The Journal of Molecular Diagnostics, Vol. 17, No. 1, January 2015

jmd.amjpathol.org

TECHNICAL ADVANCE

High-Quality Genotyping Data from Formalin-Fixed, Paraffin-Embedded Tissue on the DrugMetabolizing Enzymes and Transporters Plus Array Hanneke I. Vos,* Tahar van der Straaten,y Marieke J.H. Coenen,z Uta Flucke,x D. Maroeska W.M. te Loo,* andHenk-Jan Guchelaary

From the Departments of Pediatric Hematology and Oncology,* Human Genetics,z and Pathology,x Radboud University Medical Center, Nijmegen; and theDepartment of Clinical Pharmacy & Toxicology,y Leiden University Medical Center, Leiden, the Netherlands

Accepted for publication

C

a

P

h

August 28, 2014.

Address correspondence toD. Maroeska W.M. te Loo,M.D., Ph.D., Department ofPediatric Hematology andOncology, Radboud UniversityMedical Center, PO Box 9101,6500 HB Nijmegen, the Neth-erlands. E-mail: maroeska.teloo@radboudumc.nl.

opyright ª 2015 American Society for Inve

nd the Association for Molecular Pathology.

ublished by Elsevier Inc. All rights reserved

ttp://dx.doi.org/10.1016/j.jmoldx.2014.08.003

The Affymetrix Drug Metabolizing Enzymes and Transporters (DMET) Plus array covers 1936 markers in231 genes involved in drug metabolism and transport. Blood- and saliva-derived DNA works well on theDMET array, but the utility of DNA from FFPE tissue has not been reported for this array. As the ability touse DNA from FFPE tissue on the array could open the potential for large retrospective samplecollections, we examined the performance and reliability of FFPE-derived DNA on the DMET Plus array.Germline DNA isolated from archived normal FFPE tissue blocks stored for 3 to 19 years and matchedblood or saliva from 16 patients with osteosarcoma were genotyped on the DMET Plus array. Concor-dance was assessed by calculating agreement and the k-statistic. We observed high call rates for boththe blood- or saliva-derived DNA samples (99.4%) and the FFPE-derived DNA samples (98.9%). More-over, the concordance among the 16 blood- or saliva-derived DNA and FFPE DNA pairs was high (97.4%,k Z 0.915). This is the first study showing that DNA from normal FFPE tissue provides accurate andreliable genotypes on the DMET Plus array compared with blood- or saliva-derived DNA. This findingprovides an opportunity for pharmacogenetic studies in diseases with high mortality rates and preventsa bias in studies where otherwise only alive patients can be included. (J Mol Diagn 2015, 17: 4e9;http://dx.doi.org/10.1016/j.jmoldx.2014.08.003)

Supported by ZonMW grant 113202010 (D.M.W.M.t.L.).T.v.d.S. and M.J.H.C. contributed equally to this work.Disclosures: None declared.

The contribution of genetic variation to interindividualvariability of drug efficacy and toxicity profiles is widelyacknowledged. The Affymetrix Drug Metabolizing Enzymesand Transporters (DMET) Plus array (Affymetrix UK Ltd,High Wycombe, UK) is a dedicated assay for pharmacoge-netic applications as it genotypes 1936 markers in 231 genesinvolved in drug absorption, distribution, metabolism,excretion, and transport.1,2 The array is based on molecularinversion probe (MIP) technology for the simultaneousgenotyping of multiple polymorphisms in a single assay.Blood and saliva samples are used as the standard source ofgermline DNA for genotyping.3 However, collection andstorage of germline DNA have not routinely been performedin most clinical trials in the past. In particular, in pharma-cogenetic studies in the field of oncology with cancers withlow survival rates, the lack of stored blood or saliva can result

stigative Pathology

.

in biased patient populations for research because the poorresponders to therapy are likely to be missing. In such cases,stored formalin-fixed, paraffin-embedded (FFPE) tissue isoften the only available source for germline DNA. However,in the process of fixation and storage of FFPE samples, DNAdegradation and DNA-protein cross-linkages result inreduced DNA quality, which can be challenging for PCR-based analysis.4,5 Therefore, the use of archived tissue,which can be highly valuable for biomarker studies, firstneeds to be validated for the specific analysis.6

