Untargeted Metabolomics Reveal New Endogenous Plasma Biomarkers Associated with CYP3A Inhibition in Humans

Abstract

Reliable endogenous biomarkers for assessing CYP3A activity and clinical drug–drug interaction (DDI) risk remain limited. We characterized global metabolomic changes following itraconazole treatment in a fixed‐sequence DDI study (11 healthy volunteers) to identify novel CYP3A endogenous biomarkers. Untargeted metabolomic analysis identified 58 plasma metabolites that were significantly modulated (changed by > 1.5‐fold or < 0.67‐fold, P < 0.05), highlighting the dynamic and systemic impact of CYP3A inhibition on the human metabolome. Bile acids represented the most prominently affected analytes, whereas changes were also observed in endocannabinoids, xanthine metabolism, and food‐derived metabolites, revealing both on‐target and potential off‐target effects of itraconazole. For example, plasma levels of endocannabinoidome metabolites such as linoleoyl ethanolamide and N‐linoleoylglycine increased by 2.4‐ to 6.0‐fold (P < 0.05), while bile acids including glycoursodeoxycholate (GUDCA) and glycohyocholate (GHCA) decreased by 2.6‐ to 12.5‐fold (P < 0.01) following itraconazole treatment. Targeted quantification was subsequently performed for the bile acids identified from untargeted analysis and the emerging CYP3A biomarkers, 1β‐hydroxy‐deoxycholic acid (1βOH‐DCA) and its glycine and taurine conjugates. Notably, GHCA, GUDCA and 1βOH‐DCA conjugates exhibited pronounced reduction in plasma exposure (~10‐fold, P < 0.001) in the itraconazole treatment group. GHCA exhibited lower inter‐individual variability than others. In vitro characterization demonstrated CYP3A4‐selective hydroxylation of glycochenodeoxycholate to form GHCA, supporting its mechanistic linkage to CYP3A4 activity. Collectively, these findings expand the repertoire of sensitive and mechanistically supported endogenous biomarkers of CYP3A activity. Further evaluation of the newly identified GHCA and emerging bile acid metabolites across diverse CYP3A inhibitors can strengthen their translational utility in DDI risk assessment and drug development.

Study Highlights.

WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

Endogenous biomarkers for transporter‐mediated DDIs are effectively used in clinical development. However, current endogenous biomarkers for CYP3A activity have limitations.

WHAT QUESTION DID THIS STUDY ADDRESS?

Can untargeted metabolomics reveal sensitive and reliable endogenous plasma biomarkers of CYP3A activity in humans?

WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

This study expands the repertoire of endogenous CYP3A biomarkers by identifying bile acids, particularly glycohyocholate (GHCA), glycoursodeoxycholate (GUDCA), and 1β‐hydroxy‐deoxycholate (1βOH‐DCA) conjugates, as highly sensitive markers of CYP3A inhibition. It demonstrates how untargeted metabolomics can uncover dynamic, pathway‐relevant biomarkers beyond traditional probes, providing new insight into the systemic metabolic effects of CYP3A modulation.

HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?

This study supports the use of novel, sensitive endogenous biomarkers to assess pharmacokinetic variability associated with CYP3A modulation. Incorporation of such biomarkers can improve early DDI risk assessment, reduce reliance on dedicated DDI studies using probe drugs, and enhance translational confidence when predicting CYP3A‐mediated interactions across development and clinical settings.

Cytochrome P450 3A (CYP3A) enzymes are responsible for the metabolism of a large number of drugs; therefore, accurate assessment of CYP3A activity is critical for predicting pharmacokinetic (PK) variability, evaluating drug–drug interaction (DDI) risk, and guiding clinical dosing strategies. ¹ , ² Reliable endogenous biomarkers of CYP3A activity can reduce reliance on dedicated DDIs studies that require dosing of probe substrate drugs (e.g., midazolam), and improve translational confidence across drug development programs.

Two endogenous biomarkers of CYP3A activity have been most widely used in clinical development–plasma 4β‐hydroxycholesterol (4β‐OHC) and the urinary 6β‐hydroxycortisol/cortisol ratio. ³ , ⁴ Although 4β‐OHC is formed via CYP3A‐mediated cholesterol oxidation, its limited sensitivity to CYP3A inhibition and long half‐life (~17 days) generally restrict its utility to the assessment of CYP3A induction. Clinically, 4β‐OHC has been applied for induction detection during phase I dose escalation. ⁵ Conversely, the urinary 6β‐hydroxycortisol/cortisol ratio can reflect both induction and inhibition; however, inadequate sensitivity and specificity, high intra‐ and inter‐individual variability, and reliance on urine collection limit its practical utility. ⁶ , ⁷

A bile acid metabolite, 1β‐hydroxy‐deoxycholic acid (1βOH‐DCA), was identified as a CYP3A biomarker based on the clinical observations reported by Bodin et al. (2005). ⁸ Increased urinary excretion of 1βOH‐DCA was first reported in patients treated with the CYP3A inducer carbamazepine, which prompted mechanistic follow‐up studies confirming its formation from deoxycholic acid via CYP3A in human liver microsomes and recombinant systems. ⁸ Subsequent work published in 2016, demonstrated that CYP3A4, CYP3A5, and CYP3A7 selectively catalyze this biotransformation, establishing 1βOH‐DCA as a potential biomarker of CYP3A activity. ⁹ More recent studies (reported in 2024 and 2025) evaluated 1βOH‐DCA and its glycine and taurine conjugates in plasma, where they exhibit improved dynamic range and sensitivity to both CYP3A induction and inhibition in clinical DDI studies. ¹⁰ , ¹¹ Although these candidates show enhanced sensitivity, they remain early in their clinical validation. Plasma‐derived extracellular vesicles (EVs) have emerged as an alternative approach for assessing tissue‐specific protein expression including CYP3A4. ¹² However, EV‐based approaches require specialized sample processing and have not yet achieved the level of clinical validation, standardization, and operational simplicity needed for routine implementation in clinical studies.

