Non-parametric estimation of the age-at-onset distribution from a cross-sectional sample

Area of Science:

• Biostatistics
• Epidemiology
• Genetics

Background:

• Estimating disease age-of-onset distribution is crucial for public health and clinical research.
• Traditional methods often struggle with prevalent cases in cross-sectional studies.
• Accurate estimation is vital for understanding disease progression and risk factors.

Purpose of the Study:

• To develop a novel non-parametric method for estimating age-of-onset distributions.
• To accommodate prevalent disease cases in cross-sectional population samples.
• To provide unbiased estimates of disease onset and covariate distributions.

Main Methods:

• Estimating the joint distribution of disease onset and post-onset survival times.
• Conditioning on survival until the age at sampling to handle truncation.
• Utilizing a computationally efficient expectation-maximization (EM) algorithm.
• Marginalizing over survival time to obtain age-at-onset distribution.

Main Results:

• The proposed method provides non-parametric estimates of age-of-onset distributions.
• It accurately handles categorical covariates and yields unbiased covariate distributions.
• Simulations demonstrate good performance even with significant truncation in prevalent cases.
• Applied to BRCA1/2 mutation-associated breast cancer in the Washington Ashkenazi Study.

Conclusions:

• The developed method offers a robust approach for estimating disease age-of-onset distributions.
• It is particularly valuable for analyzing cross-sectional data with prevalent disease.
• The findings have implications for understanding genetic predispositions like BRCA mutations and breast cancer risk.