Pharmacogenetics and Pharmacokinetics of Tamoxifen in a Zimbabwean breast cancer cohort

Abstract

Tamoxifen is the most used hormonal therapy for estrogen receptor positive breast cancer. CYP2D6 is the main enzyme in the metabolic pathway of tamoxifen to endoxifen. Variations in endoxifen plasma concentrations are associated with CYP2D6 polymorphisms. This study aimed to determine the association between the CYP2D6 polymorphisms and endoxifen plasma concentrations in a cohort of Zimbabwean breast cancer patients (n = 40). TaqMan genotyping and copy number assays were done to determine CYP2D6 genotypes. Tamoxifen and metabolites were quantitated using LC-MS/MS. The population had high frequencies of the CYP2D6 reduced function alleles, *17 (15%) and *29 (18%). The median endoxifen concentration was 4.78 ng/ml and 55% of the patients, mostly intermediate metabolizers were below the endoxifen therapeutic threshold 5.97 ng/ml. The CYP2D6 phenotypes and activity scores were significantly associated with endoxifen plasma concentrations (p = 0.0151) and with endoxifen to N-desmethyl tamoxifen ratios (p = 0.0006).

1. Introduction

Breast cancer is one of the most common cancers in the world, accounting for about 25% of newly diagnosed cancers ^1,2. Mortality rates in sub-Saharan Africa remain high compared to high income countries despite lower incidence rates. Breast cancer remains the second most common cancer in Zimbabwean women after cervical cancer ³. About 60–75% of new breast cancer cases in the world are estrogen receptor (ER) positive ⁴. Tamoxifen is the commonly prescribed anti-estrogenic drug for ER positive cancer and is the drug of choice for premenopausal women with ER positive breast cancer ^5–7. Tamoxifen is a selective estrogen receptor modulator (SERM) which may reduce the mortality rate by approximately 30% ⁸ and lower the risk of recurrence by almost a half ⁹.

Tamoxifen is a pro-drug, and it is converted to its primary metabolites, N-desmethyl-tamoxifen (NDM-tamoxifen) and 4-hydroxy-tamoxifen. Secondary metabolism converts the primary metabolites into endoxifen, which is the most active metabolite ⁴. Reaching the therapeutic levels of endoxifen is critical in the efficacy of tamoxifen ¹⁰. Low serum concentrations of endoxifen have been associated with increased risk of breast cancer recurrence⁹. There is however variability in plasma concentrations of endoxifen in different breast cancer patients receiving the standard dose of tamoxifen, resulting in differences in the clinical outcome. Up to 50% of the patients do not respond to tamoxifen due to low endoxifen concentrations which is because of genetics and other factors ^11,12.

Cytochrome P450 2D6 (CYP2D6) is responsible for metabolism of approximately 25% of all prescribed drugs ¹³. Tamoxifen is primarily metabolized through N-demethylation to NDM-tamoxifen which accounts for 92% of tamoxifen metabolism ^14–16. NDM-tamoxifen if further metabolized to endoxifen, hence CYP2D6 is considered a critical enzyme in tamoxifen metabolism to this active metabolite ¹⁷. CYP2D6 gene is a highly polymorphic gene, which carries over 100 genetic variants, some of which may result in a CYP2D6 enzyme which is non-functional or has reduced function ⁹. These variants may result in variability in endoxifen plasma concentrations and may impact the success of tamoxifen therapy. Classification of patients according to the CYP2D6 activity score has been proposed as a strategy for individualizing tamoxifen therapy ¹⁸. Association between lack or decreased enzyme activity and poor treatment outcome remains unclear, with CYP2D6 genotyping not being commonly adopted in clinical practice. The aim of this study was therefore to determine the effect of CYP2D6 polymorphisms on endoxifen levels in Zimbabwean breast cancer patients on tamoxifen therapy towards rationalizing the use of pharmacogenetics guided dosing in clinical practice.