For several Affymetrix single nucleotide polymorphism(SNP) arrays, the performance of FFPE-derived DNA has

Genotyping FFPE Tissue DNA on DMET Array

been compared with DNA from fresh frozen tissue,revealing a high degree of concordance; however, thesestudies have primarily focused on copy number variationsand chromosomal aberrations in tumor tissue.7e9 In addi-tion, Affymetrix has successfully demonstrated the suit-ability of FFPE-derived DNA for MIP technology usingboth normal and tumor FFPE samples.10 DNA from FFPEtissue may also be suitable for genotyping by the DMETarray; however, the DMET Plus array differs from standardMIP arrays because it performs an initial 36-plex multiplexPCR to preamplify genes known to have pseudogenes orclose homologs. To date, no studies have reported the utilityof DNA from FFPE samples as a template for the DMETPlus array, and no comparison of genotype performancewith DNA from blood or saliva has been made. In thepresent study, the performance and reliability of FFPE-derived DNA on the DMET Plus array are assessed usinga series of matched blood or saliva and normal FFPEsamples from patients with osteosarcoma.

Materials and Methods

DNA Extraction

From 16 patients with primary, high-grade osteosarcomatreated at the Radboud University Medical Center (Nijme-gen, the Netherlands) DNA was isolated from archivednormal FFPE tissue and blood or saliva after obtainingwritten informed consent from the patients or parents. Ifclinically available, blood was used as the primary sourcefor reference DNA; otherwise, saliva was used. The studywas approved by the local ethics committee, and all tissuewas handled according to the Code Proper Use of HumanTissue in the Netherlands (Federation of Dutch MedicalScientific Societies, http://www.federa.org/codes-conduct,last accessed April 16, 2014).

DNA was isolated from fresh EDTA whole blood sam-ples (n Z 13) using the QIAamp DNA Blood Midi kit(Qiagen, Venlo, the Netherlands) according to the manu-facturer’s protocol. Saliva was collected (n Z 3) using theOragene DNA saliva collection kit (DNA Genotek Inc.,Kanata, ON, Canada), and DNA was extracted followingthe manufacturer’s instructions.

FFPE tissueederived DNA from the same patients wasobtained using stored FFPE tissue blocks from surgicalresection specimens. At time of resection, fresh tissue wasfixed in 4% buffered formalin, processed routinely, andembedded in paraffin as described previously.11 At the timeof DNA isolation, the FFPE tissue blocks had been stored atroom temperature for 3 to 19 years. Before DNA isolation,all tissue blocks were assessed by an experienced pathologist(U.F.) using hematoxylin and eosinestained slides. From aseries of tissue blocks with optically distinguishable normaland/or malignant tissue, FFPE tissue blocks that containedonly normal tissue were selected to ensure isolationof germline DNA only. All tissue blocks contained

The Journal of Molecular Diagnostics - jmd.amjpathol.org

nondecalcified, normal soft tissue, including lung, muscle,skin, lymph node, or a combination. The FFPE tissue sam-ples were sectioned using a microtome, which was cleanedwith ethanol between cases to avoid contamination. Ten 20mm tissue sections were used for isolation as described byHagleitner et al,11 with some adjustments. Samples wereincubated in 800 mL of 5% Chelex-100 (Bio-Rad, Vee-nendaal, the Netherlands) in TET lysis buffer [5 mL of 1mol/L Tris HCl, pH 8.5, 1 mL of 0.5 mol/L EDTA, pH 8.0,250 mL of 20% Tween-20 (Merck, Amsterdam, theNetherlands), and 493.75 mL of sterile water] and 80 mL ofproteinase K solution (20 mg/mL; Roche Diagnostics,Almere, the Netherlands) for 16 hours at 56�C in a Ther-momixer (350 rpm), followed by 48 hours at 37�C with theaddition of 80 mL of proteinase K at 0 and 24 hours. After asubsequent 10 minutes of incubation at 95�C, the sampleswere centrifuged for 10 minutes at 16,000� g and cleared byremoval of the supernatant twice. Addition of 8 mL of RNaseA (100 mg/mL; Qiagen catalog no. 19101) to the supernatantwas followed by 30 minutes of incubation at room temper-ature and 10 minutes at 80�C. The DNA was then purifiedusing the QIAamp DNA Micro Kit (Qiagen), with bufferAW2 and 20 mL of AE applied twice instead of once.

