# Relevance and limitations of clinical follow-up in a pharmacokinetic study on direct oral anticoagulants

## TL;DR

This study found no link between blood levels of two anticoagulants and clinical outcomes in hospitalized patients over two years.

## Contribution

The study evaluates real-world pharmacokinetic data of DOACs and their clinical relevance in a prospective cohort.

## Key findings

- No significant association was found between high drug levels and bleeding events.
- Low drug levels were not linked to increased ischemic or thromboembolic events.
- The study had limited power due to a small sample size and frequent treatment changes.

## Abstract

Direct oral anticoagulants (DOACs), including apixaban and rivaroxaban, are widely prescribed for the prevention of cardioembolic stroke and the treatment of venous thromboembolism. Although routine therapeutic monitoring is not required, real-world patients often present with multiple comorbidities or drug–drug interactions that may alter drug exposure. Observational data indicate that high DOAC plasma concentrations may increase bleeding risk, whereas low concentrations may predispose individuals to thromboembolic or ischemic events. In this study, we aimed to analyze the relationship between DOAC plasma concentrations and clinical outcomes in hospitalized patients from the OptimAT cohort.

This prospective study included 200 inpatients from Geneva University Hospitals receiving apixaban (n = 100) or rivaroxaban (n = 100). Pharmacokinetic parameters (AUC 0 h–8 h, C-max, and C-trough) were derived from plasma concentrations measured using validated LC–MS/MS quantification of dried blood spot samples. Clinical follow-up was performed for 2 years to record thromboembolic, ischemic, and bleeding events, which were classified according to the ISTH criteria. Patients were categorized into percentile-based exposure groups, defined separately for apixaban and rivaroxaban. Kaplan–Meier survival curves and Cox regression models were used to assess associations between extreme versus moderate plasma concentrations and clinical outcomes.

During the 2-year follow-up, 27 clinical events were recorded, corresponding to an incidence rate of 9.3 per 100 patient–years for any bleeding event (95% CI: 5.57–14.44) and 3.9 per 100 patient–years for ischemic or thromboembolic events (95% CI: 1.68–7.67). These included one major bleeding event (0.5%), eighteen minor bleeding events (9%), six arterial ischemic events (3%), and two venous thromboembolic events (1%). No significant association was found between high apixaban/rivaroxaban plasma concentrations and bleeding events or between low apixaban/rivaroxaban plasma concentrations and ischemic/thromboembolic events for any pharmacokinetic parameter. The study power was limited by the small sample size and a high rate of apixaban and rivaroxaban treatment modifications or discontinuations (41%), which reduced the number of evaluable patients.

In this real-world cohort, apixaban and rivaroxaban plasma concentrations were not associated with clinical outcomes. While highlighting the limitations of analyzing clinical outcomes within the framework of a pharmacokinetic study, these data are informative for future meta-analyses aimed at minimizing publication bias from neutral results.

Infographic summarizing a pharmacokinetics study on direct oral anticoagulants. The title reads “Relevance and limitations of clinical follow-up in a pharmacokinetics study on direct oral anticoagulants”. Three columns are presented. Left panel: Study design–200 patients treated with apixaban or rivaroxaban, 2-year follow-up, objective to assess the link between plasma concentrations and clinical events; hypothesis: high levels increase bleeding risk, low levels increase ischemic risk. Middle panel: Results–high interindividual variability, no significant association with clinical outcomes; 27 clinical events, including 9% minor bleeding, 0.5% major bleeding, and 4% ischemic/thromboembolic events. Right panel: Key insight–small sample size and treatment changes reduced statistical power; drug levels did not predict outcomes but were consistent with major trials.