|
Activity Number:
|
193
- Modeling, Design Strategies and Assessments of Biomarkers
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
|
Abstract #312901
|
|
|
Title:
|
A Linear Mixed Effects Model with Censored Response for Longitudinal Biomarker Data Subject to Lower Limits of Detection
|
|
Author(s):
|
Xueli Liu* and Chun Zhang
|
|
Companies:
|
AbbVie Inc. and AbbVie Inc.
|
|
Keywords:
|
Biomarkers;
Censored data;
Lower limits of detection;
Mixed effects model;
Maximum likelihood
|
|
Abstract:
|
In Immuno-oncology or early oncology development, biomarkers are often measured repeatedly over time. Such measurements are subject to lower limits of detection (LLD) due to assays sensitivity, and thus result in left-censored data. A common practice is to either substitute data below LLD with an assumed value smaller or equal to LLD or to discard. This may lead to biased estimates of parameters of interest, especially when the proportion of data below LLD is high. We treat the longitudinal biomarker data below LLD as left-censored response and fit linear mixed effects model to study time/dose effect and interaction terms. We use lmec package in R to obtain the maximum likelihood estimates of the model parameters. For comparison purpose, lme4 package in R is employed to analyze data with deletion or substitution. Simulation studies under different scenarios show that the proposed linear mixed effects model with censored response leads to more satisfactory parameter estimates as measured in terms of bias and mean squared error. A case study of PD biomarker analysis with cytokine data in an early phase oncology trial is shown to illustrate the application of the proposed model.
|
Authors who are presenting talks have a * after their name.
