Skip to content

Technology Evaluation
Center (TEC)

CYP2D6 Pharmacogenomics of Tamoxifen Treatment


Background:  Tamoxifen is prescribed as a component of adjuvant endocrine therapy to prevent endocrine receptor-positive breast cancer recurrence, as treatment of metastatic breast cancer, and to prevent disease in high-risk populations and in women with ductal carcinoma in situ (DCIS).  The cytochrome P450 (CYP) metabolic enzyme CYP2D6 has a major role in tamoxifen metabolism.  The CYP2D6 gene is polymorphic; variant DNA gene sequences resulting in proteins with markedly reduced or absent enzyme function may be associated with lower plasma levels of active tamoxifen metabolites, particularly endoxifen which is solely dependent on CYP2D6 for its production, and are thus hypothesized to have an impact on tamoxifen treatment efficacy.  Patients who have little or no CYP2D6 enzyme function are called poor metabolizers (PMs) compared to patients with 2 fully functional alleles, termed “wild type” or extensive metabolizers (EMs).  Those with enzyme activity in between are called intermediate metabolizers.

Objective:  This Assessment evaluates the evidence for CYP2D6 genotyping, compared to no testing, to direct treatment regimen choices for patients at high risk for primary breast cancer or breast cancer recurrence, and improve survival outcomes.

Search strategy:  MEDLINE® was searched (via PubMed) using the search string ("Breast Neoplasms"[MeSH®] AND "Tamoxifen"[MeSH®]) AND "Cytochrome P-450 Enzyme System"[MeSH®] through January 2011.  Clinical trials, recent reviews (2008–2011), editorials, and letters related to the pharmacogenomics of tamoxifen were retrieved.  In addition, text and reference lists of retrieved papers were examined for additional relevant articles.

Selection criteria:  Full-length, peer-reviewed papers reporting studies of postmenopausal women undergoing endocrine therapy whose treatment regimen selection is based on CYP2D6 genotyping versus usual selection methods, or studies of the association of CYP2D6 genotype with intermediate (e.g., tamoxifen active metabolite levels) or final outcomes (e.g., time to recurrence, survival) were selected for review.  Possible sources of bias and other factors that might influence the quality of the results were considered.

Main results:  One U.S. Food and Drug Administration (FDA) -cleared test for CYP2D6 genotyping has consistent evidence of analytic validity (i.e., technical accuracy and reliability). 

Evidence for clinical validity (i.e., association of CYP2D6 genotype with clinical outcomes) consists of the following elements grouped into the beginning of two possible evidence chains, A and B:


  • Association of genotype with plasma levels of active tamoxifen metabolites:  Five prospective cohort studies of adjuvant tamoxifen treatment provide consistent evidence that CYP2D6 nonfunctional variant alleles are associated with significantly reduced plasma endoxifen levels.  However, endoxifen levels overlap across all genotypes, suggesting that CYP2D6 genetic variability does not explain all variability in endoxifen levels.  Somewhat surprisingly, since generation of 4-hydroxy tamoxifen, another active tamoxifen metabolite, does not depend solely on CYP2D6, three of four studies report a significant association of low CYP2D6 function with reduced plasma 4-hydroxy tamoxifen levels.  Co-administration of a potent CYP2D6 inhibitor to CYP2D6 homozygous wild-type patients is associated with endoxifen levels near those of patients who are poor metabolizers.


  • Association of in vivo endoxifen levels with clinical outcomes:  Two studies report on the relationship between CYP2D6 genotype and active tamoxifen metabolites, and between genotype and clinical outcomes in the same patient population.  Both studies enrolled patients from Asian populations, focusing almost exclusively on the prevalent reduced function CYP2D6*10 variant in this population.  Both studies reported reduced endoxifen and/or 4-OH tamoxifen concentrations in conjunction with variant alleles, and in conjunction reported decreased disease- or recurrence-free survival.  One or both studies have study design flaws likely resulting in selection bias.  In addition both studies are small, resulting in estimates of association with wide confidence intervals.  Thus, the relationship between endoxifen (or 4-hydroxy tamoxifen) plasma concentrations and clinical outcomes has not been well-established in Asian populations, nor has it been studied in Caucasian populations with null function CYP2D6 genotypes.



