Effect of CYP3A4*22, CYP3A5*3, and CYP3A combined genotypes on tamoxifen metabolism

Background Tamoxifen is one of the cornerstones of endocrine therapy for breast cancer. Recently, the decreased activity CYP3A4*22 allele and the loss of function CYP3A5*3 allele have been described as potential factors that could help to explain the inter-patient variability in tamoxifen metabolism. The aim of this study is to investigate the effect of CYP3A4*22, CYP3A5*3, and CYP3A combined genotypes on tamoxifen metabolism. Methods DNA from 667 women enrolled in the CYPTAM study (NTR1509) was genotyped (CYP2D6, CYP3A4*22, and CYP3A5*3). Tamoxifen and metabolite concentrations were measured in serum, and metabolic ratios were calculated. The effect of the CYP3A4*22, CYP3A5*3, and CYP3A combined genotypes in addition to the CYP2D6 genotypes was examined by multiple linear regression analysis. Results CYP3A4*22 carriers reached significant higher concentrations of tamoxifen, N-desmethyl-tamoxifen, and 4-hydroxy-tamoxifen compared to non-carriers, whereas a tendency toward increased endoxifen levels was observed (p = 0.088). The metabolic ratio tamoxifen/N-desmethyl-tamoxifen was significantly higher in CYP3A4*22 individuals (0.59 vs. 0.52, p < 0.001). At the same time, CYP3A4*22 genotype contributed to improving the inter-variability [R 2 of the (log-transformed) metabolic ratio tamoxifen/N-desmethyl-tamoxifen improved from 21.8 to 23.9%, p < 0.001]. CYP3A5*3 marginally improved the explained variability of the (log transformed) metabolic ratio 4-hydroxy-tamoxifen/endoxifen (from 44.9 to 46.2%, p < 0.038). Conclusion Our data demonstrate that CYP3A genotype has a minor effect to explaining the variability between patients in tamoxifen metabolism and has no added value in addition to CYP2D6 genotype. Electronic supplementary material The online version of this article (10.1007/s00228-017-2323-2) contains supplementary material, which is available to authorized users.


Introduction
Breast cancer is the most common diagnosed cancer in women, representing nearly 25% of all cancers [1]. Approximately 60-75% of breast cancer patients have estrogen receptorpositive tumors [2], and in such cases, endocrine therapy may be indicated.
Tamoxifen has been widely prescribed to treat breast cancer patients with estrogen-receptor tumors for more than 40 years [3,4]. As a prodrug, tamoxifen is metabolized by different cytochrome P-450 enzymes to its primary metabolites, 4hydroxy-tamoxifen and N-desmethyl-tamoxifen [5] (NDM-tamoxifen). A second biotransformation from NDM-tamoxifen into endoxifen is principally regulated by CYP2D6 enzyme. At the same time, 4-hydroxy-tamoxifen also is biotransformed into endoxifen, mainly controlled by CYP3A4/5 and CYP2D6 enzymes, among others [6] (Fig. 1). Endoxifen is believed to be the most relevant tamoxifen metabolite since it is found in larger concentrations than 4-hydroxy-tamoxifen [7].
Additionally, CYP2D6 is considered the rate-limiting enzyme in tamoxifen metabolism [8] because it metabolizes the transformation of NDM-tamoxifen into endoxifen, which accounts for around 92% of tamoxifen metabolism [9]. However, it only partially explains the inter-patient variability of the metabolic ratio NDM-tamoxifen/endoxifen. According to Mürdter and colleagues [10], 68.7% of the variance in metabolic ratio of NDM-tamoxifen/endoxifen is explained by polymorphisms in CYP2D6.
Polymorphisms in genes encoding for other enzymes, such as CYP3A [11,12], have been also related to tamoxifen metabolism. CYP3A4 is implicated in the metabolization of 30-50% of common therapeutic drugs [13], whereas CYP3A5 is also known to have a relevant function in drug metabolism [14].
CYP3A4 plays a role in the transformations of 4-hydroxytamoxifen to endoxifen, tamoxifen to NDM-tamoxifen, and tamoxifen to 4-hydroxy-tamoxifen (Fig. 1). Genetic polymorphisms of CYP3A4, with some effect on tamoxifen metabolism, have been identified [15,16]. Still, there is limited information about the clinical relevance of most of these polymorphisms. However, CYP3A4*22 has been suggested to be an actionable CYP3A allele [17]. With a frequency of 5-7% in Caucasian population, CYP3A4*22 has been associated with decreased CYP3A4 activity [18]. CYP3A4*22 has been suggested to have a role in the metabolism of immunosuppressive drugs [18,19], whereas for tamoxifen, diverse evidence can be found in the literature [20][21][22]. Teft et al. suggested that CYP3A4*22 carriers were two times more likely to have higher endoxifen levels [20]. Antunes et al. proposed that CYP3A4*22 genotype is associated with increased concentrations of 4-hydroxy-tamoxifen in the presence of impaired CYP2D6 activity [21]. In a clinical setting, Baxter and colleagues described CYP3A4*22 carriers tend to have less hotflashes symptoms when compared with non-carriers [22].
CYP3A5 genetic polymorphisms are also involved in tamoxifen metabolism, but studies with tamoxifen have yielded conflicting data. Initially, Jin et al. described that carriers of non-functional CYP3A5 alleles, such as CYP3A5*3, were more likely to have higher endoxifen concentrations than individuals with a functional CYP3A5*1 allele [23]. Yet, no significant association between CYP3A5 polymorphisms with tamoxifen and its metabolites concentrations or clinical outcome has been found by other researchers [10,20].
Combined data about the effect of CYP3A4 and CYP3A5 has also been analyzed in renal [19] and heart transplantation [24]. However, little is known about this combined effect on tamoxifen metabolism. In an attempt to elucidate the factors that are related to variability in tamoxifen metabolism, we aimed to investigate the effect of CYP3A4*22, CYP3A5*3, and CYP3A combined genotypes on tamoxifen metabolism.

