The efficacy of obesity polygenic risk scores in an autistic population

Polygenic risk scores (PRS) bear great promise in understanding complex diseases and improving clinical diagnoses, but the competency of these risk scores in different populations is known to vary significantly, especially between those of European ancestry and those of other ethnic ancestries. However, the applicability of these risk scores across populations different by disease, instead of ethnicity, is poorly understood. A current and largely unexplored population for the accuracy of PRS is that of individuals with autism. Combined with the increased prevalence of obesity in autistic populations, we seek to evaluate the difference in efficacy of PRS for obesity in autistic versus non-autistic populations. We show that genetic variants strongly associated with obesity in non-autistic populations are significantly less representative of the disease in autistic populations. Rather, these cases of obesity phenocopies in patients with autism may follow a different and non-conventional mechanism of action involving the regulation of oxytocin in the brain among other potential behavioral factors. Our findings elucidate the limits of PRS across populations contrasting by disease and suggest that obesity may be regulated differently in individuals with autism as compared to those without autism.