A Systematic Review and Classification of the Effects of P-glycoprotein Inhibitors and Inducers in Humans, Using Digoxin, Fexofenadine, and Dabigatran as Probe Drugs

Abstract

P-glycoprotein is a critical efflux transporter that may significantly affect the pharmacokinetics of various drugs by influencing their absorption, distribution and elimination. While European and American regulatory guidelines provide lists of P-glycoprotein modulators, they lack specificity concerning in vivo studies and clear guidance on inducers, creating uncertainty in their clinical relevance. A systematic search on in vivo clinical studies involving healthy volunteers using fexofenadine, dabigatran and digoxin as P-glycoprotein substrates has been performed in accordance with the PRISMA guidelines. A total of 151 studies assessing the impact of P-glycoprotein modulators on the concentration–time profile of P-glycoprotein substrates were retrieved. Additionally, data on the P-glycoprotein modulators’ effect on cytochrome P450 3A4 induction or inhibition were also collected. P-gp modulators were classified as potent, moderate, weak or non-interactors for P-glycoprotein, with or without cytochrome P450 3A4 impact, on the basis of the area under the concentration–time curve ratio. This classification was adapted from the Food and Drug Administration criteria for cytochrome interactions. This systematic review identified 49 area under the plasma concentration–time curve ratio values corresponding to P-glycoprotein inhibitors, 23 to P-glycoprotein inducers and 131 to non-interactors. Of these, only 32.5% and 41.1% were classified as weak to potent, respectively. Only 0.7% of inhibitors and no inducers were classified as potent. This suggests that most P-glycoprotein modulators have a limited impact on drug exposure. The potential for interaction increases when P-glycoprotein modulators also affect cytochrome P450 3A4, which is the case for 59.9% of P-glycoprotein modulators. However, some moderate P-glycoprotein modulators may have clinically significant effects depending on the therapeutic margin of the substrate and the clinical context.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40262-025-01514-3.

Key Points

P-gp inhibitors alone tend to have a minimal effect on P-gp substrate exposure. However, when both P-gp and CYP3A4 are involved, the risk is significantly higher, especially when CYP3A4 modulation is strong.

To enhance the precision of guidance for researchers and clinicians, it is advisable to refine the classification of compounds that impact both P-glycoprotein (P-gp) and CYP3A4.

Additionally, a more detailed categorisation of compounds that act as P-gp inducers could significantly improve the accuracy in predicting drug interactions.

This refined classification would facilitate a clearer understanding of the interaction dynamics, leading to more informed decision-making in clinical settings.

Introduction

P-glycoprotein (P-gp) is an efflux transporter pump that plays an important role in drug transport. It was first extensively studied in oncology, where it plays a significant role in chemoresistance by actively expelling various xenobiotics from cells, thereby reducing their intracellular concentrations [1–3]. P-gp is expressed in a variety of organs including the small intestine, the liver, the pancreas, the kidneys [4, 5] and at blood–tissue barriers such as the blood–brain barrier, the blood–testis barrier and the placenta [6, 7]. The pharmacokinetics of numerous substrate drugs of P-gp has been shown to be influenced by P-gp inhibitors, which enhance drug absorption and reduce metabolism and elimination, resulting in an overall increase in drug exposure. Conversely, P-gp inducers have been shown to decrease drug concentrations [8]. The inhibition of P-gp-mediated transport involves various mechanisms. P-gp inhibition can occur through (i) competitive inhibition, where two substrates compete for the same binding site, preventing simultaneous binding, or (ii) through non-competitive inhibition, where inhibitors bind to distinct, functionally independent sites, modulating P-gp activity indirectly, or (iii) through the blockage of ATP hydrolysis, a process essential for P-gp’s efflux function [9]. While not exclusively the case, P-gp inhibitors and inducers often concurrently affect cytochrome P450 (CYP) 3A4 activity due to their similar physicochemical properties or shared regulatory pathways, which enable them to interact with both proteins [10–12]. The inhibition potential of P-gp and CYP 3A4 is influenced by similar structural properties: high lipophilicity and considerable molecular weight, which contribute to their binding affinity [9, 13, 14], and physical proximity of the sites of action, facilitating common interactions [10]. Structural properties of inducers include lipophilic and voluminous molecules (e.g. antibiotics (rifampicin, macrolides), glucocorticoids, oral antidiabetic drugs, ritonavir) with flexible conformations that allow interactions with the large ligand-binding pocket of nuclear receptors such as Pregnane X Receptor (PXR) and Constitutive Androstane Receptor (CAR). This enabled transcriptional activation of genes encoding transport proteins such as P-gp [15, 16]. As PXR and CAR also regulate CYP 3A4 expression [17], compounds that induce CYP 3A4 often simultaneously induce P-gp, making a CYP 3A4 induction a predictor of P-gp induction [17, 18]. Consequently, there are relatively few compounds that are specific to P-gp as substrates, inhibitors or inducers.

