Relevance of in vitro SCREENIT results for drug-induced QT interval prolongation in vivo: A database review and analysis

Introduction: The use of an isolated rabbit heart model (SCREENIT) to predict drug-induced QTc prolongation in animals was assessed using hERG and telemetry data. Purpose: We compiled data from (i) hERG assay (IC50s), (ii) SCREENIT assay (APD60) and (iii) in vivo non-rodent telemetry studies (QTc interval) and evaluated the reliability of APD60 to fit with IC50s and QTc prolongation using the ratio to free plasma level (FPL). Eighty-two compounds were separated into three classes based on hERG IC50s (class I: IC50s <= 1 mu M, n=7; class II: IC50s> 1 mu M to <= 10 mu M, n=15; class III: IC50s>10 mu M, n=60). Results: Three class I compounds did not prolong QTc at the FPL equivalent to their IC50s (43% hERG false positives). There were no false positives in SCREENIT. Six class 11 compounds prolonged the QTc interval. Results showed 40% hERG false negatives and no SCREENIT false negatives. Nine compounds had no effect on QTc, and two prolonged APD60 at an equivalent concentration/FPL (13% false positives). Three class III compounds prolonged QTc at an FPL lower than maximum SCREENIT concentrations (5% false negatives). Four other compounds generated SCREENIT false positive results (7%). Conclusion: SCREENIT increased the predictability of preclinical results for QTc prolongation without generating any false positive results in class 1(13% in class 11). Making decisions without isolated heart data increases the risk for eliminating efficient drugs displaying hERG inhibition. (C) 2008 Elsevier Inc. All rights reserved.