The Pharmacogenomics Journal (2011) 11, 100–107 & 2011 Macmillan Publishers Limited. All rights reserved 1470-269X/11 www.nature.com/tpj ORIGINAL ARTICLE

Efficacy of tamoxifen based on cytochrome P450 2D6, CYP2C19 and SULT1A1 genotype in the Italian Tamoxifen Prevention Trial

D Serrano1, M Lazzeroni1,2, The role of pharmacogenomics and tamoxifen was investigated by analyzing 3 1 several polymorphisms of cytochrome P450 and SULT1A1 in a nested C-F Zambon , D Macis , case control study from the Italian Tamoxifen Prevention Trial. This study 4 1 P Maisonneuve , H Johansson , included 182 Caucasian subjects, 47 breast cancer (BC) cases and 135 A Guerrieri-Gonzaga1, matched controls. We used the AmpliChip CYP450 Test to screen 33 alleles M Plebani3, D Basso3, of CYP2D6 and 3 of CYP2C19. One more variant for CYP2C19*17 and two J Gjerde5, G Mellgren5,6, single-nucleotide polymorphisms for the gene SULT1A1 were also performed. 4 1,7 By using the AmpliChip CYP450 Test, out of 182 subjects, we identified 8 N Rotmensz , A Decensi poor metabolizer (PM), 17 intermediate metabolizer (IM), 151 extensive and B Bonanni1 metabolizer (EM) and 3 ultrarapid metabolizer (UM). PM women allocated to the tamoxifen arm showed a higher risk of developing BC compared to the 1Division of Cancer Prevention and Genetics, remaining phenotypes (P ¼ 0.035). In an exploratory analysis, among 58 European Institute of Oncology, Milan, Italy; 2University of Rome Tor Vergata, School of women with a CYP2D6*2A allele, 9 BCs were diagnosed in the placebo arm Oncology, Rome, Italy; 3Department of and only 1 in the tamoxifen arm (P ¼ 0.0001). CYP2C19 and SULT1A1 Laboratory Medicine, University-Hospital of polymorphisms did not show any correlation with tamoxifen efficacy. Padova, Padova, Italy; 4Division of Epidemiology Tamoxifen showed reduced efficacy in CYP2D6 PMs in the chemoprevention and Biostatistics, European Institute of Oncology, Milan, Italy; 5Hormone Laboratory, Haukeland setting. Conversely, the CYP2D6*2A allele may be associated with increased University Hospital, Bergen, Norway; 6Institute of efficacy of tamoxifen. These findings support the relevance of pharmaco- Medicine, University of Bergen, Bergen, Norway genomics in tailoring tamoxifen treatment. and 7Division of Medical Oncology, E.O. Ospedali The Pharmacogenomics Journal (2011) 11, 100–107; doi:10.1038/tpj.2010.17; Galliera, Genova, Italy published online 23 March 2010 Correspondence: Dr D Serrano, Division of Cancer Prevention Keywords: CYP2D6; genotype; chemoprevention; tamoxifen; breast cancer and Genetics, European Institute of Oncology, via Ripamonti 435, Milan 20141, Italy. E-mail: [email protected]

Introduction

Tamoxifen is the most widely used drug for the prevention of -receptor- positive breast cancer (BC),1 the treatment of ductal carcinoma in situ,2 the adjuvant treatment of premenopausal BC3 and the therapy of male BC.4 Nevertheless, its clinical effectiveness varies among individuals. Cytochrome 450 (CYP450) is the major mediator of drug . It is a family of several isoenzymes, and each may have different single- nucleotide polymorphisms (SNPs) resulting in the enhancement or reduction of activity. CYP2D6 and CYP2C19 are involved in the metabolism of several different and drugs, and are among the first examples of genetic polymorphisms identified in the superfamily of CYP .5 CYP2D6 catalyzes the bioconversion of approximately 25% of commonly used drugs, such Received 10 August 2009; revised 4 February 2010; accepted 22 February 2010; as b-blockers, antiarrhythmic agents, tricyclic antidepressants, selective seroto- published online 23 March 2010 nin-reuptake inhibitors, opioids and anticancer agents, including tamoxifen.6,7 Tamoxifen activity based on genotype characterization D Serrano et al 101

