Co‐consuming green tea with raloxifene decreases raloxifene systemic exposure in healthy adult participants

Abstract

Green tea is a popular beverage worldwide. The abundant green tea catechin (−)‐epigallocatechin gallate (EGCG) is a potent in vitro inhibitor of intestinal UDP‐glucuronosyltransferase (UGT) activity (K _i  ~2 μM). Co‐consuming green tea with intestinal UGT drug substrates, including raloxifene, could increase systemic drug exposure. The effects of a well‐characterized green tea on the pharmacokinetics of raloxifene, raloxifene 4′‐glucuronide, and raloxifene 6‐glucuronide were evaluated in 16 healthy adults via a three‐arm crossover, fixed‐sequence study. Raloxifene (60 mg) was administered orally with water (baseline), with green tea for 1 day (acute), and on the fifth day after daily green tea administration for 4 days (chronic). Unexpectedly, green tea decreased the geometric mean green tea/baseline raloxifene AUC_0–96h ratio to ~0.60 after both acute and chronic administration, which is below the predefined no‐effect range (0.75–1.33). Lack of change in terminal half‐life and glucuronide‐to‐raloxifene ratios indicated the predominant mechanism was not inhibition of intestinal UGT. One potential mechanism includes inhibition of intestinal transport. Using established transfected cell systems, a green tea extract normalized to EGCG inhibited 10 of 16 transporters tested (IC₅₀, 0.37–12 μM). Another potential mechanism, interruption by green tea of gut microbe‐mediated raloxifene reabsorption, prompted a follow‐up exploratory clinical study to evaluate the potential for a green tea–gut microbiota–drug interaction. No clear mechanisms were identified. Overall, results highlight that improvements in current models and methods used to predict UGT‐mediated drug interactions are needed. Informing patients about the risk of co‐consuming green tea with raloxifene may be considered.

Study Highlights.

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

The natural product green tea has been studied extensively as a precipitant of cytochrome P450 (CYP)‐mediated and organic anion transporting polypeptide (OATP)‐mediated drug interactions, but potential interactions involving other pathways, such as the UDP‐glucuronosyltransferases (UGTs), have not been investigated in humans. A mechanistic static model predicted two abundant catechins in green tea would increase the area under the plasma concentration versus time curve of the intestinal UGT substrate raloxifene by up to six‐fold relative to baseline.

• WHAT QUESTION DID THIS STUDY ADDRESS?

Does the in vitro–in vivo prediction translate to human participants? Do mechanisms underlying the pharmacokinetic green tea–raloxifene interaction involve intestinal transporters and/or the gut microbiota?

• WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

Green tea precipitated a pharmacokinetic interaction with raloxifene that was opposite to the predicted effect, indicating the overarching mechanism does not involve inhibition of intestinal UGTs. Results implicated one or more processes involved in intestinal absorption. MRP3 was the only transporter identified that may contribute to the interaction. An exploratory follow‐up clinical study suggested green tea did not interrupt raloxifene enterohepatic recycling through inhibition of bacterial beta‐glucuronidase activity. The preexisting gut microbiota may interact with green tea to influence raloxifene stool recovery.

• HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?

CYP‐based mechanistic static models used to predict pharmacokinetic interactions involving UGTs with improved predictive performance are needed. Health care providers may consider informing their patients about potential risks when co‐consuming green tea with raloxifene.

INTRODUCTION

Next to water, tea is the most consumed beverage worldwide. ¹ Many types of tea, including black tea and green tea, are derived from the leaves of the Camilla sinensis plant but differ in chemical composition based on the degree of fermentation. ² Green tea is produced by immediately processing harvested leaves. The polyphenolic catechins are major antioxidants in green tea believed to contribute to purported health benefits, including cardioprotection, chemoprevention, and weight loss. ¹ , ² , ³ , ⁴ , ⁵ Within the last decade, increasing evidence indicates that an interaction between green tea phytoconstituents and the gut microbiota may contribute to the health benefits of green tea. ⁶ , ⁷ Although consumers are aware of these benefits, they may not be aware of potential interactions with co‐consumed drugs.

As with drug–drug interactions, pharmacokinetic natural product–drug interactions occur when the precipitant (aka perpetrator) natural product alters the absorption, distribution, metabolism, and/or excretion of the object (aka victim) drug. Common mechanisms include induction or inhibition of drug metabolizing enzymes, particularly the cytochrome P450s (CYPs), and transporters. Regarding transporters, studies showed green tea to decrease systemic area under the plasma concentration versus time curve (AUC) to the minimally metabolized drugs digoxin, fexofenadine, lisinopril, and nadolol. ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² These observations were attributed to inhibition of an intestinal uptake transporter(s) by green tea. Pharmacokinetic green tea–drug interactions involving biochemical targets in the intestine in addition to the CYPs and transporters, such as UDP‐glucuronosyltransferases (UGTs), remain understudied. Catechins and other green tea polyphenols undergo extensive glucuronidation, thus could be competitive UGT inhibitors. ¹³ , ¹⁴ In vitro evidence for two abundant catechins in green tea, (−)‐epicatechin gallate (ECG) and (−)‐epigallocatechin gallate (EGCG), demonstrated potent reversible inhibition of intestinal UGT activity (K _i  ≤2 μM) using raloxifene, a selective estrogen receptor modulator and intestinal UGT substrate. ¹⁵