In the 2000s, On the basis of the recognition that P-gp transporters are important determinants of drug exposure [19–21], the International Transporter Consortium (ITC) released a white paper outlining methodologies and proposing recommendations to guide preclinical and clinical studies on transporter-mediated drug interactions [22, 23]. Following these works, the Food and Drug Administration (FDA) and the European Medicines Agency (EMA) released in 2012 a draft guidance for drug–drug interactions (DDI) involving P-gp. These guidelines aim to determine whether a new molecular entity acts as an inhibitor or inducer, and to establish when an in vivo clinical study is necessary. The guidelines recommended digoxin as the reference P-gp probe for renal P-gp inhibition or induction, and dabigatran and fexofenadine for intestinal P-gp modulation [24, 25]. At that time, in vivo clinical studies testing for P-gp modulation were not required for drugs showing no influence on CYP3A4. In May 2023, a revised version of the FDA guidelines specified that a P-gp inhibitor is defined as a drug that demonstrates in vitro inhibition and causing a ≥ 1.5-fold increase in the area under the concentration–time curve (AUC) of dabigatran, digoxin, or edoxaban [26]. However, the revision did not provide any information on P-gp induction.

Although DDIs involving P-gp are routinely checked in clinical practice, the risk evaluation provided by DDI software or guidelines are often based on a combination of in vitro and in vivo data, and may not consider the combined effect of the interactions involving both CYP3A4 and P-gp, which can compromise the assessment of the true risk of the interaction. There is currently no exhaustive list of P-gp inhibitors and inducers categorised by the magnitude of change in exposure in humans, a resource that would be highly valuable in clinical settings. The aim of our systematic review is to compile a comprehensive list of in vivo clinical studies on the P-gp modulatory effect of various drugs, and to classify their effect on exposure of the three P-gp substrates digoxin, fexofenadine and dabigatran.

Materials and Methods

Search Strategy and Selection Criteria

This systematic review was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [27]. The protocol was designed a priori but was not submitted to public repositories and is available upon request. Systematic literature searches were conducted in PubMed, EMBASE.com and Cochrane Central Register of Controlled Trials (Wiley) on 4 April 2022, with an update on 8 March 2024 and with the assistance of a biomedical librarian. The search terms were the following: ‘dabigatran’, ‘digoxin’, ‘fexofenadine’, ‘P-glycoprotein’, associated with ‘P-gp’, ‘ABCB1’, ‘MDR protein 1’, ‘multidrug resistance protein 1’, ‘multidrug transport protein 1’, ‘multidrug transporter 1’, ‘multidrug resistance transporter 1’, ‘ATP binding cassette subfamily b’, ‘ATP binding cassette subfamily b member 1’, ‘PGY 1 protein’ and ‘area under the curve’, ‘plasma concentration-time curve’, ‘drug bioavailability’, ‘drug blood level’, ‘digoxin blood level’.

Filters to exclude animal studies and conference abstracts were applied to the strategies, but no date or language limits were applied. The strategies were peer-reviewed according to the Peer Review of Electronic Search Strategies (PRESS) checklist [28] by another librarian and are presented in Table S1. References were imported into EndNote 20 (Clarivate™, London, UK) and deduplicated. Additional publications were identified by backward citation searching from relevant reviews and systematic reviews.

Inclusion criteria were the following: (1) in vivo studies including healthy volunteers, (2) studies employing dabigatran, digoxin or fexofenadine as substrates to investigate the inhibitory or inductive potential of probe drugs and (3) studies reporting relevant outcome measures. Preclinical studies, including human in vitro experiments, animal data and dual publications were excluded.

Selection Process

One investigator (C.C.O.) conducted the study selection on the basis of titles and abstracts, using Rayyan^® (rayyan.qcri.org) [29] to facilitate this process. Subsequently, two investigators (C.C.O. and C.C.S.) independently performed the study selection process on the basis of the full texts of the articles. In case of disagreement regarding the eligibility of a study, a discussion was undertaken until a consensus was reached.

Qualitative Assessment

Two investigators (C.C.O. and C.C.S.) appraised the quality of the evaluated studies using tools selected according to the study design. The Risk of Bias In Non-Randomized Studies (ROBINS-I) tool was used for non-randomized and cross-sectional studies [30], while the Risk Of Bias 2 tool from the Cochrane collaboration was used for randomized studies (RoB 2.0) [31]. We assessed and reported the overall risk of bias for the primary outcome, which is the area under the curve ratio (AUCR). Using the ROBINS-I tool, we evaluated biases related to result reporting, outcome measurement, missing data, deviations from intended interventions, intervention classification, participant selection and confounding. For the RoB 2.0 tool, we examined biases from randomization, deviations from intended interventions, missing outcome data, outcome measurements and selection of reported result. Using the signaling questions in the Rob 2.0 and Robins-I tools, we assessed each domain and categorised them as ‘low risk of bias’, ‘some concerns’, or ‘high risk of bias’. We determined an overall risk of bias for each included study on the basis of the following criteria: low risk of bias (plausible bias unlikely to seriously alter the results) if all domains were judged are at low risk; some concerns of bias (plausible bias that raises some doubt about the results) if one or more domains were some concerns; and high risk of bias (plausible bias that seriously weakens confidence in the results) if one or more domains were judged to be at high risk. We used the ClinPK checklist to evaluate the transparency and completeness of pharmacokinetic data reporting [32]. Before application, two researchers (C.C.O. and C.C.S.) adapted the checklist for the included studies by removing seven unsuitable items. We then calculated the count and percentage of included studies that satisfied each item.

