Concordance of DMET plus genotyping results with those of orthogonal genotyping methods

Abstract

There are several hurdles to the clinical implementation of pharmacogenetics. One approach is to employ pre-prescription genotyping, involving interrogation of multiple pharmacogenetic variants using a high-throughput platform. We compared the performance of the Drug Metabolizing Enzymes and Transporters (DMET) Plus array (1,931 variants in 225 genes) with that of orthogonal genotyping methods in 220 pediatric patients. A total of 1,692 variants had call rates >98% and were in Hardy-Weinberg equilibrium. Of these, 259 were genotyped by at least one independent method, and a total of 19,942 single-nucleotide polymorphism (SNP)-patient sample pairs were evaluated. The concordance rate was 99.9%, with only 28 genotype discordances observed. For the genes deemed most likely to be clinically relevant (TPMT, CYP2D6, CYP2C19, CYP2C9, VKORC1, DPYD, UGT1A1, and SLCO1B1), a total of 3,799 SNP-patient sample pairs were evaluable and had a concordance rate of 99.96%. We conclude that the DMET Plus array performs well with primary patient samples, with the results in good concordance with those of several lower-throughput genotyping methods.

Figure 1. Performance of the DMET Plus array

The Affymetrix Drug Metabolizing Enzymes and Transporters (DMET) genotyping platform scans 1,931 different variants. The accuracy of the platform was determined for SNPs that passed two quality control criteria. The first quality control filter removed any SNPs with call rates below 98%, and the second filter removed any SNPs that were not in Hardy-Weinberg equilibrium (HWE) at a P-value < 0.001. A total of 1,692 SNPs passed QC and the concordance of genotypes was determined between the DMET array and those available from orthogonal genotyping. A total of 259 SNPs were assessed by DMET and at least one additional genotyping method.

Figure 2. Concordance of DMET MAF with HapMap CEU MAF

The minor allele frequency (MAF) of all SNPs that passed quality control was determined in patients of European ancestry (x-axis) and compared to the corresponding frequency in HapMap CEU samples (y-axis). Linear regression between the DMET and HapMap allele frequencies suggest that the values between the populations are highly correlated (R² = 0.9711).

Figure 3. Concordance of the DMET Plus array with other genotyping methods

The DMET genotyping was compared to all other sequencing and genotyping data available in 220 patients. A total of 259 SNPs were assessed by the DMET array and at least one orthogonal method. Of these, 233 SNPS were assessed by one orthogonal method, 22 SNPs were assessed by two orthogonal methods, and 4 SNPs were assessed by three orthogonal methods, yielding a total of 289 tests of concordance between an orthogonal method and a DMET SNP call. (A) 167 DMET SNPs were in common with the Affymetrix Gene Chip (orange squares), 3 SNPs with Prometheus TPMT genotyping (purple triangles), 78 with the Illumina Golden Gate chip (red circles), 8 with the Sequenom iplex Gold MassARRAY platform (green diamonds), 26 with Sanger sequencing (white circles), and 7 with Beckman Coulter GenomeLab SNPstream (blue squares). (B) A threshold of 98% was used to determine SNPs with poor concordance. SNPs below this threshold (6, 2%) were further investigated, where possible, by comparing the discordant genotypes with any other overlapping genotyping methods.