Redefining cardiovascular safety: A dual-pathway, mechanistic in silico screen for the detection of trafficking-mediated and "hidden" cardiotoxicity.

BACKGROUND

The Comprehensive in vitro Proarrhythmia Assay (CiPA) shifted cardiovascular safety assessment from binary hERG testing toward mechanistic multichannel integration. A key blind spot remains: many workflows treat ion-channel density (Gmax) as static and therefore miss compounds that prolong QT by reducing hERG surface expression rather than by acute pore block.

METHODS

We present Biosentry In Silico v1.0, a dual-pathway in silico screening framework that combines (i) acute multichannel pore block (Hill-based concentration-response) with (ii) a steady-state trafficking scalar, implemented as a closed-form equilibrium reduction in functional hERG density at the modeled exposure. Trafficking inputs are handled using explicit evidence states (parameterized, provisional, data-insufficient) within a multi-tier routing workflow; downstream AP outputs use a simplified screening qNet proxy rather than a regulatory-standard CiPA integral.

RESULTS

As a proof-of-concept, a mechanistic ten-compound set spanning balanced blockers, trafficking-dominant agents, and classical proarrhythmic exemplars showed internally consistent behavior: the framework detected strong trafficking-mediated "hidden" liability for Probucol and routed borderline Pentamidine to follow-up, while preserving low-risk labels for established balanced blockers under multichannel integration. On the evaluable CiPA reference subset (n = 21), dual-pathway labeling improved specificity for high-risk detection from 0.923 to 1.000 at fixed recall (0.500), driven by removal of one balanced-blocker false positive rather than increased sensitivity to CiPA High labels.

CONCLUSIONS

A steady-state trafficking scalar offers a practical proof-of-concept route to incorporate chronic hERG surface-density effects into CiPA-style screening without full protein-lifecycle kinetics. In the present benchmark, the empirical gain is false-positive reduction at unchanged recall; broader detection of trafficking-mediated High-risk labels remains limited by sparse IC50,traf coverage. This framework is intended for research-stage preclinical screening and assay prioritization, not as a standalone clinical predictor.