the topic were expressed at this meeting; clearly best practices have yet to emerge. One approach being considered is to “bin” variants into categories of clinical importance, then present patients with incidental genomic variations that meet “actionability” criteria according in part to personal preference. However, “actionability” is rather difficult to define. For example, is a variable “actionable” simply because patients are interested in knowing about their genotype at that location? Or is the threshold of “actionability” only met when the presence or absence of a variation indicates an immediate clinical intervention to avert an immediate health risk?

Also, “actionability” is often highly context-specific (e.g., a high-risk BRCA1 gene mutation is less actionable in an 80-year-old childless male than in a 25-year-old female with three children). How complex will the binning need to be to account for patient characteristics? One way to assess the “actionability” of incidental results would be to consider them as several billion individual screening tests performed in unselected individuals. In the case of exome or whole-genome sequencing, one could argue that the screening test is not the biochemical act of measuring and calling the sequence, but the choice to actively look for and interpret the result at a given spot in the genome.

In 1968, Wilson and Jungner of the World Health Organization published what have become classic criteria for adopting a screening test for a given condition. They said the condition should be an important health problem and should have a treatment; facilities for diagnosis and treatment should be available; there should be a latent stage of the disease; there should be a test or examination for the condition which should be acceptable to the population; the natural history of the disease should be adequately understood; there should be an agreed policy on whom to treat; the total cost of finding a case should be economically balanced in relation to medical expenditure as a whole; and case-finding should be a continuous process, not just a “once and for all” project.

With minor adaptations, most of these criteria seem sensible as a general guide for deciding whether to actively interpret and return incidental variants that are discoverable on an exome or whole-genome sequence. Clearly, in a time when patients are increasingly invested in their health care, individual patient preference will play an important role in any real-world solution to this issue. However, a degree of paternalism/maternalism may be necessary when the science does not support a conclusion that any genomic finding is more than noise.