In silico prediction of the lipid-lowering efficacy of the oral PCSK9 inhibitor enlicitide using a mechanistic computational model.

BACKGROUND AND AIMS: Regulatory agencies increasingly support model-informed drug development, including quantitative systems pharmacology (QSP) models, but confidence in model-based predictions for decision making remains limited. We previously developed a mechanistic QSP model of atherosclerosis to simulate LDL-C responses to lipid-lowering therapies, which was extensively calibrated and retrospectively validated using published data from prior clinical trials. The recent completion of two phase 3 trials of the oral PCSK9 inhibitor enlicitide decanoate, CORALreef Lipids (NCT05952856) and CORALreef AddOn (NCT06450366), provided an opportunity for prospective validation, the highest standard for establishing predictive credibility.

METHODS

The model was recalibrated using phase 1 PK/PD data and phase 2b LDL-C response data to incorporate enlicitide's effect. Predictions for both trials were generated and publicly posted, independently of the sponsor and prior to any data release.

RESULTS

For CORALreef Lipids, the predicted week-24 LDL-C reduction was -54.5% (95% PPI: -56.1%, -53.0%), closely matching the subsequently reported -55.8% (95% CI: -60.9%, -50.7%). For CORALreef AddOn, predicted week-8 LDL-C reductions were -59.4% (95% PPI: -65.7, -53.4) with enlicitide, -18.0% (-20.1, -15.8) with bempedoic acid, -24.6% (-26.6, -22.7) with ezetimibe, and -39.8% (-42.8, -36.9) with their combination; comparison with actual trial outcomes is pending trial readout.

CONCLUSIONS

This first successful prospective validation of our mechanistic QSP model against a large cardiovascular phase 3 trial, confirms the model's credibility. It supports the contribution of in silico trials to accelerate the development and evaluation of innovative lipid-lowering therapies to improve cardiovascular outcomes.