PM543. A Data Mining Algorithm for Personalized Medicine in Schizophrenia
Beom S. Lee, Roger S. McIntyre, James E. Gentle, Nan Sook Park, David A. Chiriboga, Yena L. Lee, Marie McPherson · The International Journal of Neuropsychopharmacology · 2016
Abstract Background: Despite opportunities provided by genome-wide association studies (GWAS) for personalized medicine, there exists no known method to predict who will or will not benefit from a specific antipsychotic drug. While conventional methods seek to identify significant genes or factors for prediction, this study proposes a data mining algorithm that integrates multiple genes and baseline profiles simultaneously, resulting in a mechanism to directly identify individuals who will response to a specific type of antipsychotic medication. Methods: The approach included revision of Latent Group Effectiveness Modeling (LGEM).1 The revised algorithm was applied to the data obtained from the CATIE2 study. The predictors were either (1) 13 single-nucleotide polymorphisms (SNPs) and 53 baseline variables or (2) 25 SNPs and the same 53 baseline variables, depending on the previous GWAS findings and data availability. The outcome variables were either (1) improvement in the Positive and Negative Syndrome Scale (PANSS) (Yes/No) or (2) completion of phase 1/1A (Yes/No). Each of those four predictor-outcome combinations was tried for each of the five antipsychotic drugs (Perphenazine, Olanzapine, Quetiapine, Risperidone, and Ziprasidone), leading to 20 prediction scenarios in total. Results: For 18 out of 20 prediction scenarios, all three performance measures were greater than .50 (sensitivity .51−.79, specificity .52−.79, accuracy .52−.74). Notably among several relevant predictions, the model provided a promising prediction for Ziprasidone for the case involving completion of phase 1/1A (Yes/No) predicted by 13 SNPs and 53 baseline variables (sensitivity .75, specificity .74, accuracy .74). Conclusion: The proposed algorithm simultaneously used both genetic information and baseline profiles to identify which individual person will benefit from a specific antipsychotic drug among the patients with schizophrenia. As this method employs a general algorithm applicable to a variety of diseases and medications, it can be easily adopted in many other clinical practices for personalized medicine.