UGT Glucuronidation (Phase 2)

What glucuronidation is, and why phase 2 gets left out human data

Glucuronidation transfers glucuronic acid from the cofactor UDP-glucuronic acid onto a nucleophilic group on the substrate — a phenolic or aliphatic hydroxyl, a carboxylic acid, an amine, sometimes a thiol — producing a large, acidic, highly water-soluble conjugate that is excreted in urine or bile. Quantitatively it is the single largest conjugation pathway in human drug metabolism, and it is the terminal step for a great many phase-1 metabolites as well as for drugs that need no oxidation at all. The UGT enzymes sit in the endoplasmic reticulum with their active sites facing the lumen, which means the substrate and the cofactor both have to be transported in; that topology is part of why in-vitro UGT work is technically awkward and why the kinetic literature is thinner and noisier than the CYP literature. That technical difficulty is most of the answer to why phase 2 is missing from interaction checkers. The rest of the answer is historical: the drug-interaction field grew up around CYP3A4 and inherited its vocabulary, so a UGT interaction has no potency class, no index substrate consensus, and no regulatory classification scheme comparable to the FDA CYP table. The consequence is a systematic blind spot, and it falls precisely on the drugs that escape CYP metabolism — which is to say on lorazepam, oxazepam and temazepam, on morphine, on lamotrigine, on mycophenolate, and on the cannabinoids.

• High capacity, low affinity: glucuronidation is hard to saturate, which is why it carries the bulk of a large dose.
• Glucuronides can be hydrolysed back to the parent by bacterial beta-glucuronidase in the gut, producing enterohepatic recirculation and a second exposure peak.
• Acyl glucuronides of carboxylic acids are chemically reactive and are implicated in idiosyncratic toxicity — a conjugate that is not simply inert.
• Being glucuronidated rather than oxidised is a form of interaction resistance, which is exactly why lorazepam and oxazepam are preferred in liver impairment and in polypharmacy.

Sources: Rowland A 2013 · Clarke DJ 1994 · Flockhart DA 2021

The isoforms that matter human data

Four UGTs carry most of the clinically important work, and each has a signature substrate that makes it memorable.

Sources: Rowland A 2013 · Mazur A 2009 · Innocenti F 2004 · Anderson GD 1996 · Clarke DJ 1994

Why cannabinoids are heavily glucuronidated, and what 11-OH-THC and THC-COOH-glucuronide mean human data

Δ9-THC is a lipophilic phenol, which makes it an almost ideal glucuronidation substrate once phase 1 has worked on it, and the sequence is worth following because both pharmacology and drug testing hang on it. CYP2C9 (with a CYP3A4 contribution) hydroxylates THC at the 11 position to give 11-hydroxy-THC, which is itself psychoactive — at least as potent as the parent and arguably more so at the CB1 receptor — and which crosses into the brain readily. Oral dosing produces far more 11-hydroxy-THC than inhalation does, because an oral dose passes through gut and liver before reaching the circulation, and that single fact explains most of the difference in character, onset and duration between an edible and an inhaled dose better than any claim about the plant does. 11-hydroxy-THC is then oxidised further to 11-nor-9-carboxy-THC (THC-COOH), which is not psychoactive, and THC-COOH is glucuronidated — principally by UGT1A9, with UGT1A3 and UGT2B7 contributions — to THC-COOH-glucuronide. That glucuronide is the dominant urinary species and it is what a urine cannabinoid screen is actually measuring: immunoassays are calibrated against the carboxy metabolite, and confirmatory chromatography typically requires an enzymatic or alkaline hydrolysis step to cleave the glucuronide before the free acid can be quantified. Two consequences follow that are widely misunderstood. First, a positive urine test indicates a metabolite of a metabolite and carries no information about current impairment. Second, the very long detection window in frequent users is a property of THC sequestration in adipose tissue with slow redistribution, not of the glucuronide being unusually persistent — the conjugate is cleared quickly once formed, but formation continues for as long as THC is being released from fat. Cannabidiol follows an analogous path: oxidation by CYP2C19 and CYP3A4, then extensive glucuronidation by UGT1A9, UGT2B7 and UGT2B17, with direct glucuronidation of the parent phenol as well.