In this study, we employed an untargeted metabolomics strategy using plasma samples from a clinical DDI study involving itraconazole, a recommended strong CYP3A inhibitor. ¹³ This untargeted high‐resolution mass spectrometry‐based approach allowed for a comprehensive and unbiased detection of metabolic perturbations caused by CYP3A inhibition, facilitating the identification of previously unrecognized endogenous biomarkers and novel metabolic pathways influenced by CYP3A. Prior efforts to identify endogenous CYP3A biomarkers have largely relied on predefined metabolites (e.g., 1βOH‐DCA, ¹⁰ 4β‐hydroxycholesterol, ¹⁴ targeted‐steroid panel ¹⁵ ) or untargeted analysis of urine. ¹⁶ The overall experimental design for this study is depicted in Figure 1. Following the initial discovery phase, we quantified newly identified candidates from this untargeted analysis along with selected previously reported bile acid biomarkers, 1βOH‐DCA and its glycine and taurine conjugates, in the clinical samples. Furthermore, follow‐up metabolic phenotyping assays were conducted using human liver microsomes and recombinant P450 enzymes to investigate the CYP3A selectivity for certain newly identified candidates. Collectively, this work aims to expand the repertoire of sensitive, mechanistically supported endogenous plasma biomarkers of CYP3A activity and to provide a resource for validating novel biomarker candidates across multiple metabolic pathways.

Figure 1.

Schematic overview detailing the clinical study design, untargeted metabolomics, targeted validation, and metabolic profiling and phenotyping. This approach was used to identify plasma biomarkers of CYP3A inhibition following itraconazole administration in healthy volunteers and to follow‐up on a newly discovered bile acid. This figure was created using BioRender.

MATERIALS AND METHODS

Detailed descriptions of selected methods are provided in the Supplementary Materials and Methods , with core methods/procedures outlined below.

Clinical DDI study

The effect of itraconazole (200 mg QD for 7 days), a strong CYP3A4 inhibitor, on the PKs of an investigational drug was evaluated in a phase I, open‐label, two‐period, fixed‐sequence study in healthy adults (Figure S1 ). Full details are provided in the Supplementary Materials and Methods .

Untargeted LC–MS metabolomic analysis of pooled plasma samples

An untargeted LC–MS metabolomic analysis was conducted using PK samples from the clinical DDI study. Plasma samples from 11 participants were included in this analysis. For metabolomic profiling, samples were selected to capture baseline, maximal CYP3A inhibition, and recovery phases, including pre‐dose (0 hr), pooled post‐dose plasma representing 1–4 hr (comprising 50 μL aliquots from 1, 1.5, 2, and 4 hr), and a late post‐dose sample at 96 hr (Figure 1). The pooled 1–4 hr samples were intended to reflect the period of highest itraconazole exposure and maximal CYP3A inhibition. Untargeted metabolomic profiling was performed by Metabolon Inc (Morrisville, NC) using the Global Discovery Panel. Details of instrument configuration, chromatographic conditions, mass spectrometry acquisition, and metabolite detection are provided in the Supplementary Materials and Methods .

For biomarker identification, metabolomic profiles were analyzed from samples collected during two study periods: no‐itraconazole (control) and itraconazole (inhibitor) conditions. Untargeted metabolomic data were log‐transformed prior to analysis. A paired t‐test was performed to compare the inhibitor condition with the corresponding control condition at each sampling phase. Putative CYP3A biomarkers were defined as metabolites demonstrating a > 1.5‐fold increase or decrease in abundance in the inhibitor vs. control condition, with a nominal P‐value < 0.05 observed in at least two sampling phases.

Targeted quantification and data analysis of the selected bile acid metabolites

Plasma bile acid metabolite concentrations were quantified using the LC–MS/MS method described in the Supplementary Materials and Methods . Concentrations below the limit of quantification (BLQ) were flagged in the dataset. For non‐compartmental analysis, BLQ values were set to zero for AUC integration but excluded from counts of quantifiable samples. Concentration–time data were analyzed over the itraconazole dosing interval of 0 to 24 hours post‐dose. All calculations, including non‐compartmental analysis (NCA), were conducted in R (version 2026.01.0), utilizing the PKNCA package. Below‐limit‐of‐quantification (BLQ) concentrations were treated as zero. The area under the concentration–time curve from 0 to 24 hours (AUC0‐24 h) was determined using the linear‐up/log‐down trapezoidal method. AUC0‐24 h values were summarized using geometric means and geometric coefficients of variation (GeoCV%), calculated from log‐transformed data. Pairwise comparisons between treatments (Inhibitor vs. Control) were conducted on log‐transformed AUC0‐24 h values using a paired approach. Geometric mean ratios (GMR; Inhibitor/Control) and their corresponding 90% confidence intervals (CI) were derived by exponentiation of the mean difference and confidence limits on the log scale.

In vitro characterization of glycochenodeoxycholate 6α‐hydroxylation

Enzyme kinetics of glycochenodeoxycholate (GCDA) 6α‐hydroxylation to GHCA were characterized in human liver microsomes across a range of substrate concentrations, and reaction phenotyping was conducted using a panel of 15 recombinantly expressed human P450 isoforms (CYP1A2, 1B1, 2A6, 2B6, 2C8, 2C9, 2C18, 2C19, 2D6, 2E1, 2 J2, 3A4, 3A5, 3A7, and 4F2). GHCA formation was quantified by LC–MS/MS, and recombinant P450 activities were scaled to relative hepatic abundance to estimate fractional isoform contributions; full assay conditions, kinetic models, and bioanalytical parameters are provided in the Supplementary Materials and Methods .

RESULTS

Identification of putative CYP3A biomarkers

Untargeted metabolomics of plasma samples from the clinical DDI study identified a total of 1313 metabolites; of which, 1086 (83%) had known structures (Table S1 ). Putative CYP3A biomarkers were selected based on predefined criteria, including consistent directionality of fold changes (all less than or greater than 1) across three distinct sampling time points (0 hr, pooled 1–4 hr, and 96 hr) under control and inhibition conditions. Additionally, a paired t‐test (P‐value < 0.05) with a fold‐change ratio of < 0.67 or > 1.5 was required in at least two of the three time points. The application of these selection criteria resulted in the identification of 85 putative plasma biomarkers of CYP3A inhibition, of which 59 were structurally annotated (Table 1 ). These metabolites spanned several biochemical classes, with prominent representation from food/plant‐derived components (12 metabolites), secondary bile acid metabolism (7), xanthine metabolism (7), and the endocannabinoid sub‐pathway (6) (Table 1 ). Saccharin, an inert excipient present in the itraconazole formulation, was excluded from the pool of biologically meaningful biomarkers despite exhibiting differences between dosed and control (Table 1 ). Similarly, the parent drug itraconazole was excluded as it represents exogenous components of the administered treatment rather than endogenous biochemicals. Notably, itraconazole itself did not meet the predefined biomarker selection criteria, as its fold‐change at 96 hr (1.49‐fold) was below the required > 1.5‐fold threshold. This exclusion refined the list to 58 structurally annotated metabolites of relevance to CYP3A inhibition (Figure 2 ).