Quality Control of DNA Samples

To assess the quality of the DNA samples, a qualitative sizerange PCR was performed as described previously.11 In brief,50 ng of genomic DNA (based on NanoDrop measurementsusing NanoDrop 2000; Thermo Scientific, Breda, theNetherlands) from FFPE tissue (n Z 16) and one representa-tive blood sample was included. The assay contained fiveprimer sets that amplify fragments that range from 100 to 400bp, as described previously.12MilliQ was included as negativecontrol and 50 ng of DNA from frozen tissue sections as apositive control for the assay. The PCR products were visu-alized and recorded in an ethidium bromideestained 2%agarose gel using the Imago (Isogen Bioscience, IJsselstein,the Netherlands).

Genotyping

Thirty-two DNA samples were genotyped using the DMETPlus array according to the manufacturer’s instructions(Affymetrix UK Ltd).1 All samples were normalized to 60ng/mL based on NanoDrop measurements, and 17 mL (1.02mg of DNA) per sample was used for genotyping. Genotypeswere calculated with DMET console software version 1.3(Affymetrix UK Ltd) using the Dynamic Genotype Bound-aries algorithm version 2.

In addition, the 16 blood- or saliva-derived DNA sampleswere genotyped for 381 markers, of which 13 SNPs wereoverlapping with the DMET Plus panel, using a custom-basedassay on the Illumina BeadArray platform (Illumina Inc., SanDiego, CA). Genotypes were calculated using IlluminaGenome Studio software version 2008.1 (Illumina Inc.).

5

Figure 1 Quality of DNA isolated from formalin-fixed, paraffin-embedded (FFPE) tissue. The size ranges of the PCR products of 16 DNAsamples from FFPE tissue are given. Lane B is the representative DNAsample from blood (matching with FFPE sample no. 7); lane M, 100-bpladder.

Table 1 Genotyping Call Rate and Concordance per Blood- orSaliva-Derived DNA and FFPE DNA Pair

Sampleno.

Call rate, % Genotype concordance, %

Blood orsaliva

FFPEtissue

Excludingno calls

Excludingno calls andPRAs

1 99.48 99.38 99.48 99.842 99.74 99.07 99.37 99.843 99.59 99.64 99.48 99.844 99.07 99.69 99.37 99.895 99.22 99.43 99.42 99.896 99.33 96.32 97.46 98.427 98.81 99.27 99.26 99.798 98.91 98.55 98.78 99.259 99.17 99.17 98.89 99.3710 99.48 99.38 99.16 99.6811 99.48 99.43 98.90 99.5312 99.53 99.22 98.85 99.3113 99.53 98.76 98.84 99.2614 99.90 99.17 98.90 99.3715 99.12 96.53 98.54 98.9716 99.69 99.74 99.48 99.69

Sample call rates and genotype concordance for DNA derived from blood(samples 1 to 13), saliva (samples 14 to 16), and matching FFPE tissue onthe Affymetrix Drug Metabolizing Enzymes and Transporters Plus array.FFPE, formalin-fixed, paraffin-embedded; PRA, possible rare allele.

Vos et al

Statistical Analysis

Genotyping results of the 16 DNA pairs from blood orsaliva and FFPE tissue were evaluated for call rate and SNPand copy number variation concordance. Call rates werederived from the DMET console software calculation,which is the percentage of SNPs successfully genotyped persample, excluding no calls but including possible rare allele(PRA) calls. PRA calls are assigned by the DMET consolesoftware to probe signals that fall out of the range of thetraining data because the variant alleles were rarely or notobserved in assay development. For each marker, theconcordance between the genotypes obtained from theblood- or saliva-derived DNA and the FFPE-derived DNAwas determined by calculation of the percentage of geno-type agreement and the k-statistic. The k-statistic measuresagreement beyond chance alone; a k-statistic of 0 indicateschance agreement, whereas a k-statistic of 1 indicatescomplete agreement. The overall agreement and overall k-statistics were calculated as the mean of the agreement and kof all markers. The analyses were performed using Rversion 3.0.2 (http://www.r-project.org) and included nocalls and PRA calls. In addition, genotype concordance wascalculated for each DNA pair with and without inclusion ofPRA calls (ie, the percentage of markers with the samegenotype call for DNA from blood or saliva and FFPE tissueof the total number of markers, excluding those with nocalls and/or PRA for DNA from either source). Calculationsper pair were performed using Microsoft Excel (MicrosoftCorp, Redmond, WA).

