Year of Award

2018

Document Type

Thesis

Degree Type

Master of Science (MS)

Degree Name

Pharmaceutical Sciences

Department or School/College

Biomedical and Pharmaceutical Sciences

Committee Chair

Erica L. Woodahl

Commitee Members

Richard Bridges, Keith Parker, Nigel Priestley

Keywords

Pharmacogenetics, pharmacogenomics, CYP2D6

Publisher

University of Montana

Subject Categories

Other Pharmacy and Pharmaceutical Sciences

Abstract

BACKGROUND: CYP2D6 is difficult to accurately genotype due to a large number of single nucleotide variants (SNVs), indels, and structural variation such as deletions, duplications, and CYP2D6/CYP2D7 hybrid genes. CYP2D6 targeted genotyping panels are of limited utility; clinically relevant variants that are not genotyped will be missed. Sequencing solves this problem but requires additional tools to address structural variation. The goal of our study was to determine the predictive power of Stargazer, a novel allele-calling program, which combines SNV/indel calls with structural variation identification.

METHODS: In a panel of 309 human livers, CYP2D6 diplotypes and activity scores were initially assigned manually using PGRNSeq SNV/indel data and then reassigned after inclusion of Stargazer-derived structural variation data. We determined CYP2D6 activity in human liver microsomes with metoprolol and dextromethorphan as probe substrates. Then, we used linear regression to assess the relationship between activity and activity scores assigned using SNV/indel data alone versus SNV/indel + structural variation data.

RESULTS: Without incorporating structural variation data, diplotypes were incorrectly assigned for 67 samples (22%); activity scores were incorrect for 26 samples (8.4%). Structural variants included 23 deletions, 47 duplications, and 39 hybrids. When diplotypes were assigned based on SNV/indel data alone, activity score explained 31% of the variation in CYP2D6 activity with metoprolol (R2 = 0.31, p < 0.001) and 36% with dextromethorphan (R2 = 0.36, p < 0.001). When reassigned with SNV/indel plus structural variation data, this increased to 36% for metoprolol (R2 = 0.36, p < 0.001) and 41% for dextromethorphan (R2 = 0.41, p < 0.001).

CONCLUSION: The accuracy of CYP2D6 phenotype prediction can be improved by using a next-generation sequencing approach coupled with a tool such as Stargazer to detect common and rare SNVs and indels as well as structural variation in CYP2D6.

Available for download on Saturday, January 19, 2019

Share

COinS
 

© Copyright 2018 Rachel Dalton