Table 1.

Putative CYP3A biomarkers identified in this study. This table summarizes statistically significant changes in 59 metabolites following co‐administration of itraconazole (INH) compared with control (CTRL). Values represent fold‐change ratios (INH/CTRL) at 0 hour, pooled 1–4 hour, and 96 hour

Sub pathway	Metabolite name	Fold‐change (INH/CTRL ratio)			Matched paired t‐test, INH/CTRL (0 hour)		Matched paired t‐test, INH/CTRL (1–4 hour)		Matched paired t‐test, INH/CTRL (96 hour)
		0 hour	1–4 hour	96 hour	P	q‐value	P	q‐value	P	q‐value
Androgenic steroids	11‐ketoetiocholanolone glucuronide	0.53	0.66	0.56	0.021	0.1064	0.091	0.2222	0.0303	0.0643
Benzoate metabolism	4‐methylcatechol sulfate	0.41	0.49	0.5	0.0032	0.0516	0.0124	0.0911	0.1556	0.1788
Benzoate metabolism	4‐vinylphenol sulfate	0.51	0.52	0.77	0.0384	0.1434	0.0373	0.1457	0.5922	0.3915
Chemical	2‐naphthol sulfate	0.49	0.47	0.57	0.0011	0.0327	0.0006	0.0172	0.0059	0.0218
Diacylglycerol	Palmitoyl‐arachidonoyl‐glycerol (16:0/20:4) [2]*	1.06	1.9	1.97	0.413	0.4579	0.0337	0.1422	0.0039	0.017
Endocannabinoid	Palmitoyl ethanolamide (16:0)	1.41	2	1.72	0.0199	0.1035	0.001	0.0221	0.01	0.0311
Endocannabinoid	Arachidonoyl ethanolamide (20:4)	1.52	1.98	2.36	0.0098	0.0774	0.0038	0.0423	5.8e‐05	0.0016
Endocannabinoid	Oleoyl ethanolamide (18:1)	1.85	3.23	3.29	0.0046	0.0594	0.0004	0.0118	0.0004	0.0045
Endocannabinoid	N‐palmitoyltaurine	2.16	2.52	1.59	0.0066	0.0648	0.0028	0.0368	0.1662	0.1863
Endocannabinoid	Linoleoyl ethanolamide (18:2)	2.36	4.57	4.1	0.0067	0.0648	0.0002	0.0075	0.0004	0.0044
Endocannabinoid	N‐oleoyltaurine	2.43	3.22	3.71	0.0005	0.021	0.0001	0.0054	0.0004	0.0046
Fatty acid metabolism (Acyl Carnitine, Hydroxy)	(S)‐3‐hydroxybutyrylcarnitine	0.66	0.61	1.15	0.0019	0.038	0.0147	0.0981	0.3348	0.2787
Fatty acid metabolism (Acyl Carnitine, Medium Chain)	Cis‐3,4‐methyleneheptanoylcarnitine	0.6	0.5	0.68	0.0002	0.0128	3.1e‐06	0.0003	0.005	0.0196
Fatty acid metabolism (Acyl Carnitine, Polyunsaturated)	Docosahexaenoylcarnitine (C22:6)*	1.53	1.65	1.96	0.006	0.0631	0.0032	0.0391	0.0005	0.0048
Fatty acid metabolism (Acyl Glycine)	N‐palmitoylglycine	1.97	3.19	3.15	0.0057	0.0631	2.6e‐05	0.0016	0.0003	0.0039
Fatty acid metabolism (Acyl Glycine)	N‐linoleoylglycine	2.47	5.95	4.58	0.0192	0.101	9.8e‐07	0.0002	0.0002	0.003
Fatty acid, branched	Cis‐3,4‐methyleneheptanoate	0.52	0.46	0.61	0.0008	0.0278	3.7e‐06	0.0003	5.2e‐05	0.0016
Fatty acid, dicarboxylate	Glutarate (C5‐DC)	2.59	2.15	1.82	8.5e‐06	0.0037	0.0001	0.0054	0.001	0.0073
Fatty acid, monohydroxy	9‐hydroxystearate	0.87	1.54	1.61	0.3175	0.4191	0.0359	0.1425	0.0155	0.0416
Food component/plant	Dihydrocaffeate sulfate (2)	0.29	0.29	0.74	0.0118	0.0831	0.0073	0.071	0.0357	0.0722
Food component/plant	Ethyl beta‐glucopyranoside	0.34	0.34	1.08	0.0276	0.1219	0.0384	0.1466	0.4781	0.3462
Food component/plant	S‐allylcysteine	0.5	0.5	0.67	0.0022	0.0412	0.0042	0.0464	0.0247	0.0559
Food component/plant	Alliin	0.51	0.47	0.75	0.0163	0.0942	0.0134	0.0948	0.0379	0.0751
Food component/plant	N‐acetylalliin	0.55	0.48	0.67	0.0031	0.0505	0.0009	0.0221	0.012	0.0346
Food component/plant	Methyl glucopyranoside (alpha + beta)	0.58	0.58	0.88	0.0023	0.0426	0.0326	0.1415	0.1826	0.1938
Food component/plant	Glucuronide of piperine metabolite C17H21NO3 (4)*	1.65	1.64	1.24	0.0019	0.038	0.0023	0.0345	0.0613	0.0988
Food component/plant	4‐allylphenol sulfate	1.96	2.06	1.38	0.0002	0.0139	9.2e‐05	0.0054	0.0199	0.0487
Food component/plant	Dihydroferulic acid sulfate	2.02	2.22	1.36	0.0234	0.1139	0.033	0.1418	0.0981	0.1333
Food component/plant	2‐piperidinone	2.39	2.75	1.77	2.3e‐05	0.0041	2.1e‐05	0.0014	0.0001	0.0023
Food component/plant	Dihydroferulate	2.4	2.9	1.44	0.0062	0.0632	0.0143	0.0977	0.5331	0.3673
Food component/plant	Saccharin	3.6	56.93	2.24	2.3e‐05	0.0041	2.4e‐09	1.7e‐06	0.0041	0.0174
Glycerolipid metabolism	Glycerol 3‐phosphate	1.52	1.56	1.13	0.0012	0.0331	0.0068	0.0676	0.1113	0.1452
Leucine, isoleucine and valine metabolism	Isobutyrylcarnitine (C4)	0.66	0.56	0.51	0.0007	0.025	0.0001	0.0059	2.1e‐05	0.0013
Leucine, isoleucine and valine metabolism	Isovalerate (i5:0)	0.98	1.68	2.24	0.9647	0.6545	0.0036	0.0416	8.0e‐06	0.0006
Lysine metabolism	N‐acetyl‐cadaverine	1.53	1.57	1.65	0.0098	0.0774	0.0196	0.1121	0.0663	0.1034
Methionine, cysteine, SAM and taurine metabolism	N‐methyltaurine	0.48	0.48	0.42	0.0606	0.1809	0.0361	0.1425	0.0005	0.0048
Methionine, cysteine, SAM and taurine metabolism	S‐methylmethionine	0.64	0.72	0.47	0.0103	0.0774	0.0651	0.1955	0.0107	0.0326
Polyamine metabolism	Acisoga	0.65	0.81	0.63	6.7e‐05	0.0084	0.0675	0.1994	0.0006	0.0051
Primary bile acid metabolism	Glyco‐beta‐muricholate**	0.28	0.48	0.32	0.0011	0.0327	0.0054	0.0566	0.0028	0.0138
Primary bile acid metabolism	Glycocholate	1.14	3.84	2.23	0.1253	0.2626	0.001	0.0221	0.0085	0.0278
Purine metabolism, (Hypo)xanthine/Inosine containing	Inosine 5′‐monophosphate (IMP)	1.72	1.63	2.44	0.0131	0.0861	0.0168	0.1041	0.0053	0.0201
Purine metabolism, (Hypo)xanthine/Inosine containing	Inosine	1.83	1.19	1.74	0.0017	0.038	0.1855	0.3222	0.0492	0.0886
Secondary bile acid metabolism	Glycoursodeoxycholate	0.11	0.11	0.08	8.4e‐05	0.0085	0.0001	0.0054	0.0002	0.0029
Secondary bile acid metabolism	Isoursodeoxycholate	0.17	0.19	0.22	0.0078	0.0704	0.0186	0.1099	0.0684	0.1057
Secondary bile acid metabolism	Ursodeoxycholate	0.23	0.32	0.19	0.014	0.0882	0.0909	0.2222	0.0283	0.0607
Secondary bile acid metabolism	Glycohyocholate	0.28	0.39	0.24	0.0005	0.021	0.0004	0.0117	0.0053	0.0201
Secondary bile acid metabolism	Lithocholate sulfate (1)	0.49	0.48	0.38	0.0173	0.0963	0.0145	0.0977	0.0066	0.0237
Secondary bile acid metabolism	Glycolithocholate sulfate*	0.5	0.31	0.57	0.0653	0.1845	0.0111	0.0895	0.0354	0.0719
Secondary bile acid metabolism	Taurolithocholate 3‐sulfate	0.5	0.36	0.54	0.0486	0.1627	0.0113	0.0896	0.0377	0.0751
Short chain fatty acid	Butyrate/isobutyrate (4:0)	1.1	2.17	3.76	0.4362	0.4705	0.0069	0.0685	2.8e‐07	4.6e‐05
Tocopherol metabolism	Gamma‐CEHC	0.51	0.43	0.61	0.012	0.0839	0.0012	0.0225	0.0013	0.0087
Tyrosine metabolism	Phenol sulfate	1.99	1.95	1.35	0.0216	0.1084	0.0161	0.1028	0.0464	0.0856
Xanthine metabolism	Theobromine	0.1	0.15	0.75	0.0024	0.0426	0.0092	0.0818	0.4226	0.3181
Xanthine metabolism	Theophylline	0.13	0.15	0.87	0.0027	0.046	0.0032	0.0391	0.3858	0.3023
Xanthine metabolism	5‐acetylamino‐6‐amino‐3‐methyluracil	0.15	0.19	0.76	0.001	0.0324	0.0006	0.0182	0.2265	0.2203
Xanthine metabolism	7‐methylxanthine	0.2	0.3	1.01	0.0004	0.021	0.0185	0.1099	0.8593	0.471
Xanthine metabolism	1,7‐dimethylurate	0.25	0.33	0.95	0.0081	0.071	0.0339	0.1424	0.1143	0.1475
Xanthine metabolism	3‐methylxanthine	0.31	0.37	1	0.0141	0.0882	0.0127	0.0918
Xanthine metabolism	1‐methylxanthine	0.32	0.35	0.93	0.0338	0.1352	0.0314	0.1388	0.1914	0.1975