2. Materials and Methods

2.1. Study population

We conducted an observational cross-sectional study to evaluate the effect of CYP2D6 polymorphisms on metabolism of tamoxifen in a Zimbabwean breast cancer cohort. The study was carried out in adult patients (≥18 years) being treated at Parirenyatwa Hospital in Harare. Only participants with a breast cancer histology confirmed by a pathologist were included in the study. Data was collected from questionnaire guided interviews and from medical records. Data collected included histology, stage, estrogen receptor status, duration on tamoxifen, comorbidities, comedications, surgery performed and HIV status. Sample size was calculated using the A.J. Dobson’s formula ¹⁹. A total of 40 breast cancer patients on tamoxifen therapy, were recruited into the study and followed for a period of five months. All participants gave their written informed consent.

2.2. Adherence and side effect prevalence in patients on tamoxifen therapy

Side effects:

Patients were asked if they experienced any of the side effects included on the questionnaire. A set of the most common tamoxifen side effects including hot flashes, vaginal discharge, amenorrhea, menstrual changes, oligomenorrhea, musculoskeletal pain, low libido and bone pain were included. Two dichotomous side effect variables with yes or no responses were considered for analysis.

Adherence:

The Medication Adherence Rating Scale (MARS) questionnaire ²⁰ was used. The questionnaire is an easy subjective measure of medication adherence and was used to assess the patients’ experience with tamoxifen therapy. The questionnaire results in four-item scores which are then summed up to define three adherence levels whereby high adherence = 0, medium = 1–2 score, and low adherence = 3–4 score. Adherence was also quantitatively determined by measuring the levels of tamoxifen where concentration less than 60 ng/mL tamoxifen have been associated with poor adherence to drug intake ^21,22.

2.3. CYP2D6 Genotyping

DNA was extracted from 200 μl of peripheral whole blood using the MagMAX^™ DNA Multi-Sample Ultra 2.0 Kit on the Thermofisher KingFisher^™ Flex Purification System with the MagMAX^™ Ultra 2.0–200 μl script for KingFisher Flex (Thermo Fisher Scientific, Marsiling Industrial Estate, Singapore). Extracted DNA was quantified with the Qubit 4 fluorometer using the Qubit dsDNA BR Assay Kit. Genotyping was conducted using the GenoPharm^® open array panel following manufacturers instruction. Briefly, a reaction mixture of 5 μl genomic DNA and 5 μl of TaqMan^™ Genotyping master mix (Cat. No. 4462164) was prepared per sample. The PCR mix was transferred to the GenoPharm^® custom open array panel using the automated Applied Biosystems^™ QuantStudio^™ 12K Flex OpenArray^™ AccuFill^™ System according to the manufacturer’s instructions.

A no template control (reaction mixture with all reagents but no template DNA) was included in each run. The 33 nl reaction mix was run per data point on the Applied Biosystems^™ QuantStudio^™ 12K Flex Real-Time PCR System (ThermoFisher Scientific, Marsiling Industrial Estate, Singapore). Genotypes for the samples were determined by the TaqMan^™ Genotyper Software as per the manufacturer’s instructions. Genotype calls were generated by the TaqMan^® Genotyper Software. The CYP2D6 copy number was determined using the Applied Biosystems TaqMan copy number assays for exon 9, the primary copy number assay (Assay ID: Hs00010001_cn) to quantify CYP2D6 duplications or identify CYP2D6 gene deletions (CYP2D6*5) in the samples. AlleleTyper^™ software was used to convert sample genotype information for the CYP genes interrogated to the star (*) allele nomenclature using a predefined allele translation table that maps a specified allele pattern to the star allele call.

2.4. Measurement of tamoxifen and metabolite in plasma.

Plasma concentrations of tamoxifen and its metabolites were quantified using a validated LC-MS/MS method. Plasma samples were extracted using protein precipitation where 200 μl of plasma was spiked with 10 μl of 2 μg/ml propranolol (internal standard) followed by addition of 590 μl of ice-cold acetonitrile. The mixture was vortexed for 30 sec and sonicated for 10 min. This was followed by centrifugation at 16 000 × g for 10 min. The supernatant was collected and evaporated to dryness under a gentle stream of nitrogen. The residue was reconstituted in 50 μl of mobile phase and 10 μl was injected for LC-MS/MS analysis.

