The effects of tamoxifen and its metabolites on circulating estrogen metabolites among pre- and postmenopausal women.

The study population for this analysis was drawn from the previously described Ultrasound Study of Tamoxifen [ 41 , 42 ]. In brief, participants included women aged 30-70 years who were at high risk or diagnosed with breast cancer and referred to tamoxifen therapy for clinical indications (tamoxifen-treated), and untreated women who screened negative for breast cancer. Participants were recruited from the Barbara Ann Karmanos Cancer Institute (KCI) and Henry Ford Health Systems (HFHS) in Detroit, Michigan. The analytical populations for the analyses presented herein comprised of tamoxifen-treated women (n=82), either diagnosed with (n=65), or at high risk of (n=17), breast cancer and are summarized in Fig. 1 . Women were excluded if they never started tamoxifen or had missing serum samples at the 12-month visit (n=13). Given that EM concentrations fluctuate over the menstrual cycle, and menstrual cycle phase at the time of blood draw was incomplete for premenopausal women (n=46), longitudinal analyses were restricted to women who were postmenopausal at baseline (n=23). Four women underwent menopausal transition during the 12-month study period, and these women were included in the postmenopausal cohort at 12-month for the cross-sectional analysis but were not part of the longitudinal cohort. For the cross-sectional analyses, women who discontinued tamoxifen and had no detectable tamoxifen metabolites at 12 months (n=9) were excluded ( Fig.1 ). After applying the above exclusions, final analytic samples included (1) postmenopausal women at baseline (n=23) for the longitudinal analysis, and (2) both premenopausal (n=33) and postmenopausal women (n=27) at 12 months following tamoxifen initiation for the cross-sectional analysis ( Fig.1 ). Informed consent was obtained from all participants and study procedures for the Ultrasound Study of Tamoxifen were approved by the Institutional Review Boards (IRB) of KCI, HFHS, and the National Cancer Institute (NCI). Additionally, the present analysis received an IRB exemption from the University of Maryland IRB given the use of previously collected de-identified biological specimens and data. Serial serum samples collected prior to and approximately 12 months after tamoxifen initiation ( Supplementary Fig. S1 ) were stored at −80°C and sent to the Frederick National Laboratory for Cancer Research (Frederick, MD) to measure EM. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) run on two TSQ Quantiva LC-MS systems was used to measure a comprehensive panel of 15 EMs which included the parent estrogens, estrone (E1) and estradiol (E2), and metabolites in the i) 2-hydroxylation (2-OH) pathway: 2-hydroxyestrone (2-OHE1), 2-hydroxyestradiol (2-OHE2), 2-hydroxyestrone-3-methyl ether (3-ME1), 2-methoxyestrone (2-ME1), and 2-methoxyestradiol (2-ME2); ii) 4-hydroxylation (4-OH) pathway: 4-hydroxyestrone (4-OHE1), 4-methoxyestrone (4-ME1), and 4-methoxyestradiol (4-ME2), and iii) 16-hydroxylation (16-OH) pathway: 16α-hydroxyestrone (16α-OH-E1), estriol (E3), 16-ketoestradiol (16-keto-E2), 16-epiestriol (16-epiE3), and 17-epiestriol (17-epiE3). Details of the assay have been published previously [ 43 ]. All serum samples were analyzed following enzymatic hydrolysis with β-glucuronidase and sulfatase to quantify total estrogen metabolites (i.e., conjugated and unconjugated forms). Blinded duplicate quality control samples were included within and across batches; all samples were randomized across batches. The coefficients of variation (CV, within and between batches) were <3% for all the estrogen metabolites. The limit of detection of each estrogen metabolite ranged from 0.33-0.37 pmol/L [ 43 , 44 ]; no samples had undetectable levels of EM. Concentrations of tamoxifen and tamoxifen metabolites, including 4-hydroxy-N-desmethyl-tamoxifen (endoxifen), 4-hydroxy-tamoxifen (4-OHT), and N-desmethyltamoxifen (NDT), collected at 12 months following tamoxifen initiation, were measured in serum using LC-MS/MS by the Illinois Institute of Technology Research Institute (IITRI). The limit of quantitation of the tamoxifen metabolites was as follows: tamoxifen (5 ng/ml), NDT (500 ng/ml), 4-OHT (0.5 ng/ml) and endoxifen (50 ng/ml). For 4-OHT, seven samples had levels below the lower limit of quantitation of 0.5 ng/ml. For these, the