MICE models: superior to the HERG model in predicting Torsade de Pointes

Abstract

Drug-induced block of the cardiac hERG (human Ether-à-go-go-Related Gene) potassium channel delays cardiac repolarization and increases the risk of Torsade de Pointes (TdP), a potentially lethal arrhythmia. A positive hERG assay has been embraced by regulators as a non-clinical predictor of TdP despite a discordance of about 30%. To test whether assaying concomitant block of multiple ion channels (Multiple Ion Channel Effects or MICE) improves predictivity we measured the concentration-responses of hERG, Nav1.5 and Cav1.2 currents for 32 torsadogenic and 23 non-torsadogenic drugs from multiple classes. We used automated gigaseal patch clamp instruments to provide higher throughput along with accuracy and reproducibility. Logistic regression models using the MICE assay showed a significant reduction in false positives (Type 1 errors) and false negatives (Type 2 errors) when compared to the hERG assay. The best MICE model only required a comparison of the blocking potencies between hERG and Cav1.2.

Figure 1. IC₅₀s of the 55 drugs included in the dataset for hERG, Cav1.2 and Nav1.5.

Symbols indicate mean values and lines the 95% confidence intervals of data. Values for the three channels are plotted as a function of drugs numbered in decreased order of potency. The drug order for each channel is indicated in Supplementary Table 3.

Figure 2. ETPC indexes of torsadogenic and non-torsadogenic drugs.

Terfenadine is concordant and verapamil is discordant with the hypothesis that potent hERG blockers are torsadogenic. (a) IC₅₀ values for hERG, Nav1.5 and Cav1.2 and the maximum effective free plasma concentration. (b) Plot of the ETPC Indexes for hERG (indicated at the interception of blue and orange segments), Cav1.2 (point at the end of the blue segment) and Nav1.5 (point at the end of the orange segment). Dashed lines indicate the region defined by Redfern's ETPC index criteria of 30 for torsadogenic drugs. (c) The ETPC Indexes for hERG, Cav1.2 and Nav1.5 channels of the 32 +TdP and 23 −TdP drugs included in the dataset. The profile was explained in the legend of Fig. 1. Drugs were numbered for clarity in decreased order of IC₅₀/ETPC values. +TdP drugs included: (1) ibutilide, (2) quinidine, (3) thioridazine, (4) flecainide, (5) halofantrine, (6) droperidol, (7) terodiline, (8) bepridil, (9) procainamide, (10) terfenadine, (11) nilotinib, (12) methadone, (13) sotalol, (14) moxifloxacin, (15) cisapride, (16) paliperidone, (17) haloperidol, (18) sparfloxacin, (19) astemizole, (20) dofetilide, (21) disopyramide, (22) sertindole, (23) clozapine, (24) chlorpromazine, (25) voriconazole, (26) pimozide, (27) sunitinib, (28) cilostazol, (29) solifenacin, (30) paroxetine, (31) risperidone, and (32) amiodarone. −TdP drugs included: (1) piperacillin, (2) verapamil, (3) metronidazole, (4) ceftriaxone, (5) linezolid, (6) telbivudine, (7) phenytoin, (8) ribavirin, (9) lamivudine, (10) diltiazem, (11) raltegravir, (12) saquinavir, (13) mibefradil, (14) duloxetine, (15) donepezil, (16) pentobarbital, (17) sitagliptin, (18) dasatinib, (19) diazepam, (20) mitoxantrone, (21) nifedipine, (22) nitrendipine, and (23) loratadine. Dashed lines have the same meaning as before.

Figure 3. Relationship between ETPC Indexes and the fraction of torsadogenic drugs.

Panels (a–d) show the fraction of +TdP drugs (P(+TdP)) present in each decade of −log transformed ETPC indexes for hERG, Cav1.2 and Nav1.5, and CavD respectively. Note that CavD is equivalent to log(hERG IC₅₀/Cav IC₅₀).

Figure 4. The MICE approach results in better predictive Models for TdP.

(a) Calculated torsadogenic risk under Models 1 and 5 based in cross-validated probabilities. Boxes indicated the 25, 50 and 75 percentile; whiskers indicate the 10 and 90 percentiles, and filled diamond, mean values. Arrows indicate the probabilities of 0.47 and 0.46 at the J-index for Models 1 and 5 respectively. (b) ROC analysis of Model 1 and 5. The arrows indicate the cutoff point associated with the J-index. The dashed line indicates the performance of a Model that does not discriminate. (c) 2 × 2 Contingency tables for both models. Numbers in parenthesis indicate the area under the curve ± sem. Misclassified drugs are indicated in the tables at the bottom.