|
|
||||||||
Molecular Diagnostics and Genetics |
1 Department of Pathology, and 2
Section of Clinical Pharmacology, Department of Medicine, Dartmouth Medical School and Dartmouth-Hitchcock Medical Center, Lebanon, NH.
3 Department of Pathology, Medical College of Wisconsin, Milwaukee, WI.
aAddress correspondence to this author at: Department of Pathology, Dartmouth-Hitchcock Medical Center, One Medical Center Drive, Lebanon, NH 03756. Fax 603-650-8590; e-mail jyeo{at}dartmouth.edu.
| Abstract |
|---|
|
|
|---|
Methods: We genotyped 200 anonymized human DNA samples with the Pyrosequencing® platform at the Medical College of Wisconsin and with the NanoChip platform at Dartmouth Medical School. We compared CYP2D6 genotypes and resolved samples with genotypic discrepancies with the Jurilab CYP2D6 duplication/deletion assay or with traditional DNA sequencing. The Jurilab assay is a long-range PCR assay used to evaluate sequence structures 3' of the CYP2D7 and CYP2D6 coding regions. For the NanoChip platform, we performed multipad addressing and duplicate runs to test the intra- and intercartridge precision, within- and between-run precision, and reproducibility of the defined genotypes.
Results: We used both platforms to genotype all 200 DNA samples for CYP2D6*3, *4, *5, *6, *7, *8, and gene duplication. The 2 methods showed 99.4% concordance in the genotyping results; we found only 8 discrepant genotypes among 1400 DNA analyses. Confirmatory molecular analysis of the discrepant genotypes revealed that the NanoChip assay showed better agreement. The imprecision of the NanoChip method (CV) was 8.9%17.7%.
Conclusions: This validation study of the NanoChip electronic microarraybased CYP2D6 genotyping assay revealed a CV <20% and good concordance with the Pyrosequencing method and a confirmatory sequencing method.
| Introduction |
|---|
|
|
|---|
40% of human hepatic phase I drug metabolism (1). CYP2D6 alone is responsible for the metabolism of 20%25% of prescribed drugs (2). CYP2D6 is a highly polymorphic enzyme of 497 amino acid residues, and >50 human CYP2D61
alleles have been described (3). CYP2D6*3, *4, *5, *6, *7, *8, *11, *12, *13, *14, *15, *16, *18, *19, *20, *21, *38, *40, *42, and *44 are nonfunctioning alleles, whereas *9, *10, *17, *36, and *41 reportedly have decreased, substrate-dependent activity (3)(4). Screening for CYP2D6*3, *4, and *5 identifies at least 95% of poor metabolizers in a white population (5)(6), with *4 being the most frequently found in this group (7). Pharmacogenetic tests can aid in the selection of the most appropriate drugs, dosage, and schedule for achieving therapeutic efficacy and preventing adverse drug reactions (8). Gasche et al. described a patient who developed life-threatening opioid intoxication after he was given small doses of codeine for the treatment of a cough associated with bilateral pneumonia. CYP2D6 genotyping later revealed this patient to be an ultrarapid metabolizer (UM)2 of codeine (9).
The Food and Drug Administration is encouraging the incorporation of pharmacogenomics into the clinical care of patients (10)(11). Drugs with amended labeling containing pharmacogenetic information include atomoxetine (Strattera®) (12), 6-mercaptopurine (Purinethol®), azathioprine (Imuran®), and irinotecan (Camptosar®).
Nanogen introduced the NanoChip® Molecular Biology Workstation (MBW) for concurrent detection of polymorphisms in multiple samples (13)(14)(15)(16). Roche and Affymetrix marketed the AmpliChip CYP450 test, which the Food and Drug Administration approved for the detection of 29 known polymorphisms in the CYP2D6 gene, including gene duplications and deletions, as well as 2 variations in the CYP2C19 gene (17). Although a number of commercially available genotyping platforms now exist, we are not aware of any reports that directly compare cytochrome P450 genotyping assays on 2 or more platforms.
| Materials and Methods |
|---|
|
|
|---|
We originally extracted human DNA from whole blood with the PureGene® DNA purification reagent set (Gentra Systems) and then measured the ratio of the optical densities at 260 nm and 280 nm by spectrophotometry to quantify the DNA sample and check its purity.
genotyping comparison with the nanochip mbw
We regenotyped the same 200 human DNA samples at Dartmouth with the NanoChip MBW. We performed 2 PCRs for each DNA sample with Nanogens proprietary primers. We first amplified a fragment from CYP2D6 and CYP2D8 to determine the CYP2D6 duplication/deletion status (amplicon A). Second, we amplified 2 different fragments, which were used to detect single-nucleotide polymorphisms (SNPs) in CYP2D6 (amplicon B). We desalted these amplicons with the Multiscreen® vacuum manifold (Millipore) and resuspended them in 130 µL of 50 mmol/L L-histidine.