concentration was replaced by half the limit of quantitation of 4-OHT. The CVs were <5% for each of the tamoxifen metabolites, except for 4-OHT (CV=20.7%). CYP2D6 was sequenced because tamoxifen is metabolized by the CYP2D6 enzyme, and patients with low enzyme activity may have less therapeutic efficacy from tamoxifen. Saliva samples were collected from participants at baseline using Oragene DISCOVER collection vials (DNAGenotek, Ontario, Canada). DNA extraction, genotyping and CYP2D6 sequencing and phenotype assignment were conducted at the NCI Cancer Genomics Research Laboratory as previously described [ 45 ]. CYP2D6 phenotypes were classified as poor (activity score: 0), intermediate (activity score: >0–1.0), efficient (activity score: >1.0–2.0), or ultra-metabolizers (activity score: >2.0) [ 46 ]. Information on participant characteristics, including demographics, menopausal, and reproductive history was collected via a questionnaire at baseline before tamoxifen initiation and follow-up at 12 months after tamoxifen initiation. Characteristics included self-reported endocrine symptoms (hot flashes and joint pain) in the past two weeks prior to the visit (yes/no), alcohol intake in the past 24 hours (yes/no), and current smoking status (yes/no) based on current smoking status and the date when last smoked. Height, measured at baseline, and weight, measured at each visit, were used to compute body mass index (BMI, kg/m 2 ). Women were considered postmenopausal if periods stopped for 12 months or longer due to natural causes, or surgical procedures, such as bilateral oophorectomy. Women aged 55 years or older who had a hysterectomy but not oophorectomy, and whose periods stopped due to endometriosis, uterine ablation, irregular perimenopausal cycles, Novasure endometrial ablation, medication/drug therapy, or any other procedure or therapy were classified as postmenopausal. After 12 months following tamoxifen therapy, if in addition to the above, periods stopped for 12 months or longer due to birth control shots or tamoxifen use, women aged ≥55 years were classified as postmenopausal. Four women changed menopausal status (premenopausal to postmenopausal) from the baseline to the 12-month follow-up visit. Distributions of participant characteristics by menopausal status were assessed by estimating frequencies and percentages (categorical variables) as well as means and standard deviations (continuous variables). Among postmenopausal women, paired t-tests (continuous variables) and McNemar’s tests (dichotomous variables) were used to evaluate differences in select characteristics between baseline and the 12-month follow-up visit. Parent estrogens and their metabolites were natural-log transformed to account for non-normal right-skewed data. EMs were analyzed individually, and as the sum of all 15 EMs (total EM), parent estrogens (E1 and E2), and based on metabolic pathways (2-OH, 4-OH, and 16-OH) [ 24 – 27 ]. The metabolic pathways were created by summing the individual metabolites within respective pathways. Paired t-tests were used to assess the mean change in circulating levels of EM from baseline to 12 months after tamoxifen initiation. Percent change in circulating EMs was estimated by the ratio of geometric mean (GM) difference between baseline and 12-month EM levels to baseline EM levels, multiplied by 100. GMs and 95% confidence intervals (CIs) were estimated for the metabolites of estrogen and tamoxifen at 12 months following tamoxifen initiation, by menopausal status. Spearman rank correlation coefficients were used to evaluate the association between EMs, and tamoxifen metabolites, by menopausal status. In cross-sectional analyses, multivariable linear regression was used to estimate regression coefficients (“beta coefficients”, presented in 100 ng/ml) and 95% CIs for the association between metabolites of tamoxifen and estrogen, by menopausal status. Age (quadratic), BMI (normal/overweight/obese) at the 12-month follow-up visit, and baseline parent estrogen levels (continuous) were identified as potential confounders in the relationship between estrogen and tamoxifen metabolites at 12-months following tamoxifen initiation and were included in the final models. Statistical significance was assessed at α=0.05. As these were exploratory analyses, multiple testing corrections were not applied. All analyses were conducted using SAS 9.4 (Cary, North Carolina).

