Kweekel, D.M.
Citation
Kweekel, D. M. (2009, May 26). Pharmacogenetics of irinotecan and oxaliplatin in advanced colorectal cancer. Retrieved from https://hdl.handle.net/1887/13820
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden Downloaded from: https://hdl.handle.net/1887/13820
Note: To cite this publication please use the final published version (if applicable).
COMPARISON OF GENETIC
POLYMORPHISMS IN DNA ISOLATED FROM BLOOD AND PARAFFIN EMBEDDED
COLORECTAL CANCER TISSUE
Dina M. Kweekel, Tahar Van der Straaten, Miriam Koopman, Gerrit A. Meijer, Johan W.R. Nortier, Cornelis J.A. Punt, Hans Gelderblom, Henk-Jan Guchelaar
submitted
ABSTRACT
PURPOSE:
Pharmacogenetic studies nearly always use genotypes determined in DNA from peripheral blood. Formaldehyde-fixed paraffin-embedded tissue (FFPE), which is routinely archived in many clinical trials, is potentially an interesting source of DNA for pharmacogenetic studies. However, concordance of germline drug metabolism pathway polymorphisms in DNA from peripheral blood and FFPE has not yet been investigated.
EXPERIMENTAL DESIGN:
11 genotypes relating to chemotherapy drug metabolism or efficacy (ABCB1 rs1128503, ERCC1 rs11615, ERCC2 rs13181 and rs1799793, GSTP1 rs1695, MTHFR rs1801133, TP53 rs1042522, SLC19A1 rs1051266 and XRCC1 rs25487) were determined in DNA from FFPE of colorectal tumors and blood by pyrosequencing and TaqMan. Kappa and z-statistics were calculated to assess concordance.
RESULTS:
149 paired FFPE tumor and blood DNA samples were available for comparison. Overall, 20 out of 1,420 genotypes were discordant (1.4%). None of the individual genotypes showed a discordance between tumor and blood significantly different from 0.0%, except GSTP1 (95%
CI: 0.1- 5.9%).
CONCLUSION:
This is the first study showing that, for most of the genotypes analyzed, the use of FFPE samples is a valid alternative to peripheral blood, and vice versa.
INTRODUCTION
The field of pharmacogenetics has developed rapidly over the recent years 1;2, as have techniques used for DNA isolation and single nucleotide polymorphism (SNP) genotyping.
Studies are either hypothesis-driven using a candidate-gene approach or they involve a whole genome approach with genotyping of a large number of genetic variants, without making any presumptions about the underlying mechanisms. Most studies use whole blood as a source of DNA, because it is relatively easy to obtain in a clinical setting. DNA extraction from whole blood is a simple and fast method, which yields large amounts of high quality DNA. Alternative sample types include buccal swabs or saliva which are ideal for both the use in non-clinical settings (e.g. for sampling at the patient’s home) and for shipping by mail 3-6, or (tumor) tissue. Tumor as a source of DNA is of special interest because tumor biopsies and/or resection specimens of large patient populations have been archived in pathology laboratories and are potentially useful for retrospective pharmacogenetic studies
7. Tumor tissue samples are composed of heterogeneous cell types (stroma and tumor cells), and may therefore include different DNAs within a single sample. Stroma cells contain DNA equivalent to germline DNA while tumor DNA on top of the germ line characteristics harbors somatic alterations, both in sequence and copy number, that are associated with carcinogenesis. Tumor tissue is most often banked as formaldehyde fixed paraffin-embedded (FFPE) material. Quality of DNA isolated from FFPE tissue depends on the duration of formaldehyde fixation, and whether or not the formaldehyde is buffered 8. Remnants of substances used during fixation or paraffin embedding (e.g. xylene) cause low DNA isolation yields 9. DNA damage can create problems with amplification, and primer recognition of templates. To circumvent these problems, amplicons need to be short and primers need to be designed in such a way that they are close to the region of interest. Indeed, genetic variations spanning more than a few base-pairs, such as repetitive elements in the well-known uridine- glucuronosyl transferase 1A1 enzyme (UGT1A1*28) or the thymidylate synthase (TYMS) 28 base-pair repeat, may therefore be more difficult to determine in FFPE samples as compared to blood. Indeed, pilot experiments showed that genetic variations in TYMS and UGT1A1 could not be determined in most of our FFPE samples, due to DNA destruction (data not shown). All of these arguments make peripheral blood the preferred DNA source for most pharmacogenetic studies. On the other hand, methods much more demanding on DNA quality than PCR, like comparative genomic hybridization arrays have been successfully performed using FFPE derived DNA 10;11. Also, one may argue that tumor DNA is more informative compared to peripheral blood, for instance when studying antitumor drug efficacy. Indeed, tumor DNA may have been altered during the process of tumor development
12 and this may explain a variable sensitivity of tumors to systemic treatments.
