By Brent Kious, Anna Docherty, Leslie Francis, Teneille Brown, Jeffrey Botkin, Douglas Gray, Brooks Keeshin, Louisa Stark, Brieanne Witte, and Hilary Coon
Companies that offer direct-to-consumer genetic testing (DTC-GT) will soon be able to provide scores that estimate suicide risk.
Our recent study, appearing in Genetics in Medicine, the official journal of the American College of Medical Genetics and Genomics (ACMG) raises ethical questions about how these risk scores will be understood.
The causes of suicide are complex. Many risk factors intersect: psychiatric symptoms like depression, different life stressors, family history, access to lethal firearms, and substance abuse, among many others.
Still, death from suicide is strongly heritable. Roughly 50% of the variation in the risk of death from suicide can be attributed to genetics. Like most psychiatric behaviors, however, death from suicide has only rarely been associated with single genetic variants. Risks of suicide are polygenic.
Recent genome-wide association studies have been published about the risk of death from suicide. These studies allow researchers to estimate polygenic risk scores—scores based on the tiny and interrelating ways that many genetic variants contribute to variation in suicide risk.
Polygenic risk scores (PRS) have been calculated for many traits from large genome-wide association, helping to reveal important genetic contributions to risk at the population level.
At an individual level, however, PRS can only estimate an individual’s relative increase or decrease in the genetic contribution to overall risk for a condition compared to the population. Because this genetic risk interacts with many other complex risks in ways we have yet to understand, these estimates of the contribution of genetic risk alone cannot predict risks for any particular individual.
But do individuals understand these limitations? Our study gathered focus groups of people with either a personal history of suicide attempt or a family history of suicide. We asked these individuals how they perceived the significance of these risk scores, both personally, and for their family members.
Their responses demonstrated that this information has both potential risks as well as benefits.
Many welcomed the possibility, anticipating that learning of an increased PRS would encourage seeking preventative care and avoiding other behaviors that further increase risk. Our participants also surmised that widespread availability of suicide PRS for could reduce stigma. Several thought that obtaining suicide PRS would compel family members or others to regard suicide and associated mental health concerns as medically legitimate problems.
On the other hand, some participants worried that suicide PRS might increase stigma, especially if this information became available to others such as schools, employers, or potential romantic partners. Many also raised concerns about negative psychological effects of suicide PRS: increased self-stigma, resignation toward suicide because of the perception that it is in some way fated to genetics, and, for family members of individuals with higher risk scores, increased anxiety and “helicopter parenting.”
Most importantly, participants strongly supported extensive education about the meaning of PRS reports and suggested that PRS testing should happen only with clinical guidance, such as genetic counseling.
Of notable concern, our participants viewed the information in binary black and white terms, rather than as a continuum. Given the unique concerns that suicide risk information could actually trigger suicide, there is good reason to treat this information with great care.
However, with the relatively hands-off approach of the Food and Drug Administration for genetic tests for health risks, PRS suicide scores could enter the market without FDA pre-market review, as 23andMe’s recent PRS for diabetes did. PRS scores for suicide may eventually prove useful for helping to predict risk and suggest treatment or environmental changes to prevent suicide. However, at present, they have no utility for individual clinical management. In this sense, the risk scores themselves might pose a significant risk.