P-glycoprotein (ABCB1) Efflux

What it does, and why the two consequences are separate human data

P-glycoprotein is an ATP-dependent efflux pump with broad specificity for large, lipophilic, often cationic molecules. It is not a metabolising enzyme — it changes where a molecule is, not what it is — and that is the reason it belongs on this shelf despite being neither a CYP nor a conjugating enzyme: it sets the concentration that the enzymes then act on. Its two most consequential locations do different jobs and produce different clinical outcomes. In the apical membrane of the intestinal enterocyte, it pumps absorbed substrate back into the gut lumen, which lowers oral bioavailability directly and also, by returning the molecule to the enterocyte repeatedly, gives enterocyte CYP3A4 more chances to metabolise it — a futile cycling that amplifies first-pass loss beyond what either mechanism would do alone. In the brain capillary endothelium it is a principal component of the blood-brain barrier, pumping substrates back into the blood and keeping them out of the central nervous system. The consequence of inhibiting it there is a change in BRAIN exposure that can occur with little or no change in plasma concentration, which is a genuinely different kind of interaction: a plasma-level-based monitoring strategy cannot detect it. It is also present in the biliary canaliculus and the renal proximal tubule, where it contributes to elimination, and in the placenta, where it limits fetal exposure.

• Efflux, not metabolism: it relocates a molecule rather than transforming it.
• Gut consequence: lower oral bioavailability, amplified by repeated exposure to enterocyte CYP3A4.
• Blood-brain-barrier consequence: lower brain exposure, independent of plasma level.
• Biliary and renal consequence: contributes to elimination, so inhibition raises systemic exposure of substrates cleared that way.
• Placental consequence: part of why some drugs reach the fetus and others do not.
• A plasma concentration cannot tell you whether a blood-brain-barrier interaction has occurred.

Sources: Schinkel AH 2003 · Schinkel AH 1996 · Sadeque AJM 2000 · Greiner B 1999 · Wandel C 2002*

Substrates, with digoxin as the clinical index human data

Digoxin is the reference substrate for this transporter in clinical pharmacology, for three converging reasons: it is a good P-glycoprotein substrate, it is negligibly metabolised by CYP enzymes so a change in its exposure is attributable to transport rather than to metabolism, and it has a narrow therapeutic index so the change is clinically visible. If a study wants to demonstrate that an agent inhibits or induces P-glycoprotein in humans, digoxin is what it uses. The clinically loudest substrate on the list, though, is loperamide, and it is loud for the blood-brain-barrier reason rather than the absorption one: loperamide is a potent opioid agonist that produces no central effect at normal doses precisely because P-glycoprotein keeps it out of the brain, and blocking that pump has been shown to allow central opioid effects to appear.

Sources: U.S. Food 2023 · Schinkel AH 2003 · Greiner B 1999 · Sadeque AJM 2000 · Zhu HJ 2006 · Flockhart DA 2021

Inhibitors and inducers, drug and botanical contested human data

The drug inhibitors are largely the same names as the CYP3A4 inhibitors, and the botanical inhibitors are the same names as the CYP3A4 botanical inhibitors, which is the single most useful pattern on this page. Piperine is the clearest botanical case and the mechanism is explicit in the title of the paper that characterised it: piperine inhibits human P-glycoprotein AND CYP3A4. That dual action is the entire pharmacological basis of black pepper reputation as a bioavailability enhancer — it reduces efflux and reduces first-pass oxidation simultaneously, which is why the measured effect on curcumin bioavailability in human volunteers was around twentyfold rather than the modest increase either mechanism alone would give. The inducer side is dominated by two agents that hit ABCB1 and CYP3A4 together through PXR: rifampicin, in which intestinal P-glycoprotein induction was directly demonstrated as the mechanism of the digoxin interaction, and St John wort, where reduced digoxin exposure was measured with a hypericum extract.

Contested — caveat. Rifampicin, St John’s wort, verapamil, quinidine, clarithromycin and ritonavir are established in humans with digoxin or an equivalent probe. The flavonoid and cannabinoid rows are in-vitro determinations without human quantification. Note also that in-vitro P-glycoprotein inhibition potency does not track CYP3A4 inhibition potency even for agents that do both, so a strong CYP3A4 inhibitor is not automatically a strong P-glycoprotein inhibitor.

