CYP2C9 as gatekeeper of cannabinoid 11-hydroxylation: shunt pathways are inferred, not measured

Question: How do CYP2C9 genetic polymorphisms and drug inhibitors shift metabolic clearance of cannabinoids onto alternative CYP3A and CYP2C19 pathways in human liver microsome studies?

Answer

Across the retrieved papers, the evidence supports a narrower and more qualified answer than the question's framing assumes. The only study that directly addresses the mechanism shows, in human liver microsomes, that 11-hydroxylation of Δ8-THC — the quantitatively dominant oxidation — is formed predominantly by CYP2C9 with only a minor CYP2C19 contribution, that CYP2C9 genotype-based activity scores correlate with 11-OH-Δ8-THC formation across donors, and that CYP2C9*3/*3 carriers are predicted to lose ~80% of that formation clearance, while ring-hydroxylated metabolites are formed mainly by CYP3A [1]. The same study's conclusion that clearance in the presence of CYP2C9 inhibitors or polymorphisms "is likely mediated by CYP3A4 and CYP2C19" is explicitly framed as a prediction, not a measured redistribution of flux, and it concerns Δ8-THC only [1]. Independent human data do support a functional role for the alternative route in vivo, since a polygenic score for CYP3A4 expression was associated with Δ9-THC-induced psychotomimetic effects, consistent with faster CYP3A-mediated degradation [2]. Taken together, these papers support a model in which CYP2C9 governs the high-affinity 11-hydroxylation branch and CYP3A/CYP2C19 constitute a lower-affinity buffer that becomes relatively more important when CYP2C9 activity is lost — but the buffering capacity, the fractional shift, and the applicability to cannabidiol remain unquantified and largely untested in the retrieved literature.

CYP2C9 dominates 11-hydroxylation of THC isomers, with CYP2C19 as a minor rather than a rescue enzyme

In HLM, 11-OH-Δ8-THC was the most abundant metabolite detected, and its formation was attributed predominantly to CYP2C9 with a minor contribution from CYP2C19, assigned using recombinant enzymes, inhibition studies and correlation analyses [1]. Genotype-based activity scores across individual donors correlated with 11-OH-Δ8-THC formation (p = 0.0315), and an 80% decrease in formation clearance was predicted for CYP2C9*3/*3 carriers [1]. Unbound binding constants for metabolite formation spanned 3.6 nM to >121 nM, which the authors interpret as evidence that multiple enzymes contribute rather than a single high-affinity species [1]. The designation of CYP2C19 as "minor" is important: it constrains how much compensation a CYP2C19-mediated pathway can plausibly provide when CYP2C9 is compromised [1].

CYP3A carries the ring-hydroxylated branch, making it the structurally plausible overflow route

Two ring-hydroxylated metabolites (M3, M4) were formed mainly by CYP3A, and CYP2D6, CYP1A2 and CYP3A4 activities were associated with formation of the pentyl side-chain oxidized products M1 and M2 [1]. This regiochemical division of labour is the mechanistic basis for expecting that CYP2C9 blockade would not abolish clearance but would redirect it toward alternative oxidation sites rather than simply reducing total turnover [1]. The CYP3A branch is also the one with human support: CYP3A4 expression polygenic score was significantly associated with Δ9-THC-induced psychotomimetic effects, which the authors interpret as genetic liability to produce more CYP3A4 enzyme being associated with faster Δ9-THC degradation and diminished effects [2].

Route of administration changes how much 11-hydroxylation the alternative pathways must absorb

In a controlled trial in adults aged ≥65, oral THC produced delayed, lower exposure with 11-OH-THC concentrations approximately matching parent THC, whereas vaporized THC produced rapid, THC-dominant exposure with minimal 11-OH-THC [5]. Because first-pass hepatic 11-hydroxylation is far more prominent after oral dosing, any CYP2C9-dependent shunt onto CYP3A/CYP2C19 should be most consequential by the oral route — insofar as the shunt exists [5].

Genotype-to-activity mapping and HLM-to-clearance extrapolation both carry error that propagates into shunt predictions

Predictions of the kind made for CYP2C9*3/*3 depend on converting diplotype into activity, and homology-based variant-effect predictors perform poorly for pharmacogenes: AlphaMissense placed variants in its "ambiguous" class at nearly twice the proteome-wide rate across six clinically important CYPs, and within that class its scores were essentially uninformative about measured CYP2C9 deep mutational scan activity, with no tested method agreeing significantly with clinical annotations [4]. Separately, HLM-based CYP2C9 clearance is systematically biased because cytosolic liver fatty acid binding protein alters metabolite formation kinetics, a model including FABP1–CYP2C9 protein–protein interactions best fitted the data, and CYP2C9 plus FABP1 expression together best predicted probe hydroxylation across 61 human liver S9 fractions [3]. Both results imply that a fraction-metabolized value assigned to CYP2C9 in HLM, and therefore the inferred residual flux available to CYP3A/CYP2C19, carries uncertainty that is rarely propagated [3] [4].

