Semiparametric joint modeling for biomarker trajectory before disease onset

Area of Science:

• Biomedical Research
• Statistical Modeling
• Neurodegenerative Diseases

Background:

• Understanding biomarker changes relative to disease pathogenesis is crucial.
• Analyzing temporal biomarker evolution before disease onset presents analytical challenges, including left-truncation bias due to differing time scales (natural vs. time-on-study).

Purpose of the Study:

• To propose a semiparametric joint model for analyzing biomarker temporal evolution prior to disease onset.
• To develop a flexible model accommodating natural time scales (e.g., age) and time-to-disease onset.
• To investigate brain biomarker trajectories preceding preclinical Alzheimer's disease.

Main Methods:

• Development of a semiparametric joint model.
• Utilizing a profile kernel estimating equation approach for estimating regression coefficients and baseline mean trajectory functions.
• Establishing large-sample properties of estimators and conducting simulation studies for performance evaluation.

Main Results:

• The proposed method effectively estimates biomarker trajectories before disease onset.
• Application to Alzheimer's disease revealed a decline in cortical thickness across brain regions preceding disease onset.
• APOE4 carriers exhibited lower cortical thickness levels compared to non-carriers.

Conclusions:

• The developed statistical model offers a robust approach for analyzing longitudinal biomarker data in relation to disease pathogenesis.
• Findings highlight significant changes in brain structure prior to Alzheimer's disease onset, with genetic factors (APOE4) influencing these trajectories.
• This research provides insights into early disease mechanisms and potential targets for intervention.