The CYP2C19 enzyme metabolizes more than 25 drugs, CYP2D6 genotype may be an independent predictor of including some antidepressants, proton pump inhibitors tamoxifen efficacy in women with early BC. and tamoxifen.8 Recently it has been shown that CYP2C19 This nested study is based on the Italian Tamoxifen has a role in the activation of clopidogrel, an antiaggregant Prevention Trial. Hysterectomized healthy women drug. SNP variants of this enzyme interfere with the clinical (N ¼ 5408) were randomly assigned to receive 20 mg tamox- activity of this drug, to the point where the FDA were ifen per day or placebo. After a median 81.2 month follow- induced to request a modification in the prescribing up, we diagnosed 79 subjects with BC (34 in the tamoxifen information of the drug.9 More than 70 alleles of CYP2D6 arm and 45 in the placebo arm, HR (95% CI) 0.75 and about 20 alleles of CYP2C19 have been identified with (0.48–1.18)).24 To develop a more tailored and more large interindividual and interethnic differences. The result- effective chemoprevention strategy, we have recently pub- ing CYP2D6 phenotypes can be classified into poor lished preliminary data on this study,25 indicating that metabolizers (PMs), with two nonfunctional ; inter- subjects with the CYP2D6*4/*4 genotype were less likely to mediate metabolizers (IMs), including one nonfunctional benefit from tamoxifen as a chemopreventive agent of BC in and one reduced activity, or two alleles with reduced the prevention setting. Here we present a more comprehen- activity; extensive metabolizers (EMs), with at least one sive analysis of allelic variants of CYP2D6, CYP2C19, in the normal allele and ultrarapid metabolizers (UMs) due to the same cohort, using the AmpliChip CYP450 Test (Roche presence of more than two functional alleles (http:// Diagnostics, Monza, Italy). In addition, we studied the www.cypalleles.ki.se).7,10,11 Two main CYP2C19 phenotypes CYP2C19*17 allele and two SNPs of the SULT1A1 gene to have been described: poor (two nonfunctional genes, further characterize the significance of tamoxifen pharma- namely *2 and/or *3) and EMs (at least one functional cogenetics. allele, namely *1). Recently, a novel allele, CYP2C19*17, has 12 been described: this variant is associated with an increased Materials and methods enzyme activity due to enhanced . This allele is genetically defined by the two SNPs, CYP2C19–3402 Subjects and samples C4T and CYP2C19–806 C4T, that were found to be in The subjects were selected from the Italian Tamoxifen 12 complete linkage in the populations studied by Sim et al. Prevention Trial in which 5408 healthy hysterectomized Tamoxifen is a prodrug undergoing considerable first-pass women aged 35–70 years at average risk for BC were oxidative metabolism into more potent active metabolites, randomly assigned to receive tamoxifen (20 mg daily) or such as 4-hydroxytamoxifen (4-OH-tamoxifen) and 4- placebo for 5 years. The trial subjects characteristics and 13,14 hydroxy-N-desmethyl-tamoxifen (endoxifen). They main findings have already been published.26,27 When this both have a 30- to 100-fold higher affinity to ER receptors study was planned, 79 women had developed BC and no compared to tamoxifen, respectively. Because endoxifen woman had died of BC. Two unaffected control women were reaches a steady-state plasma concentration 6- to 10-fold matched by age, center and time on study for each case. The 10,14,15 higher than 4-OH-tamoxifen, it is considered the institutional review boards at each participating center most relevant metabolite in determining the parent drug’s approved the case–control study and each subject signed clinical efficacy. CYP2D6 is one of the most significant an informed consent. 14 enzymes that modulates plasma endoxifen concentration. At the follow-up visit, we asked participants if they would The relationship of CYP2D6 activity, the level of endoxifen be willing to donate a blood sample for genetic research and the antiestrogenic activity has been recently clarified by purposes. Forty-seven BC patients and 135 of the selected 16 Wu et al. However, many other enzymes are involved in matched control agreed to donate blood samples for the the pathway, including CYP3A4, CYP2C9, CYP2B6 and study. To increase the statistical power of the study, we broke 17 CYP2C19 and may support the metabolic pathway in case the individual matching and used all available controls, of CYP2D6 PM. after verifying, however, that patients and controls had Less data are reported for CYP2C19, only recently a similar characteristics. Clinical characteristics are shown in clinical role of this enzyme for tamoxifen activity has been Table 1. shown.18,19 Moreover, the clearance, mediated by sulfo- transferase (SULT)1A1 and UDP-glucuronosyltransferase DNA preparation (UGT) 2B15 and UGT1A4, may regulate the metabolite Genomic DNA was extracted from whole blood with the concentration as reported by Nowell et al.20 QIAamp DNA Blood Kit (Qiagen, Valencia, CA, USA), So far the main focus of research has been on CYP2D6 according to the manufacturer’s instructions and was PMs, although activity enhancement may also have clinical adjusted to a concentration of 20 ng mlÀ1. The samples were relevance. Genotypically, PM patients or patients receiving stored at À80 1C. CYP2D6-inhibiting drugs have reduced plasma concentra- tions of endoxifen.14,15,21 At the clinical level, PM subjects Analysis of polymorphisms with BC on adjuvant tamoxifen tend to have a higher risk of CYP2D6 genotyping. The analysis of CYP2D6 polymorphisms relapse.22,23 This worse clinical outcome has been associated was performed by means of the AmpliChip CYP450 with a decreased prevalence of hot flashes, a putative marker Test (Roche Diagnostics) according to the manufacturer’s of tamoxifen activity.22 The findings have indicated that instructions. The test screens for 29 known polymorphisms of

The Pharmacogenomics Journal Tamoxifen activity based on genotype characterization D Serrano et al 102