Data Extraction and Synthesis

The information on clinical studies related to P-gp inhibition and induction (i.e. study type, date, duration, substrate and interacting molecule name and dose, type of interaction (inhibition/induction), % change in AUCR) was organised using Microsoft Excel ((version 16.0), Microsoft Corporation (2016)). The AUCR, representing the geometric mean ratio (AUCR = AUC_treatment/AUC_control) was retrieved from the publications. To calculate the AUCR, we used reported AUC from time zero to infinity (AUC₀–∞) for single-dose and multiple-dose studies. If AUC₀–∞ data were unavailable, the AUC calculated over a dosing interval at steady state (AUC₀–τ) was used. These metrics were chosen as they provide equivalent information for AUCR calculation. The extracted data were reported in a spreadsheet table, one interaction per row, with literature references. Extracted data are available in Tables S2 and S3, for P-gp inhibitors and inducers, respectively.

Data Analysis and Clinical Classification

The cut-off values used to classify the inhibitory or inductive levels on the basis of AUCR were aligned to the FDA DDI criteria, encompassing both the criteria of P-gp inhibition and CYP inhibition and induction criteria. Drugs were classified as weak P-gp inhibitors if the AUCR was between ≥ 1.5 to < 2.0, moderate if the AUCR was between ≥ 2.0 to < 5.0 and potent if the AUCR was ≥ 5.0. Similarly, drugs were classified as weak P-gp inducers if the AUCR was between ≤ 0.80 and > 0.50, moderate if the AUCR was between ≤ 0.50 and > 0.20 and potent if the AUCR was ≤ 0.20. When multiple clinical trials evaluated the same inhibitor or inducer with the same substrate, the AUCR values were presented as ranges. The effects of the drugs on CYP3A4 were also included and classified into three levels: potent, moderate and weak inhibitors or inducers, according to the information provided by the FDA and EMA in each drug monography. We specifically addressed substances exhibiting time-dependent modulation effects by comparing single dose studies from those observed after multiple doses, while including the timing of the perpetrator intake and the administered doses.

Results

We identified 1794 articles through our search. After removing 548 duplicates, 1246 records were screened, and 978 were excluded on the basis of the exclusion criteria. Of the 268 sought for retrieval, 5 could not be retrieved (full text not accessible). A total of 263 studies were assessed for eligibility, and 140 were included in review. Additionally, after backward citation searching of relevant reviews and systematic reviews, 11 more studies were included, bringing the total of 151 included studies (Fig. 1).

Fig. 1.

PRISMA flow diagram of clinical studies selection process (included and excluded studies)

Qualitative Assessment

Figures 2 and 3 summarise the risk of bias assessments for the randomized controlled trials (RCT) and non-randomized controlled trials (NRCT), respectively. Detailed risk of bias assessments for each included study are provided in Tables S4 and S5 for Rob 2.0 and ROBINS-I, respectively.

Fig. 2.

Methodological quality graph: authors’ judgements on each methodological quality item, presented as percentages across all included randomized controlled trials

Fig. 3.

Methodological quality graph: authors’ judgements on each methodological quality item presented as percentages across all included non-randomized controlled trials

Regarding the transparency and completeness of reporting in the 151 clinical pharmacokinetic studies, the overall assessment indicated a high level of transparency. The lowest rate was observed in study titles, with only 40.4% of studies adequately identifying the drugs and population studied. This was followed by the description of study limitations, particularly the disclosure of potential sources of bias and imprecision, which was adequately described in only 64.9% of studies. The compliance rates with ClinPK criteria in included studies are detailed in Table S6.

Overall, 49 AUCR values (23.7%) were classified as P-gp inhibitors according to the FDA criteria of AUCR > 1.5, whereas 102 AUCR values were not (49.3%). Among AUCR values classified as P-gp inhibitors, 21.2% were classified as moderate, and only one as a potent inhibitor. Four studies, which evaluated digoxin with increasing doses of inhibitors, namely itraconazole [33], atorvastatin [34], etanercept [35] and obeticholic acid [36], reported non-linear AUCR increases, indicating that P-gp inhibition reaches a saturation point at higher inhibitor concentrations. Theses results confirm previously reported data of the saturable nature of P-gp [19]. Concerning P-gp inducers, none reduced the AUCR to 0.20 or below (i.e. there was no strong P-pg inducer). Nine AUCR moderately reduced to values between 0.50 and 0.20, while the majority (58.9%) showed a maximum AUCR of 0.80. The comprehensive list of all AUCR values for P-gp inhibition and induction is presented in Tables 1 and 2, respectively. In total, 27 drugs were classified as P-gp inhibitors, and 11 as P-gp inducers. The extensive list of all AUCR values of clinical studies provided by our systematic review involving P-gp substrates when used with P-gp inhibitors or inducers classified as with and without CYP 3A4 influence, along with the reported statistical significance of AUCR values, are presented in Tables S4 and S5, respectively.