• THC to 11-hydroxy-THC: CYP2C9 dominant. The metabolite is psychoactive and the oral route generates much more of it.
• 11-hydroxy-THC to THC-COOH: not psychoactive.
• THC-COOH to THC-COOH-glucuronide: UGT1A9 principally, with UGT1A3 and UGT2B7. This is the urine analyte.
• Confirmatory testing hydrolyses the glucuronide first, which is why free and total carboxy-THC are different numbers.
• A positive urine test measures history, not impairment.
• CBD is glucuronidated by UGT1A9, UGT2B7 and UGT2B17, and is also a direct substrate as a phenol.

Sources: Mazur A 2009 · Huestis MA 2007 · Scheidweiler KB 2013 · Sachse-Seeboth C 2009 · Watanabe K 2007 · Jiang R 2011

UGT inhibition by cannabidiol and by curcuminoids contested in vitro

Both of the phase-2 inhibitors that matter most for this library are botanical. Curcuminoids were shown, in the same comparative study that characterised their CYP inhibition, to inhibit UDP-glucuronosyltransferase and sulfotransferase activity as well — which makes turmeric one of the few commonly consumed substances that perturbs phase 1 and both major phase-2 pathways at once, and means a turmeric interaction cannot be predicted from a CYP table alone. Cannabidiol inhibits UGT isoforms including UGT1A9 and UGT2B7 in vitro, which is mechanistically important for two reasons: UGT2B7 is the morphine enzyme, and UGT1A9 handles mycophenolate and propofol. Cannabidiol metabolites have themselves been characterised as inhibitors of drug-metabolising enzymes. None of these has been converted into a well-quantified clinical interaction study on the UGT side, which is exactly the blind spot this page exists to name: the in-vitro evidence that a widely consumed cannabinoid inhibits the enzyme that conjugates morphine has been available for years without a corresponding human study.

Contested — caveat. The valproate and lamotrigine, anticonvulsant-induction and oral-contraceptive rows are established in humans. The cannabidiol and curcuminoid UGT rows are in-vitro determinations with no controlled human pharmacokinetic confirmation, and because phase-2 interactions have no regulatory potency classification the class labels here are the authors reading of the sourced data rather than an FDA category.

Sources: Volak LP 2008 · Bahramsoltani R 2017 · Mazur A 2009 · Jiang R 2011 · Nasrin S 2021 · Anderson GD 1996 · Rowland A 2013 · Moore LB 2000 · Anderson GD 2016 · Stout SM 2014

Two worked examples: valproate with lamotrigine, and irinotecan with UGT1A1 human data

Valproate inhibits the glucuronidation of lamotrigine, principally at UGT1A4, and the effect is large enough to be built into prescribing: co-administration roughly doubles lamotrigine exposure and approximately doubles its half-life, so lamotrigine starting doses and escalation steps are halved in patients taking valproate. This is not a footnote — lamotrigine carries a risk of serious rash including Stevens-Johnson syndrome that is associated with rapid dose escalation, so the interaction converts a titration schedule into a safety issue. The mirror interaction also exists: carbamazepine, phenytoin, phenobarbital and oral ethinylestradiol INDUCE lamotrigine glucuronidation and roughly halve its concentration, which is why lamotrigine dosing is one of the few places where clinicians routinely think in phase-2 terms at all, and why stopping an oral contraceptive can produce lamotrigine toxicity. Irinotecan is the pharmacogenetic example. Irinotecan is converted to the potent topoisomerase inhibitor SN-38, and SN-38 is inactivated by UGT1A1 glucuronidation; reduced-function UGT1A1 genotypes, principally the star-28 promoter repeat and star-6 in East Asian populations, slow that inactivation and are associated with severe neutropenia and diarrhoea. The association was strong enough that UGT1A1 genotype information entered the irinotecan label, making it one of the earliest phase-2 pharmacogenetic markers in routine oncology. Both examples make the same point: phase 2 produces interactions and genetic variation of exactly the same clinical seriousness as phase 1, and it is absent from most interaction tools.

• Valproate plus lamotrigine: exposure roughly doubled, dosing halved, rash risk tied to escalation speed.
• Carbamazepine, phenytoin, phenobarbital or ethinylestradiol plus lamotrigine: exposure roughly halved.
• Stopping an inducer is the reciprocal event and can produce toxicity with no dose change.
• UGT1A1 star-28 and star-6: reduced SN-38 inactivation, severe irinotecan neutropenia, genotype information on the label.
• Gilbert syndrome is the benign phenotype of the same reduced UGT1A1 function, which is why an incidental mildly raised unconjugated bilirubin is a hint about this pathway.

Sources: Anderson GD 1996 · Innocenti F 2004 · Ando Y 2000* · Rowland A 2013 · Flockhart DA 2021