LC separation was performed on a Shimadzu Nexera XR HPLC (Shimadzu, Kyoto, Japan). The mobile phase consisted of 0.1% formic acid in 10mM ammonium formate solution as mobile phase A and 0.1% formic acid in acetonitrile as mobile phase B. Chromatographic separation was achieved on a Zorbax C18 2.1 X100 mm, 3.5 μm column (Zorbax Agilent, Santa Clara, CA, USA). Analytes were eluted on a using a gradient programmed as follows: 0–9 min, B 30%, 9.01–9.5 min, B 52%- and 9.5–13-min B 30% at a flow rate of 0.3 ml/min. The column was maintained at a temperature of 40 °C.

Analyte detection was achieved on a SCIEX Triple Quad^™ 3500 mass spectrometer (AB Sciex LLC, Redwood City, CA, USA). The instrument was operated in multiple reaction monitoring (MRM) mode. The Turbo VTM source was used with an electrospray ionization (ESI) probe at operated at 500 °C, curtain gas 25, ion spray voltage 5500, GS1 and GS2 gas 50 and 30 respectively. Analytes were followed using multiple reaction monitoring (m/z 372.5→ 72.2, 374.4→ 58.1, 358.4→ 58.0, and 260.3→ 183.3 for tamoxifen, endoxifen, NDM-tamoxifen and propranolol respectively). Analyst software v 1.6 was used for method development, data acquisition and processing.

2.5. Data and statistical analysis

Statistical analysis was performed using the Stata software package, version 22.0 (StataCorp LLC, Texas, USA) and GraphPad Prism 8.4.3 (GraphPad Software, San Diego, CA, USA). Analyses included descriptive statistics, paired t tests. Two-sample t tests were used to explore crude associations between categorical and continuous variables. The chi-square test (χ²) was used for associations between categorical variables. Results with p values less than .05 were deemed to be statistically significant.

3. Results

3.1. Baseline characteristics

A total of 40 hormone receptor positive patients being on tamoxifen therapy were recruited in the study. Baseline characteristics are described in Table 1. The average age of the enrolled participants was 50.2 years (± 10.4) and 55.0% were post-menopausal women. The median BMI was 29.2 kg/m² with a range of 26.2 – 33.7 kg/m². All the recruited participants were female except 1 who was male. Many of the patients were HIV negative (82.5%) with 36.5% of the participants being hypertensive, 2.5% diabetic and 12.5% having other comorbidities. Most of the breast cancer patients had a late diagnosis with at least stage 3A breast cancer at diagnosis (64.9%). Of the recruited patients, 92.5% presented with Invasive Ductal Carcinoma (IDC) breast cancer while 5.0% had Invasive Lobular Carcinoma (ILC).

Table 1:

Baseline demographics for breast cancer patients in this study (n=40 patients)

Variable		N (%), Mean (SD), Median (IQR)
Age (years)		50.2 (±10.4)
BMI (kg/m2) median (range)		29.2 (26.2–33.7)
Menopausal status n (%)	Pre-menopausal	18(45)
	Post-menopausal	22(55)
HIV Status n (%)	Positive	7(17.5)
	Negative	35(82.5)
Comorbidities n (%)	Hypertension	10(25)
	Diabetes	1(2.5)
	Other	5(12.5)
Comedication n (%)	None	18(45)
	1–2	18(45)
	3–5	4(10)
	ART	7(17.5)
Cancer stage n (%)	1	1(2.5)
	2A	4(10)
	2B	8(20)
	3A	3(7.5)
	3B	13(32.5)
	3C	3(7.5)
	4	6(15)
	Unknown	2(5)
Type of cancer n (%)	IDC	37(92.5)
	ILC	2(5)
	Unknown	1(2.5)
Surgery n (%)	Mastectomy	32(80)
	None	8(20)
Receptor status n (%)	ER+	8(20)
	PR+	2(5)
	ER+ & PR+	20(50)
	Triple positive	5(12.5)
	Unknown	5(12.5)
Duration on tamoxifen n (% frequency)	<1 month	2 (5)
	1–2 months	11(27.5)
	3 months - 1 year	18(45)
	>1 year	9(22.5)