Assessment of associations of SNPs with DNA source(blood or saliva versus FFPE tissue) was performed by lo-gistic regression analysis using PLINK software version1.07 (http://pngu.mgh.harvard.edu/purcell/plink, last accessedApril 14, 2014).13 P < 0.05 was used as significance level,without correction for multiple testing because of the explor-atory nature of this analysis.

Results

To assess the suitability of DNA isolated from FFPE tissuefor the DMET Plus array, germline DNA derived fromFFPE tissue and matched blood or saliva samples as high-quality reference material was collected from 16 patientswith osteosarcoma. The average yield of the FFPE-derivedDNA samples was 12.2 mg (range, 1.93 to 36.8 mg); hence,

6

isolation yielded sufficient DNA from all FFPE samples fordownstream array analysis.

Quality of DNA Isolated from FFPE Tissue

Prior to DMET analysis, the quality of all FFPE tis-sueederived DNA samples was evaluated by size rangePCR. The reference blood DNA revealed bands up to 400bp using gel electrophoresis, whereas the FFPE DNAsamples showed amplicon sizes ranging from 100 to 300 bp,with a reduced intensity of the largest fragment visible inmost cases (Figure 1).

Genotype Call Rates

All DNA samples from blood, saliva, and FFPE tissuewere successfully genotyped by the DMET Plus array. Themean � SD call rate for all blood- or saliva-derived DNAsamples was 99.4% � 0.30% (Table 1). For the FFPE-derived DNA samples, the mean � SD call rate was98.9% � 1.0%. Fourteen FFPE DNA samples had callrates >98%, which is the threshold recommended by themanufacturer (Table 1).1 The two FFPE DNA sampleswith lower call rates (96.32% and 96.53%) were, however,still within a range suitable for genetic association studies,in which a minimum of 95% is considered acceptable.Although these samples were also observed to have smallamplicon sizes in the size range PCR (maximum, 100 bp),indicative of reduced DNA quality, this correlation was not

jmd.amjpathol.org - The Journal of Molecular Diagnostics

Genotyping FFPE Tissue DNA on DMET Array

consistent among the other samples with similar fragmentsizes. However, a reduction in the signal intensity of the100-bp band was observed specifically in the two sampleswith lower call rates.

Genotype Concordance

To analyze the reliability and accuracy of the genotypecalling of FFPE-derived DNA samples by the DMET Plusarray, the genotype concordance was calculated for thepaired blood- or saliva-derived DNA and FFPE DNA gen-otyping results (Table 2). Of the 1931 markers, 1061showed no genetic variation in our population, which isexpected based on the known minor allele frequencies andprevious DMET studies.14e16 The genotype concordancewas calculated with all markers and with only the 870polymorphic markers. The overall agreement with allmarkers included was 97.4% with a k-statistic of 0.915. Thek-statistic calculated per marker ranged from �0.143(TPMT rs1800460, CYP21A2 rs72552757) to 1 (eg, CHST3rs4148943, FMO3 rs1736557), excluding the markers thatwere monomorphic, for which the k-statistic was alsodefined as 1. Analysis of the concordance of the 35 geneswith a very important pharmacogene (VIP) status(PharmGKB, http://www.pharmgkb.org, last accessed July14, 2014) covered by the DMET Plus array revealed anoverall concordance of 96.6% with a k-statistic of 0.884,ranging from a mean of 68.8% for the markers in SLC19A1to 100% in eg, COMT and GSTT1 (Supplemental Table S1).

