DYNAMIC RISK PREDICTION FOR CERVICAL PRECANCER SCREENING WITH CONTINUOUS AND BINARY LONGITUDINAL BIOMARKERS

Area of Science:

• Biostatistics
• Oncology
• Epidemiology

Background:

• Cervical precancer risk prediction currently relies on Pap cytology, which has limitations in accuracy and reproducibility.
• Human papillomavirus (HPV)-positive women require accurate risk stratification for effective management.
• Longitudinal biomarker data, such as HPV DNA methylation, show potential for improved precancer risk assessment.

Purpose of the Study:

• To develop a dynamic risk prediction model that integrates longitudinal biomarker data for improved cervical precancer risk estimation.
• To enhance the clinical management of HPV-positive women by providing more accurate risk stratification.
• To create a robust statistical framework for dynamic risk prediction using multiple biomarker types.

Main Methods:

• A joint modeling approach was proposed, linking continuous methylation biomarkers and binary cytology biomarkers to the time-to-precancer outcome via shared random effects.
• The time scale was discretized to enable closed-form likelihood expressions, avoiding complex high-dimensional integration.
• The method accommodates interval-censored time-to-event data, incorporates sampling weights for stratified data, and provides immediate and 5-year risk estimates.

Main Results:

• The application of the dynamic risk model to longitudinal HPV methylation data demonstrated improved risk stratification for HPV-positive women.
• The proposed joint model effectively integrates multiple biomarker types for more precise risk prediction.
• The method's ability to handle complex data features (interval censoring, sampling weights) enhances its practical utility.

Conclusions:

• The developed dynamic risk model offers a significant improvement over traditional methods for cervical precancer risk prediction.
• Integrating longitudinal biomarker data, particularly HPV DNA methylation, enhances the ability to identify high-risk individuals.
• This approach provides valuable tools for clinical decision-making in the management of HPV-positive women, potentially reducing unnecessary procedures and improving outcomes.