Integration of high-throughput genotyping data into pharmacometric analyses using nonlinear mixed effects modeling.

Area of Science:

• Pharmacogenomics
• Computational Biology
• Statistical Modeling

Background:

• High-throughput genotype (HTG) platforms enable cost-effective screening of numerous single nucleotide polymorphisms (SNPs).
• Nonlinear mixed-effects (NLME) modeling is a powerful tool for analyzing complex biological data, including pharmacokinetic relationships.
• Integrating genetic data into NLME models can enhance understanding of genotype-phenotype correlations.

Purpose of the Study:

• To develop and evaluate an algorithm for implementing HTG data within NLME modeling frameworks.
• To assess the algorithm's capability in identifying true genotype-phenotype relationships and minimizing false positives.
• To explore the synergistic combination of HTG data and NLME modeling for novel drug insights.

Main Methods:

• An algorithm involving preselection of SNPs via analysis of variance (ANOVA) and stepwise integration (forward inclusion/backward elimination) into NLME models was developed.
• The algorithm was applied to four simulated pharmacokinetic datasets, each with 300 patients and 1200 SNPs per patient.
• Statistical analysis was conducted using SAS 9.1.3 and NONMEM VI.

Main Results:

• The algorithm successfully identified all true positive genotype-phenotype relationships across the simulated datasets.
• Initially, two datasets showed no false positive (FP) relationships, while two others revealed three FP relationships.
• Subsequent application of independent evaluation datasets and backward elimination effectively reduced the FP rate to zero in all cases.

Conclusions:

• A novel algorithm facilitates the integration of HTG data into NLME models for robust genotype-phenotype relationship detection.
• Independent evaluation datasets are crucial for refining models and reducing false positive findings.
• This approach synergistically combines the strengths of NLME modeling and HTG data, offering potential for new discoveries in drug response and genotype-phenotype correlations.