In addition, the genotype concordance was calculated perblood- or saliva-derived DNA and FFPE DNA pair to assessthe potential link between concordance and FFPE DNA callrate (Table 1). Genotype concordance excluding no callsand PRAs exceeded 99.2% with the exception of the twopairs of which the FFPE DNA samples had call rates <98%.Of seven markers miscalled in more than five blood- orsaliva-derived DNA and FFPE DNA pairs, the cluster plotsof four markers were considered unreliable. Without thesemarkers, the genotype concordance would be somewhathigher; therefore, the reported concordances are slightlyunderestimated. Furthermore, the copy number calling was100% concordant across the samples.

The genotypes obtained from the blood- or saliva-derivedDNA samples were validated for a subset of 13 SNPs,which were included on the Illumina BeadArray platform aswell. Genotypes of these SNPs (rs1045642, rs1128503,rs2231142, rs1801243, rs1005695, rs3787728, rs1056892,

Table 2 Overall Genotyping Concordance

Marker Agreement, % k-Statistic

All markers 97.4 0.915Polymorphic markers 94.3 0.812

Overall agreement and k-statistic were calculated with all 1931 markersand with only the 870 polymorphic markers: markers for which the minorallele was detected in at least one sample.

The Journal of Molecular Diagnostics - jmd.amjpathol.org

rs776746, rs1138272, rs1800566, rs689453, rs10898,rs4407290) were 100% concordant with the genotypes ob-tained using the DMET Plus array.

Discordant Results

Of the 799 mismatches among the 30,896 SNP-sample paircomparisons, most (81.1%) involved no calls or PRA calls(ie, switches from a genotype call in blood or saliva DNA tono call or PRA call in FFPE tissue DNA or vice versa andPRA call to no call or vice versa). The remaining genotypeto genotype switches were dominated by switches of ho-mozygous calls to heterozygous calls (11.01%), followed byheterozygous to homozygous switches (6.26%), homozy-gous to homozygous (either common or variant allele)switches (1.13%), and allele deletions or gains (0.5%).

Of all markers, those with the 20% lowest k values (allk < 0.4) were considered poorly concordant, excludingmarkers with a low k but agreement >80% because forthose markers the k value is strongly affected by a lowminor allele frequency. Visual inspection revealed that only9 of the 42 markers have a call rate <80% across all 32samples, indicating that the marker concordance is nothighly linked to the general marker call rate. For 28 of 42markers, the cluster plots were considered unreliable(Supplemental Table S2).

In addition, we assessed whether the discordant resultswere randomly distributed by testing for associations of the1931 markers with DNA source (blood or saliva versusFFPE tissue), including genotype calls only. Thirteenmarkers had significant differences in minor allele fre-quencies between blood or saliva and FFPE DNA samples(P < 0.05). Of these markers, nine were excluded becauseof unreliable cluster plots (CYP11B1 rs4534, CYP2A6rs4079369, CYP2A7 rs4079366, CYP2B6 rs36079186,CYP2C9 rs1057910, CYP2D6 rs5030865, CYP2D6rs61736512, SULT1A2 rs1059491, TBXAS1 rs8192868),leaving four SNPs (FMO6 rs2272797, CYP2B6 rs2279343,CYP39A1 rs2277119, and GSTM1 rs74837985), with onlythe latter (agreement, 56.3%; k Z 0.420) not overlappingwith the top 42 markers with lowest concordance.

The 42 markers with the most discordant results alsoincluded markers located in genes for which the DMET Plusarray performs automated haplotype calling and phenotypetranslation. Because calling and translation automaticallyinvolve all markers per gene, discordances in haplotype andsubsequent phenotype results were observed in one or moresample pairs for CYP2B6, CYP2C9, CYP2D6, SLCOB1,TBXAS1, TPMT, and UGT1A1.

Discussion

There is a large worldwide collection of stored FFPE tissue,which can be highly valuable for providing germline DNAfor genetic studies. In this study, we examined the utility of

7

Vos et al

DNA isolated from normal FFPE tissue blocks for analysison the DMET Plus array, which is a useful tool for phar-macogenetic studies. We found that the DNA isolated fromFFPE tissue performs well on the array and yields high-quality data. Not only did we observe high call rates, butimportantly we also noted that the concordance with geno-typing results of matching blood or saliva samples was high.These data indicate that DNA from FFPE tissue can bereliably used as a template for the DMET Plus array.

