Pharmacogenomics of tamoxifen and aromatase inhibitors


  • James N. Ingle MD

    Corresponding author
    1. Division of Medical Oncology, Mayo Clinic, Rochester, Minnesota
    • Division of Medical Oncology, Mayo Clinic, 200 First St. SW, Rochester, MN 55905===

    Search for more papers by this author
    • Fax: (507) 284-1803

    • James N. Ingle serves as a consultant to Novartis.

  • Presented at Endocrine and Targeted Manipulation of Breast Cancer: Proceedings of the Sixth Cambridge Conference, April 30–May 1, 2007, Cambridge, Massachusetts.


In selection of therapy for women with breast cancer, the focus has been almost exclusively on the characteristics of the tumor, eg, estrogen receptor (ER) and HER-2. Until recently, essentially no attention has been paid to the host and her genetic makeup as it relates to the metabolism of different drugs. The first real clinical application of pharmacogenetics in breast cancer management relates to tamoxifen's biotransformation to active anticancer metabolites. New information has arisen on the metabolism of tamoxifen to the active metabolite, 4 hydroxy-N-desmethyl-tamoxifen (endoxifen). Endoxifen is a metabolite with antitumor activity and affinity for the ER that is similar to 4-hydroxy-tamoxifen, but 1 that is normally present in substantially higher concentrations. CYP2D6 plays a central role in the metabolism to endoxifen and 1 published study shows that genotypic differences in CYP2D6 and use of CYP2D6 inhibitors has an impact on outcomes of women treated with tamoxifen. The aromatase inhibitors represent a major class of drugs in the armamentarium against breast cancer. The aromatase gene has been resequenced and functional genomics have been performed on the identified nonsynonymous coding single nucleotide polymorphisms showing significant decreases in levels of activity. These findings are consistent with a hypothesis that genetic variation in the CYP19 gene might be important in the activity of aromatase inhibitors. Currently, the emphasis is on examining multiple genes (thus pharmacogenomics) in pharmacodynamic and pharmacokinetic pathways in women receiving aromatase inhibitors for breast cancer. Cancer 2008. © 2007 American Cancer Society.

Pharmacogenomics address the relation between a patient's genetic makeup and response to a given drug. The importance of pharmacogenomics for clinical practice relates to the potential to determine the right drug at the right dose for a given patient.1 Of particular note is that inherited variations related to drug effects are common.1 It has been with the advances in our knowledge of the human genome, as provided by the Human Genome Project, that pharmacogenetics has evolved into pharmacogenomics, a term defined by Weinshilboum1 as the influence of DNA sequence variation on the effect of the drug. It is clear that in addition to drug metabolism, pharmacogenomics will also be important in the consideration of drug disposition and drug targets.2 The National Institutes of Health (NIH) has recognized the importance of pharmacogenomics studies through the establishment of the Pharmacogenetics Research Network (PGRN).3

Pharmacogenomic studies are beginning to yield important findings relating to tamoxifen and the aromatase inhibitors, a newer class of drugs that has obtained a position of central importance to the care of women with breast cancer. This brief review will consider some of the recent findings and consider their impact on the management of patients.


The selective estrogen receptor (ER) modulator tamoxifen has been the most important therapeutic agent in breast cancer for the past 3 decades for those women whose tumors are potentially endocrine-sensitive as indicated by expression of the ER and/or progesterone receptor (PgR). Tamoxifen has received approval from the US Food and Drug Administration for multiple breast cancer indications. These include the treatment of; 1) postmenopausal metastatic breast cancer in 1977, 2) postmenopausal lymph node-positive women in the adjuvant setting for early breast cancer in 1986, 3) premenopausal women with metastatic breast cancer in 1989, 4) patients irrespective of menopausal status with lymph node-negative breast cancer in the adjuvant setting in 1990, 5) high-risk women (‘prevention’) in 1998, and 6) women with resected ductal carcinoma in situ in 2000. Thus, tamoxifen has received FDA approval for the full spectrum of breast cancer from metastatic disease to those women who are at higher risk of developing breast cancer. In addition, tamoxifen has been approved for treatment of men with metastatic breast cancer in 1993.