Applying a mechanistic static model, ECG and EGCG were predicted to increase raloxifene AUC by up to 1.3‐ or 6.1‐fold compared with baseline depending on the method used to estimate catechin concentrations in the intestine. ¹⁵ Following US Food and Drug Administration (FDA) guidance, ¹⁶ these predictions prompted the current work. The initial objective was to evaluate the effects of a well‐characterized green tea product on the pharmacokinetics of raloxifene, raloxifene 4′‐glucuronide, and raloxifene 6‐glucuronide in healthy adult participants upon oral administration of raloxifene. An unexpected decrease in raloxifene AUC prompted investigation of alternate mechanisms of this natural product–drug interaction that involve transporters and the gut microbiota. Results highlight that improvements in current models and methods used to predict potential UGT‐mediated natural product–drug interactions are warranted.

MATERIALS AND METHODS

Green tea study materials

One lot of a single batch of a widely consumed green tea (coded T21) was sourced, characterized, and acquired for all clinical and in vitro studies as described. ¹⁷ T21 was selected from a collection of 34 top‐selling commercially available products in the US. The chemical composition of each product was characterized using untargeted liquid chromatography mass spectrometry metabolomics analysis, and similarities were assessed based on composite scores analysis. ¹⁷ , ¹⁸ T21 closely matched the chemical composition of the National Institutes of Standards and Technology reference material. ¹⁷ The data comparing T21 with the other green tea products are freely accessible in the Center of Excellence for Natural Product Drug Interaction Research database. ¹⁹ , ²⁰

Clinical studies

Preparation of brewed green tea

Catechin content in the T21 teabags averaged 0.9 ± 0.1, 10.5 ± 0.4, 31.3 ± 2.1, 7.3 ± 0.5, and 26.4 ± 2.2 mg for (+)‐catechin, (−)‐epicatechin (EC), (−)‐epigallocatechin (EGC), ECG, and EGCG, respectively. ¹⁵ One teabag was steeped in 240 mL of water at 80°C for 3 min, removed, and the remaining liquid was compressed out of the teabag into the tea. The prepared tea was cooled to 50°C before administering to the participants.

Main study protocol

The Washington State University (WSU) Institutional Review Board approved the clinical protocol. After obtaining written informed consent, eight male and eight nonpregnant, nonlactating female adults meeting the inclusion/exclusion criteria (Figure S1) were enrolled. Participants abstained from caffeine‐containing products, fruit juices, and other dietary/herbal supplements for at least 1 week prior to study initiation and for the duration of the study.

This open‐label, fixed‐sequence, crossover study consisted of three arms: baseline (raloxifene with water), acute (green tea with raloxifene), and chronic (green tea × 4 days, then green tea with raloxifene on day 5) (Figure S2). Each arm was separated by at least 7 days. Participants fasted from midnight prior to each inpatient study day, during which blood (7 mL) was collected before and after raloxifene administration at 0.25, 0.5, 1, 1.5, 2, 3, 4, 6, 8, and 12 h. Participants continued to fast until after the 4‐h blood collection, when lunch and snacks were provided. They were discharged after the 12‐h collection; they returned for blood collection at 24, 48, 72, and 96 h. Plasma was harvested by centrifugation, mixed with 20% ascorbic acid (50:1, v/v) to minimize oxidation of the catechins, ¹⁵ , ²¹ and stored at −80°C. Blood pressure, pulse, and oxygen saturation were recorded periodically throughout the inpatient days. During the baseline arm, participants were administered one 60 mg raloxifene tablet (Teva Pharmaceuticals, Sellersville, PA, USA) with 240 mL warm water (50°C) (Figure S2). During the acute green tea arm, participants were administered 240 mL tea with raloxifene; the tea was administered again at 4 and 8 h. During the chronic green tea arm, participants were administered tea three times daily at evenly spaced intervals during the 8‐h workday for 4 consecutive days. On day 5, raloxifene was administered with tea in the same manner as described for the acute green tea arm.

Follow‐up exploratory study protocol

Results from the main study prompted a follow‐up, small exploratory study (approved by the WSU Institutional Review Board) aimed to assess the potential contribution of the gut microbiota to the observed pharmacokinetic green tea–raloxifene interaction. Healthy adult participants (three males and three nonpregnant, nonlactating females) were enrolled; four of them were retained from the main study. The study consisted of three sequential arms: raloxifene + water, raloxifene + acute green tea, and green tea alone (Figure S3). Each arm was separated by at least 7 days. Blood was collected from 0 to 48 h at the same intervals as the main study. Correlation analyses of partial AUCs with the AUC_0–96h of raloxifene indicated that a 0–48‐h collection was a reasonable surrogate of AUC_0–96h (r ² = 0.87). Each participant collected one bowel movement within 48 h prior to raloxifene/green tea administration and all bowel movements from 0–24 and 24–48 h after raloxifene/tea administration using a Fisherbrand™ Commode Specimen Collection System (Pittsburgh, PA, USA). Concurrently, they collected a pea‐size stool sample before and from 24 to 48 h using an OMNIgene•GUT kit (DNA Genotek, Ottawa, Ontario, Canada). Missed collections (due to constipation) were noted. Stool samples were stored at −20°C until further processing (described later). The pea‐size samples were shaken via OMNIgene•GUT kits and stored at room temperature.