Table 1 Subjects’ main clinical characteristics carried out by starting with 100 ng DNA in a final volume of 25 ml, as described previously.28 Data were analyzed by an Total Breast cancer Controls P-value ABIPRISM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA, USA). The CYP2C19*1, *2, *3 All women 182 47 135 and *17 alleles were defined according to the ‘Allele nomenclature for cytochrome P450 enzymes’ (http:// Age www.cypalleles.ki.se). 45–49 48 15 33 50–54 57 13 44 55–59 53 14 39 SULT1A1 genotyping. We screened for the SULT1A1*2 60+ 24 5 19 0.74 variant allele by a hybridization probe-based real-time PCR assay by using a Light Cycler (Roche Molecular Year of randomization Biochemicals, Pleasanton, CA, USA), as described 1992–1993 96 25 71 previously.29 To verify the results, we performed a PCR- 1994–1995 74 20 54 based restriction fragment length polymorphism assay as 1996–1997 12 2 10 0.89 described by others,20,33,34 and subjected the resulting PCR product of 281 bp to cleavage by the restriction Oophorectomy endonuclease HaeII (New England BioLabs, Ipswich, MA, No 70 19 51 USA). In addition, the PCR product was bidirectionally Yes 112 28 84 0.86 sequenced using Big Dye Terminator version 3.1 Cycle Hormone replacement therapy Sequencing kit (Applied Biosystems), according to the Never 124 33 91 manufacturer’s instructions, on an ABI Prism 3700 DNA Ever 58 14 44 0.86 Analyzer (Applied Biosystems). Sequencing reactions were performed with the forward and reverse primers identical to those used in the real-time PCR assay and PCR-based restriction fragment length polymorphism assay. the CYP2D6 gene (including gene duplications and deletions) Statistical analysis allowed for the identification of 33 different alleles: The Fisher’s exact test and the Mantel–Haenszel test for trend Full activity: *1, *2 and *35 were used to compare the prevalence of the different alleles in Reduced activity: *9, *10, *17, *29, *41 and *10XN, *17XN cases versus controls within the placebo group, and in the and *41XN proportion of PM with BC in the placebo group versus the No activity: *3, *4, *5, *6, *7, *8, *11, *14A, *15, *19, *20, tamoxifen group. All tests were two-sided. Analyses were *36, *40 and *4XN carried out with the SAS software (version 8.02, Cary, NC, USA). Increased activity: *1XN, *2XN and *35XN and unknown activity: *14B, *25, *26, *30 and *31 (http://www.cypalleles. Results ki.se). The AmpliChip CYP450 Data Analysis Software derived Subjects and genotype the genotype, listed the corresponding identified alleles, and Out of the 79 subjects affected by BC, 47 were willing to provided a prediction of the individual’s CYP2D6 enzymatic donate blood for the analysis. From the unaffected controls, activity by classifying subjects into four classes: UMs, EMs, 135 subjects agreed to participate. All women were Cauca- IMs, and PMs. UMs and EMs have more than two functional sians, median age was 53 years for both cases and controls. alleles and at least one functional allele, respectively; IMs The 47 cases included 37 invasive cancers and 10 cases of have one null and one reduced activity allele or two reduced intraepithelial neoplasia; 27 cases were in the placebo arm activity alleles; while PMs are characterized by having two and 20 in the tamoxifen arm. Among the 47 tumors, 63% nonfunctional CYP2D6 alleles. were receptor positive, 58% progesterone receptor positive and in 26% of the cases the HER2 was over- CYP2C19 genotyping. CYP2C19*1, CYP2C19*2 and CYP2C19*3 expressed. Following are the clinical characteristics of the polymorphisms were detected by means of the AmpliChip 182 subjects that contributed to BC risk: 69 women had CYP450 Test (Roche Diagnostics), as described in the previous bilateral oophorectomy, 43 had monolateral and 68 had section. The CYP2C19–3402 C4T polymorphism was evaluated intact ovaries (two subjects were not known). Fifty-eight by means of restriction fragment length polymorphism subjects used, at a certain point, hormonal replacement analysis: 100 ng of genomic DNA was used as the template therapy (6 only before the trial, 31 only during treatment for PCR amplification using the primer pairs 2C19-3402F and 21 always, respectively), whereas 124 had never used and 2C19-3402R (50-AATAAAGATGACCTTGATCTGG-30 and hormonal therapy. All these characteristics were balanced 50-GTCTCCTGAAGTGTCTGTAC-30,respectively).Theassays between arms (data summarized in Table 1). were performed as previously described.28 Table 2 shows the detailed CYP2D6 genotype and CYP2C19–806 C4T SNP was discriminated by means of predicted phenotype frequency distribution: 82.9% were the 50 nuclease polymerase chain reaction assay (TaqMan) EMs, 1.6% were UMs, 9.3% were IMs and 4.3% were PMs,

The Pharmacogenomics Journal Tamoxifen activity based on genotype characterization D Serrano et al 103

