Validation of a Large Custom-Designed Pharmacogenomics Panel on an Array Genotyping Platform

Abstract

Background

Pharmacogenomics has the potential to improve patient outcomes through predicting drug response. We designed and evaluated the analytical performance of a custom OpenArray^® pharmacogenomics panel targeting 478 single-nucleotide variants (SNVs).

Methods

Forty Coriell Institute cell line (CCL) DNA samples and DNA isolated from 28 whole-blood samples were used for accuracy evaluation. Genotyping calls were compared to at least 1 reference method: next-generation sequencing, Sequenom MassARRAY^®, or Sanger sequencing. For precision evaluation, 23 CCL samples were analyzed 3 times and reproducibility of the assays was assessed. For sensitivity evaluation, 6 CCL samples and 5 whole-blood DNA samples were analyzed at DNA concentrations of 10 ng/µL and 50 ng/µL, and their reproducibility and genotyping call rates were compared.

Results

For 443 variants, all samples assayed had concordant calls with at least 1 reference genotype and also demonstrated reproducibility. However, 6 of these 443 variants showed an unsatisfactory performance, such as low PCR amplification or insufficient separation of genotypes in scatter plots. Call rates were comparable between 50 ng/µL DNA (99.6%) and 10 ng/µL (99.2%). Use of 10 ng/µL DNA resulted in an incorrect call for a single sample for a single variant. Thus, as recommended by the manufacturer, 50 ng/µL is the preferred concentration for patient genotyping.

Conclusions

We evaluated a custom-designed pharmacogenomics panel and found that it reliably interrogated 437 variants. Clinically actionable results from selected variants on this panel are currently used in clinical studies employing pharmacogenomics for clinical decision-making.

Impact Statement

The custom-designed genotyping panel presented here is used in clinical studies assessing the value of testing for pharmacogenomic variants. This potentially furthers implementation of pharmacogenomics in clinical practice and may benefit a large patient population taking drugs with a pharmacogenomics component. The panel provides reliable genotypes for 437 variants in a Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory, and clinically actionable data is reported through an access-protected, web-based portal (genomic prescribing system) that predicts drug response in an easily interpretable format, i.e., a traffic-light system. The data presented add to the knowledge in the field of genotyping panels for pharmacogenomics.

Introduction

Pharmacogenomics is the study of how an individual’s genetic composition affects his or her response to medications. Genetic variants, such as single-nucleotide variants (SNVs), can result in loss-of-function of drug-metabolizing genes and duplication of certain genes may result in gain-of-function. These genetic variations may be implicated in efficacy, e.g., absorption, distribution, metabolism, and excretion (ADME), as well as safety for some medications. Taking the most extensively studied enzyme family, cytochrome P450, family 2 (CYP2), as an example, CYP2C19 loss-of-function alleles are associated with reduced formation of the active metabolite of the antiplatelet prodrug clopidogrel (1). On the other hand, individuals with more than 2 normal functional copies of CYP2D6 genes are considered ultrarapid metabolizers, potentially exhibiting symptoms of morphine overdose even with standard doses of its codeine prodrug (2).

Genotype-based guidelines for genetic variants that have sufficient evidence available for the use of pharmacogenomics information in clinical settings have been published by the Clinical Pharmacogenetics Implementation Consortium (CPIC) (3–5). To date, there are 146 gene–drug pairs published with sufficient evidence for at least 1 prescribing action to be recommended (CPIC levels A and B) (6). Genotyping panels focusing on different therapies have been established: medications for cardiovascular diseases (7), anticancer therapies (8–10), and nonsteroidal antiinflammatory drugs (11), as well as broad-based ADME panels (12–14). There are also genotyping panels for specific genes that are highly polymorphic and clinically important, such as CYP2D6 (15) and CYP2C19 (16).