The demographic and reproductive characteristics of the analytic populations are presented in Tables 1a and 1b . At baseline, postmenopausal women (n=23), 17 of whom identified as African American, seven as White and one as other race/ethnicity, had an overall mean (SD) age of 60.7 (6.2) years and BMI of 31.1 (6.9) kg/m 2 ( Table 1a ). At 12 months following tamoxifen initiation premenopausal women (n=33) had a mean (SD) age of 47.3 (5.9) years and a BMI of 29.9 (6.6) kg/m 2 , while the average (SD) age and BMI of postmenopausal women (n=27) were 60.5 (6.6) years and 31.2 (7.6) kg/m 2 , respectively ( Table 1b ). Among postmenopausal women (n=23), the highest concentrations of EM (pmol/l) prior to tamoxifen treatment were observed for circulating estradiol (GM: 190.4, 95% CI: 146.2, 248.0) followed by 2-OHE1 (GM: 95.0, 95% CI: 78.1, 115.7) and E3 (GM: 72.3, 95% CI: 60.2, 86.8). In the 12 months following tamoxifen initiation, serum levels of EM decreased as compared to measures pre-tamoxifen initiation (baseline), however, statistically significant decreases were primarily observed in the select metabolites of the 2-OH and 16-OH pathway EMs. Among postmenopausal women, total EM decreased by 13.8% (p=0.02), a decline that may be largely driven by a 15.3% reduction in the 2-OH pathway (p=0.03) and a 17.2% reduction in 16-OH pathway (p=0.003). Of the individual EMs, statistically significant decreases were observed in the 2-OH pathway metabolites 2-ME1 (20.2%, p=0.001) and 2-ME2 (18.9%, p=0.02) and the 16-OH pathway metabolite E3 (27.0%, p=0.001) ( Table 2 ). Geometric means of the circulating EMs (pmol/l) and tamoxifen metabolites (ng/ml) at 12 months following tamoxifen initiation are presented in Supplementary Table S1 . Among premenopausal women, following parent estrogens, the highest concentrations observed for EM were among 2-OHE1 (GM: 142.7, 95% CI: 101.3, 201.2) and E3 (GM: 120.0, 95% CI: 81.4, 176.8). This was similar among postmenopausal women with the highest EM concentrations observed for 2-OHE1 (GM: 81.0, 95% CI: 67.7, 96.9), and E3 (GM: 57.9, 95% CI: 47.2, 71.0). The concentration of tamoxifen and its metabolites among both pre- and postmenopausal women were highest for circulating NDT followed by serum tamoxifen levels (Table 3). Among premenopausal women, strong positive correlations were observed between parent estrogens and select metabolites within the 2-hydroxylation and 16-hydroxylation pathways, with the strongest between E1 and E2 (r s =0.95) ( Supplementary Table S2 ). Among postmenopausal women, however, fewer and more modest correlations were observed ( Supplementary Table S3 ). Of the tamoxifen metabolites, the strongest correlation was observed between tamoxifen and NDT in both premenopausal (r ds. =0.80) and postmenopausal women (r s =0.71) ( Supplementary Table S4 ). Serum concentrations of the parent drug, tamoxifen, were positively associated with levels of total EM (β=1.02, 95% CI: 0.32, 1.72), parent estrogens (β=1.29, 95% CI: 0.45, 2.12), E1 (β=1.24, 95% CI: 0.44, 2.04), and