whereas three subjects could not be evaluated for poor Table 2 Number, frequency and predicted phenotype for sample quality. All polymorphisms studied were in Hardy– each CYP2D6 genotype Weinberg equilibrium. Genotype CYP2D6 N Frequency Predicted Total Genotype and breast cancer phenotype N (%) Table 3 shows the distribution of CYP2D6 expected phenotypes between BC cases and controls. Within the *1/*1XN 1 0.005 UM placebo arm, there was no significant difference in the *1/*2AXN 1 0.005 UM *1XN/*2A 1 0.005 UM 3 (1.64) distribution of CYP2D6 phenotypes between BC cases and *1/*1 17 0.093 EM unaffected controls. However, in the tamoxifen arm, a *1/*10B 4 0.022 EM significantly higher frequency of PMs in the cancer cases *1/*2A 24 0.132 EM (15.0%) was observed compared to the controls (1.5%) *1/*2D 1 0.005 EM (P ¼ 0.035). *1/*3 5 0.027 EM Table 4 shows the distribution of CYP2D6*2A alleles *1/*35 11 0.060 EM between cases and controls. Although the distribution of *1/*41 13 0.071 EM CYP2D6*2A carriers was not different between BC cases and *1/*4A 19 0.104 EM controls in the placebo group, there was a significantly *1/*4AXN 1 0.005 EM higher frequency of *2A carriers in the tamoxifen arm in the *1/*4J 1 0.005 EM unaffected subjects (47.7%) than in the cancer cases (5.0%), *1/*5 3 0.016 EM *1/*7 1 0.005 EM (P ¼ 0.0001). This suggests that carriers of a CYP2D6*2A may *1/*9 7 0.038 EM have higher benefit from tamoxifen therapy compared with *1XN/*41 1 0.005 EM other genotypes associated with an EM phenotype. *1XN/*4A 1 0.005 EM The CYP2C19 analysis included three alleles detected by *10B/*2A 2 0.011 EM the AmpliChip CYP450 Test, *1, *2 and *3. The CYP2C19*17 *2A/*2A 5 0.027 EM variant, predicted to have a UM phenotype, was analyzed *2A/*35 8 0.044 EM separately. The genotype frequency was: 42.5% *1*1, 24.3% *2A/*41 2 0.011 EM *1*2, 23.2% *1*17, 3.8% *17*17, 4.4% *2/*17, three samples *2A/*4A 8 0.044 EM were undetectable. None of the phenotypes analyzed *2A/*4B 1 0.005 EM showed any association with BC or with tamoxifen benefit *2A/*4D 1 0.005 EM *2A/*4 3 0.016 EM (Table 5). *2A/*6A 1 0.005 EM Table 6 shows the distribution of SULT1A1 genotypes: *2AXN/*4A 1 0.005 EM 52% *1/*1 (EM), 39.6% *1/*2 (IM) and 8.5% *2/*2 (PM), *2AXN/*5 1 0.005 EM respectively. Low or absent SULT1A1 activity is predicted to *2D/*3 1 0.005 EM result in an accumulation of tamoxifen metabolites and *35/*41 1 0.005 EM consequently likely to increase efficacy. However, no *35/*4A 4 0.022 EM differences between SULT1A1 genotypes and the risk of *35/*9 2 0.011 EM developing BC were detected in our study. *35XN/*4A 1 0.005 EM 151 (82.96) *41/*41 4 0.022 IM *41/*5 1 0.005 IM Discussion *41/*6B 1 0.005 IM *41/*9 1 0.005 IM Over the last few years, the clinical impact of pharmacoge- *4A/*41 7 0.038 IM nomics has continued to increase. Tamoxifen has a complex *4D/*41 3 0.016 IM 17 (9.34) pharmacological pathway and its metabolites have a high *3/*4A 1 0.005 PM variability in serum concentrations from patient to pa- *3/*5 1 0.005 PM tient.30 Among the several enzymes involved in the *4A/*4A 2 0.011 PM metabolic process, CYP2D6 seems to be the most relevant. *4A/*5 2 0.011 PM It is highly polymorphic, and its variants have been *4A/*7 1 0.005 PM classified into four main phenotypes: PMs, IMs, EMs and *4D/*4A 1 0.005 PM 8 (4.39) Not assessable 3 0.016 3 (1.64) UMs. The blood concentrations of tamoxifen’s most potent Total 182 (100) and abundant metabolite, endoxifen, have been shown to correlate significantly with the CYP2D6 phenotype.21,29 Abbreviations: EM, extensive metabolizer; IM, intermediate metabolizer; PM, poor Furthermore, Goetz et al.22 reported that patients with metabolizer; UM, ultrarapid metabolizer. impaired CYP2D6 metabolism have nearly a twofold higher risk of BC recurrence when treated with tamoxifen in the adjuvant setting. Moreover, the subjects’ phenotype Goetz et al.,31 with a fourfold increase of BC relapse in PM appeared to result from the combination of the genotype subjects. and drugs (or nutrients) that may interact with CYP450, in In agreement with Goetz’ initial observation,22 we particular antidepressants and CYP2D6, as reported by reported that CYP2D6*4/*4 genotype was associated with

The Pharmacogenomics Journal Tamoxifen activity based on genotype characterization D Serrano et al 104

Table 3 Association between AmpliChip CYP2D6 expected phenotype, disease status (unaffected/breast cancer) and treatment allocated arm (placebo/tamoxifen)

Expected phenotype CYP2D6 Total Control Breast cancer OR (95% CI)

Total* 182 135 47 Ultrarapid metabolizer 3 2 (1.5%) 1 (2.0%) 1.54 (0.14–17.5) Extensive metabolizer 151 114 (84.4%) 37 (78.7%) 1.00 Intermediate metabolizer 17 13 (9.6%) 4 (8.5%) 0.95 (0.29–3.09) Poor metabolizer 8 4 (3.0%) 4 (8.5%) 3.08 (0.73–12.9) No call 3 2 (1.5%) 1 (2.1%) 1.54 (0.14–17.5)

Placebo arm 97 70 27 Ultrarapid metabolizer 3 2 (2.9%) 1 (3.7%) 1.20 (0.10–13.8) Extensive metabolizer 78 55 (78.6%) 23 (85.2%) 1.00 Intermediate metabolizer 10 8 (11.4%) 2 (7.4%) 0.60 (0.12–3.03) Poor metabolizer 4 3 (4.3%) 1 (3.7%) 0.80 (0.08–8.07) No call 2 2 (2.9%) — —

Tamoxifen arm** 85 65 20 Ultrarapid metabolizer — — — — Extensive metabolizer 73 59 (90.8%) 14 (70.0%) 1.00 Intermediate metabolizer 7 5 (7.7%) 2 (10.0%) 1.69 (0.30–9.61) Poor metabolizer** 4 1 (1.5%) 3 (15.0%) 12.64 (1.22–131.) No call 1 — 1 (5.0%) —

Odds ratios (ORs) and 95% confidence intervals (CIs) obtained from exact logistic regression. *P ¼ NS for poor metabolizers versus the remaining groups. **P ¼ 0.035 for poor metabolizers versus the remaining groups in the tamoxifen group.

