Cited 0 times in

Development of a longitudinal model to characterize QT interval change in moxifloxacin treatment

Development of a longitudinal model to characterize QT interval change in moxifloxacin treatment
Other Titles
Moxifloxacin 투약에 따른 QT 간격 변화를 평가하기 위한 경시적 모델의 개발
Issue Date
Graduate School, Yonsei University
Dept. of Medical Science/박사
ICH Guidance E14 on the evaluation of QT/QTc interval prolongation for non-antiarrhythmic drugs was endorsed in 2005. Standard statistical analysis of QT measurements from these studies usually compares time-matched baseline corrected QTc of placebo and active treatments. In thorough-QT studies, moxifloxacin is often used as a positive control to assess the relative risk of a test drug. Recent work has examined the time course of moxifloxacin and placebo effects. Characterizing the moxifloxacin effect over time, the influence of any covariate on this effect is important to interpret future TQT studies. With this background, this work analyzed QT interval data observed in placebo and moxifloxacin treatments obtained from a meta-analysis of subject-level data from 12 QT studies drawn from different drug programs and comprising 749 subjects of diverse demographic characteristics.The purpose of this work was to develop a longitudinal model to describe the time course of QT interval in healthy volunteers given placebo and moxifloxacin treatments and to examine the source of variability affecting QT interval with a focus on the variability in demographic factors and inter-study differences.Electrocardiograph data from placebo and 400 mg moxifloxacin treatments were analyzed using NONMEM software. A longitudinal QT interval model incorporating the variability in demographic factors and the inter-study difference was developed within a mixed effects model framework according to the following steps: individual correction, baseline correction, drug effect, inter-study difference, covariate effect, and model evaluation. The data from 12 studies were modeled together using a meta-analysis approach.- Individual correction: The exponent on RR in the QT correction factor was estimated for each individual rather than assuming fixed values as used in the Bazett and Fridericia corrections.- Baseline correction: Following an appropriate individual correction of the QT interval for heart rate, a mixed effect model was fit to the data obtained from the placebo treatment to explore the circadian effect on the baseline QT interval using cosine functions with up to three periods (24, 12, 6 or 8 hours). Interindividual variability was included in acrophase and amplitude for each period.- Drug effect: The data obtained from the moxifloxacin treatment were then analyzed by fixing the baseline structural model to the one obtained from the placebo treatment in the previous step. The drug effect on QT interval was described as a direct function of time and QT measurements was modeled and tested using diverse equations.- Inter-study difference: The variability among different studies was modeled using a fixed study-effect method on the assumption that the model parameters of the individuals follow the same distributions across studies. The method assumes that there are different unrelated model parameter values between different studies, and the study effect is introduced as a covariate for the model parameter.- Covariate effect: The covariate effect is analyzed by a stepwise covariate model building process. Categorical covariates such as sex, race and smoking and continuous covariates such as age, height, weight, BMI, or average amount of alcohol drinking were tested.- Model evaluation: Visual Predictive Checks were used to examine the model fit to the observed data.The estimated QTc interval yielded a heart-rate correction slope alpha of 0.35 with 16% (CV) inter-subject variation. A two-oscillator model with 24- and 12-hour period best described the circadian variation of baseline QT measurements. For the effect of moxifloxacin on QT interval, a Bateman function was selected to represent the elongating aspect of QT interval for the early phase followed by shortening for the later part of the observation. The difference among the 12 studies was best modeled using separate baseline mesor parameters for each study, ranging between 378 and 410. In the covariate effect, the baseline QTc interval and the magnitude of the drug effect were found to be significantly higher in women than in men. The acrophase for the 24-hour period circadian rhythm was estimated to be lower in Asians than in other ethnic groups. Age was also found to be an important covariate, showing that baseline value increased with age. When the final model was evaluated using VPC, the model was found to adequately described the observed data.The present analysis was carried out as a meta-analysis using observed data across a number of QT trials. The developed longitudinal mixed effects model well described the time course of the QT interval when the source of variability in the QT interval was appropriately included. The model provided useful information on potential covariates influencing the QT interval change such as sex, age, and race. The final model may influence future trial design and will assist in contextualizing information from future TQT studies.
Appears in Collections:
2. 학위논문 > 1. College of Medicine (의과대학) > 박사
Yonsei Authors
사서에게 알리기
교내이용자 서비스로 제공됩니다.
RIS (EndNote)
XLS (Excel)


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.