AUGUST 20, 2024
Polygenic risk scores are significantly less predictive of opioid dependence risk than environmental factors—in particular, education level and annual household income, according to new study results.
The findings indicate that genetic assays intended to predict individual risk for opioid use disorder—such as the AvertD test, developed by AutoGenomics Inc. and recently approved by the FDA—are premature and should be viewed with skepticism, and that the interplay between
Polygenic risk scores are significantly less predictive of opioid dependence risk than environmental factors—in particular, education level and annual household income, according to new study results.
The findings indicate that genetic assays intended to predict individual risk for opioid use disorder—such as the AvertD test, developed by AutoGenomics Inc. and recently approved by the FDA—are premature and should be viewed with skepticism, and that the interplay between environmental and genetic factors is likely to have far more predictive utility than genetics alone.
The authors analyzed data from 1,958 adults of European ancestry, deriving polygenic risk scores (PRS) based on a large-scale multi-trait analysis of genome-wide association studies (GWAS) of opioid use disorder. The analysis took into account 1 million single-nucleotide polymorphisms (SNPs), or genetic mutations, in arriving at PRS for opioid dependence.
Of those analyzed, a total of 420 participants had a lifetime diagnosis of opioid dependence. An analysis of the relative importance of the factors associated with opioid dependence revealed that annual household income affected one-fourth of these cases (25.9%). Education level accounted for 22%. Age and being in a partnered relationship explained 18.7% and 15.3% of the observed variance, respectively. The PRS lagged significantly behind these factors in predictive value, accounting for just 8% of the variance in the analysis.
Joseph Deak, PhD, an associate research scientist at Yale School of Medicine, in New Haven, Conn., and co-lead author of the study, said the findings were in line with expectations.
“We’ve made a lot of advancements in our understanding of the underlying biological mechanisms relating to risk for opioid dependence,” Deak said. “And because of the opioid use epidemic—fueled in large part by questionable prescription practices—there’s naturally been a keen interest in the question of whether we can predict, in advance, who will and who will not be at risk for developing dependence before prescribing opioids.
“Most experts would agree that that’s a good idea in principle, but we’re simply not there yet in terms of the science, and it could potentially be harmful to move ahead before the science can support using genetic assays to predict opioid dependence risk on the individual scale.”
Deak pointed out that this study included a PRS generated from one of the best available GWAS of opioid use disorder to date: a GWAS capturing the effects of millions of variants across the genome, while the FDA-approved AvertD test takes into account only 15 SNPs.
“Our findings clearly demonstrate that environmental factors correlated much more strongly with opioid dependence risk than a state-of-the-art PRS capturing SNPs from across the genome, so candidly, it’s hard to understand the scientific rationale for the FDA’s approval,” he said. “[AutoGenomics], in response to criticism from the medical community, has said that, ‘Well, you don’t have our proprietary algorithm.’ But even of the 15 SNPs that they’re looking at, we know that some of them aren’t necessarily meaningful to begin with.”
AutoGenomics could not be reached for comment.
Abraham Palmer, PhD, a professor and the vice chair of basic research in the Department of Psychiatry at the University of California, San Diego, where he studies the relationships between genes and behavior, praised the study’s methodology.
“People often imagine human behavior as being either genetically determined or environmentally determined, but of course both play a role,” he said. “The authors have acknowledged that in their study design, and they’re examining the relative contributions of both environmental and genetic factors to the risk for developing opioid use disorder, as well as the interactions between the two. That’s an exciting and important question.”
Palmer added that there are methodological limitations to PRS prediction models that must be considered when using them to draw conclusions, limitations that Deak and his co-authors acknowledge.
“For one thing, the data that we use to train these polygenic prediction models are ancestry group-dependent,” he said. “So, if you have a model that’s trained on data from people of European ancestry, then your scores will be most accurate in other people of European ancestry, and the quality of the prediction will degrade as you move further away from that group.”
Palmer concurred with Deak that while GWAS and PRS are potent tools for population-level research, using genetic assays of any kind to predict individual risk is premature.
“It’s foolish to try to do prediction without taking into account everything we know,” he said. “So only using genetic or only environmental information to make predictions—both methods are probably suboptimal. We need methods that can take both into account, across diverse populations. And as the authors say in this paper, we’re simply not there yet.”
The research was published in Psychological Medicine (2024;54[8]: 1779-1786).
—Ajai Srinivas