Table 4 Association between AmpliChip CYP2D6*2A expected phenotype, disease status (unaffected/breast cancer) and treatment allocated arm (placebo/tamoxifen)

Expected phenotype CYP2D6 Total Control Breast cancer OR (95% CI)

Total* 182 135 47 At least 1 *2A allele 59 48 (35.6%) 11 (23.4%) 0.60 (0.28–1.30) Other nonpoor 112 81 (60.0%) 31 (66.0%) 1.00 Poor metabolizer 8 4 (3.0%) 4 (8.5%) 2.61 (0.62–11.1) No call 3 2 (1.5%) 1 (2.0%) 1.31 (0.11–14.9)

Placebo arm** 97 70 27 At least 1 *2A allele 27 17 (24.3%) 10 (37.0%) 1.77 (0.67–4.63) Other nonpoor 64 48 (68.6%) 16 (59.3%) 1.00 Poor metabolizer 4 3 (4.3%) 1 (3.7%) 1.00 (0.10–10.0) No call 2 2 (2.9%) — —

Tamoxifen arm*** 85 65 20 At least 1 *2A allele 32 31 (47.7%) 1 (5.0%) 0.09 (0.01–0.57) Other nonpoor 48 33 (50.8%) 15 (75.0%) 1.00 Poor metabolizer 4 1 (1.5%) 3 (15.0%) 6.60 (0.63–68.8) No call 1 — 1 (5.0%) —

Odds ratios (ORs) and 95% confidence intervals (CIs) obtained from exact logistic regression. *P ¼ 0.04 and **P ¼ 0.35 for poor metabolizers versus the remaining groups. ***P ¼ 0.0001 for poor metabolizers versus the remaining groups.

reduced tamoxifen efficacy even in the prevention setting.25 additional enzymes known to affect tamoxifen metabolism, In this study, we have expanded the allele coverage for the CYP2C19 and SULT1A1, for a more comprehensive investi- CYP2D6 genotyping by including more alleles that may gation of the resulting phenotype. We found 100% con- affect tamoxifen metabolism. Furthermore, we included two cordance between the real-time TaqMan CYP2D6*4 Allelic

The Pharmacogenomics Journal Tamoxifen activity based on genotype characterization D Serrano et al 105

Table 5 Association between CYP2C19 genotype, disease status (unaffected/breast cancer) and treatment allocated arm (placebo/tamoxifen)

CYP2C19 Total Control Breast cancer OR (95% CI)

Total 182 135 47 *1/*1, *1/*2, *1-*2/*17 (EM) 175 131 (97.0%) 44 (93.6%) 1.00 *17/*17 (UM) 7 4 (3.0%) 3 (6.4%) 2.24 (0.48–10.4)

Placebo arm 97 70 27 *1/*1, *1/*2, *1-*2/*17 (EM) 94 68 (97.1%) 26 (96.3%) 1.00 *17/*17 (UM) 3 2 (2.9%) 1 (3.7%) 1.31 (0.11–15.0)

Tamoxifen arm 85 65 20 *1/*1, *1/*2, *1-*2/*17 (EM) 81 63 (96.9%) 18 (90.0%) 1.00 *17/*17 (UM) 4 2 (3.1%) 2 (10.0%) 3.50 (0.46–26.6)

Abbreviations: EM, extensive metabolizer; UM, ultrarapid metabolizer. Odds ratios (ORs) and 95% confidence intervals (CIs) obtained from exact logistic regression.

Table 6 Association between SULT1A1 genotype, disease phenotype,32 and appear to confer increased CYP2D6 status (unaffected/breast cancer) and treatment allocated activity.33 This observation needs further confirmation and arm (placebo/tamoxifen) should be investigated in parallel with the endoxifen level associated with this particular genotype. SULT1A1 Total Control Breast cancer OR (95% CI) Although inhibitors have become the standard genotype adjuvant treatment for postmenopausal positive-estrogen- receptor BC in Europe and North America,34 tamoxifen is Total 136 46 still widely used in premenopause, and at any age in Asia WT/WT 71 (52.2%) 24 (52.2%) 1.00 WT/*2 55 (40.4%) 18 (39.1%) 0.97 (0.48–1.96) and developing countries. Aromatase inhibitors are not *2/*2 10 (7.4%) 4 (8.7%) 1.18 (0.34–4.12) exempt from important adverse events, including muscu- loskeletal disorders and increased bone fracture risk. More- Placebo arm 71 26 over, a provocative computer modeling study suggested that WT/WT 37 (52.1%) 13 (50.0%) 1.00 tamoxifen could actually be more effective than aromatase WT/*2 29 (40.9%) 11 (42.3%) 1.08 (0.42–2.76) inhibitors if administered only to EM subjects.35 Genotype *2/*2 5 (7.0%) 2 (7.7%) 1.14 (0.20–6.60) information allows physicians to tailor treatment according to the patient’s genetic and metabolic profile. This aspect is Tamoxifen arm 65 20 especially relevant with a narrow therapeutic window. WT/WT 34 (52.3%) 11 (55.0%) 1.00 Warfarin is an example: including the genotype information WT/*2 26 (40.0%) 7 (35.0%) 0.83 (0.28–2.44) 36 *2/*2 5 (7.7%) 2 (10.0%) 1.23 (0.21–7.30) helps to adjust the drug dose. In the adjuvant setting it has been proposed to increase Odds ratios (ORs) and 95% confidence intervals (CIs) obtained from exact logistic the dose up to 40 mg of tamoxifen in PM subjects. With this regression. dose it has been shown that it is possible to reach similar level of endoxifen in IM and PM phenotype comparing EM Discrimination Assay and the AmpliChip CYP450 Test for with standard dose.37 In the prevention setting, it is the CYP2D6*4/*4 genotype. However, with the AmpliChip reasonable not to use tamoxifen in PM women, avoiding a CYP450 Test, three more PMs (genotypes CYP2D6*3/*4, potential toxic therapy with little or no benefit at all. CYP2D6*3/*5 and CYP2D6*4/*7) could be identified in the Conversely, in UM or even EM, the tamoxifen dose may be placebo arm, as compared to the TaqMan CYP2D6*4 test. In personalized to a lower range, for example, 5–10 mg per day, addition, three UMs were also identified in the placebo arm. because we showed that low-dose tamoxifen maintains anti- Our findings confirm that tamoxifen-treated women with proliferative activity on BC whereas toxicity may be PM phenotypes appear to have a higher incidence of BC reduced.38–40 compared with women with other phenotypes (P ¼ 0.035). CYP2D6 genotyping could be evaluated together with Interestingly, our expanded genotyping evaluation serum endoxifen levels and biomarkers of tamoxifen activity showed that having at least one CYP2D6*2A allele was and safety (for example, mammographic density, IGF-I, associated with a lower risk of developing BC under SHBG, lipids and endometrial thickness) after a short period tamoxifen treatment (P ¼ 0.0001 in Table 4) when compared of treatment. Such interaction may be a key point to to other EM CYP2D6 variants. This observation is supported personalized treatment and dose, in a manner that is similar by previous findings indicating that CYP2D6*2A alleles, in to monitoring statin therapy with levels and particularly the À1584G4C polymorphism, can rule out PM biphosphonates with bone density. Preliminary data suggest