The accepted dose of tamoxifen is 20 mg per day. We examined steady-state levels of tamoxifen and 2 of its major metabolites, 4-hydroxy-tamoxifen and N-desmethyl-tamoxifen, and found substantial interpatient variability for all 3; eg, the median level for tamoxifen after 6 weeks of therapy was 107 ng/mL with a range of 24 to 317 ng/mL.4 These data suggest that the accepted dose of tamoxifen may not be the optimal dose for all patients and, perhaps, that tamoxifen may not be an appropriate drug for some patients.

Recently, there has been new information regarding the metabolism of tamoxifen. For years, 4-hydroxy-tamoxifen has been considered the most important metabolite from the standpoint of therapeutic efficacy. Desta et al.5 performed a comprehensive examination of tamoxifen and found that the major primary metabolite of tamoxifen is N- desmethyl-tamoxifen, a reaction catalyzed by cytochrome P450 (CYP) 3A4/5 whereas 4-hydroxy-tamoxifen constitutes a minor metabolite with this reaction catalyzed by several P450 enzymes including CYP2D6. The important finding was that N-desmethyl-tamoxifen converted to 4-hydroxy-N-desmethyl-tamoxifen, a metabolite that has been given the name endoxifen. Several characteristics of endoxifen are of note. Endoxifen has been shown to have essentially the same binding affinity for ER-alpha as 4-hydroxy-tamoxifen6 and was equipotent to 4-hydroxy-tamoxifen in terms of inhibiting estrogen-stimulated growth in ER-positive breast cancer cell lines.6, 7 Of particular importance is that endoxifen is normally present in substantially higher concentrations, up to 10 times greater, than 4-hydroxy-tamoxifen in women taking the standard dose of tamoxifen for adjuvant therapy of breast cancer.8

The crucial finding in terms of metabolism is that endoxifen is produced by the hydroxylation of N-desmethyl-tamoxifen solely by the cytochrome P450 enzyme CYP2D6.5 CYP2D6 is a highly polymorphic gene with alleles with normal activity, decreased activity, or no activity and some cases of increased activity due to gene duplication.9 The impact of genotype on endoxifen levels was examined, demonstrating reduced levels with low-activity polymorphisms.10, 11

Goetz et al.12 took the next step and examined outcomes as a function of CYP2D6 genotype in a well-defined population of postmenopausal women with early-stage breast cancers whose tumors were known to be ER-positive and who were treated with tamoxifen monotherapy for 5 years. This population was drawn from 1 arm of a prospective clinical trial begun in 1989, providing substantial follow-up with a median of patients still alive at 11.4 years.13 CYP2D6*4 was the only genotype examined in the analysis but this accounts for the majority of poor metabolizers in Caucasian women. Despite the relatively small sample size (223 patients) and the inclusion of only the *4 genotype, which underestimates the frequency of decreased activity, the women homozygous for the *4 genotype (*4/*4) (13 patients, 6% of total) had a worse recurrence-free time and disease-free survival compared with those women with without this variant allele or who carried only 1, ie, a heterozygous state (*4/wildtype). Although these differences were significant in univariate analyses, significance was lost in multivariate testing. It was noted that women with the *4/*4 genotype had a significantly lower incidence of moderate or severe hot flashes, which was considered consistent with the previous observation that CYP2D6 is responsible for the metabolic activation of tamoxifen to endoxifen. That is, the *4/*4 patients might be expected to have a lower incidence of hot flashes, as they would have lower levels of the active metabolite endoxifen.

Utilizing this same study population, Goetz et al.14 went on to collect information relating to the concurrent use of medications (selective serotonin reuptake inhibitors and the serotonin and norepinephrine reuptake inhibitors) that have been shown to inhibit CYP2D6 function.8, 10, 11 When the genotype and presence or absence of the CYP2D6 inhibitor were utilized to determine a metabolizer phenotype it was found in a multivariate analysis that patients with decreased metabolism had a significantly shorter time to recurrence and worse recurrence-free survival. The conclusion from the study was that CYP2D6 metabolism, when measured by genetic variation and enzyme inhibition, is an independent predictor of outcome in postmenopausal women who have received adjuvant tamoxifen. Of note is that in the Italian chemoprevention trial with tamoxifen from Bonanni et al.15 that patients who developed breast cancer were significantly more likely than controls to have the CYP2D6 *4/*4 genotype, suggesting to the authors that women with this genotype may be less likely to benefit from tamoxifen.