Factors related to tissue fixation and storage affect thequality of the DNA that can be recovered from the spec-imen. SNP arrays based on MIP technology may be anoptimal solution for working with such degraded DNAbecause of the small target DNA sequences (minimumtarget length of 40 bp). A previous study by Wang et al10

using an Affymetrix MIP array with a panel of 50,000markers found a mean call rate of 98.9% for DNA fromnormal FFPE tissue samples, similar to the mean call rateobserved in the current study. Hence, our data for theDMET Plus array extend these observations, suggesting thateven though the DMET procedure includes a multiplex PCRstep, this MIP-based array is indeed suitable for the suc-cessful genotyping of DNA of reduced quality. The com-bination of the multiplex PCR step to preamplify regions ingenes with pseudogenes and close homologs and the MIPtechnology has previously been found to yield reliable ge-notypes in these challenging genes (eg, CYP2D6 andUGT1A1) in a validation study of the DMET Plus arrayusing blood samples.17 Our study reveals that reliable ge-notypes can be obtained for markers of these challenginggenes even in the context of FFPE-derived DNA.

In our study, we obtained reliable genotyping results fromFFPE-derived DNA samples on the DMET Plus array, evenfor the samples with degraded DNA based on the size rangePCR. Apparently, the primers for the DMET Plus array havebeen designed in such a way that they are very close to theSNP of interest. On the basis of our test panel, the pre-screening of samples based on the size range PCR does notappear to be an essential control to determine whethersamples are of sufficient quality for further DMET analysis.We observed two samples with relatively low call rates,which corresponded to a mild decrease in concordancecompared with samples with call rates >98%. Therefore, wesuggest evaluating the call rates of FFPE-derived DNAsamples before further genetic analyses because we foundthat low call rates are related to a lower concordance be-tween blood or saliva and FFPE tissue. Therefore, the callrates are related to the reliability of the data. In case callrates of a small number of samples are too low for reliableassociation analysis, these samples could still be included ina meta-analysis together with the DMET genotyped samplesby specifically genotyping the significantly associatedmarkers using other genotyping methods (eg, TaqManallelic discrimination assays).11

The analysis of the discordant results indicated that certainSNPs were nonrandom miscalled. Despite the limited

8

number of samples in the association analysis, it cannot beexcluded that genotyping results for the four SNPs that werefound to be associated with DNA source are systematicallyunreliable in FFPE DNA samples. Therefore, we suggestcaution when including these SNPs in association analyseswith the (clinical) outcome of interest. The same applies tothe reported list of markers with poor concordance, sincethese markers are prone to discrepancies as well. However, asindicated, some of these markers show unreliable genotypeclustering, which might not only explain the sensitivity tosample quality variation but also indicate the necessity toexclude them from further analyses regardless of the sampletype used as DNA source. The VIP genes covered by theDMET Plus array had good overall concordance; for thoseVIP genes that had the least agreement, this could beexplained by the presence of markers reported here to havepoor concordance. As subsequently expected, the discordantmarkers were found to affect the haplotype calling andphenotype translation as performed by the DMET consolesoftware, which is explained by absent or incorrect genotypecalls in the FFPE DNA samples. Hence, exclusion of thereported discordant markers is warranted in manual trans-lation but could still affect reliability because of the missinggenotypes. However, of the alleles that had poor concordanceand that are used in clinical guidelines, most are expected tohave a low impact because of their low allele frequency.We used only normal FFPE tissue sections to isolate

germline DNA, avoiding different genotyping results inDNA from FFPE tissue compared with blood or salivabecause of somatic alterations in tumor DNA. The issue ofusing tumor DNA in pharmacogenetic studies has beensubject of debate, and although germline DNA is preferred,recent studies have provided evidence that DNA from FFPEtumor tissue blocks may be used as a valid alternative inpharmacogenetic cancer studies when no other samples areavailable.18,19 However, in the current study, discordantresults that may be produced when using FFPE tumor tissueon the DMET Plus array are not explored.Our published optimized protocol for DNA isolation from