Note: Cell shading reflects statistical significance from paired t‐tests: red and dark blue indicate P ≤ 0.05, with red denoting significantly higher and dark blue denoting significantly lower mean values, respectively. Light red and light blue indicate 0.05 < P < 0.10, representing trends toward higher or lower mean values, respectively, whereas white cells indicate non‐significant changes. Corresponding P‐values and q‐values are reported for each time point. See Table S1 for additional details on these metabolites. An asterisk (*) next to the metabolite name indicates a compound that has not been confirmed using an authentic standard but for which Metabolon Inc. is confident in its identity. A double asterisk (**) indicates a compound for which an authentic standard is not available, but Metabolon Inc. is reasonably confident in its identity based on the available information.

Figure 2.

Heatmap of metabolite fold‐change. This heatmap visually represents the log₂ fold‐change in the abundance of 58 structurally annotated metabolites following itraconazole treatment. The data is presented across three distinct sampling windows: 0 hr, pooled 1–4 hr, and 96 hr. Color scale: Red hues indicate an increased abundance of the metabolite (positive log₂ fold‐change). Blue hues indicate a decreased abundance of the metabolite (negative log₂ fold‐change). White/gray indicates no significant change (0 log₂ fold‐change). The intensity of the color reflects the magnitude of the change, ranging from −3 (strong decrease) to 2 (strong increase) in log₂ fold‐change values. An asterisk (*) next to a metabolite name indicates a compound that has not been confirmed using an authentic reference standard; however, Metabolon is confident in the proposed identification based on available data. A double asterisk (**) indicates a compound for which an authentic reference standard is not available; however, Metabolon is reasonably confident. Of note, during follow‐up work with authentic reference standards obtained after the untargeted analysis, we determined that the peak annotated by Metabolon as glyco‐beta‐muricholoate** co‐elutes with 1β‐hydroxy‐glycodeoxycholate (1βOH‐GDCA) and most likely corresponds to the latter; see Discussion for details.