Here, we are reporting on the design and evaluation of a custom OpenArray pharmacogenomics panel (OA-PGx panel) in the setting of a Clinical Laboratory Improvement Amendments (CLIA)-certified and College of American Pathologists (CAP)-accredited laboratory. This panel currently supports preemptive pharmacogenomics clinical studies, including the African American Cardiovascular Pharmacogenomics Consortium (The ACCOuNT Consortium), the 1200 Patients Project and the Implementation of Point-of-Care Pharmacogenomic Decision Support in Perioperative Care (The ImPreSS Trial) operated through the Center for Personalized Therapeutics at the University of Chicago (17–19). For user-friendliness, interpretations of found variants are reported through an access-protected web-based portal (the genomic prescribing system, GPS), which provides a simplified user interface, including traffic-light iconography, an explanatory legend on every page, and an immediately available list of pharmacogenomics drug alternatives alongside each currently prescribed medication (20). At the time of writing of this paper, among the 437 validated variants, 113 variants on 45 genes were associated with 65 clinically actionable drugs, and therefore could be translated to patient-specific interpretations.

Materials AND Methods

Design of the OA-PGx Panel

The OA-PGx panel includes (a) variants in well-known drug-metabolizing genes, with high-level of evidence in CPIC guidelines, PharmGKB, and/or the Dutch Pharmacogenetics Working Group (DPWG), and (b) variants of clinical significance carefully selected from a comprehensive review of the literature and likely to be included in professional guidelines in the near future. Variants were selected by a process of literature review to identify polymorphisms associated with drug-related outcomes. The selection process follows a methodology previously described to identify medications and related germline markers with published pharmacogenomics evidence (20, 21). The methodology is supported by an automated literature search algorithm and integration of variants identified by these expert groups, curated by manual review by at least 2 team members to select variants with the highest level of evidence.

The OA-PGx panel is comprised of 4 customized TaqMan^® OpenArray Genotyping Plates, Format 128 (Thermo Fisher Scientific, SKU 4471116). On each genotyping plate, there are 48 subarrays arranged into 4 rows (A-D) and 12 columns (1–12). Each DNA sample is loaded into 2 adjacent subarrays, e.g., DNA sample for one individual is loaded into subarrays A1 and B1 (see Fig. 1 in the online Data Supplement). Each subarray (e.g., A1) can be individually preloaded with 64 assays arranged in 8 subcolumns (a–h) and 8 subrows (1–8). Therefore, on a single genotyping plate, a maximum of 128 assays for 24 samples including controls can be run. We decided to preload 120 assays per genotyping plate, or 60 assays per subarray, for a total of 480 assays. The panel targets 478 variants, including 2 triallelic variants. Each triallelic variant requires 2 assays for genotyping as OpenArray technology is based on allelic discrimination. Therefore, there are 480 assays on the panel.

Fig. 1.

Examples of scatter plots for satisfactory and unsatisfactory performances. (A), Satisfactory: acceptable PCR amplification and clear separation of clusters. (B), Unsatisfactory: low PCR amplifications and diffused clusters. (C), Unsatisfactory: acceptable PCR amplifications but diffused clusters.

DNA Extraction

Unless otherwise stated, DNA was extracted from whole-blood samples using a Maxwell^® 16 Blood DNA Purification Kit on a Maxwell RSC instrument (Promega). The instrument uses MagneSil^® Paramagnetic Particles to purify genomic DNA, with a typical yield of 37 µg of genomic DNA from 500 µL of whole blood. DNA samples from the Molecular Diagnostic Laboratory, University of Chicago (UC Molecular Laboratory, Chicago, IL, website: https://dnatesting.uchicago.edu/) were extracted using FlexSTAR (Autogen) with a typical yield of 80 µg genomic DNA from 1–3 mL of blood per sample. DNA concentrations were determined using a NanoDrop ND 1000 Spectrophotometer (NanoDrop Technologies). All DNA samples were stored at 2 °C to 6 °C (short-term) or −15 °C to −25 °C (long-term) until genotyping analysis.