Goetz et al.14 presented their data at a meeting of the Advisory Committee for Pharmaceutical Science to the Clinical Pharmacology Subcommittee of the US Food and Drug Administration on October 18, 2006. The consensus of the advisory committee was that the tamoxifen label should be updated to reflect the increased risk for breast cancer in postmenopausal women with ER-positive breast cancer who are CYP2D6 poor metabolizers by genotype or drug interaction. Consensus was not reached on whether CYP2D6 testing should be recommended or presented as an option. The final decision from the FDA regarding the label change is pending. However, the use of CYP2D6 testing has found its way into some clinical practices. The ethics of CYP2D6 testing is being discussed in the published literature. Hartman and Helft16 proposed that it is reasonable to do CYP2D6 testing in postmenopausal women, despite the need for additional study, as there is an attractive alternative to tamoxifen for patients who are CYP2D6 poor metabolizers, namely, the aromatase inhibitors. In premenopausal women the alternative to tamoxifen would be an aromatase inhibitor in addition to ovarian suppression and there are no randomized trials that have demonstrated the equivalence between these 2 approaches.

Wegman et al.17 reported the opposite findings from those noted above14 in that they found that in women treated with 5 years of tamoxifen that those who were homozygous or heterozygous for the CYP2D6 *4 genotype (32 patients) had better disease-free survival than those with neither *4 allele (79 patients), although this did not reach statistical significance (P = .12). The basis of the discordant findings is not clear. Additional retrospective studies are needed and are planned utilizing material from large, well-conducted prospective clinical trials with reasonable follow-up. However, it is this author's view that a strong case can be made for the conduct of prospective studies evaluating the value of CYP2D6 testing. A clear advantage of a prospective trial would be that a microarray hybridization method such as the AmpliChip18 could be used for genotyping of DNA from blood with coverage of most of the different genotypes. Such an approach would provide a more accurate determination of the CYP2D6 status than provided with the polymerase chain reaction-based approaches utilized in studies examining archival materials, as in the studies noted above,14, 15, 17 in which only 1 or 2 CYP2D6 genotypes were examined.

Aromatase Inhibitors

In the past decade the aromatase inhibitors have emerged as a powerful addition to the clinicians' armamentarium against breast cancer. They appear to be more efficacious in the advanced disease setting,19 have become clearly established in the adjuvant setting,20 and are a major focus of ongoing trials in the prevention setting in postmenopausal women.21

In reviewing the major trials of aromatase inhibitors in the adjuvant setting, one is struck by differences among patients. These differences are seen in terms of efficacy, ie, whether a patient experiences recurrence of her breast cancer, but also in terms of tolerability. Perhaps 1 of the most striking events seen with aromatase inhibitors is musculoskeletal complaints, which can be devastating in some patients, necessitating discontinuation of the aromatase inhibitor. These differences in clinical outcomes suggest that there may be pharmacogenomic differences between patients. A major work that supports this hypothesis is from Ma et al.,22 who sequenced the aromatase gene in 60 patients from each of 4 ethnic groups (Caucasian-Americans, African-Americans, Han Chinese-Americans, and Mexican-Americans) and found 88 polymorphisms resulting in 44 haplotypes. Substantial variation between the 4 ethnic groups was seen in the pattern of polymorphisms. These investigators performed functional genomic studies with 4 nonsynonymous coding single nucleotide polymorphisms (cSNP) and found a significant correlation between level of activity and immunoreactive protein. Three of the 4 cSNPs had levels of immunoreactive protein that were significantly lower than wildtype aromatase enzyme. It might be expected that patients with decreased aromatase activity would have decreased potential to benefit from the use of an aromatase inhibitor.

Currently, several large studies evaluating pharmacogenomics of aromatase inhibitors are being conducted under the auspices of the PGRN. The Mayo Clinic PGRN in collaboration with the M. D. Anderson Cancer Center and the Indiana University PGRN are conducting a study to enroll 1000 patients treated with anastrozole for adjuvant therapy of early breast cancer. The hypothesis to be tested is that inherited variation in pathways for anastrozole metabolism (pharmacokinetic pathway) and/or steroid biosynthesis, metabolism, and effect (pharmacodynamic pathways) contribute to variation between individuals in terms of anastrozole efficacy and toxicity. In that study genes will be tested for SNPs and intragene haplotypes in both the anastrozole pharmacokinetic and estrogen pharmacodynamic pathways.


