Predicting venous thromboembolism risk from endogenous thrombin potential-based normalized activated protein C sensitivity ratio assay: toward early assessment of combined oral contraceptives.

BACKGROUND

Combined oral contraceptives (COCs) increase venous thromboembolism (VTE) risk, depending on estrogen type, dose, and progestin. While epidemiological studies provide insight into these risks, they require years to complete. The normalized activated protein C sensitivity ratio (nAPCsr), a standardized assay of acquired activated protein C resistance, has emerged as a potential biomarker of COC-induced VTE risk.

OBJECTIVES

To develop a population-based in silico model predicting VTE risk associated with various COC formulations based on their mean nAPCsr values.

METHODS

We analyzed 200 plasma samples from non-COC users and 257 from users of 9 different COCs. We constructed an exponential model to correlate the mean nAPCsr of 5 COCs with their available population-based VTE relative risk, as extracted from a published meta-analysis. We assessed model performance using R 2, Spearman's rank correlation coefficient, and the root mean square error, and performed a sensitivity analysis by excluding COC nonusers. We then estimated population-based VTE risks for the 4 COCs not used in model construction.

RESULTS

The model demonstrated high predictive accuracy (R2 = .96; root mean square error = 0.21; Spearman correlation coefficient = 1) and remained robust despite group size imbalance. Predicted VTE risks for ethinylestradiol 30 μg with dienogest 2 mg, ethinylestradiol 20 μg with drospirenone 3 mg, estradiol 1.5 mg with nomegestrol acetate 2.5 mg, and estetrol 15 mg with drospirenone 3 mg were 4.36, 3.43, 1.50, and 1.45, respectively, consistent with or complementary to existing epidemiological evidence.

CONCLUSION

Our model, based on mean nAPCsr, provides a reliable, biomarker-based approach for predicting population-based COC-related VTE risk. This strategy could help shorten the time between product launch and population-based risk assessment.