Concordance of germline drug metabolism pathway polymorphisms in peripheral blood and DNA from FFPE has not yet been investigated in colorectal cancer (CRC). Therefore, the aim of the present study was to compare the genotypes of a large set of paired blood and FFPE samples in CRC patients, for a broad selection of SNPs that are frequently used in pharmacogenetic studies. SNPs in dihydropyrimidine dehydrogenase (DPD), although of interest in CRC, were not selected because of low frequencies of the mutant alleles.
METHODS
SAMPLES
DNA was obtained from Caucasian CRC patients participating in the multi-center CAIRO-1 trial after written informed consent. The inclusion criteria and the clinical results of this study have been published elsewhere 13. EDTA-blood was stored at -20 ∞C before isolation;
FFPE samples were collected from multiple pathology laboratories in The Netherlands and stored under standard conditions. Germline DNA was isolated from EDTA-blood with the Magnapure LC (Roche Diagnostics, Almere, The Netherlands) according to the manufacturer’s instructions. Tumor DNA was obtained from FFPE samples of areas containing > 70% tumor tissue with QIAmp DNA micro-kit columns (Qiagen, Venlo, The Netherlands), using a protocol that has been optimized for obtaining high quality DNA 10.
SNP Target# Sequence 5’-3 Modification*
GSTP1 rs1695 PCR-f AGGACCTCCGCTGCAAATAC
PCR-r CTGGTGCAGATGCTCACATAGTT Biotin
Sequence primer-f CTCCGCTGCAAATAC Target Sequence A/GTCTCCCTCAT Dispensation order cAGaTCTCT
TP53 rs1042522 PCR-f GAAGACCCAGGTCCAGATGAAG Biotin
PCR-r CCGGTGTAGGAGCTGCTGG
Sequence primer-r GGTGCAGGGGCCACG Target Sequence C/GGGGGAGCAGCCT Dispensation order tGCGcAGCAG
XRCC1 rs25487 PCR-f TAAGGAGTGGGTGCTGGACTGTC Biotin
PCR-r CAGGGTTGGCGTGTGAGG
Sequence primer-r CGTGTGAGGCCTTACC Target Sequence TCC/TGGGAGGGCA Dispensation order gTCTcGAGC
# f = forward orientated, r = reverse orientated
Table 1
Primers and probes for pyrosequence analysis and fragment length analysis
GENOTYPING ASSAYS
The following SNPs were determined with TaqMan 7500 (Applied Biosystems, Nieuwerkerk aan den IJssel, The Netherlands) according to the manufacturer’s protocol: ABCB1 rs1128503 and rs1045642, ABCG2 rs2231142, ERCC1 rs11615, MTHFR rs1801133, SLC19A1 rs1051266, ERCC2 rs1799793 and rs13181. ABCB1, ABCG2 and SLC19A1 are custom designed assays.
ERCC1, MTHFR and ERCC2 are pre-designed assays. GSTP1 rs1695, TP53 rs1042522 and XRCC1 rs25487, were determined with the Pyrosequencer 96MA (Isogen, IJsselstein,
The Netherlands). The pyrosequencing reactions were performed according to the manufacturers’ protocol. PCR primers, target sequence and the calculated dispensation order for each SNP are listed in Table 1. Note that the lower case nucleotides in the dispensation order are negative controls, these nucleotides will not be incorporated in the target DNA and consequently should not appear in the pyrogram. The technicians performing the analyses were blinded with respect to sample identity. As a quality control in the pyrosequencing and Taqman assays, at least 5% of samples were genotyped in duplicate. In addition negative controls (water) were used. No inconsistencies were observed. In case of discrepancies between genotypes in tumor and blood, the sample pair was re-analyzed in one run. In this run we included five randomly selected samples and water as controls.