Genotyping

DNA samples were diluted to 50 ng/µL using nuclease-free water (Ambion^® no. AM9930). For each sample to be run on a genotyping plate, 3 µL of DNA was transferred into a well of a 384-well sample plate (Thermo Fisher, catalog no. 4406947). 3 µL of Genotyping Master Mix (Thermo Fisher) was added and mixed well with the DNA. A no template control (NTC; reaction mixture with all reagents but no template DNA) was included in each run as a negative control. The 384-well sample plate was then covered with Adhesive PCR Foil (Thermo Fisher) and centrifuged on a PCR plate spinner (VWR International) for 1 min at 500g. 5 µL of sample was loaded on each subarray of the genotyping plate using OpenArray AccuFill (Thermo Fisher) according to the manufacturer’s instructions. After loading, the genotyping plate was immediately sealed with an OpenArray case lid (Thermo Fisher) using consumables provided from QuantStudio^TM 12K Flex OpenArray Accessories Kit and Plate Press 2.0 (Thermo Fisher). The genotyping plates were then placed into the QuantStudio 12 K Flex Real-Time PCR System v.1.2.2 (Thermo Fisher) for SNV genotyping experiments. Once data was acquired, the results were exported from the QuantStudio to Thermo Fisher Real-Time qPCR Genotyping App v.3.8 (Thermo Fisher Genotyping App), a cloud-based software, URL: https://apps.thermofisher.com/apps/spa for data analysis.

Real-time data (which show reporter signals from VIC and FAM dyes normalized to fluorescence signal of ROX dye, indicating alleles 1 and 2, respectively) were analyzed using autocalling on Thermo Fisher Genotyping App. Autocalling used a reference panel, with the assumption that all variants were in Hardy–Weinberg equilibrium. A reference panel covering heterozygous and both homozygous calls on the OA-PGx panel was built using reference samples that had confirmed genotypes, including Coriell Institute cell line (CCL) DNA samples and samples from the UC Molecular Laboratory [for ryanodine receptor 1 (RYR1) variants] as well as Knight Diagnostic Laboratories (CLIA-certified) at Oregon Health & Science University (OHSU, Portland, OR, website: https://knightdxlabs.ohsu.edu/).

The quality control (QC) images and scatter plots were reviewed prior to data analysis. QC images including postread ROX (using a passive reference dye present in the genotyping master mix to reveal potential technical issues), postread VIC, postread FAM, and leak-check images were reviewed. The quality of scatter plots was examined using Thermo Fisher Genotyping App to evaluate the NTC and all clusters.

Validation Studies

The validation studies consisted of accuracy, precision, and sensitivity evaluation. Accuracy studies were performed by comparing the genotypes of the variants determined by the OA-PGx panel with at least one of 2 reference genotyping methods, next-generation sequencing (NGS), and/or Sequenom MassARRAY iPLEX platform (MassARRAY). Reference genotypes for the 40 CCL samples that were used for accuracy studies were determined by accessing the 1000 Genomes Project (1KGP) database (phase 3), which was constructed using NGS. Twenty-two DNA samples extracted from whole blood were randomly chosen from 1200 Patients Project samples that were previously genotyped at OHSU, which used MassARRAY technology (17, 22). For variants that had discordant calls with the reference genotypes from OHSU, but were deemed clinically essential, we performed Sanger sequencing to confirm the genotypes. Six DNA samples were used for accuracy evaluation of RYR1 genotyping and sequences were provided by the UC Molecular Laboratory, which had determined these by NGS. A precision study was performed by genotyping 23 CCL samples in triplicate runs to assess the assay’s reproducibility and this served a dual purpose for accuracy evaluation. A sensitivity study that used 6 CCL samples and DNA extracted from 5 whole blood samples assessed the performance of genotyping assays by using 2 DNA concentrations: the manufacturer’s recommended DNA concentration, 50 ng/µL, (i.e., 125 ng/assay) and one-fifth of the recommended concentration, 10 ng/µL (i.e., 25 ng/assay). In total, 43 different CCL samples and DNA extracted from 33 whole-blood samples were used in the validation study of the OA-PGx panel. These studies on clinical pharmacogenomics were approved by the institutional review board at the University of Chicago Medical Center (IRB10-487-A and IRB17-0890).