Marked reductions were observed in several bile acids and xanthine‐related metabolites (Table 1 , Figure 2 ). The heatmap presented in Figure 2 visually corroborates the consistent directionality of these changes across the three designated sampling windows. Examples of bile acids that decreased upon itraconazole treatment included glycoursodeoxycholate (GUDCA), ursodeoxycholate, isoursodeoxycholate, glyco‐β‐muricholate (GβMCA), glycohyocholate (GHCA), glycolithocholate sulfate, taurolithocholate‐3‐sulfate, and lithocholate sulfate. Xanthine derivatives, including theobromine, theophylline, 1,7‐dimethylurate, 7‐methylxanthine, 1‐methylxanthine, and 3‐methylxanthine, also showed significant decreases, although only at the 0 hr and pooled 1–4 hr time points. In contrast, robust increases in plasma levels (multi‐fold) were observed for several endocannabinoids. Representative examples included N‐linoleoylglycine, linoleoyl ethanolamide, N‐oleoyltaurine, N‐palmitoylglycine, oleoyl ethanolamide, and palmitoyl ethanolamide. Metabolites classified as food‐ or plant‐derived components exhibited more variable responses, with some showing increases (e.g., 2‐piperidinone, 4‐allylphenol sulfate, and dihydroferulate) and others decreases (e.g., dihydrocaffeate sulfate and ethyl β‐glucopyranoside). However, the magnitude of these changes was generally smaller than those observed for metabolites in the bile acid, xanthine, and endocannabinoid‐related pathways (Figure 2 ). Among the 58 significantly altered metabolites, some have previously been reported as CYP3A4 substrates or CYP3A4‐mediated metabolites, including arachidonoyl ethanolamide ¹⁷ and glycohyocholate. ¹⁸ Arachidonate (20:4n6) and several complex lipids containing an arachidonoyl acyl‐chain were also identified (Table S1 ). However, classical downstream oxidized arachidonic acid metabolites, such as prostaglandins, thromboxanes, leukotrienes, hydroxyeicosatetraenoic acids, or epoxyeicosatrienoic acids are not detected among the reported metabolites. A comprehensive summary of all 1313 identified metabolites, including fold changes and statistical significance (P‐values and q‐values) across all three time points, is provided in Table S1 .

Targeted quantification of plasma PKs for selected bile acid metabolites

To further validate the findings, we evaluated three bile acid metabolites from our untargeted metabolomics analysis (GUDCA, GβMCA, and GHCA) across full sampling time points. We additionally included three previously evaluated bile acid metabolites ¹⁰ (1βOH‐DCA, 1βOH‐GDCA, and 1βOH‐TDCA) in this targeted multiplex LC–MS/MS based quantification. GUDCA, GβMCA, and GHCA were selected based on the rank‐ordering of inhibitor‐to‐control ratio and statistical significance (lowest P‐value) (Table 1 , Figure 2 ). Among these six analytes, GβMCA was not detectable even with a limit of detection of 5 pg/mL. Table 2 summarizes the geometric mean AUC_0–24h for the five analytes in control and inhibitor groups. One participant was excluded from the statistical analysis for 1βOH‐DCA as all PK samples post‐itraconazole administration were BLQ. Across all measurable metabolites, itraconazole treatment resulted in a marked reduction in their systemic exposure (geometric mean AUC_0‐24 ratio, 0.0897–0.105). In contrast to the higher inter‐individual variability observed for several metabolites, particularly under inhibitor condition (geometric CV ranging from 48 to 105% and 69–218% in control and inhibitor groups), GHCA showed relatively low variability (geometric CV, 50% and 69% in control and inhibitor groups) (Table 2 ). Nevertheless, the 90% confidence intervals of geometric mean AUC ratios for all analytes were well below unity. A similar result was observed when the data were analyzed for AUC_0–96 (Table S2 , Figure S2 ). The magnitude of change observed for 1βOH‐DCA and its conjugates is consistent with prior findings. ¹⁰ , ¹¹ Concentration–time profiles of the five individual bile acid metabolites and the total unconjugated and conjugated 1βOH‐DCA are shown in Figure 3 . Itraconazole was administered once daily for 7 days beginning on Day 5 (Figure S1 ). To visualize the effect of repeated itraconazole administration on baseline metabolite levels, the plasma concentrations of four bile acids were plotted for each individual subject (Figure S3 ). The concentrations were measured at baseline (Day 1 and Day 5), after three once‐daily doses of itraconazole (Day 8), and after seven once‐daily doses (Day 12). A clear and statistically significant reduction in the plasma concentrations of 1βOH‐GDCA, 1βOH‐TDCA, GUDCA, and GHCA was observed after first three itraconazole doses. This reduction in plasma metabolites was even more pronounced following seven doses. The individual data shown in Figure S3 confirm that this reduction was consistent across nearly all study participants. Exploratory analyses of baseline AUC_0‐24 did not reveal an appreciable influence of age or sex on inter‐individual variability in plasma metabolite exposure; however, these findings should be interpreted cautiously because the study was not powered to assess demographic factors, particularly given the limited number of female participants (n = 3).

Table 2.

Summary of geometric mean AUC_0‐24h for five individual metabolites following no‐itraconazole (Control) and itraconazole (Inhibitor) treatments

Metabolite	N (C/I)a	Control Geometric Mean (gCV%) AUC0‐24h [ng*h/mL]	Inhibitor Geometric Mean (gCV%) AUC0‐24h [ng*h/mL]	Geometric Mean Ratio (Inhibitor/Control) (90% CI)b
1βOH‐DCA	11/10	8.43 (48.2)	0.783 (146)	0.0972 (0.0556–0.170)c ,*
1βOH‐GDCA	11/11	215 (83.9)	21.3 (108)	0.0992 (0.0634–0.155)*
1βOH‐TDCA	11/11	83.3 (104)	8.72 (218)	0.105 (0.0543–0.202)*
1βOH‐DCA + 1βOH‐GDCA + 1βOH‐TDCA	11/11	311 (84.7)	31.6 (129)	0.102 (0.0622–0.166)*
GUDCA	11/11	1060 (105)	97.1 (127)	0.0914 (0.0514–0.162)*
GHCA	11/11	221 (50)	19.9 (68.8)	0.0897 (0.0656–0.123)*

Note: Paired t‐test P‐value. *P < 0.0001.

^a Number of participants (N) and geometric CV (%) are reported for Control (C) and inhibitor (I).