We transferred 60 µL of each amplicon A product to a 96-well plate and conducted a single genotyping run to detect CYP2D6 gene duplication/deletion. To check for intracartridge reproducibility, we electronically addressed the amplicons to 2 separate sites on a 100-site NanoChip microarray (NanoChip H2 cartridge) with the MBW Loader (Nanogen). After addressing, we saturated the microarray with high-salt buffer (500 mmol/L sodium chloride and 50 mmol/L sodium phosphate) for 1 min and then removed the buffer. We infused a set of Nanogens proprietary reporters into the microarray for the duplication/deletion assay and detected the addressed amplicons with the NanoChip Reader (see Fig. 1A in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol53/issue5). We achieved hybridization specificity for the reporters by means of a thermal reporting protocol involving an incremental reduction in temperature of the reporting mix over the microarray and a series of high-salt buffer washes at a single temperature. The addressing and reporting of amplicon A were repeated with a 2nd cartridge to check for intercartridge reproducibility of genotype identification. Forty different DNA samples were genotyped in a single cartridge.
SNP-detection assays for CYP2D6*3, *4, *6, *7, and *8 (see Table 1 in the online Data Supplement) were performed with amplicon B products (see Fig. 1B in the online Data Supplement). We used the same addressing procedure but sequentially infused specific Nanogen-provided reporters (sequence information proprietary) for each of the 5 alleles into the cartridge, with washes of low-salt buffer (50 mmol/L sodium phosphate) between reporter infusions. The addressing and reporting of amplicon B products were repeated with a 2nd cartridge to check for intercartridge reproducibility of genotype identification. We performed a new set of PCRs after using the amplicons in the first PCR set and repeated the loading and reporting procedures to check the reproducibility of the results.
resolution of discrepancies
We compared the CYP2D6 genotyping results on the 2 platforms for each DNA sample. Samples with genotype discrepancies in the duplication/deletion assays were retested with a commercially available CYP2D6 duplication/deletion PCR assay (Jurilab; see in the online Data Supplement). This assay used long-range PCR amplification and gel-based detection to determine CYP2D6 duplications and deletions. In brief, we used 27.6 µL of CYP2D6 duplication/deletion PCR mixture (Jurilab) and 0.4 µL of HotStarTaq DNA polymerase (Qiagen) per sample and added
40 ng of DNA sample to each PCR reaction. Amplicons were then run on an agarose gel (6 g/L) and imaged with an AlphaImager 2000 (Alpha Innotech Corporation). We resolved samples with discrepancies in CYP2D6*3, *4, *6, *7, and *8 polymorphisms via bidirectional sequencing of the appropriate genomic DNA fragments.
genotyping precision of the nanochip mbw
We conducted independent precision studies for the duplication/deletion and SNP assays with the NanoChip methodology. We amplified and addressed 3 DNA samples from Coriell Cell Repositories [1 CYP2D6 wild-type (WT) sample (NA11277), 1 deletion sample (NA07552), and 1 duplication sample (NA07441)] onto pads on a NanoChip H2 cartridge. Each sample was separately aliquoted and amplified to obtain 20 replicates. After sample addressing, we infused the set of reporters for the duplication/deletion assay into the cartridge and analyzed the genotypes with the NanoChip Reader. The experiment was repeated twice to check reproducibility. We followed the same procedure for the SNP assay with 1 CYP2D6 WT DNA sample (NA11277), 1 *4 heterozygous variant (HT) DNA sample (NA12960), and 1 *4 homozygous variant (HM) DNA sample (NA11281) from Coriell Cell Repositories.
| Results |
|---|
|
|
|---|
genotyping precision and comparison
The precision studies for replicate runs on 2 cartridges with the NanoChip CYP2D6 duplication/deletion and SNP assays yielded CVs of 8.9%17.7%. A typical run (n = 40) of a known sample with duplication in the duplication/deletion assay showed a mean green-to-red signal ratio (2 SDs) of 1.51 (0.32). An outlier showed a 1.16 green-to-red ratio, and a known sample heterozygous for CYP2D6*4 showed a green-to-red ratio of 0.87 (0.3) (Table 1
).
|
There was no change in CYP2D6 genotype for any of the 200 human DNA samples for assays conducted on different NanoChip cartridges, indicating excellent intercartridge reproducibility. We detected no genotyping discordance for the same sample in repeat runs on the NanoChip MBW (n = 8); this result signified 100% reproducibility in genotyping results with this system.