STATISTICS
Concordance of genotypes determined in DNA from FFPE tumor samples and EDTA blood was tested with the k statistic, which tests the agreement between two paired results. A k
> 0.80 indicates a good agreement. We also calculated 95% confidence intervals (CI) of the percentage of disagreement using the z test for a single proportion 14. P-values of <0.05 were considered statistically significant. All statistical analyses were performed using SPSS v14.
RESULTS
Both EDTA-blood and FFPE tumor samples were available of 149 patients with advanced CRC who participated in a prospective randomized study 13. Tumor and peripheral blood genotypes were determined for a total of 11 SNPs from 8 different chromosomal regions.
The SNPs were selected based on their frequent determination as candidate genes in pharmacogenetic studies in CRC 15. Depending on genotype, results were obtained in 87- 100% of peripheral blood samples; FFPE genotypes were obtained in 77-97% of samples.
Genotype distributions (Table 2) were in accordance with previously published results (http://www.ncbi.nlm.nih.gov/SNP, accessed Oct, 2008 16-19). All genotypes were in Hardy- Weinberg Equilibrium (HWE), except for XRCC1 (both in EDTA-blood and in FFPE, HWE p-values of 0.025 and 0.046, respectively), GSTP1 (EDTA-blood only, p-value 0.048) and ABCG2 (FFPE only, p-value 0.034).
Paired tumor and peripheral blood genotypes were obtained for 77-95% of patients, depending on genotype. We found an excellent agreement between the peripheral blood genotypes and the genotypes determined in corresponding FFPEs (all k ≥0.95, Table 3). In total, 20 out of 1,420 (1.4%) genotype pairs were discordant in the current sample set. More specifically, 16 heterozygotes appeared homozygous in the FFPE tissue, and the remaining 4 discordant SNPs were genotyped for homozygotes.
For each individual SNP, a maximum of 3.0% of the pairs showed different results. The 95% confidence levels of genotype discordance lie within the intervals of 0.0- 4.9% (ABCB1 rs1128503), 0.0- 3.6% (ERCC1 rs11615), 0.0- 3.8% (ERCC2 rs13181), 0.0- 5.1% (ERCC2 rs1799793), 0.1- 5.9% (GSTP1 rs1695), 0.0- 4.0% (MTHFR rs1801133), 0.0- 4.1% (TP53 rs1042522), 0.0- 2.3% (SLC19A1 rs1051266) and 0.0- 2.1% (XRCC1 rs25487). Therefore, with the exception of GSTP1, none of the discordances is significantly different from 0%.
gene rs No. % success blood
genotype distribution blood*
% success FFPE
genotype distribution FFPE*
ABCB1 1128503 95.3% 17.6-50.0-32.4% 91.9% 18.2-48.9-32.8%
ABCB1 1045642 96.0% 28.0-48.3-23.8% 96.0% 28.7-46.2-25.2%
ABCG2 2231142 94.6% 66.7-32.6-0.7% 95.3% 69.7-30.3-0.0%
ERCC1 11615 91.9% 39.4-48.9-11.7% 96.6% 40.3-47.9-11.8%
ERCC2 13181 97.3% 35.9-49.7-14.5% 85.9% 41.4-41.4-17.2%
ERCC2 1799793 91.9% 41.6-48.2-10.2% 91.3% 44.1-42.6-13.2%
GSTP1 1695 100.0% 38.3-53.0-8.7% 90.0% 44.0-48.5-7.5%
MTHFR 1801133 87.2% 42.3-50.8-6.9% 91.9% 43.1-47.4-9.5%
TP53 1042522 100.0% 45.6-48.3-6.0% 77.2% 49.6-44.3-6.1%
SLC19A1^ 1051266 99.3% 33.8-50.0-16.2% 85.2% 31.5-51.2-17.3%
XRCC1 25487 100.0% 34.2-55.7-10.1% 94.6% 35.5-54.6-9.9%
* frequency distribution is shown as: homozygote wildtype – heterozygote – homozygote mutant; FFPE formalin fixed paraffin embedded tissue; ^ alternatively called: reduced folate carrier (RFC)
Table 2
Genotype distributions and genotyping success rates.
