Assessing the importance of rare genetic variants for drug response

Abstract

Inter-individual variability in drug response is commonly observed in pharmacological treatment, resulting in 40-70% of patients suffering from low drug efficacy or adverse drug reactions. These negative therapeutic effects significantly contribute to patient morbidity and mortality and constitute an important reason of post-market drug withdrawal. It is estimated that 20-30% of such variability can be explained by genetic factors, i.e., genetic variations in genes that involved in pharmacokinetics, pharmacodynamics and immune-related idiosyncratic drug response.

In this thesis, frequencies of common alleles in clinically important pharmacogenes were analyzed based on Next Generation Sequencing (NGS) data from major human populations. Specifically, frequencies of 176 star alleles in 12 clinically relevant CYP genes across European, African, East Asian, South Asian and Latino populations were evaluated and translated into population-specific functional variability data (Paper I). This variability specifically in the genetically isolated Ashkenazi Jewish population was further studied using an extended pharmacogene list and the unique pharmacogenetic profile of Ashkenazi Jews compared to the European population was revealed (Paper II). In addition, frequencies of four common HLA risk alleles (HLA-B*57:01, HLA-B*15:02, HLA-A*31:01 and HLA-B*58:01) that are strongly associated with drug hypersensitivity reactions were analyzed in up to 74 countries. Based on frequency data, a global cost-effectiveness model was established to inform conditionally cost-effectiveness of preemptive HLA risk allele genotyping in all studied countries (Paper III).

Next, my projects focused on identifying and functional characterizing rare variants in pharmacogenes using NGS data. Specifically, I and my colleagues established a computational prediction framework by first, optimizing classification thresholds of 18 current prediction algorithms using 337 experimentally characterized variants across 44 pharmacogenes, and second, obtaining the combination of optimized algorithms that generates result with highest prediction performance. The resulted optimized framework can quantitatively predict functional effect of pharmacogenetic variants, particularly the rare ones, with predictive power outperformed all other algorithms (Paper IV). This optimized prediction framework was further employed to evaluate variants in two extremely conserved pharmacogenes DPYD and TPMT. While the predictive accuracy being demonstrated using a benchmark dataset containing 70 in vivo characterized DPYD and TPMT variants, this framework was used to analyze genetic encoded functional complexity of the two genes as well as profile population-scale phenotype frequencies using NGS data (Paper V). In addition, this framework was applied to interrogating functionality of variants in a whole set of pharmacogenes (n=208). This study demonstrated that rare variants significantly contribute to pharmacogenomic variability in more than half of studied pharmacogenes and are important in predicting abnormal drug response (Paper VI). Finally, genetic variability of drug targets targeted by all FDA-approved drugs were studied. Particularly, it was revealed that one in six individuals carries at least one missense variant locating in drug-target binding sites and demonstrated that such variants are functionally important and can modulate pharmacological response and guide development of new candidate drugs (Paper VII).

Overall, the findings in this thesis provide a valuable resource for population pharmacogenetic variability and can guide the optimization of population-specific sequencing strategies. Furthermore, I and my colleagues developed a computational tool to interpret the functional impact of pharmacogenetic variability and demonstrated that rare variants in pharmacokinetic and pharmacodynamic genes significantly contribute to variability in drug response. Combined, these findings underscore the importance of incorporating NGS-based pharmacogenomic interpretation into clinical decision making to refine personalized medicine.