We found 99.4% concordance (8 genotypic discrepancies among 1400 DNA analyses) between the Pyrosequencing and NanoChip platforms in the CYP2D6 genotyping results (Table 2
). We attempted to resolve discrepancies in the results of the assay for duplication/deletion of the CYP2D6 gene by means of the Jurilab CYP2D6 assay. Expected DNA amplification products for WT, *5 HT, *5 HM, and CYP2D6 duplication/multiplication were obtained (see Table 2 and Fig. 2 in the online Data Supplement).
|
We could not unambiguously resolve 2 samples (DMS 4 and 7) with respect to CYP2D6 duplication/deletion (Table 2
). The Jurilab assay reported that DMS 4 was *5 WT, in agreement with the Pyrosequencing results, but the fluorescence signals for the NanoChip assay indicated a deletion (Fig. 1
). Because DMS 4 was also homozygous for the *2 allele in the NanoChip assay (Fig. 1
), the overall call is consistent with either *2 HM/*5 WT, or *2 HT/*5 HT. For the DMS 7 sample, the Jurilab assay found no duplication, in agreement with the Pyrosequencing result, but both the NanoChip duplication/deletion assay and the *2 SNP assay showed a red-to-green ratio of fluorescence signals indicative of CYP2D6 duplication, with 1 copy of the *2 allele and 2 copies of the *1 (WT) allele (Fig. 1
). In evaluating CYP2D6 duplication/deletion, such discrepancies tend to be more difficult to resolve because no available method can conclusively define the correct genotype. Further investigation is required to confirm the genotypes of these 2 samples, given the limitation of long-range PCR analysis.
|
comparison of assay throughput
The Pyrosequencing platform and the NanoChip MBW use different technologies for analyzing CYP2D6 SNPs and gene duplication/deletion. The total time required for the analysis was 9 h 48 min on the Pyrosequencing platform and 11 h 18 min on the NanoChip MBW (see Table 3 in the online Data Supplement). Required operator times were 3 h 30 min for the NanoChip MBW and 5 h 35 min for the Pyrosequencing platform.
| Discussion |
|---|
|
|
|---|
Pyrosequencing assays can be run in the simplex mode (1 sequencing primer) or the multiplex mode (up to 3 sequencing primers). In the multiplex mode, the presence of a rare mutation within the reading frame of one of the sequencing primers may interfere with the assay, thereby causing unexpected reference-peak heights that produce a "check" call, requiring the operator to verify the result. Manual interpretation of such check calls may be incorrect, and confirmation requires repeat testing in the simplex mode. For the DMS 136 sample, a manual Pyrosequencing call of *6 HT was made, but the NanoChip assay reported it as *6 WT. Sequencing of the appropriate DNA fragment revealed *6 WT, but a SNP at position 1704 (1704C>G, CYP2D6*28) (see Fig. 4 in the online Data Supplement) might have contributed to the *6 HT call in the Pyrosequencing multiplex analysis, because the SNP at position 1704 would contribute signal to some of the Pyrosequencing peaks in the multiplex analysis used for making the genotyping call for *6 (1707delT). When this sample was run as an individual simplex assay with Pyrosequencing, all SNPs were scored correctly, however. Thus, there is some risk that an unknown mutation may interfere with Pyrosequencing genotyping in the multiplex format because of the assays 3-SNP detection limit.
Discrepancies in DMS 109, 150, and 167 were due to buffer contamination or operator error during the Pyrosequencing runs, and the results of repeat runs were consistent with the corresponding NanoChip runs.
The Jurilab long-range PCR/gel-based assay was used to resolve duplication/deletion discrepancies; however, we discovered a limitation of the Jurilab assay in the course of this study. One study sample (DMS 85) was called as *5 HT in the NanoChip duplication/deletion assay, but Pyrosequencing scored the signals as a check call. The operator was inclined to interpret the result as *5 HT. The Jurilab assay called it as *5 WT; however, further workup with an improvised PCR/gel assay and sequencing at Nanogen revealed the sample to be *16 HT, which is a CYP2D7/CYP2D6 hybrid (20). Because CYP2D6*16 is a hybrid of CYP2D7 and CYP2D6 that includes the CYP2D7 frameshift that renders the protein product nonfunctional, this allele is appropriately called a CYP2D6 deletion. CYP2D6*16, however, is not detected as a CYP2D6 deletion by common long-range PCR assays, including the Jurilab assay. The Nanogen duplication/deletion assay, which compares the relative fluorescence signals from CYP2D6 and CYP2D8, will genotype the *16 allele as *5 HT. Because CYP2D6*16 is a nonfunctioning allele similar to *5, the predicted phenotype does not change, regardless of whether this sample is called *1/*5 or *1/*16. The potential for long-range PCR assays like the Jurilab assay to miss some relatively rare CYP2D6 deletions and duplications has previously been reported (21)(22).