^b Geometric mean ratio represents the Inhibitor‐to‐Control ratio of AUC_0‐24h. Geometric means and corresponding 90% confidence intervals (CI) are reported for each treatment.

^c Geometric mean ratio based on 10 participants. Detailed LC‐MS/MS analytical conditions, including chromatographic gradients and mass spectrometric parameters for quantification of these metabolites, are provided in Table S4–S7 within the Supplementary Materials and Methods.

Figure 3.

Effects of itraconazole on plasma bile acid exposure. Plasma concentration–time profiles of bile acids measured during Period 1 (control) and Period 2 (co‐administration with itraconazole, a CYP3A inhibitor). Shown are concentrations of 1β‐hydroxy‐deoxycholate (1βOH‐DCA), 1β‐hydroxy‐glycodeoxycholate (1βOH‐GDCA), 1β‐hydroxy‐taurodeoxycholate (1βOH‐TDCA), total 1βOH‐DCA and its conjugates (sum of 1βOH‐DCA, 1βOH‐GDCA, and 1βOH‐TDCA), glycoursodeoxycholate (GUDCA), and glycohyocholate (GHCA). Blue filled circles represent the control condition (Period 1), and red filled squares represent the itraconazole treatment condition (Period 2). Data are presented as mean ± standard deviation (SD) (n = 11 per group, except for 1βOH‐DCA in the itraconazole inhibitor group where n = 10 due to one participant with all values below the lower limit of quantification (BLQ). Concentrations below the lower limit of quantification were excluded from the plots. Figure S2 shows plasma bile acid exposure from 0 to 96 hours; however, interpretation in Period 2 is limited due to repeated itraconazole dosing (24, 48, and 72 h).

Confirmation of glycochenodeoxycholate (GCDCA) as a CYP3A4 substrate and GHCA as a CYP3A4 product

Incubation of GCDCA with recombinant CYP3A4 resulted in formation of a monohydroxylated metabolite (theoretical mass: 464.3018; observed mass: 464.3018). This monohydroxylated metabolite was confirmed to be GHCA using an authentic standard of GHCA, which eluted at the same retention time (tR = 3.2 min, Figure S4 ). Other possible hydroxylated products of glycochenodeoxycholic acid (i.e., glyco‐α‐muricholic acid and glyco‐ß‐muricholic acid) exhibited different retention times than the monohydroxylated metabolite generated by CYP3A4. Together these data support the assignment of GHCA as a metabolite of GCDA formed by CYP3A4.

Furthermore, GCDA 6α‐hydroxylation to GHCA in human liver microsomes was best described by a substrate activation (allosteric sigmoidal) kinetic model, yielding fitted V_max and S₅₀ values of 87.2 pmol/min/mg protein and 193 μM, respectively (Figure 4 a ). This was further confirmed by visual inspection of the corresponding transformed Eadie‐Hofstee plot, which exhibited the characteristic curvature associated with positive allosteric behavior. Phenotyping studies using recombinant P450s demonstrated that GHCA formation was predominantly catalyzed by CYP3A4, with a minor contribution from the neonatal isoform CYP3A7 (Figure 4 b ). On the contrary, negligible GHCA formation was observed with the other P450 isoforms tested (Figure 4 b ) and in non‐transfected insect control (data not shown). When reaction velocities were scaled using reported relative hepatic abundances of CYP3A4 and CYP3A7, GCDA 6α‐hydroxylation was predicted to be catalyzed almost entirely by CYP3A4 (97.4%), with CYP3A7 contributing only a small fraction (2.6%) (Figure 4 c,d ).

Figure 4.

(a) Enzyme kinetics of glycochenodeoxycholate 6α‐hydroxylation in human liver microsomes, along with the corresponding Eadie‐Hofstee transformation (inset). Formation of GHCA was best described by a substrate activation (allosteric sigmoidal) kinetic model with fitted V_max and S₅₀ values of 87.2 pmol/min/mg protein and 193 μM, respectively. (b) Measured GHCA formation rates from glycochenodeoxycholate across a panel of 15 human recombinant P450 enzymes. (c) GHCA formation rates adjusted for the relative hepatic abundance of each P450 enzyme. (d) Fractional contribution of individual P450 isoforms to glycochenodeoxycholate 6α‐hydroxylation, calculated from the abundance‐scaled rates shown in panel C, indicating predominant catalysis by CYP3A4 (97.38%) with a minor contribution from CYP3A7 (2.62%). Each data point in panels A–C represents the mean ± S.E. of triplicate incubations.

DISCUSSION

Untargeted metabolomics using high‐resolution mass spectrometry has increasingly demonstrated value in the discovery of endogenous biomarkers when applied to PK samples from DDI studies involving well‐characterized index inhibitors. Previous studies have successfully leveraged this approach to identify putative plasma biomarkers for drug‐metabolizing enzymes such as CYP2D6, ¹⁹ as well as, clinically relevant drug transporters, including OATP1B1/1B3, ²⁰ OAT1/3, ²¹ , ²² and OCT2/MATEs. ²³ Although Kim et al. ¹⁶ applied untargeted metabolomics approach in a clinical study using ketoconazole and rifampicin (CYP3A inhibitor and inducer, respectively) to identify urinary biomarkers of CYP3A activity, to our knowledge, the present study represents the first application of untargeted metabolomics to identify putative endogenous plasma‐based biomarkers for the activity of CYP3A, a prominent drug‐metabolizing enzyme. This work enriches clinical biomarkers in the CYP3A domain, where current biomarkers such as 4β‐hydroxycholesterol and the urinary 6β‐hydroxycortisol/cortisol ratio have clear limitations, and there is emerging interest to progress more sensitive plasma biomarkers. ¹⁰ , ¹¹