E2 (β=1.39, 95% CI: 0.34, 2.30) ( Fig. 2 , Supplementary Table S5 ). Significant positive associations were also observed between circulating endoxifen with levels of total EM (β=8.96, 95% CI: 2.46, 15.5), parent estrogens (β=10.6, 95% CI: 2.76, 18.5), E1 (β=10.1, 95% CI: 2.53, 17.8), and E2 (β=12.4, 95% CI: 3.17, 21.8) ( Fig. 3 , Supplementary Table S5 ). The tamoxifen metabolite, 4-OHT was positively associated with levels of total EM (β=37.1, 95% CI: 1.80, 72.4) and E1 (β=43.2, 95% CI: 2.18, 84.2) ( Supplementary Table S5 ). Of the 2-OH pathway metabolites, tamoxifen was positively associated with levels of 2-OHE1 (β=0.72, 95% CI: 0.02, 1.41) and 2-ME1 (β=1.14, 95% CI: 0.30, 1.99), while endoxifen was positively associated with levels of 2-OHE1 (β=6.57, 95% CI: 0.21, 12.9), 2-OHE2 (β=5.82, 95% CI: 0.90, 10.7) and 2-ME1 (β=8.58, 95% CI: 0.52, 16.6) ( Fig. 2 and Fig. 3 , Supplementary Table S6a ). 4-OHT, a tamoxifen metabolite, was positively associated with levels of 2-OHE1 (β=37.1, 95% CI: 4.92, 69.3), 2-ME1 (β=42.5, 95% CI: 0.61, 84.4) and the 2-OH pathway (β=32.8, 95% CI: 2.04, 64.8) ( Supplementary Table S6a ). Among the 4-OH pathway metabolites, a statistically significant association was only observed between the tamoxifen metabolite, 4-OHT and levels of 4-ME1 (β=27.9, 95% CI: 6.05, 49.8) ( Supplementary Table S7 ). However, tamoxifen was statistically significantly associated with the 16-OH pathway (16-OH pathway (β=0.89, 95% CI: 0.26, 1.52) and individual metabolites ( Fig. 2 , Supplementary Table S8a ) . Specifically, tamoxifen was positively associated with levels of 16α-OH-E1 (β=1.10, 95% CI: 0.40, 1.79), E3 (β=0.89, 95% CI: 0.03, 1.72), 16-keto-E2 (β=0.85, 95% CI: 0.16, 1.54), 16-epiE3 (β=0.78, 95% CI: 0.05, 1.51), and 17-epiE3 (β=0.74, 95% CI: 0.03, 1.45) ( Fig. 2 , Supplementary Table S8a ). Positive associations were also observed between endoxifen and levels of 16α-OH-E1 (β=7.96, 95% CI: 1.15, 14.8), E3 (β=12.6, 95% CI: 5.87, 19.3), 16-epiE3 (β=10.3, 95% CI: 4.30, 16.3), and the 16-OH pathway (β=9.0, 95% CI: 3.40, 14.6) ( Fig. 3 , Supplementary Table S8a ). The metabolite 4-OHT was positively associated with levels of E3 (β=50.8, 95% CI: 12.5, 89.0) and the 16-OH pathway (β=33.8, 95% CI: 2.20, 65.3) ( Supplementary Table S8a ). Finally, the tamoxifen metabolite, NDT was positively associated with levels of 16α-OH-E1 (β=0.56, 95% CI: 0.01, 1.11) only ( Supplementary Table S8a ). Among postmenopausal women, a positive association was observed between NDT and levels of 3ME1 (β=0.18, 95% CI: 0.03, 0.34) ( Supplementary Table S6b ). No other cross-sectional associations between the metabolites of tamoxifen and estrogen were detected among postmenopausal women. In our study sample, we examined the distributions of CYP2D6 metabolizer status at baseline and at 12 months post-tamoxifen treatment ( Supplemental Table S9 ). The majority were classified as either intermediate or efficient CYP2D6 metabolizers with a very low proportion classified as low metabolizers (n=3 premenopausal, n=1 postmenopausal).