Quantification of raloxifene and raloxifene glucuronides in plasma and stool

Plasma was processed and analyzed for raloxifene and raloxifene 4′‐ and 6‐glucuronide by ultra‐high‐performance liquid chromatography/tandem mass spectrometry (UHPLC/MS/MS) as described ¹⁵ , ²² with modifications. In brief, plasma (400 μL) was mixed with methanol (800 μL) containing 0.1% formic acid and the internal standards raloxifene‐d₄, raloxifene‐d₄‐4′‐glucuronide, raloxifene‐d₄‐6‐glucuronide, and ethyl gallate (Toronto Research Chemicals, Toronto, Canada) then centrifuged (16,900  g × 10 min). The supernatant (950 μL) was evaporated to dryness using a centrifugal vacuum concentrator (Eppendorf, Hamburg, Germany) and reconstituted with 50:50 water: methanol (28 μL) containing 0.1% formic acid.

Stool samples were mixed with a 1:2.5 ratio of Milli‐Q water by weight and emulsified to uniform consistency with a hand‐held immersion blender. Aliquots were transferred to 1.5 mL Eppendorf tubes and stored at −80°C pending analysis by UHPLC–MS/MS. Bristol Stool Chart ratings were recorded on a 1–7 scale (1: separate hard lumps; 7: watery, no solid pieces) ²³ and verified by two individuals for the 0–24 and 24–48 h stool samples. Emulsified stool (100 μL) was mixed with methanol (400 μL) containing 0.1% formic acid and internal standards. After centrifugation (16,900  g × 10 min), supernatants were transferred to clean tubes, and 1×, 10×, and 100× dilutions (final volume 300 μL) were prepared to account for the large interindividual variability in fecal raloxifene content. Dilutions were evaporated to dryness and reconstituted with 50:50 water:methanol (100 μL) containing 0.1% formic acid. After centrifugation, reconstituted dilutions were quantified for the various analytes as described for plasma using an adjusted calibration curve (0.06–400 nM).

In vitro transporter studies

Preparation of green tea extract

An extract was prepared by combining 200 mg T21 leaf material from a tea bag with 20 mL methanol in 20 mL scintillation vials. Mixtures were shaken overnight at room temperature, filtered, dried under a stream of nitrogen, and the resulting extract was stored at room temperature until analysis. One gram of the extract contained 64.96 mg EGCG. ¹⁷ A stock solution of this extract was prepared to achieve a concentration of 10 mM EGCG by adding 141.7 μL DMSO to a 10‐mg aliquot of extract.

Solubility of raloxifene and green tea extract in buffer

The solubility of raloxifene (hydrochloride salt, BIOTANG, Lexington, MA, USA) and the green tea extract (normalized to EGCG content) in buffer was assessed as described ²⁴ to determine the appropriate concentration range for conducting the transporter assays (Table S1).

Raloxifene as a substrate for uptake transporters

Raloxifene was tested as a substrate for the following uptake transporters using transporter‐expressing cell lines or inside‐out membrane vesicles (HEK293 or MDCKII) and established probe substrates and inhibitors as described ²⁴ : sodium/taurocholate cotransporting polypeptide (NTCP), organic anion transporter (OAT)1, OAT3, organic anion transporting polypeptide (OATP)1B1, OATP1B3, OATP2B1, organic cation transporter (OCT)1, and OCT2 (Table S2). Two concentrations of raloxifene (1, 10 μM) and transport times (2, 20 min) were used. For each transporter, positive control substrates and inhibitors and negative control cell lines (empty vector transfection or untransfected cells) were included to ensure each system was functioning properly (Table S2).

Green tea extract as an inhibitor of uptake and efflux transporters

The green tea extract (normalized to EGCG content) was tested as an inhibitor of the uptake transporters (1, 10 μM) listed above, along with the efflux transporters (2, 20 μM) bile salt export pump (BSEP), breast cancer resistance protein (BCRP), multidrug and toxin extrusion protein (MATE)1, MATE2‐K, multidrug resistance‐associated protein (MRP)2, MRP3, and P‐glycoprotein (P‐gp) (Table S2) as described. ²⁴ For inside‐out vesicles containing ATP‐dependent transporters (BCRP, BSEP, P‐gp, MRP2, MRP3), incubations were conducted in the presence of 4 mM ATP or AMP to distinguish between transporter‐mediated uptake and passive diffusion into the vesicles. Assay performance was evaluated using reference inhibitors and negative control cell lines (empty vector transfected or untransfected cells). If ≥50% inhibition by the green tea extract relative to the solvent control was observed, the inhibitor concentration corresponding to 50% of control activity (IC₅₀) was determined using standard inhibition assays at seven concentrations (Table S3) as described. ²⁴

