The Last Step of THC Metabolism: What Glucuronidation Explains About Drug Tests and Drug Interactions

Phase I oxidation turns THC into 11-hydroxy-THC and then into 11-nor-9-carboxy-THC. None of those are water soluble enough to be excreted efficiently. Conjugation with glucuronic acid by UDP-glucuronosyltransferase (UGT) enzymes is what converts them into something the kidney and the bile can move.

In human enzyme work published in Drug Metabolism and Disposition, THC itself was a poor direct substrate for the UGTs tested. The molecule generally has to be oxidized first. That dependency is why phase I and phase II are not two separate stories but one sequence.

Patients ask two questions constantly: how long will this show up on a test, and will it interfere with my other medications. Both answers run through glucuronidation, and both are commonly answered with folklore.

The in vitro evidence that cannabinoids inhibit UGT enzymes is now reasonably consistent. Whether that inhibition translates into clinically meaningful interactions in patients is a separate question, and the honest answer today is that it has been modeled rather than measured.

The most direct characterization of this pathway comes from Mazur and colleagues, publishing in Drug Metabolism and Disposition in 2009. They screened twelve recombinant human UGT enzymes against a panel of classic cannabinoids: cannabinol, cannabidiol, delta-8-THC, delta-9-THC, 11-hydroxy-THC, and 11-nor-9-carboxy-THC.

The pattern that emerged is more selective than the textbook picture. Cannabinol was the most widely recognized substrate, handled by hepatic UGT1A9 and by the extrahepatic enzymes UGT1A7, UGT1A8, and UGT1A10. Oxidation of THC to 11-hydroxy-THC created a substrate for UGT1A9 and UGT1A10. Further oxidation to 11-nor-9-carboxy-THC then removed that activity and instead created a substrate recognized by UGT1A1 and UGT1A3.

That last handoff matters, because the glucuronide of 11-nor-9-carboxy-THC is the principal cannabinoid metabolite found in urine. The enzymes that form it are hepatic. The authors framed their own conclusion carefully: glucuronidation of cannabinoids depends on upstream processing by enzymes such as CYP2C9 and CYP3A4. Phase II cannot compensate for what phase I has not yet produced.

Controlled human dosing studies confirm that the conjugates, not the parent compounds, are what persist. Desrosiers and colleagues, writing in Clinical Chemistry in 2014, had frequent and occasional cannabis smokers consume a single 6.8 percent THC cigarette on a closed research unit and quantified every urine void by liquid chromatography tandem mass spectrometry.

No urine sample contained measurable free THC, 11-hydroxy-THC, cannabidiol, or cannabinol. What was measurable was 11-nor-9-carboxy-THC, THC-glucuronide, and 11-nor-9-carboxy-THC-glucuronide, present in every frequent smoker’s urine. The presence of THC-glucuronide is worth noting alongside the enzyme work: direct conjugation of THC is inefficient relative to conjugation of its oxidized metabolites, but it is not absent, and the conjugate is detectable in vivo. THC-glucuronide concentrations peaked between 0.6 and 7.4 hours after smoking, which is why the research group proposed it as a candidate marker of recent intake rather than of cumulative exposure.

Route changes the ratio. Newmeyer and colleagues, in Clinical Chemistry in 2016, dosed occasional and frequent users with smoked, vaporized, and oral cannabis in a controlled crossover. Smoked and vaporized cannabis produced broadly similar blood cannabinoid profiles. Oral cannabis produced significantly greater concentrations of both 11-nor-9-carboxy-THC and its glucuronide, which is the pharmacokinetic fingerprint of first-pass metabolism feeding the conjugation pathway more substrate.

A population pharmacokinetic analysis by Sempio and colleagues in Drug Testing and Analysis in 2022 modeled urine collected over 168 hours from six subjects who smoked low-dose (15.8 mg) and high-dose (33.8 mg) THC cigarettes. Total urinary excretion of 11-nor-9-carboxy-THC was estimated at 2.3 percent of the dose. The authors put the working figure at roughly 2 to 3 percent of bioavailable THC leaving as that metabolite and its glucuronide in urine.

That is a small fraction, and it reframes what a urine screen is measuring. A positive result reflects a trace output stream from a much larger body burden, filtered through whatever hydration state the person happens to be in. The same analysis found that urinary creatinine excretion rate was a significant covariate for urinary clearance of the metabolite, and concluded that correcting drug concentrations for creatinine or specific gravity may be more problematic than previously appreciated.

The older mass-balance work points the same direction. Wall and colleagues, in Clinical Pharmacology and Therapeutics in 1983, gave THC intravenously and orally to men and women and tracked recovery. At 72 hours, cumulative urinary excretion ranged from 13 to 17 percent of the total dose across both routes and both sexes. Cumulative fecal excretion was 25 to 30 percent after intravenous dosing and 48 to 53 percent after oral dosing. The gut, not the kidney, is the main exit.

Enterohepatic recirculation appears in nearly every popular explanation of why THC lingers. It is worth being precise about where that claim comes from. The quantitative evidence is Garrett and Hunt’s 1977 work in the Journal of Pharmaceutical Sciences, performed in three dogs, two of them biliary cannulated, given radiolabeled THC intravenously.

In that model, 55 percent of the dose appeared in bile within five days, 40 to 45 percent in feces, and 14 to 16.5 percent in urine. The authors concluded that their data were consistent with enterohepatic recirculation of 10 to 15 percent of the metabolites. They also concluded something less often quoted: that the rate-limiting step in THC elimination was the return of THC from tissue, not the recycling loop.

No equivalent cannulated mass-balance study exists in humans. The human evidence for substantial biliary output is indirect, resting on the large fecal recovery Wall and colleagues measured. Describing enterohepatic recirculation of THC as an established human mechanism overstates what has been shown. Describing it as a plausible contributor supported by animal data and consistent with human fecal recovery is accurate.

