An Exploratory, Retrospective Study of the CYPRI Score for Pharmacogenetic Testing Among Mental Health Patients

Background/Objectives: Pharmacogenetic (PGx) testing is gaining importance in optimizing psychiatric pharmacotherapy, yet routine use remains limited due to cost and unclear patient selection criteria. The CYP Pharmacogenetic Risk Score (CYPRI) is a clinical tool designed to identify psychiatric patients most likely to benefit from PGx testing, based on medication profile, adverse drug reactions (ADRs), and therapeutic drug monitoring (TDM) results. This study aimed to evaluate the clinical relevance of the CYPRI by identifying its weaknesses and gaps in a clinical setting, propose targeted modifications to address those limitations, and assess the applicability of the improved version in a routine clinical setting. Methods: In a retrospective analysis, data from 92 patients with depression at Frankfurt University Hospital were evaluated using the CYPRI score. Its association with the clinical impact of PGx testing, measured by the IMPACT score, was analyzed using ordinal regression and Receiver Operating Characteristic (ROC) analysis. Based on the findings, a revised version of CYPRI was developed and applied to the retrospective cohorts of Frankfurt and Prague. Results: The original CYPRI score was significantly associated with increased IMPACT score, suggesting its clinical value in detecting non-normal CYP2D6 and/or CYP2C19 metabolizers. However, the corrected version (hereafter referred to as CYPRI_cor), which emphasized clinically relevant pharmacokinetic factors, showed improved clinical specificity while maintaining similar discriminative performance. In the Frankfurt cohort, the area under the curve (AUC) for CYPRI_cor was 0.68 (95% CI 0.56-0.79), and in the Prague cohort, the AUC for CYPRI_cor was 0.71 (95% CI 0.60-0.81). While the overall discriminative ability in Frankfurt was slightly lower, CYPRI_cor achieved a specificity of 0.69, enabling more precise identification of patients most likely to benefit from PGx testing. A CYPRI Cut-off of ≥4 was determined to indicate clinical impact. Conclusions: The CYPRI_cor score was designed to optimize and to rule out potential limitations of the original score, particularly regarding the attribution of ADRs and the weighting of TDM results. Although the modifications did not improve discriminative performance in the Frankfurt dataset, the proposed changes remain meaningful. Prospective clinical studies need to verify the clinical utility of the CYPRI_cor.