gene genetic variation chromo- some
rs No. k Asymp- totic error
No. of evaluable pairs (% of 149)
No. of different calls (% #)
Confidence interval *
ABCB1 C1236T, Gly412Gly 7q21 1128503 0.96 0.021 131 (88%) 3 (2.3%) 0.0- 4.9%
ABCB1 T3435C, Ile1145Ile 7q21 1045642 1.00 0.000 137 (92%) 0 (0.0%) - ABCG2 C421A, Lys141Gln 4q21 2231142 1.00 0.000 134 (90%) 0 (0.0%) - ERCC1 T496C, Asn118Asn 19q13.3 11615 0.98 0.018 132 (89%) 2 (1.5%) 0.0- 3.6%
ERCC2 A2251C, Lys751Gln 19q13.3 13181 0.97 0.018 124 (83%) 2 (1.6%) 0.0- 3.8%
ERCC2 G965A, Asp312Asn 19q13.3 1799793 0.96 0.023 125 (84%) 3 (2.4%) 0.0- 5.1%
GSTP1 G342A, Ile105Val 11q13 1695 0.95 0.026 134 (90%) 4 (3.0%) 0.1- 5.9%
MTHFR C677T, Val222Ala 1p36.3 1801133 0.97 0.020 120 (81%) 2 (1.7%) 0.0- 4.0%
TP53 G466C, Arg72 Pro 17p13.1 1042522 0.97 0.022 115 (78%) 2 (1.7%) 0.0- 4.1%
SLC19A1 † G80A, Arg27His 21q22 1051266 0.99 0.013 126 (85%) 1 (0.8%) 0.0- 2.3%
XRCC1 G1301A, Arg399Gln 19q13.3 25487 0.99 0.013 142 (95%) 1 (0.7%) 0.0- 2.1%
†Also designated RFC. # percentage of evaluable pairs that is discordant. * 95% confidence interval limits of the percentage of different calls. By definition, the lower limit of the 95% confidence interval of the percentage of different calls cannot be lower than 0.0%.
Table 3
Concordance of polymorphisms in paired tumor and blood samples (n= 149).
DISCUSSION
This is the first study investigating concordance of a large set of genotypes in peripheral blood and tumor in patients with CRC, using formaldehyde fixed, paraffin embedded tumor tissue-derived DNA. The results of the current study show that genotyping a patient using material from FFPE and EDTA-blood yields comparable results in nearly all cases. Only a small number of samples (maximum 3.0% depending on genotype) showed discordant results, and with the exception of GSTP1, none of these discordances is statistically different from zero. This may be an expected finding, since none of these SNPs is thought to be associated with (colorectal) tumorigenesis, and therefore somatic mutations in these genes are unlikely. We found a number of discordant results, but there was no apparent bias toward a specific allele in any of the assays. All discordant tissues (both EDTA-blood and FFPE) were re-analyzed to exclude genotyping errors. DNA isolation errors could not be excluded because there was no material left to perform new isolation from FFPE samples.
In case of GSTP1, we calculated that germline genotype differs from tumor genotype in 0.1-5.9% of patients. Although statistically significant, this may be a clinically irrelevant difference, resulting from small genotyping errors or representing a false-positive finding due to multiple testing. If we would have used the Bonferroni correction to adjust for multiple testing, the confidence interval would have included zero (CI: 0.0-7.2%) and shown nonsignificance. On the other hand, this small difference may also indicate the possibility that in a limited number of tumors the GSTP1 gene is affected. In the genetically unstable tumor environment, stochastic mutations or deletions may occur in any gene, and genetic variations that confer a cellular growth benefit may even be selected. This may be the case for GSTP1, which is a part of the JNK-pathway and as such is involved in cell cycle regulation and apoptosis 20. The GSTP1 rs1695 SNP encodes an amino acid substitution that lies within the JNK protein binding site 21. Knock-down experiments with a CRC cell line have shown that GSTP1 function is essential for in vivo growth of xenografts 22. In the current analysis, we observed a transition of the heterozygote to the homozygote wild-type GSTP1 genotype in 4 of 69 (6%) tumors of heterozygote patients, indicating loss of heterozygosity (LOH).
LOH of the GSTP1 genotype was reported to be as high as 20% in heterozygote CRC patients in a smaller study 23. However, these resuls seem somewhat unlikely since GSTP1 is located on a chromosomal region which is not frequently affected in CRC 24.