Gut microbiota composition and analysis

Pea‐size stool samples were shipped within 4 months after collection to AKESOgen (Peachtree Corners, GA, USA) for 16S rRNA gene analysis of variable region 3 (V3) and variable region 4 (V4). Library preparation and sequencing of the V3/V4 region of the 16S gene were conducted using the Illumina MiSeq platform and associated kits. The microbiota data were analyzed using Nephele ²⁵ and the pipeline Mothur. ²⁶ Amplified DNA sequences were preprocessed using a Phred quality score of 19 (99% base pair accuracy). Demultiplexed reads were clustered into operational taxonomical units using an open reference approach by comparing results with the Greengenes database, allowing sequences clustering at 99% similarity.

Data analysis

Pharmacokinetic analysis

The pharmacokinetics of raloxifene and glucuronides were determined via noncompartmental analysis using Phoenix WinNonlin (v7.0, Certara, Princeton, NJ, USA) as described. ²⁴ Specifically, the first maximum concentration (C_max,1), time to reach C_max,1 (t _max,1), terminal half‐life (t _1/2), AUC_0–48h, AUC_0–96h, and AUC_0–inf were recovered.

Statistical and power analysis

The primary endpoint for the main clinical study was the green tea/baseline ratio of log‐transformed raloxifene AUC_0–96h, with a predefined no effect range of 0.75–1.33. ²⁷ The sample size for the main study was determined using SAS (v9.2; SAS Institute, Cary, NC, USA), assuming a type I error of 0.05, a power of 0.80 to detect a 25% change in the primary end point, and 28% intraindividual variability in raloxifene AUC. ²² Secondary end points (e.g., green tea/baseline ratio of C_max,1, t _1/2, t _max,1) for raloxifene, as well as green tea/baseline ratios of pharmacokinetic outcomes for the glucuronides, were evaluated using a paired two‐tailed Student's t‐test on log‐transformed data (green tea vs. baseline) or Wilcoxon signed‐rank test as appropriate. A p‐value <0.05 was considered statistically significant.

The abundance of each taxa in the stool collections was analyzed using the phyloseq package in R. ²⁸ Microbial alpha diversity was assessed using the Shannon and Chao1 diversity indices, which address species richness and evenness. ²⁹ Compositional heterogeneity of the microbial community in each sample from each arm was visualized at the genus level using principal coordinate analyses and the Bray–Curtis dissimilarity index. ²⁹ Beta diversity was statistically evaluated using the adonis function through the vegan package in R. ³⁰ This statistical approach was repeated for the putative functional data of the microbiota expressed through the Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) method. ³¹ PICRUSt is a computational method used to predict the metatranscriptomics for each sample based on 16S data.

RESULTS

Effects of green tea on raloxifene disposition in healthy adults

Study participants

All participants enrolled in the main study completed the three arms. Raloxifene and green tea were well tolerated, and no severe adverse events were reported. Participants self‐identified as White/non‐Hispanic (seven males, seven females) or Asian/non‐Hispanic (one male, one female). The median (range) age was 25 (22–42) and 29 (18–56) years for male and female participants, respectively. None of the participants reported taking concomitant medications or natural products known to alter raloxifene pharmacokinetics. The most common mild adverse events were intravenous site bruising and headache. No adverse events resulted in study discontinuation.

Raloxifene pharmacokinetics

The objective of the main study was to compare raloxifene pharmacokinetics between raloxifene administered with water (baseline) and upon acute or chronic green tea administration (Figure 1). Compared with baseline, the primary end point, geometric mean green tea/baseline raloxifene AUC_0–96h ratio, lay outside the predefined no effect range (0.75–1.33) upon acute (0.57) and chronic (0.66) green tea administration (Figure 1, Table 1). Likewise, the AUC_0–48h ratio (0.53–0.56) and C_max,1 ratio (0.37–0.39) decreased upon acute and chronic green tea administration (Table 1). Both median t _max,1 (~2 h) and geometric mean terminal half‐life (~24 h) of raloxifene were unchanged upon acute and chronic green tea administration. Both glucuronides followed similar trends as raloxifene (Figure 1, Table 1). The shorter apparent half‐lives of the glucuronides relative to raloxifene likely reflected extensive first‐pass formation of the glucuronides in the intestine.

FIGURE 1.

Concentration versus time profiles for (a) raloxifene, (b) raloxifene 4′‐glucuronide (raloxifene 4′‐G), and (c) raloxifene 6‐glucuronide (raloxifene 6‐G) following oral administration of raloxifene (60 mg) to 16 healthy adult participants. Symbols and error bars denote geometric means and upper limits of the 90% confidence interval, respectively. Black symbols and lines denote baseline (240 mL water 3× daily × 1 day), solid green symbols and lines denote acute green tea (240 mL 3× daily × 1 day), and open green symbols and dashed lines denote chronic green tea (240 mL 3× daily × 4 days). Insets depict each data set from 0 to 12 h with error bars excluded for visual clarity.

TABLE 1.