The Pharmacogenomics Journal Tamoxifen activity based on genotype characterization D Serrano et al 106

that low-dose tamoxifen differently modulates some surro- tamoxifen. A well-designed clinical trial is required to gate biomarkers in accordance with the genotype.41 A unequivocally validate such an approach. multicentric clinical study aimed at validating this approach is ongoing in Italy. Our study has some limitations. First of all, the retro- Conflict of interest spective nature of the design is a weakness shared with all other available studies. The small sample size and low PM The authors declare no conflict of interest. number gives a low statistical power and does not allow one to study the interactions between other hormonal variants, Acknowledgments such as hormone replacement therapy use or oophorect- omy, and genotype, that may influence BC risk; the This study was supported by grants from the Italian National distribution of these variables is, however, balanced between Research Council, the Italian Association for Cancer Research, the cases and controls in this study sample. Antidepressants, in American-Italian Cancer Foundation, the Italian League Against particular paroxetine and fluoxetine, known to be potent Cancer and the Grieg Foundation. We gratefully acknowledge the CYP2D6 inhibitors, are commonly prescribed with tamox- contributions of Marcel Fontecha, Lothar Weiczorek. ifen in the treatment of hot flashes or depression.15 However, in Italy, antidepressants are not commonly References prescribed for management of hot flashes, and our data show that the overall prevalence of these drugs in the trial 1 Fisher B, Costantino JP, Wickerham DL, Cecchini RS, Cronin WM, was below 1% (data on file). Nevertheless, the effect of Robidoux A et al. Tamoxifen for the prevention of breast cancer: current status of the National Surgical Adjuvant Breast and Bowel Project P-1 CYP2D6 inhibitors on the clinical outcomes cannot speci- study. J Natl Cancer Inst 2005; 97: 1652–1662. fically be addressed in this study. 2 Fisher B, Land S, Mamounas E, Dignam J, Fisher ER, Wolmark N. Many other enzymes are involved in tamoxifen metabo- Prevention of invasive breast cancer in women with ductal carcinoma lism and the redundancy may mask or balance if one in situ: an update of the national surgical adjuvant breast and bowel project experience. Semin Oncol 2001; 28: 400–418. is not functional. Recently, it has been shown that CYP2C19 3 Colleoni M, Gelber S, Goldhirsch A, Aebi S, Castiglione-Gertsch M, may be a risk factor per se, or that it can be important Price KN et al. Tamoxifen after adjuvant chemotherapy for premeno- for tamoxifen response and is further enhanced by its pausal women with lymph node-positive breast cancer: International association with CYP2D6.19 Our data do not confirm this Breast Cancer Study Group Trial 13-93. J Clin Oncol 2006; 24: 1332–1341. observation, possibly because of the relatively low number 4 Fentiman IS, Fourquet A, Hortobagyi GN. Male breast cancer. Lancet of subjects with these SNPs variants. It has also been 2006; 367: 595–604. reported that SULT1A1 may influence tamoxifen metabo- 5 Kimura S, Umeno M, Skoda RC, Meyer UA, Gonzalez FJ. The human lites’ concentration.29,42,43 However, previous studies15,29 debrisoquine 4-hydroxylase (CYP2D) locus: sequence and identification of the polymorphic CYP2D6 gene, a related gene, and a pseudogene. have not shown significant associations between SULT1A1 Am J Hum Genet 1989; 45: 889–904. genotypes and the serum levels of 4OH-tamoxifen or 6 Wilkinson GR. Drug metabolism and variability among patients in drug endoxifen per se. Our results did not show any differences response. N Engl J Med 2005; 352: 2211–2221. between SULT1A1 genotype and BC risk, suggesting that 7 Sistonen J, Sajantila A, Lao O, Corander J, Barbujani G, Fuselli S. CYP2D6 worldwide genetic variation shows high frequency of altered activity SULT1A1 may not be important in explaining tamoxifen variants and no continental structure. Pharmacogenet Genomics 2007; preventive efficacy. 17: 93–101. A challenge in interpreting the existing literature on 8 Wedlund PJ. The CYP2C19 enzyme polymorphism. Pharmacology 2000; CYP2D6 and tamoxifen is that the classifications used to 61: 174–183. 9 Ellis KJ, Stouffer GA, McLeod HL, Lee CR. Clopidogrel pharmacoge- identify different CYP2D6 phenotypes have differed be- nomics and risk of inadequate platelet inhibition: US FDA recommenda- tween studies and the allele coverage has been very limited. tions. Pharmacogenomics 2009; 10: 1799–1817. In this rapidly evolving field, it is very difficult to nail down 10 Johansson I, Lundqvist E, Bertilsson L, Dahl ML, Sjoqvist F, Ingelman- a definitive classification, because many SNPs and many Sundberg M. Inherited amplification of an active gene in the cytochrome P450 CYP2D locus as a cause of ultrarapid metabolism of enzymes concur to the final effect. Furthermore, the debrisoquine. Proc Natl Acad Sci USA 1993; 90: 11825–11829. redundancy of enzymes in the tamoxifen pathway may 11 Rodriguez-Antona C, Ingelman-Sundberg M. Cytochrome P450 phar- reduce the effect of a single enzyme. Some doubts have been macogenetics and cancer. Oncogene 2006; 25: 1679–1691. raised, lately, on the final clinical relevance of CYP2D6 and 12 Sim SC, Risinger C, Dahl ML, Aklillu E, Christensen M, Bertilsson L et al. 44 A common novel CYP2C19 gene variant causes ultrarapid drug tamoxifen activity. metabolism relevant for the drug response to proton pump inhibitors The use of an automatic system such as the AmpliChip and antidepressants. Clin Pharmacol Ther 2006; 79: 103–113. CYP450 may help to speedup the analytical process with 13 Jordan VC. Metabolites of tamoxifen in animals and man: identification, minor limitations in specificity, such as misinterpretation pharmacology, and significance. Breast Cancer Res Treat 1982; 2: 123–138. between two SNPs, one EM and the other IM. The goal of 14 Stearns V, Johnson MD, Rae JM, Morocho A, Novielli A, Bhargava P et al. pharmacogenomics is to identify biomarkers to be Active tamoxifen metabolite plasma concentrations after coadministra- applied in the clinical setting and to implement the new tion of tamoxifen and the selective serotonin reuptake inhibitor paradigm of tailored medicine. Our findings support the paroxetine. J Natl Cancer Inst 2003; 95: 1758–1764. 15 Jin Y, Desta Z, Stearns V, Ward B, Ho H, Lee KH et al. CYP2D6 genotype, notion that CYP2D6 genotype can be an appropriate antidepressant use, and tamoxifen metabolism during adjuvant breast fingerprint to design truly tailored treatment with cancer treatment. J Natl Cancer Inst 2005; 97: 30–39.