Only a few studies have compared genetic variations in diseased (tumor) and non-diseased (blood) tissues 23;25-27. Rae et al compared SNPs in CYP-enzymes and ABCB1 using breast tumor tissue and peripheral blood of 10 patients; a complete accordance (100%) was obtained for all samples 26. Weiss et al studied 23 SNPs in 80 patients with acute myeloid leukemia, using buccal swabs and cryopreserved marrow or blood. The genes that were studied include ABCB1, ABCG2, GSTP, CYP-enzymes, ERCC2 and XRCC1. Overall, 12 discordant genotypes were found among 1,765 evaluable data pairs (0.7%) 27. In the present study, 20 differences were found in 1,420 comparisons (1.4%). This is similar to the results reported by Marsh et al (13 differences in 1,193 evaluable data pairs, 1.1%) 25. In that study, 28 genetic variants were studied in blood and fresh frozen tumor tissue obtained from 44 patients with colorectal cancer (CRC). The genotypes they investigated include SNPs in ABCB1, ABCG2, CES, ERCC2 en XRCC1. However, their overall genotyping efficiency was slightly higher
(93%) compared to the current study (87%). This is likely due to the fact that instead of FFPE, the investigators were able to use fresh frozen tumor tissue. Although FFPE is more widely available, it generally yields DNA of poorer quality due to the fixation process, as opposed to fresh frozen tumor. The only study to date that investigated SNPs in tumor and germline FFPE material in CRC patients reports genotyping data on only two SNPs in the ERCC2 and GSTP1 genes 23.In this study, Le Morvan et al reported LOH in 14 out of 32 ERCC2 heterozygote patients (44%), a finding which is not confirmed by the present study (2 out of 53 heterozygotes was found homozygote in the tumor; 4%).
We showed that, at least for a number of SNPs that are frequently studied in pharmacogenetic CRC trials, genotypes determined in colorectal tumor FFPE are concordant with germline genotypes. This is an important finding, since in many clinical studies conducted in the previous years, DNA or peripheral blood was not archived, let alone frozen tissue specimens.
In contrast, FFPE samples were routinely collected in many cases, and still provide a valuable source of DNA. Although we showed excellent concordance for all studied genetic variations, it is unknown if this concordance can be extrapolated to other genes or genetic variations that may be of interest, or other types of cancer. The current study does also not account for other genetic variations such as amplification, methylation and copy number variation that could be pharmacologically relevant. We conclude that many SNPs in pharmacological candidate genes show excellent concordance between tumor and peripheral blood in CRC patients, and that for these specific SNPs, EDTA-blood and FFPE are equally valuable sources of DNA.
REFERENCE LIST
1. Swen JJ, Wilting I, de Goede AL et al.
Pharmacogenetics: from bench to byte. Clin Pharmacol Ther 2008; 83:781-7.
2. Swen JJ, Huizinga TW, Gelderblom H et al.
Translating pharmacogenomics: challenges on the road to the clinic. PLoS Med 2007; 4:e209.
3. Andrisin TE, Humma LM, Johnson JA.
Collection of genomic DNA by the noninvasive mouthwash method for use in pharmacogenetic studies. Pharmacotherapy 2002; 22:954-60.
4. Etter JF, Neidhart E, Bertrand S, Malafosse A, Bertrand D. Collecting saliva by mail for genetic and cotinine analyses in participants recruited through the Internet. Eur J Epidemiol 2005;
20:833-8.
5. Thompson MD, Bowen RA, Wong BY et al.
Whole genome amplification of buccal cell DNA: genotyping concordance before and after multiple displacement amplification. Clin Chem Lab Med 2005; 43:157-62.
6. Cao W, Hashibe M, Rao JY, Morgenstern H, Zhang ZF. Comparison of methods for DNA extraction from paraffin-embedded tissues and buccal cells. Cancer Detect Prev 2003; 27:397- 404.
7. Casparie M, Tiebosch AT, Burger G et al.
Pathology databanking and biobanking in The Netherlands, a central role for PALGA, the nationwide histopathology and cytopathology data network and archive. Cell Oncol 2007;
29:19-24.