Pharmacokinetics of raloxifene and primary glucuronides in the absence and presence of green tea in the 16 healthy adults who participated in the main clinical study.

Metric	Geometric mean (CV%)			Green tea/baseline ratio (90% CI)
	Baseline	Green tea		Acute	Chronic
		Acute	Chronic
Raloxifene
AUC0–96h (nM‐h)	14.9 (55)	8.5 (54)*	9.8 (52)*	0.57 (0.46–0.70)	0.66 (0.54–0.81)
AUC0–48h (nM‐h)	11.2 (67)a	5.91 (109)	6.26 (48)a	0.53 (0.43–0.64)	0.56 (0.43–0.72)
Cmax,1 (nM)	0.36 (66)	0.13 (54)	0.14 (69)	0.37 (0.28–0.50)	0.39 (0.24–0.64)
t 1/2 (h)	24.9 (37)b	23.4 (32)c	22.7 (22)a	0.95 (0.81–1.12)	0.89 (0.75–1.04)
t max,1 (h)d	1.25 (0.25–4)	1.5 (0.45–3)	1 (0.25–2)	NA	NA
Raloxifene 4′‐glucuronide
AUC0–96h (nM‐h)	2350 (45)	1350 (69)**	1520 (76)*	0.57 (0.45–0.73)	0.65 (0.51–0.82)
AUC0–48h (nM‐h)	1990 (48)	1080 (55)	1134 (56)	0.54 (0.43–0.68)	0.57 (0.43–0.76)
Cmax,1 (nM)	139 (71)	57.9 (62)	64.1 (61)	0.42 (0.31–0.57)	0.46 (0.31–0.70)
t 1/2 (h)	17.0 (57)b	15.9 (66)	15.6 (48)b	0.91 (0.72–1.16)	0.92 (0.72–1.18)
t max,1 (h)d	1.25 (0.25–4)	1.5 (0.5–6)	1.25 (0.5–3)	NA	NA
Metabolite‐to‐parent AUC0–96h ratio	160 (34)	161 (36)	159 (34)	NA	NA
Raloxifene 6‐glucuronide
AUC0–96h (nM‐h)	535 (48)	324 (79)*	360 (78)	0.60 (0.47–0.77)	0.67 (0.53–0.86)
AUC0–48h (nM‐h)	444 (56)	255 (68)	261 (57)	0.57 (0.45–0.74)	0.59 (0.43–0.79)
Cmax,1 (nM)	18.0 (92)	6.85 (83)	9.0 (67)	0.38 (0.23–0.63)	0.50 (0.35–0.73)
t 1/2 (h)	18.9 (40)b	20.5 (36)b	17.8 (46)c	1.08 (0.90–1.31)	0.92 (0.76–1.12)
t max,1 (h)d	1.5 (1–4)	1.5 (0.5–4)	1.5 (1–6)	NA	NA
Metabolite‐to‐parent AUC0–96h ratio	37 (45)	39 (43)	37 (44)	NA	NA

Note: *p < 0.05 or **p < 0.01, paired two‐tailed Student's t‐test on log‐transformed data (baseline vs. green tea). Data are from 16 participants except when there were insufficient datapoints to recover a robust value (^a n = 12, ^b n = 15, ^c n = 14). ^dMedian (range).

Abbreviations: AUC_0–96h, area under the plasma concentration–time curve from 0 to 96 h; AUC_0–48h, area under the plasma concentration–time curve from 0 to 48 h; CI, confidence interval; C_max,1, first maximum plasma concentration; CV, coefficient of variation; NA, not applicable; t _1/2, terminal elimination half‐life; t _max,1, time to reach C_max,1.

Pharmacokinetic results from the follow‐up study generally reproduced results from the main study with respect to raloxifene AUC_0–48h in the absence (4.2–35.1 vs. 4.8–14.4 nM‐h) and presence (2.4–10.1 vs. 3.8–15.3 nM‐h) of acute green tea (Figure 2, Table 2). However, the magnitude of the decrease in the follow‐up study was lower than that in the main study (10% vs. ~40%). Degradation of the clinical green tea product over time was ruled out based on reanalysis of the product, in which catechin content was comparable to the initial analysis (data not shown).

FIGURE 2.

Individual raloxifene area under the plasma concentration–time curve from 0 to 48 h (AUC_0–48h) values for the (a) main and (b) exploratory follow‐up clinical study. The four individuals who participated in both studies are color‐coded. The two new participants in the follow‐up study are indicated in black.

TABLE 2.

Pharmacokinetics of raloxifene and primary glucuronides and raloxifene stool recovery in the absence and presence of green tea in the six healthy adults who participated in the follow‐up clinical study.