Table 1.

Summary of P-gp inhibition levels and associated AUCR values, including the primary intestinal or renal site of action and the CYP 3A4 modulation

Inhibitor	Intestinal (I) or renal (R)	Ratio AUC (AUCR) [interval] for CYP3A4 modulation FDA category	Ratio AUC (AUCR) [interval] for P-gp inhibition
Potent P-gp inhibitors
Indinavir/ritonavir	R	> 5	5.00 (AC) 4.20 (SS)
Moderate P-gp inhibitors
Clarithromycin	I	> 5	2.00
	R		[1.35–1.70]
Cobicistat	I	[2–5]	[2.10–2.27]
Darunavir/cobicistat	I	[2–5]	2.64 (day 4) 1.88 (day 18)
Glecaprevir and pibrentasvir	I	–	2.38
Indinavir	I	> 5	2.90
Itraconazole	I	> 5	[2.09–4.00]
Ketoconazole and fluvoxamine	I	> 5	3.16
Lopinavir and ritonavir	I	> 5	[2.00–3.70]
Quinidine	I	–	2.31
	R		2.66
Rifampicin	I	≤ 0.20	[2.08–4.56] (SD) [2.40–3.13] (MD)
Ritonavir	I	> 5	[2.80–3.70] (AC) [1.40–2.00] (SS)
Rolapitant	R	–	2.27
Saquinavir and ritonavir	I	> 5	[1.41–2.19]
Valspodar	R	–	1.74 (SD) 3.05 (MD)
Verapamil	I	[2–5]	[1.30–2.87]
	R		[1.00–1.50]
Weak P-gp inhibitors
Azithromycin	I	–	1.67
Darunavir/ritonavir	I	> 5	1.72 (day 4) 1.18 (day 18)
Diosmin	I	–	1.67
Fluvoxamine	I	[1.25–2]	1.78
Itraconazole	R	> 5	1.68
Piperine	I	–	1.68
Probenecid	I	–	1.53
Quercetin	I	[2–5]	1.56
Ritonavir	R	> 5	[1.16–1.86]
TBAJ-876	R	–	1.51
Telaprevir	R	> 5	1.85
No P-gp inhibition
Ambrisentan	R	–	1.09
Anacetrapib	R	–	1.07
Aprepitant	R	[2–5]	0.93
Atorvastatin	R	–	[1.03–1.15]
BFE1224 (pro-drug of Ravuconazole)	R	–	1.18
Black Cohosh	R	–	1.05
Bosutinib	I	–	1.01
Breviscapine	I	–	0.99
Calcitriol	I	–	1.16
	R		1.02
Carvedilol	R	–	1.20
CC-90001	R	–	1.16
Cimetidine	I	[1.25–2]	1.08
	R		1.26
Citalopram	R	–	1.01
Combination of daclatasvir, asunaprevir and beclabuvir	R	[0.20–0.50]	1.23
Cremophor EL	I	–	1.28
	R		1.22
Darexaban	R	–	1.11
Dipyridamole	R	–	1.08
Echinacea	R	[0.20–0.50]	1.03
Edoxaban	R	–	1.08
Elagolix	R	[0.50–0.80]	1.26
Eliglustat	R	–	1.49
Entrectinib	R	–	1.18
Etanercept	R	–	[0.99–1.33]
Etravirine	R	[0.20–0.50]	1.19
Ezogabine and retigabine	R	–	1.08
Fedratinib	R	[2–5]	1.11
Felodipine	R	–	1.01
Fermented red ginseng	I	–	1.30
Fostamatinib	R	–	1.37
Futibatinib	R	–	1.00
Gingko biloba	R	[0.20–0.50]	1.22
Glecaprevir and pibrentasvir	R	–	1.48
Goldenseal	R	–	1.01
Grapefruit Juice	R	[2–5]	1.10
Isavuconazole	R	[2–5]	1.25
Ivacaftor	R	≤ 0.20	1.32
Kava Kava	R	–	1.02
Lacosamide	R	–	1.02
Lanabecestat	I	–	1.14
Lenalidomide	R	–	1.07
Levetiracetam	R	–	1.04
Linagliptin	R	[1.25–2]	1.01
Losartan	R	–	1.03
Maraviroc	R	–	1.00
Maribavir	R	–	1.22
Metronidazole	I	–	1.00
Midazolam	R	–	1.12
Midostaurin	R	–	1.23
Mirabegron	R	[1.25–2]	1.27
Nelfinavir	I	> 5	1.02
	R		[1.20–1.23]
Obeticholic acid	R	–	[1.01–1.07]
Omecamtiv mecarbil	R	–	1.06
Paroxetine	I	–	1.38
Pexidartinib	R	[0.20–0.50]	1.08
Probenecid	R	–	1.07
Radix Astragali extract granules	I	–	0.99
Rifampicin	R	≤ 0.20	[1.25–1.31] (SD)
Rivaroxaban	R	–	1.08
Roflumilast	R	–	1.01
Saquinavir and ritonavir	R	> 5	1.49
Seralutinib	R	–	1.11
Sotorasib	R	[0.20–0.50]	1.21
St John's wort	I	≤ 0.20	1.28
Talinolol	R	–	1.18
Telmisartan	R	–	1.22
Tezacaftor and ivacaftor	R	≤ 0.20	1.30
Ticagrelor	R	–	1.28
Tolvaptan	R	–	1.18
Troglitazone	R	–	1.04
Trospium chloride	R	–	1.11
Tucatinib	R	[2–5]	1.46
Turmeric crude extract	R	–	1.27
Vandetanib	R	–	1.23
Velpatasvir	R	–	1.27
Venetoclax	R	–	1.09
Voclosporin	R	–	1.25
Zanubrutinib	R	[0.50–0.80]	1.11

