Search

Background. Direct-to-consumer genetic testing companies provide consumers' genetic risk scores for common diseases based on genotype. Single nucleotide polymorphism (SNP)-associated risk estimates published by genome wide association studies are the most common source of genotype-driven risk information for common diseases. However, the risk estimate of any given SNP varies depending on the source population and study design of the original publication. An important factor in establishing clinical validity for genetic testing of common disease is the consistency in genetic risk scoring between direct-to-consumer companies. Such an association however, has not been well described. While small-scale studies looking at individual sample results between direct-to-consumer companies have been performed, to our knowledge, no large-scale studies aiming to measure the consistency in risk scoring have been reported. Methods. A genotyped cohort of 834 individuals was used to calculate the equivalent genetic risk score that would be produced by the direct-to-consumer genetic testing companies 23andMe and DeCODE Genetics for two diseases, type 2 diabetes (T2D) and age-related macular degeneration (AMD). These scores were compared to comprehensive academic SNP panels to look at the consistency between the risk scoring of different companies and academic literature. Results. Our results showed that although the genetic risk scores calculated based on different SNP risk panels (23andMe, DeCODE Genetics and academic) were significantly correlated, (r2 = 0.46-0.66 for T2D and r2 = 0.30-0.70 for AMD), the levels of correlation were far from appropriate to establish the clinical utility of these SNP-based genetic scores. In addition, the ranges of the estimated genetic scores varied substantially among these three different SNP risk panels with a greater number of SNPs utilized roughly correlating with increased range in risk score. Conclusion. Significant differences in the number of SNPs used to calculate risk score as well as selection of SNP risk estimates are the primary causes of inconsistency in risk scoring between direct-to-consumer companies. To improve consistency direct to consumer genetic testing companies need to incorporate into their calculations more recently published SNP associations, use consistent SNP effect sizes and use similar numbers of SNPs.