There were cases where the OA-PGx panel failed to provide genotyping calls due to either low amplification or poor separation of genotypes observed in scatter plots. For each variant genotyping assay, the individual assay and overall call rates were determined as the percentage of samples for which calls were successfully made.

Any variants for which all samples assayed met the following 3 criteria were considered validated: (a) concordant calls with reference genotypes in the accuracy study, (b) reproducible calls in the precision study, and (c) also demonstrated satisfactory performance during the validation, including sufficient amplification, clearly separated clusters and last, no amplification in the NTCs. Figure 1 shows examples of scatter plots of assays with satisfactory and unsatisfactory performances.

Results

Accuracy Studies

Assay accuracy was assessed by comparing the OA-PGx panel’s calls against the calls from at least one reference method and the results are listed in Table 1. The sources of reference genotypes are described in the Materials and Methods, and are illustrated in Fig. 2.

Table 1.

Concordance between the OA-PGx panel and reference methods for accuracy evaluation.

Reference genotyping method (source)	Number of variants with available reference genotypes	Number of samples genotyped	Experimental call rate	Number (percentage) of variants with at least one discordant genotype	Number (percentage) of variant with perfect concordance with reference method
NGS (1KGP)	429	23a	99.1%	6 (1.4%)	423 (98.6%)
NGS (1KGP)	429	17	99.1%	8 (1.9%)	421 (98.1%)
Total NGS (1KGP)b	429	40b	99.1%	13 (3.0%)	416 (97.0%)
Sequenom MassARRAY (OHSU)	342	22	98.9%	23c (6.7%)	319c (93.3%)
NGS (UC Molecular Lab)	35	6	100%	0 (0%)	35 (100%)
Overall	474	68	99.1%	34 (7.2%)	440 (92.8%)
Overall accuracy with Sanger sequencing confirmation of 4 variants				30 (6.3%)	444 (93.7%)

^a23 CCL samples were analyzed in triplicate.

^bCombined results of triplicate run using 23 CCL samples and single run using 17 CCL samples.

^cGenotypes of 15 samples for 4 discordant variants by MassARRAY were subsequently analyzed by Sanger sequencing and OA-PGx panel results were confirmed accurate.

Fig. 2.

Venn diagram overlap between the reference genotypes for 474 variants. Of 478 variants, 4 variants on the panel had no reference genotype available. OHSU: Oregon Health & Science University; MassARRAY: Sequenom MassARRAY iPLEX platform; 1KGP: 1000 Genomes Project. ^a22 patient DNA samples; ^b40 CCL samples and 22 patient DNA samples; ^c40 CCL samples; ^d40 CCL samples and 6 patient DNA samples analyzed for a single variant in RYR1; ^e6 patient DNA samples analyzed for 34 variants in RYR1.

For the 429 variants for which reference genotypes were available from the 1KGP database, we assayed 40 CCL samples from 10 ancestries (see Supplemental Table 1). Twenty-three of the CCL samples were analyzed in triplicate to also serve the purpose of precision evaluation, which will be discussed later, with the remaining 17 analyzed once. For the 40 CCL samples analyzed, the percentage of variants with perfect concordance with the reference genotypes in 1KGP database was 97.0% (416/429) (Table 1).