FFPE tissue for TaqMan SNP assays could also optimallybe used in the current study for the generation of samplessuitable for downstream use on the DMET Plus array.11 Forsingle SNP assays, not only nondecalcified but also decal-cified FFPE tissue has been shown to be useful. However,for the DMET analysis 1.02 mg of DNA (60 ng/mL) isneeded per sample, which is often difficult to obtain fromtissues decalcified with formic acid because this treatmentmakes the recovery of DNA even more challenging.Therefore, no samples from decalcified FFPE tissue couldbe included in the current study. A similar limitation of theuse of FFPE-derived DNA for the DMET array may alsoarise when researchers are restricted to tissues with lowabundance of nuclei, from which the necessary quantity ofDNA cannot be recovered. Whole genome amplificationbased on multiple displacement amplification has been re-ported to provide accurate results on the DMET Plus array

jmd.amjpathol.org - The Journal of Molecular Diagnostics

Genotyping FFPE Tissue DNA on DMET Array

and may provide a solution, although caution should betaken when using input material of low quality.20 To date,there are no reports on the reliability of whole genomeamplification of FFPE-derived DNA for use on the DMETPlus array.

In conclusion, we found that high-quality genotype datacan be obtained using the DMET Plus array on DNA iso-lated from FFPE samples. This finding provides opportu-nities for pharmacogenetic studies to make use of archivaltissue resources to analyze individuals not otherwise avail-able for sampling.

Acknowledgments

We thank Marco Tiller for technical assistance, RogierDonders for the data analysis in R, Frank van Leeuwen andHans Gelderblom for contributions to discussion of the data,and Peter Hoogerbrugge, Winette van der Graaf, and BartSchreuder for contributions to patient inclusion.

Supplemental Data

Supplemental material for this article can be found athttp://dx.doi.org/10.1016/j.jmoldx.2014.08.003.

References

1. Burmester JK, Sedova M, Shapero MH, Mansfield E: DMET micro-array technology for pharmacogenomics-based personalized medicine.Methods Mol Biol 2010, 632:99e124

2. Sissung TM, English BC, Venzon D, Figg WD, Deeken JF: Clinicalpharmacology and pharmacogenetics in a genomics era: the DMETplatform. Pharmacogenomics 2010, 11:89e103

3. Hu Y, Ehli EA, Nelson K, Bohlen K, Lynch C, Huizenga P,Kittlelsrud J, Soundy TJ, Davies GE: Genotyping performance be-tween saliva and blood-derived genomic DNAs on the DMET array: acomparison. PLoS One 2012, 7:e33968

4. Chalkley R, Hunter C: Histone-histone propinquity by aldehyde fixa-tion of chromatin. Proc Natl Acad Sci U S A 1975, 72:1304e1308

5. Srinivasan M, Sedmak D, Jewell S: Effect of fixatives and tissueprocessing on the content and integrity of nucleic acids. Am J Pathol2002, 161:1961e1971

6. Simon RM, Paik S, Hayes DF: Use of archived specimens in evalua-tion of prognostic and predictive biomarkers. J Natl Cancer Inst 2009,101:1446e1452

7. Alvarez K, Kash SF, Lyons-Weiler MA, Kim HJ, Peterson LE,Mathai B, Hagenkord JM, Monzon FA: Reproducibility and perfor-mance of virtual karyotyping with SNP microarrays for the detection ofchromosomal imbalances in formalin-fixed paraffin-embedded tissues.Diagn Mol Pathol 2010, 19:127e134

8. Lyons-Weiler M, Hagenkord J, Sciulli C, Dhir R, Monzon FA: Opti-mization of the Affymetrix GeneChip Mapping 10K 2.0 Assay forroutine clinical use on formalin-fixed paraffin-embedded tissues. DiagnMol Pathol 2008, 17:3e13

The Journal of Molecular Diagnostics - jmd.amjpathol.org

9. Monzon FA, Hagenkord JM, Lyons-Weiler MA, Balani JP,Parwani AV, Sciulli CM, Li J, Chandran UR, Bastacky SI, Dhir R:Whole genome SNP arrays as a potential diagnostic tool for thedetection of characteristic chromosomal aberrations in renal epithelialtumors. Mod Pathol 2008, 21:599e608