Association of genotype with clinical outcomes:  an ideal study would compare tamoxifen-treated women vs. those not receiving tamoxifen, with stratification by CYP2D6 genotype to see if PMs derive less benefit from tamoxifen than EMs.  One group* conducted such a study retrospectively, on archived samples from a randomized controlled trial of tamoxifen treatment.  Paradoxically, they found that EMs treated with tamoxifen received no statistically significant clinical benefit compared to EMs not treated with tamoxifen, and that carriers of a CYP2D6*4 nonfunctional variant allele obtained significant benefit from tamoxifen treatment.  There were several limitations to this study such that results are questionable. 

The other included studies enrolled only tamoxifen-treated women and evaluated outcomes by CYP2D6 genotype.  A few separately evaluated a non-tamoxifen treated control population but without direct comparison.  Seven small studies in Asian populations focused on the CYP2D6*10 reduced function allele and four reported significant results for the association of CYP2D6 genotype with outcomes of tamoxifen treatment, but may be affected in unpredictable ways by different types of bias.  Two studies that reported no association may have less potential for bias.  Twelve studies evaluated samples from primarily Caucasian patients. Of the Five largest studies, four reported no significant association between CYP2D6 reduced or null activity genotype and time to breast cancer recurrence.  Two of the negative studies were retrospective analyses of clinical trial samples and one was a matched case-control study nested within a well-documented breast cancer registry.  All three were designed to minimize the potential for bias; their sizes allowed comparison of homozygous nonfunctional CYP2D6 genotypes (PMs) with fully functional wild type genotypes (EMs), i.e. the most extreme comparison and most likely to reveal a true association.

The fourth negative study was a cohort study.  The fifth and largest study reported significant results; this study combined samples from different sources, some of which had already been analyzed for this hypothesis.  In addition, it is not clear from the report whether nearly half of the samples were obtained from patients who had survived and were available at a time distant from their diagnosis and surgery, a type of selection bias that can unpredictably affect results.  The remaining seven small studies report a variety of significant and nonsignificant results; no pattern of bias, genotyping or group scheme, or accounting for CYP2D6 inhibitor use (among possibilities) explains the differences in results.  The heterogeneity of results across all studies and clear results of no genotype-tamoxifen treatment outcome in three large studies with the least apparent potential for bias suggests lack of support for clinical validity.

There is no direct evidence of clinical utility (whether use of CYP2D6 genotype testing for endocrine therapy regimen selection improves recurrence and survival outcomes).  Two indirect evidence chains, A and B (above), can be constructed; the final element for both in postmenopausal women would be evidence that CYP2D6 PMs treated with aromatase inhibitors (AI) alone have outcomes at least as good as CYP2D6 EMs treated with AI alone or AI plus tamoxifen.  In premenopausal women who are CYP2D6 PMs, ovarian ablation or suppression, which confers a significant benefit compared to no therapy, might be added to or might replace tamoxifen.  Evidence chain A depends on demonstrating a significant association between in vivo endoxifen levels and clinical outcomes; this evidence is insufficient.  Evidence chain B depends on the association of genotype with clinical outcomes (clinical validity), which the best evidence does not support.  Without evidence of clinical validity, there can be no basis for considering a change in management for patients with specific genotypes to improve outcomes (clinical utility).

Author’s conclusions and comments:  The hypothesis examined in this Assessment is that patients with CYP2D6 gene variants that result in markedly reduced or absent enzyme function have reduced tamoxifen metabolism and lower endoxifen levels compared to genotypic wild type extensive metabolizers, and as a direct result have poorer clinical outcomes.  This hypothesis is based on the assumption, not yet supported by evidence, that some level of endoxifen is sufficient and necessary for tamoxifen efficacy, and that this level is not achieved in patients with markedly reduced or no CYP2D6 enzymatic function.  However, because tamoxifen metabolism is complex and CYP2D6 does not appear to account for all variability in endoxifen levels, it is conceivable that polymorphisms in other tamoxifen metabolic pathway enzymes may affect active metabolite levels, and in theory direct measurement of the metabolite(s) itself might be the better predictor of benefit from tamoxifen treatment.  However, measuring metabolite levels is not practical for clinical applications.

Whether lower endoxifen levels can affect the pharmacodynamics of tamoxifen and tamoxifen metabolites interaction with estrogen receptors, and ultimately tamoxifen efficacy, is unclear.  The dissociation constants of even the more weakly binding molecules, including tamoxifen itself, are reportedly still sufficient to effectively block estrogen binding.  Moreover, it is estimated that at doses used for adjuvant treatment, which is intended to saturate the estrogen receptor, more than 99% of estrogen receptors are bound by tamoxifen and its metabolites.