Study population and study design
Blood and serum samples were used from individuals enrolled in the CYPTAM study (NTR 1509) [25]. The aim of the CYPTAM study was to correlate CYP2D6 predicted phenotypes and endoxifen with relapse-free survival, disease-free survival, and overall survival. In brief, from February 2008 till December 2010, patients with early breast cancer receiving adjuvant tamoxifen were recruited in the multicenter prospective CYPTAM study in The Netherlands and Belgium.
All the enrolled patients signed an informed consent. Women with a history of a previous malignancy within the last 5 years, with the exception of patients appropriately treated for an in situ cervix carcinoma or basal cell carcinoma, were excluded. Other exclusion criteria were pregnancy, breastfeeding, or an unwillingness to sign the informed consent. The CYPTAM study was approved by the Medical Ethical Committee of the Leiden University Medical Center in Leiden (The Netherlands). After inclusion in the CYPTAM study, and having used tamoxifen for more than 2 months but less than 1 year, both whole blood and serum samples were collected for genotyping and measurement of tamoxifen and metabolites concentrations, respectively. Trough levels were obtained 12 h after the last intake of tamoxifen.

Metabolite measurements
Steady-state concentrations of tamoxifen and its metabolites (NDM-tamoxifen, 4-hydroxy-tamoxifen, and endoxifen) were measured in serum with high-performance liquid chromatography-tandem mass spectrometry (HPLC/MS/ MS). This assay was developed and validated at the laboratory of Clinical Pharmacy and Toxicology at the Leiden University Medical Center and was described in detail earlier [26]. Genotyping CYP2D6 genotyping CYP2D6 genotyping was performed using the Amplichip CYP450 test (Roche Diagnostics, Indianapolis, USA) to test the major CYP2D6 alleles in DNA isolated from blood. All CYP2D6 genotypes were translated to predicted phenotypes according to Schroth and colleagues [27]. The considered CYP2D6 predicted phenotypes are as follows: ultra-rapid (UM, duplication of active alleles), extensive (EM, two fully functional alleles), heterozygous extensive (hetEM, one normal active allele with a non-functional allele), intermediate (IM, one non-functional allele with one decreased activity allele or two alleles with decreased activity), and poor metabolizers (PM, two non-functional alleles).