Sources: Bhardwaj RK 2002 · Shoba G 1998 · Volak LP 2008 · Johne A 1999 · Moore LB 2000 · Greiner B 1999 · U.S. Food 2023 · Zhu HJ 2006 · Bailey DG 2013 · Wandel C 2002* · Flockhart DA 2021

Why P-glycoprotein and CYP3A4 overlap so much, and what it costs human data

The overlap is not a coincidence and it has four independent causes stacked on top of each other. They are co-located: enterocyte CYP3A4 and apical P-glycoprotein are in the same cell, positioned as sequential barriers to the same absorbing molecule. They are co-regulated: both ABCB1 and CYP3A4 are PXR target genes, so an inducer of one is usually an inducer of the other. They share substrate preferences: large, lipophilic molecules that fit the accommodating CYP3A4 active site are the same molecules the pump recognises. And they share inhibitors for the same structural reason. The consequence is that a great many real-world interactions are dual-mechanism, and the effect is more than additive on oral exposure. Inhibiting the pump means more molecule enters the enterocyte and less is returned to the lumen for another metabolic pass; inhibiting the enzyme means more of what enters survives. The two effects multiply through the first-pass step. This is the mechanistic reason that grapefruit, ritonavir, clarithromycin and St John wort produce disproportionately large effects on oral bioavailability, and it is why a checker that models only CYP3A4 will systematically underpredict the magnitude of these interactions. The cost of the overlap, for anyone trying to attribute a mechanism, is that it is genuinely difficult to separate the two in a clinical observation — which is exactly why digoxin and dabigatran, as non-CYP substrates, and midazolam, as a non-P-glycoprotein substrate, are the probes used to pull them apart.

• Co-located in the enterocyte, co-regulated by PXR, overlapping substrate space, overlapping inhibitor space.
• Dual inhibition multiplies through the first-pass step rather than adding.
• Probes that isolate the mechanisms: digoxin and dabigatran for transport, midazolam for CYP3A4.
• Modelling only CYP3A4 underpredicts the size of dual-mechanism interactions.
• A single botanical (piperine, curcumin) or a single drug (ritonavir) hitting both is the highest-leverage perturbation in oral pharmacokinetics.

Sources: Bhardwaj RK 2002 · Wandel C 2002* · Greiner B 1999 · Moore LB 2000 · Thummel KE 1996 · Paine MF 2006

Clinical consequence, concretely human data

Digoxin plus clarithromycin, amiodarone, verapamil or ciclosporin: nausea, vomiting, confusion, visual disturbance with yellow-green haloes, bradycardia and arrhythmia, arriving days after the new drug started — and note that hypokalaemia makes digoxin more toxic at an unchanged level, so a diuretic in the mix compounds it by a completely separate route. Digoxin plus rifampicin or St John wort: the mirror event, loss of rate control or worsening heart failure with nothing visibly changed. Dabigatran plus a P-glycoprotein inhibitor: bleeding, with no INR to warn anyone because these agents are not monitored that way. Colchicine plus clarithromycin: severe, sometimes fatal toxicity from a drug with a very narrow margin, driven by simultaneous loss of both clearance routes. Loperamide plus quinidine or another pump inhibitor: central opioid effects from an over-the-counter antidiarrhoeal, which is a pharmacokinetic demonstration that a peripherally restricted drug is only peripherally restricted for as long as the barrier holds. And for this library the reading that generalises: any claim that a botanical increases the absorption of something is a claim about this transporter and about intestinal CYP3A4, and the same mechanism that raises the absorption of the intended substance raises the absorption of everything else in the medication list that shares the pathway.

Sources: Greiner B 1999 · Johne A 1999 · Sadeque AJM 2000 · U.S. Food 2023 · Flockhart DA 2021

Polymorphism contested human data

ABCB1 is polymorphic and the synonymous 3435C>T variant, along with 1236C>T and 2677G>T/A and the haplotypes they form, has been studied extensively for associations with digoxin exposure, antiepileptic response, immunosuppressant dosing and antidepressant outcome. The literature is inconsistent, with associations reported in one direction, the other, and not at all across different substrates and populations, and no ABCB1 genotype has reached clinically actionable status comparable to CYP2D6 or CYP2C19. Part of the difficulty is that the variant is synonymous and its effect, if any, is on messenger RNA stability or on co-translational folding rather than on the amino acid sequence, which makes a consistent functional effect harder to establish.

Contested — caveat. ABCB1 genotype associations are inconsistent across studies and substrates, with no established clinical guideline. Reported here as an active research question and explicitly not as an actionable marker.

Sources: Schinkel AH 2003 · Zanger UM 2013