Metric	Geometric mean (CV%)		Green tea/baseline ratio (90% CI)
	Baseline	Acute green tea
Raloxifene
AUC0–48h (nM‐h)	8.25 (40)	7.46 (48)	0.90 (0.77–1.03)
Cmax,1 (nM)	0.32 (56)	0.28 (40)	0.87 (0.72–1.07)
t max,1 (h) a	1.25 (1–3)	1 (1–3)	NA
Raloxifene 4′‐glucuronide
AUC0–48h (nM‐h)	1473 (50)	1368 (50)*	0.93 (0.81–1.06)
Cmax,1 (nM)	94.4 (66)	75.0 (51)	0.79 (0.68–0.93)
t max,1 (h) a	1.25 (1–3)	1 (0.5–1)	NA
Metabolite‐to‐parent AUC0–48h ratio	179 (23)	184 (25)	NA
Raloxifene 6‐glucuronide
AUC0–48h (nM‐h)	189 (46)	166 (61)	0.88 (0.73–1.06)
Cmax,1 (nM)	9.39 (33)	6.85 (27)*	0.73 (0.60–0.88)
t max,1 (h) a	1.25 (1–3)	1 (1–3)	NA
Metabolite‐to‐parent AUC0–48h ratio	23 (47)	22 (50)	NA
Stool recovery
Raloxifene dose recovery0–48h (%)	10.1 (115)	10.6 (103)	1.05 (0.45–2.43)

Abbreviations: AUC_0–48h, area under the plasma concentration–time curve from 0 to 48 h; C_max,1, first maximum plasma concentration; CI, confidence interval; CV, coefficient of variation; NA, not applicable; t _max,1, time to reach C_max,1; raloxifene dose recovery_0–48h, percentage of raloxifene recovered in the stool from 0 to 48 h relative to the dose administered.

^a Median (range).

^* p < 0.05, paired two‐tailed Student's t‐test on log‐transformed data (baseline vs. green tea).

In vitro transporter assays

Raloxifene was deemed an unlikely substrate for any of the tested uptake transporters based on the fold‐accumulation remaining below the standard two‐fold cutoff at all tested conditions. Specifically, the fold‐accumulation of raloxifene ranged from 1.14–1.33 (NTCP), 0.72–0.91 (OAT1), 0.75–1.03 (OAT3), 1.04–1.19 (OATP1B1), 1.56–1.70 (OATP1B3), 0.82–1.00 (OATP2B1), 1.13–1.52 (OCT1), and 0.90–1.15 (OCT2) across the two concentrations and two incubation times. Of the uptake and efflux transporters tested, BCRP and OATP1B1 were the most sensitive to inhibition by the green tea extract (normalized to EGCG), with IC₅₀s < 1 μM (Table 3). OCT1, MATE1, MATE2‐K, MRP2, OATP1B3, OATP2B1, and P‐gp were the next most sensitive, with IC₅₀s ranging between 1 and 5 μM.

TABLE 3.

Inhibition of uptake and efflux transporters by the green tea extract.

Transporter	Type	IC50 (μM) a	95% CI
NTCP	Uptake	–	–
OAT1	Uptake	9.2	6.1–14
OAT3	Uptake	–	–
OATP1B1	Uptake	0.6	0.47–0.68
OATP1B3	Uptake	1.9	1.0–3.3
OATP2B1	Uptake	2.7	1.3–5.5
OCT1	Uptake	4.9	3.6–6.7
OCT2	Uptake	–	–
BCRP	Efflux	0.37	0.24–0.57
BSEP	Efflux	12	11–14
MATE1	Efflux	–	–
MATE2‐K	Efflux	–	–
MRP2	Efflux	2.2	1.1–4.6
MRP3	Efflux	10	5.9–18
P‐gp	Efflux	3.3	2.5–4.5

Abbreviations: BSEP, bile salt export pump; BCRP, breast cancer resistant protein; CI, confidence interval; IC₅₀, inhibitor concentration corresponding to 50% of control activity; MATE, multidrug and toxin extrusion protein; MRP, multidrug resistance‐associated protein; NTCP, sodium/taurocholate cotransporting polypeptide; OAT, organic anion transporter; OATP, organic anion transporting polypeptide; OCT, organic cation transporter; P‐gp, P‐glycoprotein.

^a Normalized to EGCG concentration quantified in the extract. –, not determined because ≤50% inhibition was observed at the highest concentration tested during screening.

Gut microbiota composition

One hypothesis for the unexpected decrease in raloxifene AUC was that green tea changed the composition of the gut microbiota. Comparing the total mass of raloxifene recovered in the stool from 0 to 48 h at baseline and in the presence of green tea, the participants clustered into three distinct groups: relative increase, no change, and relative decrease in percentage raloxifene recovery (Table 4). Analysis of microbial taxa abundance and predicted metabolic profiles of microbial genes identified two factors that differed between clusters: the taxon Alistipes and predicted changes in carbohydrate absorption and digestion genes. For all subsequent analyses, microbiota data from Arm 3 (green tea alone) were used because no differences in Alistipes counts and carbohydrate absorption and digestion functional genes, as well as the Firmicutes to Bacteroidetes ratio, were observed within each participant between pre‐ and post‐administration of raloxifene and/or green tea or between arms (Figure S4). Alistipes was approximately two‐fold higher in the participant cluster with a relative decrease in raloxifene recovery compared with the other two clusters (Figure 3a). Carbohydrate absorption and digestion functional genes were lower in the cluster with a relative increase in raloxifene recovery compared with the other two clusters (Figure 3b). Observations from the gut microbiota analysis are summarized (Figure 3c). No correlations or relationships were evident between gut microbiota composition and raloxifene plasma concentrations in the follow‐up clinical study (data not shown). The Shannon and Chao1 indices showed no difference in alpha diversity within each participant between pre‐ and post‐administration of raloxifene/green tea or between arms (data not shown).