In this exploratory analysis of the effect of tamoxifen therapy on circulating EMs, a decrease in levels of select metabolites of the 2-OH and 16-OH EM pathways was observed among postmenopausal women within the first year of treatment. Furthermore, among premenopausal women, circulating tamoxifen and its active metabolites, endoxifen and 4-OHT, were positively associated with parent estrogens, select metabolites of the 2-OH pathway (2-OHE1, 2-OHE2, 2-ME1), and 16-OH pathway metabolites. No such pattern was noted among postmenopausal women. Of the 4-OH pathway EMs, a positive association was observed only between 4-OHT and 4-ME1 among premenopausal women. The observed menopausal status-specific effect of tamoxifen, specifically on the 2-and 16-OH pathway EMs, markers of breast cancer risk, warrants further investigation. The effectiveness of tamoxifen in reducing breast cancer incidence and progression is attributed to its role in estrogen signaling by competing with endogenous estrogens for ER [ 5 , 6 ]; however, tamoxifen’s effect on circulating estrogens and estrogen metabolites remains unclear. In this study, we explored whether tamoxifen and/or its metabolites are associated with changes in estrogen metabolism, given the hypothesized role of these pathways in breast carcinogenesis. Limited studies have focused mainly on associations between tamoxifen use and circulating parent estrogens, estradiol and estrone, with conflicting associations reported [ 36 – 39 ]. It should also be noted that these prior studies used radioimmunoassay to assess EMs, which is traditionally characterized by limited accuracy and precision, especially when quantifying low metabolites concentrations [ 47 ]. In our analyses, we utilized the highly reliable and sensitive LC-MS/MS assay to assess serum EM which is especially important because of low estrogen concentrations in postmenopausal women [ 43 , 47 , 48 ]. Although the observed reduction in estradiol or estrone levels 12 months after tamoxifen initiation was not statistically significant in our study, the effect of tamoxifen in lowering circulating estrogen levels is consistent with findings from two prior longitudinal studies [ 37 , 39 ]. Among postmenopausal tamoxifen-treated breast cancer patients, Levin et al. and Lønning et al. observed reductions in circulating parent estrogen levels ranging from 6-23% [ 37 , 39 ] after 12 months of tamoxifen therapy. In contrast, Lum et al. reported increases in mean serum estradiol and estrone levels among postmenopausal breast cancer patients after two years of tamoxifen treatment [ 36 ]. Recent studies suggest that endogenous estrogen metabolism profiles, relative to concentrations of parent estrogens, may differentially influence breast cancer risk [ 24 – 29 ]. Estrogen metabolites are hypothesized to affect breast carcinogenesis either via estrogen-receptor cell signaling or genotoxic effects, depending on specific metabolites in 2-, 4-, or 16-OH pathways [ 11 , 12 ]. The estrogenic and anti-estrogenic properties of the 2-OH [ 20 , 21 ] and 16-OH pathway metabolites [ 22 , 23 ], respectively, translate to inhibitory and stimulatory effects on breast tumor proliferation. Prior to our analysis, the influence of tamoxifen therapy on estrogen metabolism in circulation, particularly across all three EM pathways, was largely unknown. We observed a 13-27% decline in the levels of select circulating EMs in postmenopausal women after 12 months of tamoxifen therapy, with statistically significant reductions observed for the 2-OH pathway and the individual metabolites, 2-ME1 and 2-ME2 as well as the 16-OH pathway and the metabolite, E3. Given the exploratory nature of our study, these findings require further investigation in future studies with larger study populations. Data supports mitogenic effects of 16-OH pathway metabolites on breast tumor proliferation [ 17 ], with E3 demonstrating the strongest proliferative effect on breast cancer cells compared with other metabolites [ 49 ], while the 2-OH metabolite, 2-ME2, shows the most potent inhibitory effect. Additionally, while elevations of the 2-OH pathway metabolites have been associated with reduced breast cancer risk [ 26 – 29 ], it remains unclear whether tamoxifen’s ability to reduce this pathway contributes to its effectiveness in lowering breast cancer risk. Further examination of tamoxifen’s effects on estrogen metabolism in large-scale longitudinal studies is needed to understand the possible roles of these different metabolic pathways in influencing the prevention