Lacking the appropriate mechanistic evidence, it remains to examine the evidence, the bulk of which addresses clinical validity, the association between CYP2D6 genotype and tamoxifen treatment outcome, and a necessary link in the construction of an indirect chain of evidence of clinical utility.  As noted, there is heterogeneity of results across studies.  Heterogeneity in effect estimates, both in size and significance or lack thereof, is likely due to the lack of power in most studies, and to potential sources of bias in some.  The analysis of archived samples from two large completed clinical trials was undertaken in order to achieve adequate power, more fully evaluate CYP2D6 genotype, evaluate aromatase inhibitor-treated control populations in tandem, and to avoid potential sources of bias.  Additionally, a well-powered case-control study nested within a population-based registry was carefully designed to avoid bias and compare results from estrogen receptor-positive, tamoxifen-treated patients, stratified by CYP2D6 genotype, with those from estrogen receptor-negative, non-tamoxifen treated patients, similarly stratified.  That the results of these three studies have to date shown  no evidence of association between CYP2D6 genotype and either tamoxifen- or aromatase inhibitor-treated patient outcomes has suggested that using the results of CYP2D6 genetic testing to influence decisions about tamoxifen treatment is not currently warranted.

Based on the available evidence, the Blue Cross and Blue Shield Medical Advisory Panel made the following judgments about whether CYP2D6 genotyping for directing endocrine therapy regimen selection for women at high risk for primary breast cancer or breast cancer recurrence meets the Blue Cross and Blue Shield Association Technology Evaluation Center (TEC) criteria.

1.            The technology must have final approval from the appropriate governmental regulatory bodies.

The Roche AmpliChip CYP450 Test is cleared by the U.S. Food and Drug Administration (FDA) and is “intended to identify a patient's CYP2D6 and CYP2C19 genotype from genomic DNA extracted from a whole blood sample. “Information about CYP2D6 and CYP2C19 genotype may be used as an aid to clinicians in determining therapeutic strategy and treatment dose for therapeutics that are metabolized by the CYP2D6 or CYP2C19 gene product”
( ).

CYP2D6 genotyping assays are also available as laboratory-developed tests (LDT).  Clinical laboratories may develop and validate tests in-house and market them as a laboratory service; laboratories offering LDTs as a clinical service must meet the general regulatory standards of the Clinical Laboratory Improvement Act (CLIA) and must be licensed by CLIA for high-complexity testing.

The FDA has been considering updating the label for tamoxifen (brand and generics) with information or recommendations regarding CYP2D6 genotyping and impact on tamoxifen efficacy.  On October 18, 2006, the FDA held an Advisory Committee meeting to answer specific questions regarding the evidence and recommendations for the label update.  Since the Advisory Committee meeting, AstraZeneca, the brand name (Nolvadex®) manufacturer, has ceased producing tamoxifen and is no longer maintaining the prescribing information. As of the date of this Assessment, no direction has come from the FDA regarding revised labeling of generic versions of tamoxifen to include CYP2D6 genotyping information.

2.            The scientific evidence must permit conclusions concerning the effect of the technology on health outcomes.

There is no direct evidence of clinical utility.  Two indirect evidence chains can be constructed.  One depends on demonstrating a significant association between endoxifen and clinical outcomes; this evidence is insufficient.  The other depends on the association of genotype with clinical outcomes; there are several limitations to the overall body of evidence, but the largest, most well-designed studies do not support a significant association.

3.            The technology must improve the net health outcome; and

4.            The technology must be as beneficial as any established alternatives.

Because the best available evidence does not support a significant association between CYP2D6 genotype and tamoxifen treatment outcome, an indirect evidence chain supporting the clinical utility of CYP2D6 genotyping for directing endocrine therapy regimen selection for women at high risk for or with breast cancer cannot be constructed.

5.            The improvement must be attainable outside the investigational settings.

Whether or not the use of CYP2D6 genotyping for directing endocrine therapy regimen selection for women at high risk for or with breast cancer improves health outcomes has not been demonstrated in the investigational setting.

Based on the above, CYP2D6 genotyping does not meet the TEC criteria for directing endocrine therapy regimen selection for women at high risk for primary breast cancer or breast cancer recurrence.

* Wegman P, Vainikka L, Stal O et al.  (2005).  Genotype of metabolic enzymes and the benefit of tamoxifen in postmenopausal breast cancer patients.  Breast Cancer Res, 7(3):R284-90.