Abstract 17484: Genome Wide Association Study of Statin-Induced Myalgias using Natural Language Processing
Introduction Genetic polymorphisms conferring increased risk of myopathy to statins therapy have been identified by performing genome-wide association studies (GWAS). Identifying cases of myopathy from EMR involves meticulous review of hundreds of records for thousands of patients for a typical GWAS.
Hypothesis We hypothesized that natural language processing (NLP) is an efficient tool to detect genetic variation and common genetic variants are associated with statin induced myalgias.
Methods We conducted an EMR based GWAS of statin-related myalgias. We developed an electronic phenotyping algorithm to detect cases and controls that included billing codes, lab data and NLP of unstructured clinical text using predictive and negative key terms. Algorithm was validated by manual review of sample cohort of patient to achieve 89.9%, 90.9, 98.6%, and 81.7% of sensitivity, specificity, negative predictive value and accuracy respectively for detection of myopathy cases. Validated myopathy algorithm was implemented on 6922 patients with 1.76 million individual EMR notes. NLP was able to narrow the probable cases to 1239 patients with 4003 records. True cases were identified after manual review of probable case records identified by NLP. Manual review was facilitated by highlighting the NLP key terms and this represented only 0.2% of the initial individual EMR notes. Study covariates and controls on statins therapy were also electronically ascertained. Genetic association analysis was performed using PLINK on 568 cases and 2685 controls on statin therapy adjusting for age and sex.
Results and Conclusions Our results highlight 9 SNP which reached a statistical significance (P<5×10−6) mediating various cellular functions such as membrane transport. Our results highlight the potential application of NLP as an efficient and reliable tool identifying statin induced myopathy from EMR linked DNA biorepository to perform large-scale GWAS with reduced cost and time.
- © 2012 by American Heart Association, Inc.