Abstract
According to the FDA Guidance for Industry on Clinical Drug Interaction (DDI) Studies with Combined Oral Contraceptives (COCs), sponsors are expected to conduct dedicated clinical DDI studies if in vitro findings suggest weak or moderate CYP3A induction, while concomitant use of COCs with strong inducers should be avoided. The guidance further suggests that a negative DDI result for drospirenone (DRSP) may be extrapolated to other progestins that are less sensitive to CYP3A modulation, such as levonorgestrel (LNG). This approach assumes that DDI-mediated changes in exposure directly translate into clinical efficacy across progestins. To evaluate the validity of this assumption, we established a quantitative link between dose, exposure, and response (Pearl Index [PI] and ovulation rate [OR]) via an integrated model-based meta-analysis, physiologically based pharmacokinetic, and pharmacokinetic/pharmacodynamic (PK/PD) modeling and simulation approach using data from 51 clinical studies in 36,040 women receiving LNG or DRSP. COCs containing LNG and DRSP were selected because they represent clinically relevant progestins at the lower and the upper end of the fraction metabolized via CYP3A4. The results of our analysis show a moderate correlation (Pearson's r = 0.52, 95% CI 0.46-0.58, P < 0.001) between PI and OR, which enables the use of OR as an ethically measurable endpoint, even at subtherapeutic doses/exposures, to predict efficacy outcomes. They further show that DDI-induced changes in exposure do not directly translate into clinical response. Therefore, DDIs with COCs should be interpreted in a PK/PD rather than a PK-only context. The quantitative framework developed in this study can serve as the scientific basis to do so.