AUCR, area under the concentration curve ratio; P-gp, P-glycoprotein; CYP, cytochrome P450; SD, single dose; MD, multiple doses; AC, acute concentration = AUCR after acute dose of the interacting drug; SS, steady-state = AUCR at steady-state dosing of the interacting drug; I, acts mainly at the intestinal level with dabigatran and fexofenadine as probe P-gp substrates; R, acts mainly at the renal level, with digoxin as probe P-gp substrate

Table 2.

Summary of P-gp induction levels and associated AUCR values, including the primary intestinal or renal site of action and the CYP 3A4 modulation

Inducer	Intestinal (I) or renal (R)	Ratio AUC (AUCR) [interval] for CYP3A4 modulation FDA category	Ratio AUC (AUCR) [interval] for P-gp induction
Moderate P-gp inducers
Carbamazepine	I	≤ 0.20	[0.40–0.61]
Rifampicin*	I	≤ 0.20	[0.33–0.59]
Weak P-gp inducers
Rifampicin	R	≤ 0.20	[0.70–0.98]
Avasimibe	R	–	0.64
Danshen ethanol extract	I	–	0.54
Green tea catechins	R	–	0.67
Phenytoin	R	≤ 0.20	0.77
Ritonavir	I	> 5	[0.71–1.15]
St John’s wort	R	≤ 0.20	[0.74–1.01]
No induction
Alfentanil	I	–	0.91
Aliskiren	R	–	0.85
Apararenone	R	–	0.89
Apple juice*	I	–	[0.27–0.73]
Bosentan	R	[0.20–0.50]	0.87
Darolutamide	I	–	0.91
Deferasirox	R	[0.50–0.80]	0.91
Echinacea purpurea	I	–	0.98
Ginkgo biloba	I	[0.20–0.50]	0.83
Grapefruit juice*	R	[2–5]	1.01
	I	[2–5]	[0.37–0.70]
Hawthorn (Crataegus oxyacantha)	R	–	0.92
Honey	R	–	0.96
Letermovir	R	[2–5]	0.87
Milk thistle	R	–	0.91
Odanacatib	R	–	0.95
Orange*	I	–	0.31
Panax ginseng	I	–	0.91
Red ginseng	I	–	0.89
Sertraline	I	–	0.84
St John’s wort	I	≤ 0.20	[0.86–0.97]
Tea	R	–	0.99

AUCR, area under the concentration curve ratio; P-gp, P-glycoprotein; CYP, cytochrome P450; SD, single dose; MD, multiple doses; AC, acute concentration = AUCR after acute dose of the interacting drug; SS, steady-state = AUCR at steady-state dosing of the interacting drug; I, acts mainly at the intestinal level, dabigatran and fexofenadine are P-gp substrates; R, acts mainly at the renal level, digoxin is a P-gp substrate

*The observed AUCR reductions are primarily attributed to the OATP uptake inhibition rather than to the P-gp efflux induction

The distribution of AUCR values collected from the 151 clinical studies for P-gp substrates with P-gp modulators, categorised by inhibition or induction potency and CYP influence, is detailed in Table 3. The substances exhibiting short-term inhibitory and time-dependent inductive effects—rifampicin, ritonavir, St. John’s wort, darunavir/cobicistat and darunavir/ritonavir—are detailed in Table 4. Rifampicin initially act as a P-gp inhibitor, increasing AUCR values of digoxin (1.31–1.46) and dabigatran (2.08–2.22) probe substrates. However, after multiple doses, a reduced AUCR was observed, ranging from 0.68–1.00 for digoxin to 0.33–0.59 for dabigatran. In contrast, we observed with fexofenadine a constant inhibition after multiple doses (AUCR 2.40–3.48). This effect is predominantly due to its inhibition of renal influx transporters and OATP-mediated intestinal and hepatic uptake, which outweighs its induction properties on P-gp [37]. However, a staggered administration of rifampicin with fexofenadine leads to a decrease in AUCR of 0.49, reflecting a diminished inhibitory effect and the predominance of rifampicin’s inductive effect. Similar to darunavir/cobicistat and darunavir/ritonavir, ritonavir exhibits a progressive decrease in AUCR over time, although its inhibitory effect remains predominant (AUCR 2.80 after a single dose, decreasing to 1.40 after multiple doses). However, when administered with a delay, its inductive effect becomes more pronounced (AUCR 1.15 with simultaneous multiple doses, decreasing to 0.71 with delayed administration). Ritonavir inhibits OATP uptake; however, this effect is less pronounced compared with its P-gp inhibition and induction [38]. Concerning St. John’s Wort, a transient inhibitory effect is observed following a single dose (AUCR 0.99–1.28), whereas inductive properties become predominant upon repeated administration (AUCR 0.75–1.01).

