FDA Researchers Reassess Cardiac Ion-Channel Protocols and Active Metabolites
New patch-clamp measurements better aligned selected drug profiles with reported clinical electrical effects. Adding active metabolites substantially narrowed the hERG safety margins for astemizole and risperidone, under an explicit competitive-binding assumption.
Revisiting Mismatches Between Channel Data and Clinical Findings
FDA researchers reassessed cardiac ion-channel measurements for drugs whose published nonclinical profiles did not readily explain reported electrocardiographic changes or proarrhythmia. The study tested whether more physiologically relevant recording conditions and attention to active metabolites could resolve particular discrepancies. It reports research findings and the authors’ interpretation; the paper explicitly states that it does not represent FDA views or policies.
The work sits within the Comprehensive in vitro Proarrhythmia Assay, or CiPA, approach, which considers several cardiac currents together. Blocking hERG potassium channels can delay repolarization, but concurrent inhibition of inward calcium or late sodium currents may modify the resulting electrical effects. A drug’s profile therefore depends on both the potency and relative balance of its channel interactions.
The investigators used engineered HEK293 cells expressing human hERG or sodium channels and CHO cells expressing human calcium channels. Manual whole-cell patch clamp was performed near physiological temperature, at 37 ± 2°C. The protocols incorporated elements consistent with existing ICH S7B questions-and-answers best practices. They did not create a new regulatory requirement.
Protocol and Exposure Checks Changed Selected Drug Profiles
The first comparisons covered chloroquine, lopinavir and ritonavir. Chloroquine’s nominal hERG half-maximal inhibitory concentration was 0.92 μM, versus 6.9 μM in a cited earlier study. It also inhibited calcium and peak sodium currents at concentrations closer to its hERG effect than the earlier profile suggested. The authors related this broader pattern to reported QT, PR and QRS changes.
For lopinavir and ritonavir, the new measurements showed stronger inhibition of the calcium current during the repolarizing ramp than of hERG. This differed from the ordering in the earlier comparison. These channel experiments generally used four to seven cells per concentration. The clinical findings were drawn from published studies rather than a newly recruited cohort.
Vanoxerine provided another test of the balance between currents. Its nominal hERG inhibitory concentration was 6.7 nM, compared with 81.5 nM for the calcium step current and 154.4 nM for late sodium current during the ramp. This preferential hERG effect was more consistent with the reported clinical proarrhythmia than a nearly balanced hERG–calcium profile would suggest. Verapamil and ranolazine supplied low-torsade-risk comparisons with different overlapping inward-current effects.
Analytical measurements also showed that nominal concentration could overstate exposure reaching the cells. Vanoxerine lost approximately 58% in the hERG perfusion system; using measured concentrations reduced its hERG inhibitory concentration to 2.2 nM. Lopinavir and ritonavir required attention to precipitation as well as loss. Concentrations were not verified for every drug in every channel assay, so the hERG loss estimates cannot simply be applied to the calcium and sodium measurements.
A Metabolite Can Change the Safety-Margin Interpretation
The study separately tested astemizole, risperidone and clarithromycin with their major metabolites. These hERG concentration-response measurements used three to nine cells per concentration. The analysis compared inhibitory concentrations with reported unbound peak clinical concentrations to calculate exposure-based safety margins.
Astemizole alone had a reported margin of 210, while its metabolite desmethylastemizole had a margin of 11. Although the metabolite was slightly less potent at hERG, its higher clinical exposure made its contribution consequential. Combining the parent and metabolite reduced the estimated margin to 3.
For risperidone, the corresponding estimates were 160 for the parent, 46 for paliperidone and 11 for their combination. Clarithromycin differed: its parent margin was 29 and the metabolite margin 505, with a combined value of 23. Thus, adding a metabolite did not produce the same degree of change for every drug.
The combination calculation assumed that each parent–metabolite pair competed for overlapping channel-binding sites. It was a model-based estimate of their joint effect, rather than a clinical outcome measurement or a direct mixture experiment. The available clinical concentrations also did not permit margins to be standardized to the concentration producing a 10-millisecond QT change.
The findings identify specific sources of translational disagreement: recording conditions, channel state, assay exposure and metabolite activity. Remaining uncertainties include assay-to-assay variability, stimulation-frequency dependence and how much inward-current inhibition is needed to offset hERG-associated risk. The authors also discuss how immature ion-channel expression in stem-cell-derived cardiomyocytes can change responses to multichannel drugs.