and management of breast cancer. Findings from our cross-sectional analyses among pre- and postmenopausal women support menopausal status-specific associations between circulating metabolites of tamoxifen and parent estrogens. Contrary to one prior study that observed significant but weak positive correlations (r s =0.29) [ 50 ] between tamoxifen metabolites and parent estrogens in serum among postmenopausal women, we did not observe an association between TAM and parent estrogens among postmenopausal women, potentially due to smaller sample size. However, among premenopausal women, for whom tamoxifen is the primary adjuvant endocrine therapy [ 51 ], circulating tamoxifen and its active metabolites, endoxifen and 4-OHT were positively associated with parent estrogens. Of the tamoxifen metabolites, both endoxifen and 4-OHT have approximately a 30-100-fold binding affinity towards estrogen receptors, when compared to tamoxifen [ 30 , 34 , 35 , 40 ]. This is consistent with our findings among premenopausal women, where stronger associations with EMs were noted with endoxifen and 4-OHT compared to tamoxifen. Though endoxifen has equally potent antiestrogenic activity as 4-OHT, the plasma concentration of endoxifen is 5-10-fold higher than that of the 4-OHT, making it the more clinically relevant tamoxifen metabolite [ 52 – 56 ]. While the sample size of our study was modest, post hoc power analyses indicated ≥80% power to detect longitudinal changes in EM concentrations ranging from 0.31 to 1.22 pmol/l. Observed absolute decreases in EMs 12-months after tamoxifen initiation among postmenopausal women, including total EM, the 2-OH pathway and 16-OH pathway, 2-ME1, 2-ME2, and E3 ranged from 2.8-86.0 pmol/l, exceeding the detectable range and supporting our study’s ability to detect meaningful longitudinal changes. For cross-sectional analyses, statistical power to detect associations between tamoxifen metabolite concentrations and EMs (per 1 ng/ml increase) ranged from 67%-98% in premenopausal women and 55%-91% in postmenopausal women. These estimates support the interpretation of our findings, suggesting that the observed associations in premenopausal women were detected with moderate to high power, whereas the null associations observed among postmenopausal women, particularly for parent estrogens, may reflect insufficient statistical power rather than a true absence of association. The increased potency of the bioactive tamoxifen metabolites suggests stronger antiestrogenic activity which is of importance for adjuvant and chemopreventative therapy for ER dependent breast cancer [ 33 ]. Tamoxifen is a primary treatment among premenopausal women and is also clinically indicated among postmenopausal breast cancer patients [ 57 ]. Yet, little is known about the relationship between the metabolites of tamoxifen and estrogen in circulation among pre- and postmenopausal women. In our cross-sectional analyses of premenopausal women, tamoxifen and its active metabolites endoxifen and 4-OHT were positively associated with 2-OH and 16-OH pathway metabolites. It is worth noting that the association between 4-OHT and EMs across all pathways were accompanied by wide confidence intervals and these relationships should be interpreted with caution. However, among postmenopausal women, only circulating NDT was positively associated with the 2-OH pathway metabolite, 3-ME1; no other associations were observed. Further, postmenopausal women had higher concentrations of tamoxifen metabolites compared to premenopausal women. Prior evidence suggests that hormonal status may be an important predictor of pharmacokinetics of tamoxifen and its metabolites, where levels of metabolites may vary based on menopausal status [ 58 ]. Both tamoxifen and estrogen metabolism are mediated via cytochrome CYP enzymes [ 13 , 14 , 30 – 32 , 59 ]. Differences in CYP enzymatic activity among breast cancer patients have been shown to influence the therapeutic effect of tamoxifen on treatment outcomes [ 60 ]. This may, in part, explain the observed variability in the relationship between tamoxifen metabolites and EM across different pathways and subsequent variations in outcomes reported among breast cancer patients [ 61 ]. Due to the limited sample size and variation in metabolizer status in our study, further analyses could not be performed. However, the limited variability in CYP2D6 metabolizer status reduces the potential for observed findings to be due to large differences in enzymatic activity. This study has some limitations to