Table 3.

Distribution of AUCR values collected from the 151 clinical studies using digoxin, fexofenadine and dabigatran in the presence of P-gp inhibitors and inducers, categorised by potency and CYP3A4 influence

AUC ratio (AUCR)	Digoxin		Fexofenadine		Dabigatran		Total
	w CYP3A4 n (%)	w/out CYP3A4 n (%)	w CYP3A4 n (%)	w/out CYP3A4 n (%)	w CYP3A4 n (%)	w/out CYP3A4 n (%)	N (%)
Inhibition
≥ 5.0 (potent)	0 (0.0)	0 (0.0)	1 (100.0)	0 (0.0)	0 (0.0)	0 (0.0)	1 (0.7)
≥ 2.0 to < 5.0 (moderate)	0 (0.0)	0 (0.0)	24 (75.0)	1 (3.1)	6 (18.8)	1 (3.1)	32 (21.2)
≥ 1.5 to < 2.0 (weak)	6 (37.5)	1 (6.3)	3 (18.8)	3 (18.8)	3 (18.8)	0 (0.0)	16 (10.6)
< 1.5 (no inhibition)	29 (28.4)	55 (53.9)	6 (5.9)	8 (7.8)	1 (1.0)	3 (2.9)	102 (67.5)
Total	35 (23.2)	56 (37.1)	34 (22.5)	12 (7.9)	10 (6.6)	4 (2.6)	151 (100.0)
Induction
≤ 0.20 (potent)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
≤ 0.50 to > 0.20 (moderate)	0 (0.0)	0 (0.0)	4 (50.0)	2 (12.5)	3 (37.5)	0 (0.0)	9 (16.1)
≤ 0.80 to > 0.50 (weak)	7 (46.7)	2 (13.3)	2 (13.3)	1 (13.3)	2 (13.3)	0 (0.0)	14 (25.0)
> 0.80 (no induction)	16 (48.5)	7 (21.2)	3 (9.1)	5 (15.2)	1 (3.0)	1 (3.0)	33 (58.9)
Total	23 (41.1)	9 (16.1)	9 (16.1)	8 (14.3)	6 (10.7)	1 (1.8)	56 (100.0)

W and w/out CYP3A4 = drug with CYP3A4 or w/out CYP3A4 activity; percentages represent the proportion AUCR in each category relative to the total number of AUCR

Table 4.

Summary of P-gp inhibition or induction levels and associated AUCR values for drugs exhibiting time-dependent modulatory effects

P-gp modulator	Substrate	Substrate dose (mg/day)	Inhibitor resp. inducer dose (mg/day)	Administration Schedule	AUCR after single dose	AUCR after multiple doses
Rifampicin (RIF)	FEX(R,S)a	60	450	FEX: day 7 RIF: days 1–7	N/A	(S): 3.48 (R): 3.10	[37]
	FEX (R,S)a	60	600	FEX: days 1, 3, 6 RIF: days 1–6	(S): 4.56 (R): 3.86	(S): 3.13 (R): 2.40	[43]
	FEX	25	600	FEX: day 7 RIF: days 1–7	N/A	0.49*	[61]
	DIG	0.25	600	–	1.31	N/A	[57]
	DIG	0.25	600	DIG: days −1, 7 RIF: days 1–6	N/A	0.75	[64]
	DIG	0.4	600	DIG: days 1, 7 RIF: days 1–7	N/A	0.84	[62]
	DIG	0.5	600	DIG: days −1, 7 RIF: days 1–7	N/A	0.84	[63]
	DIG	0.5	600	DIG: days 1, 14 RIF: days 8–21	N/A	1.25 (sim) 0.81 (del)	[58]
	DIG	0.5	600	DIG: day 28 RIF: days 1–28	1.46	N/A
				DIG: day 35 RIF: days 1–28	N/A	0.68	[66]
				DIG: day 42 RIF: days 1–28	N/A	1.00
	DIG	1	600	DIG: days 2, 18 RIF: days 8–23	N/A	0.70	[65]
	DAB	0.375	600	–	2.08	N/A	[59]
	DAB	0.750	450	–	2.22	N/A	[60]
	DAB	75	10	DAB: days 1, 19, 37 RIF: days 9–18	N/A	0.59	[18]
	DAB	75	75	DAB: days 1, 19, 37 RIF: days 27–36	N/A	0.38	[18]
	DAB	75	600	DAB: days 1, 19, 37 RIF: days 27–36	N/A	0.33	[18]
	DAB	150	600	DAB: days 1, 9, 16, 23 RIF: days 2–8	N/A	0.33	[42]
Ritonavir (RIT)	FEX	60	600 (7 days) then 800	FEX: days 3, 14 RIT: days 1–14	2.80	1.40	[38]
	DIG	0.4	200	DIG: days 1, 14 RIT: days 1–15	1.22	N/A	[68]
	DIG	0.5	600	DIG: day 3 RIT: days 1–11	1.86	N/A	[67]
	DIG	0.5	800	DIG : day 14 RIT : days 1–14	N/A	1.37 (sim) 1.16 (del)	[58]
	DAB	150	100	DAB: days 1, 14, 15 RIT: days 1–14	N/A	1.15 (sim) 0.71 (del)	[69]
St John’s wort (SJW)	FEX	25	500	FEX: days 1, 8, 17 SJW: days 8–16	N/A	1.01	[73]
	FEX	60	900	FEX: days 1, 3, 17 SJW: days 2–17	1.28	0.86	[70]
	DIG	0.25	120	DIG: days −2, −1, 1–11 SJW: days 2–10	N/A	0.84	[71]
	DIG	0.25	900	DIG: days 0, 14 SJW: days 1–14	N/A	0.77	[64]
	DIG	0.5 (2 days), then 0.25	900	DIG: days 1–15 SJW: days 6–15	0.99	0.75	[72]
	DIG	0.5	20 mL	DIG: days 1–21 SJW: days 8–21	N/A	0.92	[74]
			90			0.94
			500			1.01
			1000			0.98
			2000			0.82
			2500			0.95
			4000			0.74
			4950			0.75
Darunavir (DAR)/cobicistat (COB)	DAB	150	800/150	DAB: days 1, 4, 18 DAR/COB: days 4, 5–20	N/A	2.64 (day 4) 1.88 (day 18)	[75]
Darunavir (DAR)/ritonavir (RIT)	DAB	150	800/100	DAB: days 1, 4, 18 DAR/RIT: days 4, 5–20	N/A	1.72 (day 4) 1.18 (day 18)	[75]