The clinically interesting direction is not how cannabinoids are conjugated but what they do to the conjugation of other drugs. Nasrin and colleagues, in Drug Metabolism and Disposition in 2021, tested THC, cannabidiol, and cannabinol against recombinant UGTs and against microsomes from human liver and kidney. Cannabidiol was the strongest inhibitor, with binding-corrected IC50 values of 0.12 micromolar against UGT1A9, 0.22 against UGT2B4, 0.40 against UGT1A6, and 0.82 against UGT2B7. THC and cannabinol inhibited UGT1A9 at 0.45 and 0.51 micromolar respectively.

The kidney finding deserves attention. The authors noted that all three major kidney UGTs were inhibited, and raised the possibility of larger interaction effects with drugs cleared renally, and in patients with impaired hepatic or renal function.

Two follow-up studies from the same laboratory took a specific drug pair. Coates and colleagues, in Pharmaceutics in 2024, found that THC and cannabidiol inhibited UGT2B7-mediated morphine glucuronidation with Ki values of 0.37 and 0.16 micromolar. A 2025 study in Drug Metabolism and Disposition repeated the exercise for hydromorphone, with Ki values from 0.068 to 1.01 micromolar after correction for nonspecific binding, and physiologically based pharmacokinetic modeling predicting a 20 to 30 percent increase in hydromorphone exposure from cannabidiol in healthy and cirrhotic individuals.

Every one of those numbers comes from a test tube and a model. None comes from dosing patients. That is the correct place to stop.

The enzymology is solid. Identifying which recombinant UGT isoforms conjugate which cannabinoid metabolites is a well-controlled in vitro question, and the Mazur work has held up. The human excretion data are also strong, because controlled-dosing studies on closed research units with mass spectrometric quantification are about as clean as human pharmacokinetics gets.

The interaction evidence is a different tier. In vitro inhibition constants and physiologically based pharmacokinetic predictions are legitimate science and are the accepted first step in regulatory drug interaction assessment. They are not clinical outcomes. No published trial has dosed patients with cannabis and a UGT substrate and measured what happened.

Recombinant enzyme systems overstate selectivity. Real hepatocytes contain competing pathways, transporters, and endogenous substrates that a transfected cell line does not. Inhibition constants measured against a single isoform in isolation can look more dramatic than the same chemistry does in a whole liver.

Cannabinoids bind nonspecifically to plasticware and to microsomal protein, which is why the better studies report binding-corrected values. Older work that did not correct for this tends to understate potency, and comparing across studies that handled it differently is a real source of confusion.

The urine excretion percentages come from small samples. The Sempio population analysis modeled six subjects. The Wall mass-balance study was conducted in 1983 with the analytical methods of that era. Both are the best available data on their questions, which is not the same as being precise.

This literature does not show that cannabis use causes clinically significant harm through UGT inhibition in patients. It shows that the inhibition occurs in vitro at concentrations that modeling suggests are reachable, which is a reason to watch rather than a reason to conclude.

It does not establish how much interindividual variability in UGT activity, whether genetic or acquired, changes cannabis clearance. UGT polymorphisms are common, but no study has connected a specific UGT genotype to a measured difference in THC metabolite handling in cannabis users.

It does not support any of the claims circulating about accelerating glucuronidation to clear a drug test faster. Nothing in this pathway has been shown to be usefully manipulable by diet, supplements, or fluid loading.

Cannabis pharmacology discussions tend to stop at CB1 and CB2 receptors, or at CYP2C9 if they get as far as metabolism. Conjugation is the unglamorous back half of the process, and it is where the practical questions actually resolve: why a urine screen stays positive for weeks in a daily user, why oral and inhaled routes leave different metabolite signatures, and why the kidney matters.

The UGT inhibition work also sits inside a larger shift in how cannabis drug interactions are assessed. For years the conversation was entirely about cytochrome P450. Extending it to phase II enzymes is appropriate and overdue, particularly for patients on opioids, where morphine and hydromorphone are both UGT2B7 substrates and co-use is common.

The question I get most often about this pathway is not asked in these words. It is asked as: why did I fail a drug test three weeks after I stopped. The answer is that you were not tested for THC. You were tested for a conjugated acid metabolite that your liver keeps producing as THC slowly leaves fat, and that metabolite is a lagging indicator of exposure, not a measure of impairment.

The interaction piece I treat differently. When a patient on morphine or hydromorphone tells me they are also taking a meaningful daily dose of CBD, I pay attention. Not because I have outcome data, because I do not. Because the in vitro potency is high enough and the substrate overlap specific enough that I would rather be watching for sedation than be surprised by it. That is a clinical posture, not a claim about proven risk.

What I try not to do is dress up mechanism as evidence. A patient deserves to know which part of this is measured chemistry and which part is inference.

Glucuronidation is the step that makes THC metabolites excretable, and it explains the two things patients ask about most. Urine testing detects a conjugated acid metabolite, not THC, and accounts for only about 2 to 3 percent of the dose, which is why it tracks exposure history rather than current effect. Cannabinoids inhibit several UGT enzymes potently in vitro, so a patient on a UGT2B7 substrate such as morphine or hydromorphone warrants closer monitoring, even though no human interaction study has been done.

Keep three things and discard one. Keep that conjugation, not oxidation, is what allows excretion. Keep that the gut carries more of the eliminated dose than the kidney does. Keep that in vitro UGT inhibition by cannabidiol is potent and specific enough to justify monitoring. Discard the confident claim that enterohepatic recirculation of THC is an established human mechanism, because the quantitative data behind it come from three dogs.