TABLE 4.

Percentage of raloxifene dose recovered in the stool.

Participant no.	Raloxifene stool recovery (%)		Percentage change (%)
	Baseline	Acute green tea
8	1.7	0.49	−70 ↓
11	0.27	1.46	440 ↑
12	93.8	21.1	−80 ↓
16	18.0	71.9	300 ↑
17	17.4	19.1	10 ↔
18	69.8	60.3	−14 ↔

Note: Participants clustered as a relative increase (↑), relative decrease (↓), or no change (↔) in raloxifene stool recovery from the raloxifene arm (baseline) to the raloxifene + green tea arm based on a ±15% margin from 0.

FIGURE 3.

Gut microbiota analysis from the exploratory follow‐up clinical study. Data from the green tea arm were analyzed for (a) Alistipes relative abundance and (b) carbohydrate digestion and absorption functional gene count. Data were grouped based on relative change in percentage of raloxifene dose recovered in the stool in the baseline arm versus green tea arm. Black bars, relative increase (Inc); dark grey bars, no change (NC); light grey bars, relative decrease (Dec). (c) Summary of graphs (a) and (b) noting the change in raloxifene stool recovery relative to high or low Alistipes abundance and carbohydrate digestion and absorption genes. Symbols and error bars denote mean and standard deviation, respectively.

DISCUSSION

Green tea is one of the most popular beverages consumed worldwide, with global sales projected to exceed $19.8 billion USD by 2030. ³² Increasing sales, along with the purported health benefits of green tea, raise concerns for potential pharmacokinetic green tea–drug interactions. Although clinical studies have shown minimal interaction risk between green tea and CYP probe drugs, ³³ the effects of green tea on other major drug metabolism pathways, including UGTs, have not been evaluated. Raloxifene was selected as the object drug because it undergoes extensive first‐pass glucuronidation, predominantly in the intestine, after oral administration (oral bioavailability <2%). ³⁴ , ³⁵ Applying a mechanistic static model, the predicted increase in raloxifene AUC in the presence of green tea exceeded the FDA‐recommended cutoff, ¹⁵ prompting the present work.

Unexpectedly, whether green tea was administered acutely or chronically, the geometric mean AUC_0–96h and C_max of raloxifene decreased to below the predefined no effect range (Figure 1, Table 1). The unchanged raloxifene and glucuronide terminal half‐lives, combined with unchanged ratios of glucuronide‐to‐raloxifene AUC_0–96h in the presence of green tea, suggested inhibition of intestinal UGT activity by green tea was not the primary mechanism underlying the observed interaction. The greater decrease in raloxifene C_max,1 relative to AUC_0–96h further suggested that green tea primarily altered processes in the intestine that influence raloxifene absorption, including drug dissolution, passive permeability, active transport, and intestinal microbe‐mediated enterohepatic recycling. The latter two processes were subsequently investigated.

We hypothesized that the decreased raloxifene AUC resulted from green tea inhibiting an uptake transporter(s) in the intestine. Green tea extract inhibited only two of these transporters, specifically OATP2B1, which is expressed on the luminal membrane of enterocytes, ³⁶ and OCT1, the localization of which remains equivocal. ³⁷ However, raloxifene was deemed an unlikely substrate for any of the uptake transporters tested. Green tea extract also inhibited the efflux transporter MRP3, which is expressed on the basolateral membrane of enterocytes, ³⁶ and would be expected to result in a decrease in raloxifene AUC. Although raloxifene was not tested as a substrate for MRP3, raloxifene was shown to interact with MRP3. ³⁸ The concentration of EGCG in 240 mL of brewed tea prepared from the green tea product (coded T21) ¹⁵ was 20‐fold higher than the IC₅₀ determined for MRP3 (10 μM; Table 3), suggesting that concentrations in enterocytes near or exceeding the IC₅₀ are achievable. Chronic administration of green tea could decrease raloxifene AUC by inducing uptake transporters expressed on the basolateral membrane or efflux transporters expressed on the luminal membrane (e.g., P‐gp). However, induction of transporters other than P‐gp is uncertain. ³⁹ The similar effects observed with chronic and acute green tea administration further suggest an induction mechanism is unlikely. Besides MRP3, other untested or unknown intestinal transporters, as well as untested transporter mechanisms (e.g., stimulation), ⁴⁰ could contribute to the decreased raloxifene AUC.

