Personalizing tamoxifen therapy in adjuvant therapy: a brief summary of the ongoing discussion

KEYWORDS:

• Breast cancer
• CYP2D6
• endocrine therapy
• endoxifen
• pharmacogenetics
• pharmacokinetics
• tamoxifen
• therapeutic drug monitoring

Breast cancer still remains the most common form of cancer diagnosed among women [1], and it represents 15% of all cancer deaths in female patients [2]. Around 60% of the new breast cancer cases are estrogen-receptor positive (ER) [3]. In these patients endocrine therapy in the form of tamoxifen or an aromatase inhibitor (AI) is frequently used depending on the menopausal status of the patients [4,5]. For pre-menopausal women, tamoxifen for at least five years is the standard-of-care, while for post-menopausal women a switch to an aromatase inhibitor after two or three years of tamoxifen use is usually prescribed in the adjuvant setting [4,5]. Particularly, this sequential therapy has shown to be a well-tolerated treatment [4,5]. In the last couple of years, an extended duration of tamoxifen treatment up to ten years seems more advantageous in decreasing breast cancer recurrence and fatalities [4–6]. Nevertheless, almost 30% of early-breast cancer patients still will have a breast cancer relapse in the following fifteen years [7]. These figures fueled studies aimed at further improving tamoxifen and AI therapy in the adjuvant setting of breast cancer.

Tamoxifen is a weak anti-estrogenic drug that monopolized the endocrine therapy in both adjuvant and metastatic setting for more than 40 years [8,9]. The classical description of tamoxifen metabolism follows two parallel pathways in which tamoxifen is transformed into 4-hydroxy-tamoxifen and N-desmethyl-tamoxifen. Thereafter, a second conversion from both metabolites into endoxifen takes place. Of all tamoxifen metabolites, 4-hydroxy-tamoxifen and endoxifen have shown to have the strongest anti-estrogenic activities, which means roughly 30 to 100 times more active compared to tamoxifen alone [10]. However, normally endoxifen is considered the most active metabolite of tamoxifen because it is found in 5- 10-fold higher concentrations in comparison to 4-hydroxy-tamoxifen [11].

In this complex metabolic pathway, the Cytochrome P450 2D6 (CYP2D6) enzyme in the liver has frequently been reported as the limiting factor of tamoxifen metabolism [12]. The CYP2D6 is a highly polymorphic gene for which more than 100 polymorphisms have been described [13]. These variants can encode for an CYP2D6 enzyme that is nonfunctional, fully functional or has decreased activity. According to the combination of the CYP2D6 alleles, individuals can classically be classified into four CYP2D6 phenotypes: poor metabolizer (PM), intermediated metabolizer (IM), normal metabolizer (NM), and ultra-rapid metabolizer (UM). Another manner to categorize patients is the CYP2D6 gene activity score [14] in which a score of 2 represents a fully functional enzyme, whereas a score of 0 represents a non-active CYP2D6 enzyme.

Goetz and colleagues published in 2005 the hallmark study that started an ongoing controversy regarding the association between tamoxifen efficacy and the CYP2D6 phenotype [15]. In this study it was shown that CYP2D6 PM and IM patients reach lower endoxifen concentrations compared to NM as a result of the absent or decreased CYP2D6 enzymatic activity. As a consequence, these patients are assumed to be at higher risk of breast cancer relapse due to the lower anti-estrogenic exposure, e.g. lower endoxifen concentrations. Since the publication of this landmark study, many studies have been published in the last 15 years claiming positive and negatives outcomes regarding this relationship. Still, no general consensus have been achieved, mainly due to a great number of differences across the different studies which make comparisons even more challenging. A few examples might be the source of DNA for CYP2D6 genotyping (e.g. tumor-tissue or blood), the studied endpoints (relapse-free survival, overall survival, disease-free survival, etc.) or study design (e.g. retrospective or prospective), the genotype to phenotype translation.

In 2014, the International Tamoxifen Pharmacogenomics Consortium carried out a meta-analysis as an attempt to give a definitive answer to this topic [16]. In this study, 4973 breast cancer patients who received adjuvant endocrine therapy with tamoxifen were studied by testing three predefined inclusion criteria. Of the three analysis, only in the most limited and restricted criterion a significantly poorer survival result in PM was observed (HR: 1.25, 95% CI: 1.06–1.47, p-value: 0.009). While for many researchers this meta-analysis was accepted as a conclusive proof of the association between improved clinical outcomes and the role of CYP2D6 polymorphisms in breast cancer patients treated with tamoxifen, many other authors have been highly critical with these conclusions. An important and relevant limitation was the exclusion of studies such as ATAC [17], TEAM [18] or BIG1-98 [19], mainly due to loss of heterozygosity as observed in these studies. In theory, patients might be categorized in the ‘incorrect’ predicted phenotype due to the loss of heterozygosity. However, a meta-analysis by Ahern et al. evaluated the clinical relevance of the loss of heterozygosity in tamoxifen-treated patients [20]. Authors concluded that, despite the strong arguments for the relevance of loss of heterozygosity, it might lack clinical relevance. In any case, since this putative association between tamoxifen efficacy, CYP2D6 genotyping, and clinical outcomes still remains unclear, many other approaches have been suggested as alternatives for potentially guiding tamoxifen efficacy.