DIG, digoxin; FEX, fexofenadine; DAB, dabigatran; RIF, rifampicin; RIT, ritonavir; DAR, darunavir; COB, cobicistat; day X, duration of inducer and substrate co-administration; sim, simultaneous intake; del, delayed intake; N/A, no available value

*Decrease in AUCR observed after staggered rifampicin administration; ^aR and S enantiomers

Our results revealed that in the CYP3A4-influenced group, percentage of the AUCR ≥ 2.0 or ≤ 0.5 was 38.0% and 18.4% for inhibitors and inducers, respectively, and 2.8% respectively. 11.1% for the group without CYP 3A4 influence.

Discussion

To the best of our knowledge, this study is the first systematic review to comprehensively list all P-gp inhibitors and inducers investigated in clinical trials involving healthy volunteers and to classify these interactors on the basis of their AUCR.

Our review of 151 clinical studies, including 69 RCTs and 82 NRCTs, revealed that most P-gp inhibitors and inducers have a limited impact on drug exposure. The majority (97%) of P-gp inhibitors only act as weak to no interactors in human studies. A minority of compounds with a sole effect on P-gp were classified as moderate inhibitors, resulting in a maximum AUCR of 3.05; the AUCR was, however, greater than 5.0 for compounds inhibiting both P-gp and CYP 3A4. With regard to P-gp induction, no AUCR reductions for sole P-gp inducers were below 0.20. Only two tested compounds were classified as moderate inducer (AUCR ranging from 0.20 to 0.50) and the remaining majority showed no inductive properties.

Our findings are consistent with other studies suggesting the moderate influence of drugs on P-gp inhibition, with AUCR values increasing from 1.2 to 2.0 [39, 40]. Likewise, the limited influence of P-gp induction found in our study—a maximal decrease in AUCR to 0.27—is in good agreement with the study by Elmeliegly et al., who reported a modest decrease in digoxin AUCR ranging from 0.80 to 0.33 with rifampicin and from 0.88 to 0.58 with carbamazepine [41]. These results indicate that P-gp induction typically reduces substrate exposure by one category less than CYP3A4 induction if based on the same classification system [41]. As previously mentioned, P-gp induction is regulated by transcriptional activation through nuclear receptor that also regulates CYP 3A4 expression [17]. This shared regulatory pathway explains why CYP3A4 inducers are often co-inducers of P-gp, although the magnitude of induction may differ [18]. Limited in vivo data suggest that transporters such as P-gp are generally less inducible than enzymes, including CYP3A4 [18, 42].

A particular case was observed with rifampicin, which appears to act as both a strong inhibitor and inducer, depending on the timing and context of its administration. Kusuhara et al. [43] demonstrated that a single or simultaneous administration of rifampicin with fexofenadine increased fexofenadine AUCR by more than three-fold, whereas separate or multiple administration of rifampicin led to a reduction in fexofenadine’s exposure. The simultaneous administration of rifampicin and fexofenadine primarily drives the intestinal P-gp inhibition, causing an increase in fexofenadine’s bioavailability and AUC. Rifampicin inhibition may also be related to inhibition of both intestinal and renal P-gp as well as organic anion-transporting polypeptides (OATP), responsible for the reduction of drug absorption, metabolisation and excretion. After multiple dosing and induction of the Pregnane X Receptor, rifampicin induction of P-gp and OATP dominates, shifting its role from an inhibitor to an inducer.⁴³ Fexofenadine, also known to be a substrate of OATP transporters [44], is particularly affected by these modulation mechanisms. The same observation, although to a lesser extent, was observed for ritonavir [45], which demonstrated a more substantial inhibitory effect compared with its inducing effect [46], St John’s wort (Hypericum perforatum) and darunavir/cobicistat and darunavir/ritonavir [47–49].