The Pharmacogenomics Journal Tamoxifen activity based on genotype characterization D Serrano et al 107

16 Wu X, Hawse JR, Subramaniam M, Goetz MP, Ingle JN, Spelsberg TC. 31 Goetz M, Ames M, Gnant M, Filpits M, Jakesz R, Greil R et al. The tamoxifen metabolite, endoxifen, is a potent antiestrogen that Pharmacogenetic (CYP2D6) and gene expression profiles (HOXB13/ targets estrogen receptor alpha for degradation in breast cancer cells. IL17BR and molecular grade index) for prediction of adjuvant endocrine Cancer Res 2009; 69: 1722–1727. therapy benefit in the ABCSG 8 trial. Paper presented at 30th Annual 17 Boocock DJ, Brown K, Gibbs AH, Sanchez E, Turteltaub KW, White IN. SABCS 10–14 December 2008; San Antonio, TX 2008. (Abstract 57). Identification of human CYP forms involved in the activation of 32 Gaedigk A, Ryder DL, Bradford LD, Leeder JS. CYP2D6 poor metabolizer tamoxifen and irreversible binding to DNA. Carcinogenesis 2002; 23: status can be ruled out by a single genotyping assay for the -1584G 1897–1901. promoter polymorphism. Clin Chem 2003; 49: 1008–1011. 18 Justenhoven C, Hamann U, Pierl CB, Baisch C, Harth V, Rabstein S et al. 33 Zanger UM, Fischer J, Raimundo S, Stuven T, Evert BO, Schwab M et al. CYP2C19*17 is associated with decreased breast cancer risk. Breast Comprehensive analysis of the genetic factors determining Cancer Res Treat 2009; 115: 391–396. expression and function of hepatic CYP2D6. Pharmacogenetics 2001; 19 Schroth W, Antoniadou L, Fritz P, Schwab M, Muerdter T, Zanger UM et 11: 573–585. al. Breast cancer treatment outcome with adjuvant tamoxifen relative 34 Winer EP. Optimizing endocrine therapy for breast cancer. J Clin Oncol to patient CYP2D6 and CYP2C19 genotypes. J Clin Oncol 2007; 25: 2005; 23: 1609–1610. 5187–5193. 35 Punglia RS, Burstein HJ, Winer EP, Weeks JC. Pharmacogenomic 20 Nowell S, Sweeney C, Winters M, Stone A, Lang NP, Hutchins LF et al. variation of CYP2D6 and the choice of optimal adjuvant endocrine Association between sulfotransferase 1A1 genotype and survival of therapy for postmenopausal breast cancer: a modeling analysis. J Natl breast cancer patients receiving tamoxifen therapy. J Natl Cancer Inst Cancer Inst 2008; 100: 642–648. 2002; 94: 1635–1640. 36 Sconce EA, Khan TI, Wynne HA, Avery P, Monkhouse L, King BP et al. 21 Borges S, Desta Z, Li L, Skaar TC, Ward BA, Nguyen A et al. Quantitative The impact of CYP2C9 and VKORC1 genetic polymorphism and patient effect of CYP2D6 genotype and inhibitors on tamoxifen metabolism: characteristics upon warfarin dose requirements: proposal for a new implication for optimization of breast cancer treatment. Clin Pharmacol dosing regimen. Blood 2005; 106: 2329–2333. Ther 2006; 80: 61–74. 37 Irvin WJ, Carey LA, Dees EC, Lange L, Chiu WK, Evans JP et al. Validating 22 Goetz MP, Rae JM, Suman VJ, Safgren SL, Ames MM, Visscher DW et al. CYP2D6 genotype-guided tamoxifen therapy for a multiracial US Pharmacogenetics of tamoxifen biotransformation is associated with population. San Antonio Breast Cancer Symposium 2009. (Abstract 410). clinical outcomes of efficacy and hot flashes. J Clin Oncol 2005; 23: 38 Decensi A, Robertson C, Viale G, Pigatto F, Johansson H, Kisanga ER et 9312–9318. al. A randomized trial of low-dose tamoxifen on breast cancer 23 Goetz MP, Knox SK, Suman VJ, Rae JM, Safgren SL, Ames MM et al. The proliferation and blood estrogenic biomarkers. J Natl Cancer Inst impact of cytochrome P450 2D6 metabolism in women receiving 2003; 95: 779–790. adjuvant tamoxifen. Breast Cancer Res Treat 2007; 101: 113–121. 