8. Coura R, Prolla JC, Meurer L, Ashton-Prolla P.
An alternative protocol for DNA extraction from formalin fixed and paraffin wax embedded tissue.
J Clin Pathol 2005; 58:894-5.
9. Santos MC, Saito CP, Line SR. Extraction of genomic DNA from paraffin-embedded tissue sections of human fetuses fixed and stored in formalin for long periods. Pathol Res Pract 2008.
10. Carvalho B, Postma C, Mongera S et al. Multiple putative oncogenes at the chromosome 20q amplicon contribute to colorectal adenoma to carcinoma progression. Gut 2008.
11. van den IJssel P, Tijssen M, Chin SF et al. Human and mouse oligonucleotide-based array CGH.
Nucleic Acids Res 2005; 33:e192.
12. Rooney PH, Murray GI, Stevenson DA, Haites NE, Cassidy J, McLeod HL. Comparative genomic hybridization and chromosomal instability in solid tumours. Br J Cancer 1999;
80:862-73.
13. Koopman M, Antonini NF, Douma J et al.
Sequential versus combination chemotherapy with capecitabine, irinotecan, and oxaliplatin in advanced colorectal cancer (CAIRO): a phase III randomised controlled trial. Lancet 2007;
370:135-42.
14. Petrie A, Sabin C. Medical Statistics at a Glance.
58-60. 2005. Oxford, Blackwell Science.
15. Pander J, Gelderblom H, Guchelaar HJ. Insights into the role of heritable genetic variation in the pharmacokinetics and pharmacodynamics of anticancer drugs. Expert Opin Pharmacother 2007; 8:1197-210.
16. de Jong FA, Marsh S, Mathijssen RH et al.
ABCG2 pharmacogenetics: ethnic differences in allele frequency and assessment of influence on irinotecan disposition. Clin Cancer Res 2004;
10:5889-94.
17. Cote JF, Kirzin S, Kramar A et al. UGT1A1 polymorphism can predict hematologic toxicity in patients treated with irinotecan. Clin Cancer Res 2007; 13:3269-75.
18. Malmer BS, Feychting M, Lonn S et al. Genetic variation in p53 and ATM haplotypes and risk of glioma and meningioma. J Neurooncol 2007;
82:229-37.
19. Stoehlmacher J, Park DJ, Zhang W et al. A multivariate analysis of genomic polymorphisms:
prediction of clinical outcome to 5-FU/
oxaliplatin combination chemotherapy in refractory colorectal cancer. Br J Cancer 2004;
91:344-54.
20. Adler V, Yin Z, Fuchs SY et al. Regulation of JNK signaling by GSTp. EMBO J 1999; 18:1321-34.
21. Adler V, Pincus MR. Effector peptides from glutathione-S-transferase-pi affect the activation of jun by jun-N-terminal kinase. Ann Clin Lab Sci 2004; 34:35-46.
22. Dang DT, Chen F, Kohli M, Rago C, Cummins JM, Dang LH. Glutathione S-transferase pi1 promotes tumorigenicity in HCT116 human colon cancer cells. Cancer Res 2005; 65:9485-94.
23. Le Morvan V, Smith D, Laurand A et al.
Determination of ERCC2 Lys751Gln and GSTP1 Ile105Val gene polymorphisms in colorectal cancer patients: relationships with treatment outcome. Pharmacogenomics 2007; 8:1693-703.
24. Rooney PH, Murray GI, Stevenson DA, Haites NE, Cassidy J, McLeod HL. Comparative genomic hybridization and chromosomal instability in solid tumours. Br J Cancer 1999;
80:862-73.
25. Marsh S, Mallon MA, Goodfellow P, McLeod HL. Concordance of pharmacogenetic markers in germline and colorectal tumor DNA.
Pharmacogenomics 2005; 6:873-7.
26. Rae JM, Cordero KE, Scheys JO, Lippman ME, Flockhart DA, Johnson MD. Genotyping for polymorphic drug metabolizing enzymes from paraffin-embedded and immunohistochemically stained tumor samples. Pharmacogenetics 2003;
13:501-7.
27. Weiss JR, Baer MR, Ambrosone CB et al. Concordance of pharmacogenetic polymorphisms in tumor and germ line DNA in adult patients with acute myeloid leukemia.
Cancer Epidemiol Biomarkers Prev 2007;
16:1038-41.
