Enhancing Disease Risk Prediction through Electronic Health Records and Polygenic Scores

Recent research indicates that phenotype-based disease risk prediction utilizing electronic health record (EHR) data can effectively complement polygenic risk scores (PRS) for various common health conditions. The study, led by Andrea Ganna from the University of Helsinki, highlights that EHR-based risk scores can transfer across different healthcare systems and provide independent information from polygenic scores, thereby improving disease risk prediction. The research analyzed data from 845,929 individuals aged 32 to 70 from Finland's FinnGen, the UK Biobank, and the Estonian Biobank, employing 234 phenotype codes over two decades.

Disease Risk Prediction from Electronic Health Records Complements Polygenic Risk ScoresNatureNatureNatureNature

The study focused on 13 common conditions, including knee osteoarthritis, asthma, diabetes, and various cancers. The findings suggest that combining EHR-derived phenotype risk scores (PheRS) with PRS can yield significant benefits in predicting disease risk. The authors noted that both PheRS and PRS are independent of age and sex, providing additional layers of risk assessment.

Disease Risk Prediction from Electronic Health Records Complements Polygenic Risk Scores

The predictive performance of PheRS varied depending on the specific condition and cohort. For instance, while PheRS for asthma and knee osteoarthritis showed reduced predictive capabilities in the UK Biobank, they provided valuable cross-cohort insights for conditions like major depressive disorder and type 2 diabetes. Conversely, the PheRS for colorectal, prostate, and breast cancers demonstrated poor predictive capabilities, attributed to the limited number of affected individuals in the cohorts studied.

Disease Risk Prediction from Electronic Health Records Complements Polygenic Risk Scores

The authors advocate for further enhancement of PheRS by incorporating a broader range of data types from EHRs and examining longitudinal data across larger, more diverse patient groups. They emphasize the potential for integrating diagnosis history into clinical practice, which could lead to more accurate risk estimations.

Disease Risk Prediction from Electronic Health Records Complements Polygenic Risk Scores

The research team concluded that their findings underscore the importance of integrating EHR data with genetic information for disease risk prediction. They stated, "Information already available from the EHR provides orthogonal information to polygenic scores and could be a cost-effective approach for risk estimation." This perspective aligns with the growing interest in personalized medicine, where tailored risk assessments can lead to improved patient outcomes.

Disease Risk Prediction from Electronic Health Records Complements Polygenic Risk Scores

Despite the promising results, some experts caution against over-reliance on these predictive models. Critics argue that while EHR data can enhance risk prediction, it may not capture all relevant clinical nuances, particularly in diverse populations. Additionally, the variability in predictive performance across different cohorts raises concerns about the generalizability of the findings.

Nature

The study's limitations include the variability in predictive performance across different cohorts and conditions, which may affect the applicability of the findings in broader clinical settings. Moreover, the reliance on existing EHR data may overlook important factors that could influence disease risk, necessitating further research to validate these models in diverse populations.

Disease Risk Prediction from Electronic Health Records Complements Polygenic Risk Scores

Future research is expected to focus on refining PheRS by integrating more comprehensive EHR data and exploring the dynamics of patient health over time. As the field moves towards the clinical application of polygenic scores, the integration of EHR data will likely play a crucial role in enhancing disease risk prediction strategies.

Disease Risk Prediction from Electronic Health Records Complements Polygenic Risk Scores