Conflicts and gaps

The retrieved literature contains exactly one study that quantifies cannabinoid CYP2C9 pharmacogenetics in HLM, and it is a preprint on Δ8-THC, not Δ9-THC or cannabidiol [1]. Nothing retrieved tests a CYP2C9 inhibitor in HLM with cannabinoid substrates, so the "inhibitor" arm of the question has no direct evidence here. Two tensions are visible. First, the same study reports that C11 hydroxylation is a more important pathway for Δ8-THC than for Δ9-THC yet that CYP2C9 catalytic efficiency is lower toward Δ8-THC than Δ9-THC [1]; if the isoform's intrinsic efficiency differs between isomers, the magnitude of the CYP2C9-loss effect cannot be transferred between them without new data. Second, HLM attributes the dominant THC oxidation to CYP2C9 [1] whereas human in vivo genetic evidence implicates CYP3A4 expression in THC response [2]; catalytic efficiency and enzyme abundance are different determinants of clearance, and no retrieved paper reconciles the two. No retrieved paper reports CBD fraction metabolized, CBD 7-hydroxylation isoform assignments, or CBD genotype–exposure relationships.

What would change the answer

A direct test would measure pathway-resolved formation clearances (11-OH, ring-hydroxylated, side-chain oxidized) in the same HLM panel under vehicle versus selective CYP2C9 inhibition (for example sulfaphenazole) and selective CYP3A inhibition, reporting the fractional redistribution rather than inferring it from isoform assignment [1]. Repeating this in genotyped HLM panels with proteomic quantification of CYP2C9, CYP2C19, CYP3A4 and FABP1, following the design used across 61 human liver S9 fractions, would anchor the genotype-to-flux step and expose whether FABP1 binding blunts the expected shunt [3]. A clinical crossover in CYP2C9 poor versus normal metabolizers given oral THC or CBD, with 11-OH-THC/THC and 7-OH-CBD/CBD metabolite ratios as endpoints, would establish whether the alternative pathways actually absorb the lost flux in vivo, complementing the CYP3A4-expression approach already used with Δ9-THC infusion [2] [5]. Finally, replacing diplotype-implied activity scores with deep-mutational-scan-calibrated CYP2C9 activity would test whether the 80% loss prediction for *3/*3 survives better variant-effect modelling [4].

References

[1] Simon, K., Wen, Y. W., Dong, Y. E., Authement, A., Chen, J., Thummel, K. E., Isoherranen, N.. "Characterization of Cytochrome P450-Mediated Metabolism of Δ8-tetrahydrocannabinol in vitro". bioRxiv. 2026-10-19T00:00:00.000Z. URL: https://doi.org/10.64898/2026.10.12.758575

[2] Bright, U., Ganesh, S., Levey, D. F., Gupta, P., the Yale THC Studies Consortium,, Ranganathan, M., the IOP THC Studies Consortium,, Murray, R. M., DiForti, M., Morrison, P., D'Souza, D. C., Gelernter, J.. "Genomic Variation Predicts Real-Time Δ⁹-tetrahydrocannabinol Response in Humans". medRxiv. 2026-07-01T00:00:00.000Z. URL: https://doi.org/10.64898/2026.06.26.26356685

[3] Wen, Y. W., Friedman, E., Zelter, A., Yabut, K. C., Nath, A., MacCoss, M. J., Isoherranen, N.. "Liver Fatty Acid Binding Protein (FABP1) Affects Cytochrome P450 2C9 (CYP2C9)-Mediated Metabolism in Human Liver Models". bioRxiv. 2026-10-19T00:00:00.000Z. URL: https://doi.org/10.64898/2026.10.12.758325

[4] Xu, H., Samori, I., Nayar, G., Altman, R. B.. "Homology-Based Variant-Effect Predictors Break Down on Cytochrome P450 Pharmacogenes". bioRxiv. 2026-10-02T00:00:00.000Z. URL: https://doi.org/10.64898/2026.07.30.741616

[5] Costa, G. P. A., Asnes, S., Meyerovich, J., Eid, T., Nadim, H., Dwy, S., Gueorguieva, R., Riggs, M. M., Sofuoglu, M., Matthews, S., Nunes, J. C., De Aquino, J. P.. "Pharmacokinetics and Pharmacodynamics of Oral and Vaporized Δ9-Tetrahydrocannabinol in Older Adults". medRxiv. 2026-08-21T00:00:00.000Z. URL: https://doi.org/10.64898/2026.08.18.26360706