Population Pharmacokinetics of Apixaban in Asian Patients With Atrial Fibrillation

Overview

The goal of this observational study is to learn about the variability of apixaban blood levels and identify the clinical factors that influence them in Asian patients with atrial fibrillation. The main questions it aims to answer are: 1. What are the specific pharmacokinetic parameters and their variabilities for apixaban in this patient population? 2. Which clinical factors, such as kidney function or body weight, are the primary predictors of how apixaban is cleared and distributed in the body? Participants will be asked to give consent to allow researchers to use their existing electronic outpatient medical records and previously collected blood concentration data for analysis.

This study focuses on characterizing the "population pharmacokinetics (PopPK) of apixaban specifically within Southeast Asian and Taiwanese patient populations. While apixaban is a primary choice for stroke prevention in non-valvular atrial fibrillation (NVAF), there is a significant gap in understanding the inter-individual variability of the drug in Asian patients, who often exhibit different sensitivities and physiological characteristics compared to Western cohorts. * Technical Protocol and Analytical Approach This research utilizes "Non-linear Mixed Effects Modeling (NONMEM)" to analyze drug disposition. Unlike traditional pharmacokinetic studies that require frequent sampling from a few individuals, this PopPK approach allows for the integration of "intensive data" (multiple samples from 24 Thai patients) with "sparse data" (occasional samples from approximately 200 patients). The study aims to evaluate and compare 1-compartment and 2-compartment structural models to better describe the distribution phase of apixaban in this population. * Quality Assurance and Data Management 1. Quality Assurance Plan: The study follows a multi-center protocol involving King Chulalongkorn Memorial Hospital (KCMH) and National Taiwan University Hospital (NTUH). Data validation involves a rigorous screening of Electronic Medical Records (EMR) to ensure patients were in a steady-state condition (receiving a stable dose for at least 3 days) before data extraction. 2. Data Checks: Extracted data undergo internal consistency checks. Predefined rules are applied to verify the relationship between dosing times and sampling times to ensure the physiological plausibility of the resulting pharmacokinetic profiles. 3. Source Data Verification: Accuracy and completeness are assessed by comparing the electronic case report forms (eCRF) against the original external medical records. This process ensures that variables such as creatinine clearance (CrCL) and concomitant medications are recorded precisely. 4. Data Dictionary: A standardized data dictionary is utilized to ensure consistency across the different study sites (Thailand and Taiwan). This includes uniform coding for demographics, laboratory results (e.g., Scr, Hb, Hct), and medications known to interact with CYP3A4 and P-gp. * Operational and Statistical Standards 1. Standard Operating Procedures (SOPs): The protocol defines clear procedures for retrospective data extraction, patient recruitment for face-to-face consent, and the handling of biological data to maintain high ethical and scientific standards. 2. Sample Size Assessment: The target of 200 patients (including 200 Thai and 200 Taiwanese) was determined to provide sufficient statistical power for multivariate analysis. This size allows the model to reliably identify the impact of at least six primary clinical covariates (age, weight, sex, renal function, race, and drug interactions) on drug clearance and distribution. 3. Plan for Missing Data: A dedicated data-cleaning phase precedes modeling. If critical variables (e.g., exact dosing time or sampling time) are uninterpretable or missing beyond a specific threshold, those profiles are excluded from the primary analysis to prevent bias. Sensitivity analyses are also planned to evaluate the impact of any data exclusions. 4. Statistical Analysis Plan: Descriptive statistics will be performed using STATA. The primary PopPK analysis will use NONMEM version 7.5 with the First-Order Conditional Estimation with Interaction (FOCE-I) method. Model selection will be based on the objective function value (OFV) and goodness-of-fit plots. The final model will be validated using a data-split approach (80% index set/20% validation set) and bootstrapping techniques.