10. Wang Y, Carlton VE, Karlin-Neumann G, Sapolsky R, Zhang L,Moorhead M, Wang ZC, Richardson AL, Warren R, Walther A,Bondy M, Sahin A, Krahe R, Tuna M, Thompson PA, Spellman PT,Gray JW, Mills GB, Faham M: High quality copy number and geno-type data from FFPE samples using Molecular Inversion Probe (MIP)microarrays. BMC Med Genomics 2009, 2:8

11. Hagleitner MM, Coenen MJ, Jeuken JW, Flucke U, Schreuder HW, teLoo DM, Hoogerbrugge PM: TaqMan genotyping assays can be usedon decalcified and paraffin-embedded tissue from patients with oste-osarcoma. Pediatr Blood Cancer 2011, 56:35e38

12. van Dongen JJ, Langerak AW, Bruggemann M, Evans PA,Hummel M, Lavender FL, Delabesse E, Davi F, Schuuring E, Garcia-Sanz R, van Krieken JH, Droese J, Gonzalez D, Bastard C, White HE,Spaargaren M, Gonzalez M, Parreira A, Smith JL, Morgan GJ,Kneba M, Macintyre EA: Design and standardization of PCR primersand protocols for detection of clonal immunoglobulin and T-cell re-ceptor gene recombinations in suspect lymphoproliferations: report ofthe BIOMED-2 Concerted Action BMH4-CT98-3936. Leukemia2003, 17:2257e2317

13. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA,Bender D, Maller J, Sklar P, de Bakker PI, Daly MJ, Sham PC:PLINK: a tool set for whole-genome association and population-basedlinkage analyses. Am J Hum Genet 2007, 81:559e575

14. Uchiyama T, Kanno H, Ishitani K, Fujii H, Ohta H, Matsui H,Kamatani N, Saito K: An SNP in CYP39A1 is associated with severeneutropenia induced by docetaxel. Cancer Chemother Pharmacol 2012,69:1617e1624

15. de Graan AJ, Elens L, Smid M, Martens JW, Sparreboom A,Nieuweboer AJ, Friberg LE, Elbouazzaoui S, Wiemer EA, van derHolt B, Verweij J, van Schaik RH, Mathijssen RH: A pharmacogeneticpredictive model for paclitaxel clearance based on the DMET platform.Clin Cancer Res 2013, 19:5210e5217

16. Ten Brink MH, Swen JJ, Bohringer S, Wessels JA, van der Straaten T,Marijt EW, von dem Borne PA, Zwaveling J, Guchelaar HJ: Explor-atory analysis of 1936 SNPs in ADME genes for association withbusulfan clearance in adult hematopoietic stem cell recipients. Phar-macogenet Genomics 2013, 23:675e683

17. Fernandez CA, Smith C, Yang W, Lorier R, Crews KR, Kornegay N,Hicks JK, Stewart CF, Kawedia JD, Ramsey LB, Liu C, Evans WE,Relling MV, Broeckel U: Concordance of DMET plus genotypingresults with those of orthogonal genotyping methods. Clin PharmacolTher 2012, 92:360e365

18. Dezentje VO, van Schaik RH, Vletter-Bogaartz JM, van derStraaten T, Wessels JA, Kranenbarg EM, Berns EM, Seynaeve C,Putter H, van de Velde CJ, Nortier JW, Gelderblom H, Guchelaar HJ:CYP2D6 genotype in relation to tamoxifen efficacy in a Dutch cohortof the tamoxifen exemestane adjuvant multinational (TEAM) trial.Breast Cancer Res Treat 2013, 140:363e373

19. van Huis-Tanja L, Kweekel D, Gelderblom H, Koopman M,Punt K, Guchelaar HJ, van der Straaten T: Concordance of geno-type for polymorphisms in DNA isolated from peripheral blood andcolorectal cancer tumor samples. Pharmacogenomics 2013, 14:2005e2012

20. He YJ, Misher AD, Irvin W Jr, Motsinger-Reif A, McLeod HL,Hoskins JM: Assessing the utility of whole genome amplified DNA as atemplate for DMET Plus array. Clin Chem LabMed 2012, 50:1329e1334

9