Our untargeted metabolomics analysis identified 58 metabolites that were significantly increased or decreased (> 1.5‐ or < 0.67‐fold, P < 0.05) upon CYP3A inhibition compared with control conditions, highlighting the dynamic influence of CYP3A on the human metabolome. Marked reduction in plasma levels were observed for several metabolites within the bile acids and xanthine metabolism pathways following CYP3A inhibition with itraconazole (Table 1 , Figure 2 ). Notably, decreased bile acids included GUDCA, GHCA, GβMCA, ursodeoxycholate, isoursodeoxycholate, glycolithocholate sulfate, taurolithocholate 3‐sulfate, and lithocholate sulfate. The reduced exposure of these bile acids suggests that their formation may, at least in part, be mediated by CYP3A. Supporting this interpretation, a recent study demonstrated that the conversion of GCDA to GHCA is CYP3A4‐mediated, although enzyme selectivity has not been fully characterized. ¹⁸ CYP3A4 has also been shown to catalyze the conversion of lithocholate to ursodeoxycholate, ²⁴ providing a mechanistic basis for the reduction of plasma ursodeoxycholate observed with itraconazole treatment. However, the conversion of chenodeoxycholate to ursodeoxycholate is known to be primarily mediated by human intestinal microorganisms via formation of the intermediate 7‐keto‐lithocholate through bacterial 7β‐hydroxysteroid dehydrogenase, ²⁵ , ²⁶ , ²⁷ indicating that the net circulating pool of ursodeoxycholate reflects contributions from both CYP3A4 and microbial pathways. ⁸ , ²⁸ , ²⁹ The potential contribution of gut microbiota should therefore be considered when interpreting GUDCA and GHCA as bile acid biomarkers, as both may be influenced by host–microbial bile acid metabolism. ³⁰ , ³¹ Future work in this direction is warranted. Although the literature on CYP3A4‐mediated regulation of circulating bile acids has focused predominantly on ursodeoxycholate, glycoursodeoxycholate (GUDCA) is the obligate hepatic glycine conjugate of ursodeoxycholate, ³² and its plasma levels are therefore expected to covary with the upstream ursodeoxycholate pool. Overall, CYP3A4‐mediated oxidative reactions may contribute to the formation of several of these bile acid metabolites. However, to our knowledge, no study has directly examined the effect of CYP3A4/A5 inhibition on circulating GUDCA as an isolated endpoint, representing a gap in the understanding. Conversely, robust increases (multi‐fold elevations) were observed for several endogenous endocannabinoid and acyl‐glycines (Table 1 ). Previous studies have shown that CYP3A4 is one of the major cytochrome P450s involved in the metabolism of the endocannabinoid anandamide (AEA); in particular, arachidonoyl ethanolamide is metabolized by CYP3A4 ³³ and other CYP enzymes. ¹⁷ Given that itraconazole is a relatively selective CYP3A inhibitor, the observed pattern of changes suggests a strong biochemical impact attributable to CYP3A inhibition. In addition to CYP‐mediated metabolism, endocannabinoids such as oleoyl ethanolamide and palmitoyl ethanolamide can be hydrolyzed by fatty acid amide hydrolase (FAAH). ³⁴ Ketoconazole has been shown to inhibit FAAH, ³⁵ however, there is no clear evidence if itraconazole inhibits FAAH at clinically relevant concentrations and thereby contribute to increased endocannabinoid levels. Furthermore, itraconazole is also an inhibitor of P‐glycoprotein (P‐gp). One study demonstrated that inhibition or knockdown of P‐gp reduced luminal secretion of N‐acyl ethanolamines, including arachidonoyl ethanolamide, oleoyl ethanolamide (OEA), and palmitoyl ethanolamide (PEA), ³⁶ suggesting that altered transporter activity may additionally contribute to the observed increases. Collectively, plasma levels increase observed in metabolites within the endocannabinoid and acyl glycine pathways likely reflect a combination of on‐target CYP3A inhibition and potential off‐target effects involving FAAH and P‐gp. Further mechanistic work is needed, including evaluation of these metabolites as substrates of CYP3A, FAAH, and P‐gp. In addition, studies using CYP3A inhibitors with minimal impact on P‐gp or FAAH (e.g., grapefruit juice) would help determine whether metabolites in this pathway are suitable as selective biomarkers of CYP3A activity.

To further validate the findings from metabolomic analysis, we successfully developed and applied an HPLC–MS/MS multiplex assay to quantify the plasma exposure of five bile acids, including our newly identified candidates, GUDCA and GHCA, alongside recently proposed CYP3A biomarkers (1βOH‐DCA, 1βOH‐GDCA, and 1βOH‐TDCA). Particularly, GHCA demonstrated a geometric mean AUC ratio that was highly comparable to those of 1βOH‐DCA and its conjugates (Table 2 , Figure 3 ). Among the bile acids evaluated, baseline concentrations were lowest for 1βOH‐DCA (geometric mean, 0.396 ng/mL), followed by 1βOH‐TDCA (2.05 ng/mL), GHCA (6.00 ng/mL), 1βOH‐GDCA (6.23 ng/mL), and GUDCA (25.9 ng/mL). Notably, 1βOH‐DCA and its glycine/taurine conjugates are not included among the > 5400 metabolites of the Metabolon library, and therefore these analytes were not picked in the original 1313 metabolites identified in our metabolomic analysis. However, retrospective studies via metabolomics method using reference standards for 1βOH‐GDCA and 1βOH‐TDCA, revealed that these compounds co‐elute with glyco‐β‐muricholate (GβMCA) and tauro‐β‐muricholate (TβMCA), respectively. Interestingly, GβMCA was identified as one of the top hits in our untargeted metabolomic analysis (Table 1 , Figure 2 ). However, given the lack of GβMCA detection in the targeted HPLC–MS/MS assay, we infer that its signal in untargeted metabolomics corresponds to 1βOH‐GDCA. Importantly, this highlights a key strength of our study–utilizing an agnostic, untargeted approach not only allowed us to discover novel biomarker candidates like GHCA and GUDCA, but it also successfully captured and reaffirmed previously reported CYP3A signals without prior bias. In contrast, the lack of a distinct untargeted signal for 1βOH‐DCA and 1βOH‐TDCA in the metabolomic analysis is likely attributable to their much lower circulating concentrations, as confirmed by our targeted quantitative measurements.