The field of pharmacogenomics represents an important mechanism by which personalized medicine will become a reality. Whereas this applies to all fields of medicine, it is very clear that it will be of particular importance in oncologic management. There is the opportunity to identify therapies with the greatest potential for benefit and minimize toxicities.

In the case of breast cancer, the data with respective CYP2D6 metabolism status and its relation to ER-positive early breast cancer treatment with tamoxifen have been demonstrated in a retrospective analysis of prospective studies. Of particular note is that the use of concurrent medications that inhibit CYP2D6 offer the potential to decrease the efficacy of tamoxifen by virtue of CYP2D6 inhibition. The available data suggest that determination of the CYP2D6 genotype may be of value in selecting adjuvant hormonal therapy. Validation in existing datasets is under way but a strong case can be made that prospective study is needed.

Pharmacogenetic and pharmacogenomic studies in women receiving aromatase inhibitors are under way but are at an earlier stage of development. The demonstration of polymorphisms with low levels of activity and the clinical variability in tolerance of the aromatase inhibitors (AIs) indicate that pharmacogenomic factors may be identified that will help select therapy for patients.


The questions and discussion below come from the oral presentation given at the Sixth Cambridge Conference on Endocrine and Targeted Manipulation of Breast Cancer and do not correspond directly to the written article, which is a more general review.

Dr. Aman Buzdar: What was the FDA's recommendation? Was it that this should be done?

Dr. James Ingle: There was a unanimous agreement that the label should be modified. That was in October. This is April. We were told it takes 6 to 9 months to get an FDA label change.

Dr. Eric Winer: You need to do another study, several analyses on large, prospectively gathered cohorts would be okay. That'd make all of us believe better.

Dr. Jo Anne Zujewski: AIs and tamoxifen extensive metabolizers have similar hazard ratios over time. So would you anticipate that if you were going to do a prospective clinical trial that tamoxifen would not be inferior to an AI?

Dr. Ingle: No, the protocol was written as a superiority trial for the switching strategy of tamoxifen followed by an AI. You can actually do calculations where it is clear that the switching strategy is superior to AI monotherapy.

Dr. Zujewski: If you start out with an AI, or if you start out with tamoxifen in extensive metabolizers, wouldn't the hazard curves be the same?

Dr. Ingle: No, they're not. The hazard rates for the extensive metabolizers on tamoxifen in our experience are lower than what was shown in ATAC with the AI.

Dr. Buzdar: In spite of all the discussion…, I am not sure whether things are clearer or murkier. I thought I knew the answers and now, these SNP data, do you think this is ready for practice? I get phone calls every day, because the assays are commercially available, so the patients say that, oh my profile is this, what should I be doing? and doctors want to know, what should they be doing? What are the recommendations from the people who are running these sorts of assays? Maybe we should have some sort of guidelines with the data we currently have. Because assays are apparently available to everyone in the community.

Dr. Ingle: This is why we shared the data with the FDA when it came out. When they invited presentation to the subcommittee, we thought that was important, because life would be easier if the FDA took a position on it. In the absence of a label change, I think it is reasonable to inform postmenopausal women in the adjuvant setting of the data. We don't do it in DCIS, I mean there's no alternative, we don't know what we'd say. We fully inform the patient and if she wants it done, we do it.

Dr. William Miller: What if the patient says “you're the expert, what do you think I should do?”

Dr. Ingle: There is compelling data that a CYP2D6 genotype status is related to outcome, and there is compelling data that you should not take CYP2D6 inhibitors if you are going on tamoxifen. It is very interesting in our JCO paper in 2005 (Goetz MP et al. J Clin Oncol 2005;23:9312–938). It's the women who convert tamoxifen to endoxifen who have the bad hot flashes. I've told women who have bad hot flashes that this suggests the drug is working. Now some are happy if they get hot flashes.

Dr. Zujewski: One recommendation you could make today is to avoid antidepressants that are CYP2D6 inhibitors.

Dr. Ingle: The 1 thing that's more accepted than anything else is that if somebody is on tamoxifen you should not use CYP2D6 inhibitors. I honestly think pharmacogenomics is going to be important. It's basically important in every drug we give. And it's particularly important in these anticancer drugs. Even in our best drugs, there's room for better response rates and perhaps less toxicity. You can have 2 women who look exactly the same, put both on anastrazole, 1 doesn't know if she's taking anything and the other can't move because of all the musculoskeletal problems. So, side effects really differ.