MIXED MODELING APPROACH FOR CHARACTERIZING THE GENETIC EFFECTS IN A LONGITUDINAL PHENOTYPE

Area of Science:

• Biostatistics
• Genetics
• Cancer Epidemiology

Background:

• Longitudinal phenotypes are underutilized in estimating individual genetic effects.
• Current methods often focus on single time points, neglecting trajectory variations.
• Understanding genetic influences on longitudinal biomarkers is crucial for personalized medicine.

Purpose of the Study:

• To develop and apply a mixed modeling approach for estimating genetic effects on both baseline and slope of longitudinal trajectories.
• To characterize prostate-specific antigen (PSA) trajectories in prostate cancer-free individuals using this novel method.
• To assess the role of genetic factors in PSA variability and their implications for prostate cancer screening.

Main Methods:

• A mixed modeling approach incorporating genetic and individual-specific random effects.
• Utilizing an Average Information Restricted Maximum Likelihood (AI-ReML) algorithm for variance component estimation.
• Analyzing longitudinal PSA data from participants in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial.

Main Results:

• Significant genetic contributions were identified for both initial PSA levels and their rate of change over time.
• The method successfully identified individuals with high genetic predisposition to elevated PSA baseline and trajectories.
• Genetic factors influence PSA variability in individuals who are prostate cancer-free.

Conclusions:

• Incorporating genetic factors into longitudinal PSA monitoring enhances prostate cancer detection accuracy.
• Genetic insights can help identify individuals at risk for falsely screening positive due to PSA trajectory.
• This approach offers a more nuanced understanding of genetic influences on cancer biomarkers.