39 Guerrieri-Gonzaga A, Robertson C, Bonanni B, Serrano D, Cazzaniga M, 24 Veronesi U, Maisonneuve P, Rotmensz N, Costa A, Sacchini V, Travaglini Mora S et al. Preliminary results on safety and activity of a randomized, R et al. Italian randomized trial among women with hysterectomy: double-blind, 2 Â 2 trial of low-dose tamoxifen and fenretinide for tamoxifen and hormone-dependent breast cancer in high-risk women. breast cancer prevention in premenopausal women. J Clin Oncol 2006; J Natl Cancer Inst 2003; 95: 160–165. 24: 129–135. 25 Bonanni B, Macis D, Maisonneuve P, Johansson H, Gucciardo G, 40 Decensi A, Robertson C, Guerrieri-Gonzaga A, Serrano D, Cazzaniga M, Oliviero P et al. Polymorphism in the CYP2D6 tamoxifen-metabolizing Mora S et al. Randomized double-blind 2 Â 2 trial of low-dose gene influences clinical effect but not hot flashes: data from the Italian tamoxifen and fenretinide for breast cancer prevention in high-risk Tamoxifen Trial. J Clin Oncol 2006; 24: 3708–3709. premenopausal women. J Clin Oncol 2009; 27: 3749–3756. 26 Veronesi U, Maisonneuve P, Costa A, Sacchini V, Maltoni C, Robertson 41 Bonanni B, Serrano D, Gandini S, Guerrieri-Gonzaga A, Johansson H, C et al. Prevention of breast cancer with tamoxifen: preliminary findings Macis D et al. Randomized biomarker trial of anastrozole or low-dose from the Italian randomised trial among hysterectomised women. tamoxifen or their combination in subjects with breast intraepithelial Italian Tamoxifen Prevention Study. Lancet 1998; 352: 93–97. neoplasia. Clin Cancer Res 2009; 15: 7053–7060. 27 Veronesi U, Maisonneuve P, Rotmensz N, Bonanni B, Boyle P, Viale G et 42 Wegman P, Elingarami S, Carstensen J, Stal O, Nordenskjold B, Wingren al. Tamoxifen for the prevention of breast cancer: late results of the S. Genetic variants of CYP3A5, CYP2D6, SULT1A1, UGT2B15 and Italian Randomized Tamoxifen Prevention Trial among women with tamoxifen response in postmenopausal patients with breast cancer. hysterectomy. J Natl Cancer Inst 2007; 99: 727–737. Breast Cancer Res 2007; 9:R7. 28 Zambon CF, Fasolo M, Basso D, D’Odorico A, Stranges A, Navaglia F et 43 Nowell SA, Ahn J, Rae JM, Scheys JO, Trovato A, Sweeney C et al. al. Clarithromycin resistance, tumor necrosis factor alpha gene Association of genetic variation in tamoxifen-metabolizing enzymes polymorphism and mucosal inflammation affect H. pylori eradication with overall survival and recurrence of disease in breast cancer patients. success. J Gastrointest Surg 2007; 11: 1506–1514. Breast Cancer Res Treat 2005; 91: 249–258. 29 Gjerde J, Hauglid M, Breilid H, Lundgren S, Varhaug JE, Kisanga ER et al. 44 Goetz MP, Berry DA, Klein TE. Adjuvant tamoxifen treatment outcome Effects of CYP2D6 and SULT1A1 genotypes including SULT1A1 gene according to cytochrome P450 2D6 (CYP2D6) phenotype in early stage copy number on tamoxifen metabolism. Ann Oncol 2008; 19: 56–61. breast cancer: findings from the International Tamoxifen Pharmacoge- 30 Osborne CK. Tamoxifen in the treatment of breast cancer. N Engl J Med nomics Consortium. San Antonio Breast Cancer Symposium 2009. 1998; 339: 1609–1618. (Abstract 33).

The Pharmacogenomics Journal