be considered. Premenopausal women were excluded from longitudinal analyses due to incomplete menstrual cycle data. Given the considerable variation in estrogen concentrations across the menstrual cycle, the non-uniform timing of premenopausal blood draw, and the limited sample size, their inclusion could have contributed to substantial variability in EM levels. The timing of daily tamoxifen intake among women in the Ultrasound Study of Tamoxifen was not known and thus, differences in tamoxifen intake may contribute, in part, to variability in estrogen metabolite levels. However, the half-life of tamoxifen is 5-7 days and reaches its highest concentration in serum approximately five hours after intake [ 33 , 62 ]. As patients are advised to take tamoxifen at the same time each day [ 63 ], the within-individual variability in the EM concentrations after tamoxifen initiation is expected to be minimal. Additionally, the specific tamoxifen dose prescribed to each participant was not collected, and it is unknown whether all women in the analytic population received the standard dose. While this may have contributed to between-subject variability in the cross-sectional analyses, it is unlikely to have affected the longitudinal results, which assessed within-person changes over time. The exclusion of women who discontinued tamoxifen and had no detectable tamoxifen metabolites at 12 months may have introduced bias by excluding those who experienced treatment side effects or poor adherence. Although estrogen metabolites are biologically correlated, variance inflation factors (VIFs) for all variables across models were below three, indicating that multicollinearity was not a major concern. However, given the small sample size and the exploratory nature of the study, the findings should be further examined in larger studies. Although power analyses indicated sufficient ability to detect several key longitudinal changes, certain metabolites, specifically, 3-ME1, 4-ME1, and 4-ME2 exhibited absolute changes below the detectable threshold, and null results for these should be interpreted with caution. Similarly, limited statistical power in cross-sectional analyses among postmenopausal women may have contributed to the lack of observed associations for some metabolites. Additionally, data on time since menopause among postmenopausal women was not available. Finally, the levels of EM measured in serum may not represent the absolute levels, as some of these metabolites are excreted via urine. Despite these limitations, this study has several notable strengths. The use of the highly sensitive and specific LC-MS/MS assay, a powerful analytical technique, the changes in levels of a panel of 15 circulating EM, and their summary measures across three different pathways (2-OH, 4-OH, and 16-OH) were assessed in relation to tamoxifen use. The use of the LC-MS/MS enabled a comprehensive assessment of EM and detection of low levels of circulating estrogen, which is especially important for postmenopausal women [ 43 ]. Furthermore, as circulating tamoxifen metabolites were measured 12 months after treatment initiation, a significant strength of this study was the ability confirm adherence to tamoxifen treatment among women in the analytic population. Our analysis furthers knowledge on relationships between circulating metabolites of tamoxifen and estrogen, an understudied area of importance in breast cancer etiology. The longitudinal design allowed assessment of changes in circulating EM from baseline and 12 months following tamoxifen treatment, and this approach of measuring EM among the same postmenopausal women before and after 12 months of tamoxifen initiation, controlled for confounding arising from invariant personal factors, including age, race, BMI, reproductive factors, etc. Finally, while our study population includes over 50% African American women, future studies among larger, racially and ethnically diverse populations are needed. Our study highlights that the biomechanisms of tamoxifen and/or its metabolites on circulating estrogen metabolites may extend beyond estrogen signaling to estrogen metabolism. These preliminary findings warrant further investigation. Given the importance of estrogen metabolism pathways in the etiology of breast carcinogenesis, larger prospective racial and ethnically diverse studies with longer follow-up periods are needed to gain further insight on the relationship of tamoxifen and/or its metabolites with estrogen metabolism and assess differential effects by menopausal status.