3.2. Adherence and side effect prevalence in patients on tamoxifen therapy

A self-reported method was used to assess tamoxifen adherence. Many of the patients were adherers (76.6 %), while 20.9 % and 2.5 % were medium and non-adherers respectively. A biochemical method of assessing adherence was also used, using tamoxifen concentrations to predict the rate of adherence, with high adherence (>60 ng/ml), moderate adherence (14–60 ng/ml) and poor adherence (<14 ng/ml). Based on this biochemical assessment, 10% of the patients had poor adherence, 22.5% had moderate adherence and 67.5% were high adherers. Patients on tamoxifen treatment reported at least 1 side effect (56.1%) (Fig 1). Common side effects reported by the participants were hot flashes (31.7 %), musculoskeletal pain (14.6 %) and vaginal discharge (14.3 %) while 19.5 % reported having other side effects such as fatigue, neuropathy, and drowsiness (Fig 1) that were not present on the questionnaire.

Figure 1:

Prevalence of tamoxifen related side effects in the breast cancer study cohort

3.3. Genotype and phenotype frequencies

The allele frequencies observed in this study were compared to other allele frequencies observed in other studies and other populations (Table S1). The *1 allele, which is the wild type was 32%, comparable to the 39% in the previous study done in healthy Zimbabweans. The *17 and *29 alleles occurred at frequencies of 18.9% and 16.9% respectively. These alleles are almost absent in other populations, with frequencies below 0.9% in Asians, Europeans, and Americans. The CYP2D6 phenotypes were predicted according to the CYP2D6 diplotype to phenotype table available on the Clinical Pharmacogenetics Implementation Consortium (CPIC) website (https://cpicpgx.org). The common phenotype group was that of the normal metabolizers (NM), which was 67.5%. However, intermediate metabolizers (IMs) had quite a significant frequency of 27.5%, while the frequency of ultrarapid metabolizers (UMs) was 5%. There were no CYP2D6 poor metabolizers in this study.

3.4. Association between endoxifen concentrations and CYP2D6 predicted phenotype groups.

Tamoxifen, endoxifen and NDM-tamoxifen were quantified in patient plasma. The lower limit of quantification for the analytes was 0.01 ng/ml standard curve linear in the range between 0.01–1000 ng/ml. The median for endoxifen concentration was 4.78 ng/ml with a range of 3.88 – 8.49 ng/ml with 55% of the patients were below endoxifen therapeutic threshold (5.97 ng/ml). The median for the endoxifen/NDM-tamoxifen ratio was 0.0579 with a range of 0.03 – 0.1. There was a statistically significant difference in the endoxifen concentrations between CYP2D6 phenotype groups (p = 0.0151) (Figure 2a). The difference in endoxifen concentrations between NMs and IMs was significant (p = 0.0206) while the difference between NMs and NMs/UMs was not significant (p = 0.1478). The intermediate metabolizer group had concentrations below the therapeutic threshold (except 1 outlier). There was a wide spread of endoxifen concentrations in the normal metabolizer group, with concentrations ranging from 0.5–16 ng/ml.

Figure 2:

Association between (a) endoxifen concentrations and (b) endoxifen/NDM-tamoxifen metabolic ratio (MR) with CYP2D6 phenotype groups (intermediate metabolizers (IM) (n=11), normal metabolizers (NM) (n=27) and ultra-rapid metabolizers (UM) (n=2)). The red line depicts the endoxifen minimum therapeutic concentration of 5.97 ng/ml while the red diamonds are outliers.

3.5. Association between endoxifen/n-desmethyl-tamoxifen ratio and CYP2D6 predicted phenotype groups.

The association between CYP2D6 phenotypes and endoxifen/NDM-tamoxifen ratio was also determined (Figure 2a). There was a significant difference in metabolic ratios between the phenotype groups, with an overall p-value of 0.0006. IMs had lower endoxifen concentrations and metabolic ratios compared to NMs (p = 0.0018). The patients were further categorized according to their individual CYP2D6 activity scores (AS). Figure 2b shows the high inter-individual variability of the endoxifen/NDM metabolic ratios observed between individuals with the same activity score. The phenotype groups were further categorized into the activity scores (AS) that determine the phenotypes i.e., IMs = 0.25–1, NMs = 1.25–2 and UMs >2 ^18,23.