Untargeted metabolomics and subsequent quantification revealed GHCA as robust biomarker with relatively low inter‐individual variability. In vitro profiling and CYP phenotyping demonstrated that formation of GHCA from GCDCA is predominantly mediated by CYP3A4, with a minor contribution from CYP3A7 and no detectable contribution from CYP3A5 (Figure 4 , see Figure S5 metabolic pathway). This finding provides direct mechanistic support for the CYP3A4 dependence of GHCA formation and further strengthens the interpretation of bile acid changes observed under itraconazole treatment. Previous studies have shown that conversion of deoxycholate (DCA) to 1βOH‐DCA is predominantly mediated by CYP3A4, with additional contributions from CYP3A7, ⁹ , ²⁹ , ³⁷ whereas formation of 1βOH‐GDCA from glycodeoxycholate (GDCA) and 1βOH‐TDCA from taurodeoxycholate (TDCA) involves both CYP3A4 and CYP3A7. ²⁹ , ³⁷ Of note, the quantitative contribution of post‐hydroxylation conjugation of 1βOH‐DCA (via BAAT) vs. pre‐conjugation 1β‐hydroxylation of GDCA/TDCA to the circulating pools of 1βOH‐GDCA and 1βOH‐TDCA has not been formally characterized and represents an area for future mechanistic investigation. The observed reductions in GHCA, GUDCA, and 1βOH‐DCA‐related metabolites are unlikely to be primarily driven by itraconazole inhibition of major drug transporters. Although reduced OATP1B1/1B3‐mediated hepatic uptake of DCA, GDCA, or TDCA could theoretically decrease intrahepatic substrate availability for CYP3A‐mediated formation of 1βOH‐DCA‐related metabolites, itraconazole is not generally considered a clinically meaningful inhibitor of OATP1B1/1B3. ³⁸ In addition, P‐gp is not a major bile acid transporter, ³⁹ and efflux inhibition is not expected to account for the magnitude and direction of these decreases.

In contrast to these 1β‐hydroxylated bile acids, the present findings indicate a greater specificity of CYP3A4 for GHCA formation. Notably, the in vitro observation that GHCA formation is primarily mediated by CYP3A4 is corroborated by metabolomic genome‐wide association studies, in which circulating GHCA levels are significantly associated with reduced‐function variants in CYP3A4 (e.g., CYP3A4*22), but not with variants in CYP3A5 or CYP3A7 (references in Table S3 ). In contrast, GβMCA and TβMCA have been reported to strongly associate with polymorphisms in CYP3A7 but not CYP3A4, in previous metabolomic GWAS (references in Table S3 ). As discussed above, we speculate that these analytes are misidentified and likely correspond to 1βOH‐GDCA and 1βOH‐TDCA, respectively. Collectively, these findings highlight distinct CYP3A isoform contributions to bile acid metabolism and further support GHCA as a CYP3A4‐selective endogenous biomarker. In contrast to GHCA, the specific CYP3A isoform contribution to the GUDCA pathway is not well‐characterized. CYP3A4 contributes to ursodeoxycholate formation from lithocholate, ²⁴ although the intestinal microbial pathway from chenodeoxycholate ²⁵ , ²⁶ , ²⁷ likely dominates the circulating ursodeoxycholate/GUDCA pool. Consistent with this, circulating GUDCA showed no significant association with variants in CYP3A4, CYP3A5, or CYP3A7 in metabolomic GWAS (references in Table S3 ), indicating that GUDCA's isoform selectivity is less defined than that of GHCA.

4β‐Hydroxycholesterol (4β‐OHC), formed through CYP3A‐mediated cholesterol metabolism, is widely used in clinical DDI studies as an endogenous marker of CYP3A activity. ⁴ , ¹⁴ , ⁴⁰ Although 4β‐OHC shows good sensitivity to CYP3A induction (approximately 5–10‐fold increases), ¹⁴ , ⁴⁰ its utility for detecting CYP3A inhibition is limited, as changes in plasma exposure are modest (typically ≤ 30%) and require prolonged inhibitor dosing due to its long half‐life (approximately 17 days). ⁵ , ⁴¹ Consequently, 4β‐OHC is primarily used for exploratory assessments and to complement, rather than replace, formal DDI evaluations with probe drugs such as midazolam. ¹⁴ , ⁴² , ⁴³ In contrast, the plasma biomarkers identified in the present study are substantially more sensitive to CYP3A inhibition. Moreover, previous studies have demonstrated robust responses of 1βHO‐DCA conjugates to CYP3A induction following multiple‐dose rifampin treatment. ¹⁰ Collectively, inclusion of GHCA, GUDCA and 1βHO‐DCA conjugates in the study protocols (e.g., first‐in‐human dose‐escalation studies) has the potential to improve early DDI risk assessment and reduce reliance on dedicated, time‐consuming probe‐drug DDI studies.

There are certain limitations to this study. First, among the broader panel of putative endogenous biomarkers identified, only GHCA and GUDCA were further evaluated along with 1βOH‐DCA metabolites using a targeted LC–MS/MS. Based on the favorable effect size (inhibitor‐to‐control ratios) and P‐value in the metabolomic analysis, we followed GHCA and GUDCA specifically to readily position these two sensitive and robust biomarkers for clinical application. Additional studies will be needed to evaluate whether the remaining candidate analytes exhibit sufficient selectivity and robustness to serve as reliable biomarkers of CYP3A‐mediated DDIs. Second, the current findings are based on a single strong CYP3A inhibitor and a relatively limited sample size (n = 11 heathy participants per group), which may restrict generalizability. Although the present dataset allowed exploratory assessment of baseline inter‐subject and intra‐subject variability (Table S8 ), the sample size was insufficient to formally identify determinants of variability. Consequently, potential covariate effects, including age, sex, CYP3A4 genotype, diet, gut microbiome composition, diurnal variation, and hepatic function, require evaluation in larger, prospectively designed studies. Broader evaluation across multiple clinical CYP3A modulators with differing potency, as well as larger and more diverse study populations (e.g., cancer patients), will be important to position the clinical utility of these proposed biomarker candidates. Further collaborative studies are currently underway in this direction.

CONCLUSIONS

In conclusion, this study applied untargeted metabolomics to clinical DDI samples to uncover sensitive endogenous biomarkers of CYP3A activity. Itraconazole‐mediated inhibition revealed pronounced and pathway‐specific metabolic changes, with bile acid metabolites emerging as the most robust indicators. GHCA and GUDCA, further supported by targeted quantification, represents CYP3A4‐selective plasma biomarkers that can strengthen translational DDI assessment.

FUNDING

No funding was received for the work presented.

CONFLICT OF INTEREST

All authors are full‐time employees of Pfizer, Inc. The authors declared no competing interests for this work.

AUTHOR CONTRIBUTIONS

S.W.Y., M.V.S.V., M.V., E.P.K., L.W.T.T., M.A.C., R.S.O. wrote the manuscript. S.W.Y., M.V.S.V., M.A.C., M.D.T., M.V. designed the research. E.P.K., L.W.T.T., M.A.C. performed the research. S.W.Y., M.V.S.V., M.V., E.P.K., L.W.T.T., M.A.C. analyzed the data.

Supporting information

Data S1.

Data S2.

Data S3.

DATA AVAILABILITY STATEMENT

The authors declare that all the data supporting the findings of this study are contained within the paper.