CYP3A combined genotypes
In order to investigate the combined effect of CYP3A4*22 and CYP3A5*3, genotype clusters were formed as follows:

Statistical analysis
Metabolic ratios were determined as concentration of substrate divided by concentration of metabolite. To analyze differences between metabolic ratios, a two-sided Student's t test was used. To compare the concentrations of tamoxifen and its metabolites among CYP3A clusters, one-way ANOVA tests were used. For comparisons between tamoxifen and metabolite concentrations by CYP3A4 and CYP3A5 groups, a twosided Student's t test was performed. A multiple linear regression analysis was carried out to analyze the contributions of CYP3A4*22, CYP3A5*3, and the CYP3A combined genotypes to explain the total variability of the (log-transformed) metabolic ratios and concentrations of tamoxifen and its metabolites among treated patients. Statistical analyses were carried out with IBM SPSS for Windows, version 23.0. In analyses, test with p values <0.05 was considered to be statistically significant.

Patient characteristics
A total of 667 female patients were enrolled in the CYPTAM study from February 2008 till December 2010 from 19 participating hospitals in The Netherlands and 6 hospitals in Belgium. The mean age of included patients was 56.4 years and in 79.5% were progesterone receptor-positive tumors.  Table 2. Genotype distributions were in Hardy-Weinberg equilibrium and no linkage disequilibrium was observed between the CYP3A4*22 single nucleotide polymorphism (SNP) and the CYP3A5*3 allele (LD < 0.1).

CYP3A5 genotypes
Frequencies and distribution in the study population are listed in Table 2. The most frequent genotype was CYP3A5*3/*3, followed by CYP3A5*1/*3 and CYP3A5*  Association of tamoxifen and its metabolites to CYP3A4 genotype, CYP3A5 genotype, and CYP3A4/5 combined genotypes A substantial variation in the metabolic ratios of tamoxifen and its metabolites between individuals was observed. An overview of the mean and standard deviations (SD) of tamoxifen and its metabolite metabolic ratios by CYP3A4, CYP3A5 genotypes and CYP3A clusters is presented in Table 3. The metabolic ratio tamoxifen/NDM-tamoxifen was statistically different (p < 0.001) between CYP3A4*22/*22 and CYP3A4*1/*22 or CYP3A4*1/*1 individuals, whereas other metabolic ratios (tamoxifen/4-hydroxy-tamoxifen, 4hydroxy-tamoxifen/endoxifen, and NDM-tamoxifen) did not show any difference. The metabolic ratios of tamoxifen did not show any difference between CYP3A5*1/*3 or CYP3A5*1/*1 and CYP3A5*3/*3 individuals (p > 0.05). Figure 2 shows the comparisons of tamoxifen and its metabolite metabolic ratios stratified by the CYP3A4 and CYP3A5 genotypes.
In a third linear regression, the combined role of CYP2D6 and CYP3A clusters (C1, C2, C3, and C4) together was tested. Still, no significant improvements in the explained variability (R 2 ) were observed. A summary of CYP3A4, CYP3A5, and CYP3A covariate analysis is presented in Table 4.