Elevated levels of endogenous estrogens are strong risk factors for postmenopausal breast cancer[ 1 ], which act on target breast cells by binding to estrogen receptors (ERs) and forming an estrogen-ER complex that induces cell proliferation [ 2 ]. Tamoxifen, a selective estrogen receptor modulator (SERM), is one of the most widely used endocrine therapies for ER-positive breast cancer [ 3 , 4 ]. Tamoxifen’s mechanism of action involves the competitive inhibition of estrogen binding to ERs, thus blocking the formation of the estrogen-ER complex and inhibiting the proliferation of breast cells [ 5 , 6 ]. Studies have additionally suggested the role of tamoxifen in inducing apoptosis in breast cancer cells [ 7 – 9 ]. The hydroxylated derivatives of estrogens are hypothesized to play a role in ER-mediated tumor cell proliferation [ 2 , 10 – 12 ]. Estrone and estradiol are irreversibly hydroxylated by cytochrome P450 (CYP) enzymes at the C-2, C-4, or C-16 positions of the steroid ring resulting in metabolites in the 2-hydroxy (2-OH), 4-hydroxy (4-OH), and 16-hydroxy (16-OH) pathways [ 13 , 14 ]. These metabolites have differing affinities for ER binding [ 15 – 17 ]; estradiol binds equally to both ERα and ERβ, while estrone and 2-hydroxyestrone mainly bind to ERα, and 16α-hydroxyestradiol primarily binds to ERβ. Most methylated catechol estrogens show minimal binding affinity for both ER subtypes [ 15 , 16 ]. More recently, it has been hypothesized that catechol estrogens in the 2-OH and 4-OH pathways further oxidize to mutagenic quinone products leading to DNA damage and tumor formation [ 2 , 11 , 12 , 18 , 19 ]. The estrogenic and genotoxic activity of estrogen metabolites (EMs) are suggested to vary by the estrogen metabolism pathway [ 20 – 23 ]. Recent epidemiological studies have examined EM in relation to pre- [ 24 , 25 ] and postmenopausal [ 26 – 29 ] breast cancer risk, with higher levels of 2-OH pathway associated with reduced risk and 16-OH pathway metabolites associated with an increased risk of, primarily, postmenopausal breast cancer [ 24 – 29 ]. In particular, a pooled analysis that used data from four published cohort studies reported that higher levels of the 2-OH:16-OH ratio were associated with approximately a 30-40% reduction in postmenopausal breast cancer risk, after adjusting for parent estrogens and both parent and total estrogen [ 27 ]. Similar to estrogens, tamoxifen is primarily metabolized by the CYP enzymes into its active metabolites, 4-hydroxy-N-desmethyl-tamoxifen (endoxifen), 4-hydroxy-tamoxifen (4-OHT), and N-desmethyltamoxifen (NDT) [ 30 – 33 ]. Compared with tamoxifen, the hydroxylated derivatives have a far greater binding affinity towards the ER [ 33 – 35 ], suggesting that tamoxifen metabolites and genetic differences in metabolism may also play an important role in breast cancer prevention and management. Despite the plausibility of an association between tamoxifen and circulating estrogens [ 36 , 37 ], there is limited epidemiological evidence that tamoxifen therapy modulates endogenous estrogen and estrogen metabolite levels, with inconsistent findings reported by menopausal status [ 36 – 39 ]. In longitudinal analyses, tamoxifen therapy has been associated with an increase in circulating estradiol among premenopausal women [ 36 , 38 ], while conflicting findings have been reported among postmenopausal women [ 36 , 37 , 39 ]. Moreover, little is known about the effect of tamoxifen therapy on estrogen metabolite concentrations. Despite the importance of tamoxifen metabolites in estrogen signaling [ 34 , 35 , 40 ] and the commonality of the CYP enzymes that mediate the metabolism of both estrogens and tamoxifen to their respective metabolites [ 13 , 14 , 30 – 32 ], the relationship between circulating tamoxifen and estrogen metabolites remains unclear, warranting further investigation. This analysis aimed to explore relationships between tamoxifen, its metabolites and estrogen metabolism, as measured by a comprehensive panel of 15 estrogens and EMs. Specifically, we conducted the following exploratory analyses: (1) a longitudinal analysis to assess within-person changes in circulating levels of parent estrogens and EMs among postmenopausal women from baseline to 12 months after tamoxifen initiation, and (2) a cross-sectional analysis to evaluate associations between circulating tamoxifen and estrogen metabolites among pre- and postmenopausal women at the 12-month follow-up.