3.6. Other association studies

The chi-square association test was done to determine if there is any association between the genotype and these side effects and with the total number of side effects observed in each patient. There was no significant association observed except with menstrual changes, which had a p-value of 0.024. We explored the association between endoxifen concentration and side effects. There was also no significant association observed between endoxifen concentration and side effects (p = 0.277). The Pearson’s chi-square test showed no association between rate of adherence and polypharmacy (p = 0.399). High adherers had the highest number of side effects compared to low ones. This was however not statistically significant (p = 0.117). None of the comorbidities had a statistically significant association with the rate of adherence except for hypertension (p = 0.043), where hypertensive patients had lower tamoxifen adherence. There was no association between individual side effects and adherence, except for menstrual changes (p = 0.017).

The endoxifen/NDM-tamoxifen threshold value used in this study was 0.03. There were 15 patients (37.5%) below this threshold which were classified as slow metabolizers while the 62.5% above this ratio were classified as normal metabolizers ²⁴. Out of the slow metabolizers, 60% were IMs and 40% were NMs according to the CPIC standard classification. Additionally, 84% of those classified as normal metabolizers according to the metabolic ratio were NMs as per CPIC standard classification.

4. Discussion

In this study we show that CYP2D6*17 and *29 variants which have a high frequency in African populations have a clinically significant effect on the levels of endoxifen, the active metabolite of tamoxifen. In the 40 patients studied, 55% had endoxifen concentration below the therapeutic level of 5.97 ng/ml and where carriers of CYP2D6 diplotypes predictive of the NM or IM status. These data point to the need for clinical studies on how the CYP2D6*17 and *29 could be integrated in clinical pharmacogenetic guidelines in the use of tamoxifen in breast cancer patient of African ancestry.

There was no significant association between genotype and side effects with the exception with menstrual changes (p = 0.024), despite 56% of the patients on tamoxifen reporting at least one side effect. As expected, hot flashes were the most common side effect ²⁵. The side effects may be associated with lower estrogen levels resulting from tamoxifen induced conformational changes in the estrogen receptors. Late-stage disease presentation was another notable observation among the patients. This was expected as late presentation with locally advanced or metastatic disease has been one of the reasons associated with high breast cancer mortality in low-income countries ³. A higher prevalence of HIV (17.6 %) was also noted among the participants. This however had no association with genotype or endoxifen concentrations despite it being reported as a predictor of poor tamoxifen adherence ²⁶. This could be explained by the ART combination they were on. Most of the patients where on TenolamD (tenofovir, lamivudine and dolutegravir). These drugs do not interact with CYP enzymes at clinically relevant concentrations. Only 2 of the HIV positive patients were on TenolamE (tenofovir, lamivudine and efavirenz). The number was too low to make any meaningful association, even though efavirenz is a known inducer of CYP3A4 ²⁷. Hypertension was associated with adherence in agreement with other studies ²⁸.

Polymorphisms in the CYP2D6 gene impact the concentrations of endoxifen. Studies have demonstrated a significant gene-dose effect of CYP2D6 polymorphism for tamoxifen metabolism ²⁹. This study reports a significant association between the CYP2D6 phenotypes, endoxifen concentrations and the metabolic ratio (p = 0.0151) with significant endoxifen concentration difference between the IMs and the NMs (p = 0.0206). We observed subtherapeutic endoxifen plasma concentration (4.78 ng/ml) in the IMs. This tallied with the endoxifen/NDM-tamoxifen metabolic ratio which is a better predictor of CYP2D6 activity as it shows the rate at which NDM is being converted to endoxifen. It is also the only route on the tamoxifen pathway that is only metabolized by CYP2D6 ⁴.

We predicted activity scores ranging between 1.5 – 2 for the NMs, with endoxifen concentrations ranging between 0.5 – 16 ng/ml. This high interindividual variety in patients within the same activity group/metabolizer status points to the possible existence of genetic variants not covered by the testing panel which affect CYP2D6 activity. Another possible explanation could be unreported co-medications that are potent inhibitors of CYP2D6 resulting in phenocopying the IM and PM status in patients identified as NM. Based on the CPIC guidelines, a standard dose of 20 mg/ml is recommended for normal metabolizers ¹⁸ since they are expected to reach therapeutic endoxifen concentrations. This means that some of these patients may not fully benefit from this dose and may end up experiencing recurrence even though they are normal metabolizers. There was also potential clinical failure in 34.3% of patients in this study because of their IM status. CPIC guidelines recommends alternative hormonal therapy or an aromatase inhibitor for post and premenopausal women respectively ¹⁸. A higher FDA approved tamoxifen dose of 40 mg/day is recommended where use of an aromatase inhibitor use is contraindicated.