For the 342 variants for which reference genotypes were available through MassARRAY, their accuracies were assessed using DNA extracted from 22 whole-blood samples. For 23 variants, the genotype of at least one sample on the panel was discordant with that on MassARRAY. Some of these variants are implicated in the metabolism of commonly prescribed medications, such as clopidogrel or warfarin. For 4 of these variants, we performed Sanger sequencing to definitively determine their genotypes (see Supplemental Table 2). These 4 variants were selected because of their particular potential importance in informing the use of multiple commonly-used or high-profile medications (rs12248560 is CYP2C19*17; rs1061622 is in TNFRSF1B; rs1042713 is in CYP2C9; and rs1042713 is in ADRB2). Sanger sequencing confirmed that the results from the OA-PGx panel were accurate. The percentage of variants which showed concordance with MassARRAY was 93.3% (319/342); however, considering OA-PGx results for 4 out 23 discordant variants that were confirmed by Sanger sequencing, the total number of variants that “passed” this part of the validation was 323 (94.4%).

The 2 triallelic variants, rs2032582 and rs7900194, had reference genotypes available in the 1KGP database and also from OHSU. For each triallelic variant, results from 2 assays were needed to determine the genotype (Table 2). The principle is that an assay will only generate signals when at least one of the base pairs that it is testing for is present (23). Using the variant rs2032582 as an example, both genotypes CC and CT generate CC calls in an A/C assay, so a C/T assay is needed to differentiate them. Interpreted results according to Table 2 were 100% concordant with both 1KGP and OHSU.

Table 2.

Interpretations for the 2 triallelic variants rs2032582 and rs7900194.

rs2032582 [C/A] call	rs2032582 [C/T] call	Final genotype
AA	No amplification	AAa
CA	CC	CA
CC	CC	CC
CC	CT	CT
No amplification	TT	TTa
AA	TT	AT
rs7900194 [G/A] call	rs7900194 [G/T] call	Final genotype
GG	GG	GG
AG	GG	AG
AA	No amplification	AAa
AA	TT	AT
No amplification	TT	TTa
GG	TT	GT

^aNeeds Sanger sequencing confirmation to distinguish between a true call where no amplification is expected for one assay and a technical failure.

For the 35 variants on our panel assessing the RYR1 gene, only rs118192172 was available in the 1KGP database. Therefore, we assayed 6 samples from the UC Molecular Laboratory where these 35 RYR1 variants were sequenced by NGS. The OA-PGx panel had a 100% concordance with their respective genotypes provided by the UC Molecular Lab (and also 1KGP, only for rs118192172).

In total, reference genotypes were available for 474 variants and their accuracies could be assessed. Discordant calls were seen for 34 variants (7.2%); however, as mentioned before, for 4 of these variants, Sanger sequencing confirmed that the OA-PGx panel results were correct and thus results for 444 out of 474 variants (93.7%) were considered accurate (Table 1). For the 68 samples assayed in the accuracy studies, the overall call rate was 99.1% (Table 1 and Supplemental Table 3).

Precision Studies

The precision of assays on the OA-PGx panel was tested using the dual-purpose triplicate runs with 23 CCL samples mentioned previously in the accuracy study. The overall call rate of the triplicate run was 99.2% (Supplemental Table 3) and 6 assays failed to make reproducible calls, hence 98.8% (474/480) of the assays made reproducible calls.

Sensitivity Studies

The sensitivity study was performed using 6 CCL samples and DNA extracted from 5 whole-blood samples. Genotyping was performed on the OA-PGx panel using a DNA concentration of 50 ng/µL, as recommended by the manufacturer, and a DNA concentration of 10 ng/µL in the same run, hence allowing direct comparison of the call rates. For the experiment using 10 ng/µL DNA, 42 out of 5280 assays (11 samples × 480 assays) failed to make calls and the overall call rate was 99.2%. For 50 ng/µL DNA, 18 out of 5280 assays failed to make calls and the overall call rate was 99.6% (Supplemental Table 3).