A critical disadvantage of the use CYP2D6 as a marker for tamoxifen clinical outcomes might rely on the complexity of tamoxifen metabolism and how individuals are classified in only four predicted phenotypes. While CYP2D6 is its main limiting factor, it only partially explains the observed inter-patient variability, and much ‘information’ might be lost in this categorical translation between genotypes and phenotypes. In an effort to improve our knowledge regarding this inter-variability of tamoxifen metabolism due to CYP2D6 genotypes and phenotypes, a study by Lee and colleagues [21] using CYP2D6 long read sequencing and machine learning was carried out. Authors developed a continuous scale to predict CYP2D6 enzymatic activity instead of the classical separation in categories. Interestingly, the inter-individual variability improved from 54% to 79% of the explained variance of the concentrations of endoxifen concentrations. While this approach might help us to gain knowledge regarding CYP2D6 gene and the best manner to interpret the genotype and phenotype translation, more research is still awaiting to clarify the potential role for the clinical practice.

As a consequence, there is currently a tendency to take the focus off CYP2D6 and alternatives such as metabolite concentrations are proposed to predict tamoxifen efficacy, e.g. endoxifen or 4-hydroxy-tamoxifen concentrations. In theory, these active metabolites are supposed to be ‘closer to the pharmacological effect’ of tamoxifen than CYP2D6 genotypes or phenotypes only. Among the three principal tamoxifen active metabolites, endoxifen and 4-hydroxy-tamoxifen concentrations have been proposed as feasible alternatives to guide tamoxifen efficacy. In the case of 4-hydroxy-tamoxifen only a few studies have showed a positive association clinical outcome [1], but many others failed to find such a relationship. On the contrary, endoxifen concentrations have been suggested as a better predictor to guide tamoxifen efficacy. Three retrospective studies have obtained a positive association between improved clinical outcomes with higher endoxifen concentrations. Madlensky and colleagues investigated 1370 breast cancer patients and described a ‘threshold’ for endoxifen of 5.97 ng/ml above which individuals had a 26% lower chance of breast cancer relapse [22]. In the same line, Saladores et al. [23] and Helland and colleagues [24] also described even lower ‘cut-off’ values of 5.2 ng/ml and 3.3 ng/ml, respectively. While the aforementioned studies were based on retrospective data, only a few recent studies based on prospective datasets based on the metastatic [25,26], neoadjuvant [25], and adjuvant [27,28]settings have been analyzed, and authors have failed to validate a strong and clinically relevant relationship.

One could argue that a potential relation between higher endoxifen concentrations and better clinical outcomes may exist. Consequently, therapeutic drug monitoring (TDM) based on endoxifen concentrations may be of relevance to predict tamoxifen effectiveness. Still, such a relationship has already been studied in the adjuvant setting and in a prospective study, and no clear concentration-effect association was found [29]. The main problem with this conclusion may then rely on how high endoxifen concentrations should be at the expense of adverse effects. While several studies have shown the feasibility of higher tamoxifen dosages (e.g. 40–120 mg of tamoxifen daily) for short periods of time [30,31], there is no evidence of the long-term potential benefits of these clinical interventions and whether the use of higher doses could be harmful to patients in the long run. In contrast, TDM might potentially be of added value to assure adherence to tamoxifen treatment.

In any case, and despite the high volume of studies published on tamoxifen efficacy, the remaining question still is what is the best strategy to individualize tamoxifen therapy. To this end, new research approaches to personalize tamoxifen treatment are extremely necessary. A potential strategy might be the use of large datasets for establishing a predictive algorithm in which a combination of clinical and genetic information is implemented. In recent times, population pharmacokinetic models based on endoxifen concentrations and CYP2D6 genotypes were studied [32]. While this approach might be useful, if multiple genetic information, clinical characteristics, and long follow-up could be used, potential new manners based on novel approaches such as machine learning methods could be explored. Still, this strategy would require even more research in order to translate these outcomes to the clinical practice. Another approach would be the use of endoxifen itself as an oral drug, but its current place in therapy remains unclear and more research is required.

In conclusion, endocrine treatment with tamoxifen has been widely prescribed for many years and many studies have been carried out in order to find a valid and solid manner to individualize tamoxifen treatment. However, no general consensus has been reached out, and personalizing tamoxifen therapy still remains an ongoing discussion.

Additional information

Funding

This paper was not funded.