Studies have demonstrated the feasibility of dose escalation to achieve desired endoxifen concentrations ^30–32. Correlation of CYP2D6*10 with activity score has been successfully used to predict doses that would achieve endoxifen therapeutic concentrations ³². This could be achieved with CYP2D6*17 as the two enzymes have been shown to have comparable activity in vitro ³³. The activity of CYP2D6*29 is comparable to that of CYP2D6*17 in the metabolism of NDM-tamoxifen to endoxifen in vitro ³⁴ and in breast cancer patients ³⁵. Both CYP2D6*17 and *29 have higher activity compared to CYP2D6*10 ³⁴. Taking into account Kanji’s findings that estimated the activity score of CYP2D6*17 to be 0.3 ³⁶ and the revised activity score of CYP2D6*10 of 0.25 ³⁷. We therefore expect the CYP2D6*19 activity score to be around that of CYP2D6*17, of 0.30.

Our previous study proposed a dose increase of tamoxifen to 40 mg/day for patients homozygous for CYP2D6*17 ³⁶. The study also proposed a revised activity score for CYP2D6*17 from the CPIC assigned 0.5 ¹⁸ to 0.34 ³⁶. Re-evaluation of the CYP2D6*10 variant from an activity score of 0.5 to 0.25 resulted in improved prediction of endoxifen by CYP2D6 in Asian populations ³⁸. Dose escalation may therefore be useful in our Zimbabwean population, given the observed subtherapeutic levels in our breast cancer patients.

Other factors which affect treatment outcome such as adherence, comorbidities, and potential drug-drug interactions because of polypharmacy were interrogated. Poor adherence results in reduced plasma concentrations of endoxifen, which may result in poorer clinical outcomes, increased mortality rate and higher health care costs ³⁹. Polymorphisms in the CYP3A4/5 genes or the presence of drug inhibitors in some patients may interfere with tamoxifen metabolism, resulting in a lower endoxifen concentrations ⁴. We however, observed adherence rates (67%) with minor discrepancies between the biochemical method and self-reporting. The biochemical method identified a higher proportion of non-adherence (10%), which had been underestimated by self-reports, and this is like what has been observed before³³. There are however other factors that may alter the plasma concentration of tamoxifen such as drug-drug interactions, rate of tamoxifen absorption and genetic variants in drug transporter genes. Polymorphisms in tamoxifen metabolizing genes such as CYP3A4, CYP2C9, CYP2C19 and other Cytochrome P450 genes that were not interrogated in this study may also impact the plasma concentration of tamoxifen^4,9,19. In addition, we could not find associations between endoxifen concentration and comorbidities and/ or any potential drug-drug interactions.

In conclusion, this study reported a clinically significant effect CYP2D6*17 and *29 on the levels of endoxifen. If CPIC guidelines for CYP2D6 genotype guided use of tamoxifen are to be implemented in the Zimbabwean breast cancer patients, patients with predicted IM phenotype could benefit from a tamoxifen dose increase to 40 mg/day. A non-inferiority study with respect to safety of the 20 and the 40 mg/day tamoxifen should however be done to inform patients and clinicians on the risk such treatment modification could carry.

What is already known about this subject?

• CYP2D6*17 and *29 variants have a high frequency in African populations.

• CYP2D6 is an important enzyme in tamoxifen metabolism to the active endoxifen.

• Genetic polymorphisms in the CYP2D6 gene have been associated with variations in endoxifen plasma concentrations.

What this study adds

• Effect of CYP2D6*17 and *29 on the levels of endoxifen is clinically significant.

• there is a strong association between CYP2D6*17 and *29 genotypes with low plasma concentrations of endoxifen.

• Polymorphisms in CYP2D6 have implications for treatment in the use of tamoxifen in breast cancer patient of African ancestry.

8. Data Availability Statement

Data is contained within this article or Supplementary Material. Additional data that support the findings of this study are available on request from the corresponding author. The data are not publicly available because of privacy and ethical restrictions.