Discussion
In the present study, the contribution of CYP3A4*22, CYP3A5*3, and combined genotypes to the metabolism of tamoxifen and the formation of the active metabolite endoxifen was investigated. Our data show that CYP3A4*22 genotype slightly contributes to explaining the pharmacokinetic variability between patients receiving tamoxifen, but the effect is small. CYP3A5*3 genotype and CYP3A4/5 combined genotypes do not significantly help to improve the explained variability in tamoxifen metabolism. The explained variability (R 2 ) of (log-transformed) endoxifen concentrations due to CYP2D6 predicted phenotypes was 42.3%, while 57.0% of the variability in metabolic ratio NDM-tamoxifen/endoxifen was explained by CYP2D6. Previously, Mürdter and colleagues reported a high, 68.7% variance in metabolic ratio of NDM-tamoxifen/endoxifen due to genetic variations in CYP2D6 genotype [10]. In our study, we observed a lower variance of metabolic ratio of NDM-tamoxifen/endoxifen (57.0%); however, the data demonstrate that CYP2D6 genotype alone only partially explains the variability between patients using tamoxifen.
In this study, when the CYP3A4*22 genotype was taken into account, in addition to the CYP2D6 genotype, the explained variability (R 2 ) of (log-transformed) endoxifen concentrations slightly improved from 42.3 to 42.8% (p < 0.001), whereas the explained variability of the metabolic ratio NDM-tamoxifen/endoxifen did not significantly increase (from 57.0 to 57.4%, p = 0.375). Interestingly, the explained variability of (log-transformed) metabolic ratio tamoxifen/ NDM-tamoxifen was found to be slightly increased if the CYP3A4*22 genotype was added to the analysis (improvement from 21.8 to 23.9%, p < 0.001). A higher metabolic ratio tamoxifen/NDM-tamoxifen was also noted in CYP3A4*22 carriers (0.59 vs. 0.52, p < 0.001). At the same time, our data showed that CYP3A4*22 carriers have a statistically significant higher mean concentration of tamoxifen, 4-hydroxy-tamoxifen, and NDM-tamoxifen (p < 0.05), while a trend toward higher endoxifen concentrations was observed (p = 0.088).
Our results are in line with the previous conclusions by Teft et al. [20] and Antunes and colleagues [21]. In both studies, higher mean concentrations of tamoxifen and its metabolites were unexpectedly measured in CYP3A4*22 carriers. At first glance, a decreased CYP3A4 activity may lead to a diminished transformation of tamoxifen into its active metabolites, and consequently, lower concentrations could be expected. On the contrary, higher concentrations of tamoxifen and its metabolites were found.
A potential explanation for these findings could be due to decreased CYP3A4 activity and a larger intestinal and hepatic bioavailability of tamoxifen in the CYP3A4*22 individuals [20]. According to Teft and colleagues [20], CYP3A4*22 carriers would have a reduced intestinal CYP3A4 activity and higher tamoxifen bioavailability, which would result in higher levels of unmetabolized tamoxifen. At the same time, a diminished CYP3A4 action at hepatic level would mean a diminished hepatic first-pass metabolism of tamoxifen, which would be translated in higher remaining concentrations of tamoxifen available for further transformations into 4hydroxy-tamoxifen and NDM-tamoxifen. Moreover, Antunes et al. suggested that the reduced tamoxifen metabolism resulting from CYP3A4*22 is probably compensated by other enzymes, whereas the transformation from tamoxifen into 4-hydroxy-tamoxifen would be more relevant in CYP3A4*22 carriers when CYP2D6 activity is decreased [21]. Although this hypothesis appears plausible, we did not observe any significant difference in metabolic ratios tamoxifen/4-hydroxy-tamoxifen and 4-hydroxy-tamoxifen/ endoxifen between CYP3A4*22 and CYP3A4*1 carriers after adjustment for CYP2D6 activity.
In the present study, the CYP3A5*3 genotype does not significantly contribute to explaining the inter-variability among patients treated with tamoxifen. Only CYP3A5*3 marginally improved the explained variance of the (logtransformed) metabolic ratio 4-hydroxy-tamoxifen/endoxifen (from 44.9 to 46.2%, p < 0.038). However, we did not find any statistically significant differences in mean concentrations of tamoxifen and its metabolites, nor in the mean metabolic ratios between CYP3A5*3 and CYP3A5*1 individuals. Jin and colleagues found that CYP3A5*3 carriers treated with tamoxifen reached higher endoxifen concentrations than CYP3A5*1 individuals [23]. Our results, however, are in line with the results of Tucker et al., who did not see significant variations in tamoxifen and its metabolite concentrations among CYP3A5*3 and CYP3A5*1 carriers [30]. In a clinical context, several conflicting results have been published, showing disparate findings. According to Wegman and colleagues, CYP3A5*3 homozygous carriers tend to have an increased risk of recurrence, albeit not statistically significant [31].
In the same way, our findings suggested that CYP3A combined genotypes do not significantly contribute to explaining the variability between individuals treated with tamoxifen, with the exception of the (log-transformed) metabolic ratio tamoxifen/NDM-tamoxifen (p < 0.001). The slow metabolizer C1 group, consisting of CYP3A4*22 carriers and the non-functional CYP3A5*3 allele, showed higher metabolic ratios of tamoxifen/NDM-tamoxifen compared to the other groups (C2, C3, and C4). These results might be clarified by the previously described difference in metabolic ratio in the CYP3A4*22 individuals and therefore in the CYP3A combined genotypes.
A potential limitation of our analysis might be due to the use of CYP3A4/5 inhibitors during the study, as CYP3A4/5 activity can be influenced. Unfortunately, information about concomitant medicines was not systematically evaluated and consequently available data were too sparse for analysis.
In conclusion, our data demonstrated that CYP3A genotype slightly contributes to explaining the variability between patients in tamoxifen metabolism; however, the effect is small, and therefore, it is unlikely to have any significant clinical relevance for the efficacy of tamoxifen.
Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.