When 10 ng/µL DNA was used, 99.8% (479 out of 480 assays) of calls were consistent with their respective calls when 50 ng/µL DNA was used. Only 1 assay had an inconsistent call for a CCL sample (rs6265, a variant in the gene that codes for brain-derived neurotrophic factor). Its reference genotype was available in the 1KGP database, and we verified that the call was correct when 50 ng/µL DNA was used.

Validated Variants

The OA-PGx panel is a laboratory-developed molecular genetics test and we have set acceptable criteria before the commencement of validation as described in Materials and Methods. The OA-PGx panel targeted 478 variants; for 4 variants there was no reference genotype available, so their accuracy could not be assessed. Out of the 474 variants for which reference genotypes were available, 443 variants showed excellent concordance with their reference genotypes (or were confirmed to be correct by Sanger sequencing) and demonstrated reproducibility for all assayed samples. Use of 10 ng/µL DNA resulted in an incorrect call for a single sample for a single variant. However, this variant is still considered validated since 50 ng/µL DNA will be used.

The software Thermo Fisher Genotyping App automatically flags results that are not close to the center of any cluster nor reference in the scatter plots, and no calls are made for these cases. However, there were cases for which the software made automated calls for results located in-between clusters; these were considered invalid calls during manual review. There were 6 variants for which all calls were concordant with the reference genotypes and demonstrated reproducibility but showed unsatisfactory performance, i.e., low PCR amplification and/or poor separation of genotypes in scatter plots (Fig. 1, B and C), during the validation. Therefore, we considered these 6 variants to be not validated. In total, 437 variants were validated on the OA-PGx panel (see Supplemental Tables 3 and 4).

For 39 validated variants, only the major allele was observed during the validation: 31 of these were in the RYR1 gene. The minor allele frequencies of the remaining 8 variants are 0.0007%–0.038% in NCBI single-nucleotide polymorphism database build 153 (dbSNP) (24), similar to the variants on the RYR1 gene (0.0004%–0.1%). For these 39 variants, the first call for the alternative allele in the future will be confirmed by Sanger sequencing. The heterogeneity per sample type is listed in Supplemental Table 5.

Discussion

Testing for pharmacogenomic variants has the potential to improve efficacy and/or safety for a significant number of drugs. Preemptive testing does not delay initiation of therapy, as opposed to traditional reactive testing; however, it does require relatively large, carefully designed panels. Here, we describe the analytical validation of a large custom-designed pharmacogenomics panel on the TaqMan OpenArray genotyping platform (the OA-PGx panel), which is currently used in clinical studies.

The OA-PGx panel targets 478 variants using 480 assays. According to the manufacturer, the TaqMan OpenArray Genotyping System can achieve 99.7% concordance with the reference method (data generated on an Applied Biosystems 7900HT Fast Real-Time PCR System), 99.8% reproducibility and an overall call rate of 99.9% (25, 26). Our results showed that 98.8% (474/480) of the assays on the OA-PGx panel demonstrated reproducibility and the overall call rates were >99% throughout the validation (Supplemental Table 3), which met our expectations. The observed overall call rate for the OA-PGx panel was also comparable to those of other panels using OpenArray technology as well as other genotyping platforms such as the DMET Plus array, the VeraCode ADME Core Panel and an NGS-based panel (all of them reported overall call rates >97%) (8, 27–29). Ang et al. had also shown that the OpenArray platform could achieve >97% call rate using DNA extracted from buccal swab (sponge-tipped) samples (30).