The NanoChip assay correctly genotyped 6 of the 8 discrepancies; identification of the genotypes for the other 2 samples (DMS 4 and 7) requires further investigation. The discrepancies observed in this study demonstrate that interpreting genotyping results is very important in such assays. Some platforms may have software capable of converting fluorescence signals to genotypic calls, but manual interpretation by skilled operators is still very important, especially when the signals are not confidently scored or are low in intensity. Although bidirectional sequencing of the appropriate DNA fragment can resolve SNP discrepancies, there is currently no gold standard for verifying CYP2D6 duplication or deletion.
We used published genotype-phenotype relationships to place individuals into 4 phenotypic subgroups: ultrarapid metabolizers (UMs), extensive metabolizers (EMs), intermediate metabolizers (IMs), and poor metabolizers (PMs) (23)(24)(25). In this study, UMs carry at least 3 functional CYP2D6 alleles (CYP2D6*1), and EMs, IMs, and PMs have 2, 1, and 0 CYP2D6*1 alleles, respectively (Table 3
). On the basis of genotype-phenotype relationships, 2 (1%) of 200 samples genotyped with the NanoChip platform were UMs, 118 (59%) were EMs, 70 (35%) were IMs, and 10 (5%) were PMs, with *4 being the most frequently observed nonfunctional allele. This distribution of CYP2D6 genotypes is consistent with the frequencies in cohorts of Northern European ancestry, but not in those of Asian, Middle Eastern, African, or Mediterranean ancestry. Of the 8 samples with discrepant CYP2D6 genotyping, 7 would have differences in predicted phenotypes according to the genotypes obtained on the 2 platforms (Table 4
). Six of the 7 samples had a phenotypic change from IM to EM or vice versa. DMS 7 had an EM genotype by Pyrosequencing and a UM genotype by the NanoChip method. Incorrect calls clearly have implications for the dosing of CYP2D6 drug substrates, because UMs often require higher drug doses for therapeutic efficacy (except for CYP2D6 prodrugs). Such genotypic discrepancies thus would have clinical implications if these samples were clinical samples and if the patients predicted drug-metabolism phenotypes were being used to make drug and/or dosing decisions, particularly with drugs with a narrow therapeutic index. It is critical, therefore, to validate genotyping accuracy and to understand the limitations of an assay platform in the clinical setting. Our findings suggest that all assays with ambiguous genotyping results be repeated and verified with an alternative technology if the results remain ambiguous.
|
|
Our assay times and throughput data (see Table 3 in the online Data Supplement) reflect the practices in the Dartmouth and Wisconsin laboratories that performed the CYP2D6 genotyping assays on these 2 platforms. Two thermocyclers were used with the Pyrosequencing platform for practical reasons, so the total assay time with a single thermocycler would be
2 h longer. Although the NanoChip MBW appeared to require a longer assay time than Pyrosequencing, it required less operator time. Amplicon loading onto the microarray cartridge was fully automated once the amplicons were pipetted into the wells of a 96-well plate. The only labor-intensive part of the NanoChip assay was the reporting of individual SNPs, because each set of WT and variant reporters for a single SNP required manual infusion into the cartridge before the fluorescence signals could be reported. Because the operator time required for the NanoChip assay was less than with the Pyrosequencing assay, some laboratories may find this feature of the NanoChip MBW platform more desirable. It is likely that the demand for pharmacogenetic testing will increase, and genotyping platforms therefore will require greater automation and higher throughputs.
| Acknowledgments |
|---|
Financial disclosures: None declared.
Acknowledgements: The authors thank Drs. Ray Radtkey and Matthew Harris and the research team at Nanogen for developing and providing the PCR primers and reporter constructs for this study.
| Footnotes |
|---|
2 Nonstandard abbreviations: UM, ultrarapid metabolizer; MBW, molecular biology workstation; SNP, single-nucleotide polymorphism; WT, wild type; HT, heterozygous variant; HM, homozygous variant; EM, extensive metabolizer; IM, intermediate metabolizer; and PM, poor metabolizer. ![]()
| References |
|---|
|
|
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |