Recursive partitioning for monotone missing at random longitudinal markers

Area of Science:

• Virology
• Biostatistics
• Genetics

Background:

• HIV drug resistance mutations diminish antiretroviral therapy efficacy.
• Cross-resistance within drug classes is a significant clinical challenge.
• Existing methods often analyze virologic response at a single time point.

Purpose of the Study:

• To develop a statistical method for analyzing longitudinal virologic response with numerous genetic covariates.
• To extend this method to accommodate monotone missing data using inverse probability weighting.
• To investigate viral genetic mutations linked to reduced abacavir efficacy.

Main Methods:

• Utilized a recursive partitioning approach for continuous longitudinal data.
• Employed the kernel of a U-statistic as the splitting criterion, avoiding parametric assumptions.
• Incorporated inverse probability weights to handle missing longitudinal measurements.

Main Results:

• The proposed method effectively analyzes longitudinal data with multiple covariates.
• Simulations demonstrated the method's robust performance.
• Applied to real-world data, it identified genetic mutations associated with abacavir treatment response.

Conclusions:

• The novel statistical approach provides a powerful tool for HIV resistance research.
• It enables a more comprehensive understanding of genotype-phenotype relationships in HIV.
• This method can improve the interpretation of clinical trial data and guide treatment strategies.