Another hypothesis for the unexpected decrease in raloxifene AUC was that green tea interrupted the gut microbe‐mediated deconjugation of raloxifene glucuronides during the enterohepatic recycling of raloxifene. ⁴¹ , ⁴² Accordingly, a follow‐up small clinical study was conducted to explore the influence of the gut microbiota on raloxifene pharmacokinetics. Only acute green tea administration was considered because acute and chronic administration showed comparable effects on raloxifene pharmacokinetics. Consistent with results from the main study, green tea decreased raloxifene AUC_0–48h in four of the six participants (Figure 2, Table 3), but the magnitude of decrease was less compared with the main study (10% vs. ~40%). Degradation of the clinical green tea product over time was ruled out based on reanalysis of the product, in which the catechin content was approximately equivalent to the initial analysis. In addition, raloxifene AUC_0‐48h overlapped between the main and follow‐up studies (baseline: 4.2–35.1 vs. 4.8–14.4 nM‐h; green tea: 2.4–10.1 versus 3.8–15.3 nM‐h). The collective observations suggested that the lower magnitude of raloxifene AUC observed in the follow‐up study was due to population variability and the smaller sample size.

Gut microbiota composition varied greatly among the six participants but was unchanged across the different arms and the pre‐ and post‐raloxifene/green tea administration time points within each participant (Figure S3), indicating that raloxifene and green tea did not influence microbiota composition. Raloxifene stool recovery was markedly variable among the participants, ranging from <1% to >90% of the dose, and green tea did not change recovery in a specific direction. Although both raloxifene glucuronides were measured in stool samples, their recovery accounted for <1% relative to raloxifene in all participants (data not shown). Most of the raloxifene glucuronides were presumably hydrolyzed back to raloxifene via beta‐glucuronidase in the distal region of the gut. ⁴³ , ⁴⁴ , ⁴⁵ Nevertheless, participants clustered into one of three distinct categories with respect to the raloxifene arm relative to the raloxifene + green tea arm: increase, no change, and decrease in stool recovery (Figure 3).

Closer inspection of these clusters related to the microbiota identified the taxa Alistipes and carbohydrate absorption and digestion functional genes as factors that could have contributed to the effects of green tea on raloxifene stool recovery. Alistipes may be a marker for beta‐glucuronidase, ⁴⁶ and green tea inhibits bacterial beta‐glucuronidase. ⁴⁷ Accordingly, an individual high in Alistipes is postulated to have higher microbial beta‐glucuronidase activity and a greater extent of raloxifene glucuronide deconjugation, thus extensive enterohepatic recycling of raloxifene (i.e., higher raloxifene AUC) and lower stool recovery relative to an individual with low Alistipes. That is, individuals high in Alistipes would be more sensitive to the inhibitory effects of green tea. Consistent with this working hypothesis, the cluster high in Alistipes showed a decrease in raloxifene stool recovery in the presence of green tea (Figure 3). A larger sample size is needed to test this hypothesis. The role of carbohydrate digestion and absorption gene counts is unclear but may counteract Alistipes, as suggested by the lack of change in raloxifene stool recovery in the cluster with low Alistipes and high gene count. Despite these observations, the lack of correlation between stool recovery and AUC suggests that the preexisting gut microbiota did not contribute to the green tea–raloxifene interaction.

In summary, a well‐characterized green tea product decreased the AUC of the intestinal UGT substrate raloxifene to below the predefined no effect range. This observation was opposite to the in vitro–in vivo prediction, indicating inhibition of intestinal UGT by green tea was not the primary mechanism. Because the pharmacokinetic data indicated that the interaction occurred in the intestine, other potential mechanisms of green tea included changes in raloxifene active transport and interruption of microbe‐mediated raloxifene reabsorption. However, results from the in vitro transport studies and follow‐up clinical study did not identify a definitive mechanism. Future studies assessing the effects of green tea on the solubility, dissolution, and permeability of raloxifene will help elucidate the underlying mechanism(s). A more elaborate mechanistic static model that incorporates one or more of these intestinal processes, including those addressed in the current work, is needed to improve the predictability of potential intestinal UGT‐mediated interactions. Development of physiologically based pharmacokinetic models may provide a more mechanistic approach to the modeling of this complex interaction. Practically, health care providers may consider informing their patients about the possible risk of co‐consuming green tea with raloxifene.

AUTHOR CONTRIBUTIONS

J.D.C, S.M.J., D.‐D.T., T.O.K., and M.F.P. wrote the manuscript. S.M.J., D.‐D.T., T.O.K., K.E.T., J.S.M., D.D.S., and M.F.P. designed the research. S.M.J., D.‐D.T., T.O.K., R.S.T., A.M.‐P., M.H., M.E.L., J.R.W., and N.B.C. performed the research. J.D.C., S.M.J., D.‐D.T., T.O.K., R.S.T., and M.F.P. analyzed the data. N.B.C. contributed new reagents.

FUNDING INFORMATION

This work was supported by the National Institutes of Health National Center for Complementary and Integrative Health, specifically the Center of Excellence for Natural Product Drug Interaction Research (U54 AT008909).

CONFLICT OF INTEREST STATEMENT

The authors declared no competing interest for this work.

Supporting information

Figure S1

Figure S2

Figure S3

Figure S4

Table S1

Table S2

Table S3

Clarke JD, Judson SM, Tian D‐D, et al. Co‐consuming green tea with raloxifene decreases raloxifene systemic exposure in healthy adult participants. Clin Transl Sci. 2023;16:1779‐1790. doi: 10.1111/cts.13578