In this study, we chose to classify P-gp inhibitors and inducers into three levels of intensity, as proposed by the FDA guidelines to be aligned with the CYP categorisation. Given the modest effect of P-gp inhibitors alone on drug exposure, the binary classification of the FDA for P-gp inhibitors—AUCR < 1.5 and ≥ 1.5—could be appropriate. Yet, for compounds that inhibit or induce both P-gp and CYP3A4, a more nuanced three-tiered classification system would provide a more accurate quantification of the interaction potential. Additionally, neither the EMA nor the FDA provide any specific recommendations for P-gp induction studies or categorisation guidelines. On the basis of the present results, we could suggest using an AUCR cut-off of ≤ 0.50 to indicate a significant (moderate to major) P-gp induction, while an AUCR of > 0.50 would represent a minor P-gp induction. A guidance for P-gp and/or CYP induction classification would ensure a better alignment with the classifications used for inhibitors and offer clearer guidance for both researchers and clinicians.

From a clinical perspective, P-gp inhibitors alone tend to have minimal effect on P-gp substrate exposure, but when both P-gp and CYP3A4 are involved, the risk is significantly higher, especially when CYP3A4 modulation is strong. P-gp and CYP3A4 work synergistically to limit the intracellular accumulation of xenobiotics. The efflux of substrates by P-gp reduces the risk of exceeding the metabolic capacity of intracellular CYP enzymes, facilitating repeated substrate entry and efflux, thereby creating an efficient protective cycle. This mechanism, regulated by the PXR and retinoid X receptor (RXR), highlights the co-evolution of this vital defense system against potentially toxic compounds [10].

The clinical relevance of DDIs depends on the magnitude of the DDI effect, but also on additional factors. They include pharmacokinetic determinants such as protein binding, route of administration, drug–receptor interactions and therapeutic range. In addition, patient-specific variables, such as organ function (e.g. hepatic and renal impairment), age and comorbidities, play a critical role in determining the clinical impact of DDIs. Genetic polymorphisms in the ABCB1 gene, which encodes P-gp, may also impact the expression of P-gp transporters [50]. The complex interplay among these factors must be carefully considered in clinical practice to ensure a precise evaluation of the risk–benefit balance of a given interaction [51].

Our findings indicate that, for three drugs evaluated with both intestinal and renal P-gp substrates (clarithromycin, verapamil and grapefruit juice), AUCR values—and consequently, the classification of modulation—varied depending on the site of P-gp interaction. This complexity arises because a modulator acting predominantly at the intestinal level may have a limited impact on a renal P-gp substrate, and vice versa. Although the affinity of substrates and modulators for intestinal or renal P-gp should also be considered when interpreting the overall clinical impact of P-gp-mediated interactions, such information may not always be readily accessible in routine clinical settings.

Our comprehensive search of DDI involving P-gp has several limitations that should be mentioned. First, our study was limited to clinical studies using dabigatran, digoxin and fexofenadine, as previously recommended by the FDA and EMA guidelines at the time of data collection [24, 25]. Since 2023, the FDA guidelines recommend using edoxaban instead of fexofenadine as a P-gp probe due to fexofenadine's lack of specificity for P-gp, as it is also a substrate of OATP [44]. DDI evaluation in healthy volunteers using edoxaban has recently been conducted, indicating that, according to our study classification, dronedarone (AUCR 1.85) and quinidine (AUCR 1.77) are weak P-gp inhibitors and amiodarone (AUCR 1.40) has no influence on P-gp [52]. Second, several well-recognised P-gp inhibitors are missing from our systematic review, primarily due to the absence of published studies in healthy volunteers within the primary literature or because clinical studies were conducted using substrates not recognised by the FDA guidance. These include ciclosporin with a reported AUCR of 1.78 indicating modest P-gp inhibition [53] and fluoxetine [54], erythromycin [55] and the P-gp inducer valproic acid [56], for which no exposure values are available in the literature.

In conclusion, our systematic review provides a substantial basis for quantifiying P-gp interactions, which remains to be mostly clinically relevant for drugs inhibiting both P-gp and cytochromes. The proposed categorisation of the interaction potential based on clinical exposure assessed in robust clinical trials can be useful to inform safer co-prescription practices.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Funding

Open access funding provided by University of Geneva.

Conflict of interest

Claire Coumau and Chantal Csajka declare that they have no potential conflicts of interest that might be relevant to the contents of this manuscript.

Author contributions

C.C.O. and C.C.S. contributed to the concept and design of the systematic review. C.C.O. performed the literature search, data extraction, analysis, risk of bias assessment of the studies, and writing. C.C.S. contributed to the risk of bias assessment of the studies, and revised the manuscript critically. All authors contributed to the article and approved the submitted version.

Data availability statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.

Ethics approval

Not applicable.

Code availability

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.