NEW YORK — Polygenic risk scores (PRS) for schizophrenia have a similar penetrance in veteran and civilian US populations, despite differences in disease prevalence, a new study has found.
Using data from the US Million Veteran Program (MVP) and the related Cooperative Studies Program (CSP) #573, researchers from SUNY Downstate Health Sciences University and elsewhere examined whether PRS for neuropsychiatric conditions developed with civilian data were also associated with risk of those conditions in a veteran cohort that is generally older, male, and has higher rates of neuropsychiatric and chronic conditions.
"Building on our previous reports that published GWAS results are robustly generalizable to the US veteran population, we have demonstrated that the penetrance of schizophrenia PRS is equivalent across VA and civilian healthcare systems, despite marked differences in absolute prevalence," SUNY Downstate's Tim Bigdeli and colleagues wrote in their paper, which appeared Wednesday in JAMA Psychiatry.
Using data from CSP #573, the researchers first examined whether phenotypes gleaned from billing data within electronic health records could be used to determine whether a person had been diagnosed with schizophrenia or bipolar 1 disorder. The CSP #573 cohort included nearly 4,000 individuals who met DSM-IV criteria for schizophrenia or bipolar 1 disorder. Of these individuals, more than 95 percent could be correctly identified as a schizophrenia or bipolar 1 disorder case based on two or more relevant phecodes.
The researchers then applied that phecode strategy to identify individuals with schizophrenia or bipolar 1 disorder within the larger MVP cohort. The MVP cohort, the researchers noted, had a higher prevalence of neuropsychiatric disease than the wider US population, as expected.
Still, the researchers found that PRS generated using data from the Psychiatric Genomics Consortium could perform similarly in the MVP cohort.
Among MVP participants of European ancestry, individuals in the top decile of schizophrenia risk as determined by PRS had 2.4-fold higher odds of being diagnosed with schizophrenia than all other risk groups, and individuals in the top decile of bipolar risk were 1.7 times more likely to be diagnosed with the condition.
The effect sizes the researchers reported fell within the range of those reported by the PsycheMERGE consortium, suggesting that the penetrance of the PRS in civilian and veteran cohorts were similar despite different recruitment approaches and exposures.
MVP participants of African ancestry, meanwhile, had higher rates of schizophrenia and major depression as compared to participants of European ancestry, which the researchers noted could reflect biases in diagnoses, self-selection for using the veteran health system, or other factors. They added that that higher prevalence was not reflected by individual-level PRS-generated risk. They noted that only MVP participants of European ancestry in the highest schizophrenia PRS decile had an absolute risk similar to that of MVP participants of African ancestry in the lowest decile.
Still, the findings suggested to the researchers that "individual-level PRS informed by large-scale genetic studies are portable across US healthcare systems and have emergent potential for risk stratification, albeit with disparate specificity across ancestries."
The researchers additionally explored whether being at increased risk of neuropsychiatric disease affected veterans' risk of other conditions and illnesses. They found high PRS for schizophrenia were also associated with an increased risk of other psychiatric-related problems like anxiety, as well as with increased risk of respiratory problems. High PRS for schizophrenia also had protective associations with hearing loss and osteoarthritis.