Genetics professionals are skeptical about many tests that claim to predict complicated, genetically-mediated outcomes like what sport your child will excel at or which common illness is most likely to kill you in the end. There’s broad agreement that some tests being marketed today lack validity. It’s the geneticists’ reaction to this issue that divides the field. Those more concerned with protecting the public, even from themselves, look at tests with uncertain value and ask, “what’s the benefit?” Those more concerned with access and opposed on principle to regulation ask, “what’s the harm?”
Genetic tests for susceptibility to opioid addiction are a case in point. These tests claim to identify which individuals are most likely to become addicted to those pills. Aggressive marketing of such opioid addiction risk assessments has popped up in response to the growing concerns over the addiction epidemic. But many of these tests may be taking advantage of opioid fears rather than sound science.
In 2016, opioids killed more than a hundred people a day, and overdoses displaced car accidents as the most likely cause of accidental death in the U. S. This national disaster arose hand-in-hand with a huge increase in the use of prescription narcotics for pain control. While this is well documented, physicians still struggle to find the balance between denying relief to people in pain and feeding an epidemic that destroys lives. That’s the rationale behind the marketing of genetic testing as a way of sorting the population into those who should and those who should not be exposed to narcotics.
At best, this approach is no panacea, because the tests only claim to identify who is at greater risk of addiction, not that they can pinpoint exactly who will and who won’t get addicted. Actual outcomes, as stands to reason, are based on more than just genetic risk factors. Most people would agree that personality, upbringing, and experience also play a role in who ends up misusing their pain medications, while epidemiology suggests that the quality of medical care and access to opioids are important as well. So even in the best case scenario, genetic testing for addiction risk takes a complex decision and reduces it to the role that genes play — or more precisely, to what we currently understand about the role that genes play — which is not that much. And that’s when the test does what it claims to do. That’s the best case scenario.
In 2016, Proove Biosciences made $28 million selling its opioid risk assessment test, despite complaints from the genetics community that there was no evidence to show the test was either reliable or effective. The company also faced allegations that its marketing strategy amounted to a kickback scheme, is now being restructured under a court order, and may have criminal liability. This sordidness distracts from what is arguably a more important question: Do we need regulation to restrict sales of tests that have not been validated, or should we avoid red tape that is going to slow down healthcare innovation? Can we rely on the market to sort this out?
Here to raise that question again is Prescient Medicine, which claims that its LifeKit® Predict test can identify those at increased risk with 97 percent sensitivity and 88 percent specificity using an algorithm based on 16 genes. Those numbers sound impressive, although they are also somewhat misleading, as the paper they cite as evidence itself suggests. Most of the people tested in the study were addicts, but practically speaking, it is not much use to test opioid addicts to see if they are susceptible to opioid addiction (spoiler: yes, they are). In the real world, where most people tested are not addicts, there will be more false positives — more people identified incorrectly as at risk — and these individuals may be stigmatized and even denied treatment as a result.
But this is taking the results at face value, and there are good reasons to be more skeptical than that. The study was done by doctors affiliated with the company, and tested in a sample that was way too small, as many commentators have pointed out. Furthermore, the risk score is based on 16 genes, none of which has been shown to have more than a small effect on the risk of addiction. In a recent article, Prescient Medicine’s CEO Keri Donaldson credited the test’s effectiveness to a proprietary algorithm that assesses all 16 genes in combination. Here’s a little back-of-the-envelope math on that: Looking at one variable spot in each of 16 genes gives you more than 43 million possible combinations. Prescient’s algorithm, however, was constructed by looking at results from a grand total of just 37 individuals.
Hypothetically, two things are possible. One, the genes don’t interact, and each increases or reduces risk in a straightforward fashion, like adding up a column of numbers. In that case, you could use a modest sample (but not 37; 37 is ridiculous). However, the overall value of the test would be limited, because none of these genes has a large impact. Two, the genes do interact, in which case you’ve looked at 37 out of 43 million possible scenarios and … well, you have more work to do. Either way, it is difficult to believe that the results they are describing will be borne out in wider use.
All of this brings us back to our original question. Should we ask, “what’s the benefit?” which places the burden of proof on the company and creates a higher bar to entry, discouraging innovation and reducing access for those who feel this test might be worth trying. Or should we ask, “what’s the harm?” which looks to restrict only the most egregious offenders, and leaves the market to rid itself of crappy or unreliable tests.
The truth is that it is a false choice. We have to ask both questions, at the same time, all the time. We have to be open to a certain level of experimentation and let people decide how much risk they are willing to assume. But frightened families and overworked doctors should not be left on their own to sort out the value of tests like this, where the science is confusing and the consequences are clear.
“What’s the harm?” is not always a rhetorical question. Resources to fight addiction are scarce; misuse of resources costs lives. What’s more, the effect of labeling someone incorrectly as at risk of opioid addiction has never been measured, but it could have serious and lifelong implications. We don’t need to forbid it, but we do need to offer some information alternatives that don’t come straight from sales and marketing.