In the accuracy study, 92.8% (440/474) of the assays had concordant calls with NGS or MassARRAY (Table 1). This was significantly lower than the observed concordance by the manufacturer (99.7%) and other previously described OpenArray-based platforms, which demonstrated 95%–100% concordance with their orthogonal methods (25, 26, 28, 31, 32). Moreover, studies have shown that the DMET Plus array and the NGS-based PGRNseq panel achieved 99.9% and 99.8% concordance with their orthogonal methods, respectively (27, 33). The percentage of assays for which the OA-PGx panel had perfect concordance with the reference genotypes from the 1KGP database and the UC Molecular Lab (Table 1) —both used NGS—was 97% (416/429) and 100% (35/35), respectively. Among the 342 variants for which reference genotypes were available through MassARRAY, 6.7% (23/342) of the assays on the OA-PGx panel showed discordance (Table 1). The reference genotypes of these 23 variants were also available in the 1KGP database for the 40 CCL samples and the OA-PGx panel showed concordance for 21 of them. The genotypes for 4 of these variants were confirmed by Sanger sequencing and the results were also concordant to the OA-PGx panel. Because we considered variants with one or more discordant calls with at least 1 of the reference methods not validated unless confirmed by Sanger sequencing, the overall number of variants that passed the accuracy evaluation was 444. Thus, the lower-than-expected percentage of concordance is predominately due to discordance between the OA-PGx panel and MassARRAY.

The OpenArray platform is high-throughput, relatively inexpensive, and customizable, hence it perfectly suits the needs of our large-scale clinical studies. Ideally, a broadly inclusive pharmacogenomics panel should include variants of well-known drug-metabolizing genes, variants with high-level evidence as evaluated by CPIC, PharmaGKB, and/or DPWG and clinically important variants expected to gain this high-level evidence in the near future (17). The goal is to include variants associated with medications a person is taking as well as medications they will potentially take in the future. Moreover, the variants included on the panel have to be reviewed and modified on regular basis to keep it up to date. Although the OpenArray is an allelic discrimination platform and cannot detect novel variants, it is appropriate for a clinical setting evaluating well-studied variants. The other limitation is the genotyping for triallelic variants, which requires interpretation of a combination of 2 assays. However, triallelic variants are uncommon. It has been reported that there are 0.18% triallelic variants registered in dbSNP (23, 24). In a study that explored 382 901 variants, 2002 (0.52%) triallelic sites were found (34). To the best of our knowledge, there are only 2 triallelic variants out of 478 variants (0.42%) on our OA-PGx panel, so this level of (manual) interpretation is acceptable. We believe that the OpenArray genotyping platform is a suitable option for preemptive pharmacogenomics clinical studies. Our OA-PGx panel is complemented by an assay for CYP2D6 as this gene has a highly complex pattern of genetic variants and it encodes a major drug-metabolizing enzyme. It has been reported that standard genotyping approaches may not be able to reliably genotype some of the variants in CYP2D6 (35, 36). To address this issue, we have previously validated and reported on an extensive CYP2D6 assay that is based on Invader and TaqMan copy number assays (15).

In conclusion, we evaluated a custom-designed pharmacogenomics panel and found that it reliably interrogated 437 variants, of which 113 variants on 45 genes were associated with 65 clinically actionable drugs. Clinically actionable results from selected variants on this panel are currently used in clinical studies employing pharmacogenomics for clinical decision-making (17–20).

Supplemental Material

Supplemental material is available at The Journal of Applied Laboratory Medicine online.

Nonstandard abbreviations

OA-PGx panel

OpenArray pharmacogenomics panel

SNV

single-nucleotide variant

CCL

Coriell Institute cell line

ADME

absorption, distribution, metabolism, and excretion

CPIC

Clinical Pharmacogenetics Implementation Consortium

CLIA

Clinical Laboratory Improvement Amendments

UC Molecular Lab

Molecular Diagnostic Laboratory, University of Chicago

OHSU

Oregon Health & Science University

MassARRAY

Sequenom MassARRAY iPLEX platform

1KGP

1000 Genomes Project

NTC

no template control

QC

quality control

Human genes

CYP2C19

cytochrome P450 family 2 subfamily C member 19

CYP2D6

cytochrome P450 family 2 subfamily D member 6

ADRB2

 

HLA-B

major histocompatibility complex, class I, B

RYR1

ryanodine receptor 1