Can CYP2C9 or CYP2C19 Genotyping Explain THC Blood Results After CBD-Rich Cannabis? A 27-Person Study

Short answer: Not reliably on the evidence of this study. In a 2026 analysis of 27 people who had participated in controlled experiments involving vaporized high-CBD, low-THC cannabis, the investigators linked genotype-predicted CYP2C9 and CYP2C19 activity to serial whole-blood measurements of THC, 11-hydroxy-THC (THC-OH), free THC-COOH, and CBD. The single-consumption group showed no statistically significant CYP2C9 or CYP2C19 group differences in the examined concentrations or metabolite-to-parent exposure ratios. Among just eight people in the repeated-consumption group, some individual CYP2C9 group comparisons at particular sampling times were significant, but the study did not establish a dependable genotype-based correction for THC concentration, timing of use, or impairment.

This finding comes from the complete original study by Schulte and colleagues (2026) in the International Journal of Legal Medicine. It is a small retrospective pharmacogenetic analysis of specimens and cannabinoid measurements collected for an earlier controlled study—not a trial designed to determine whether particular drivers were impaired.

What forensic question did the experiment address?

When a forensic laboratory measures THC and metabolites in a blood specimen, one possible source of biological variation is the person’s drug-metabolizing enzymes. CYP2C9 and CYP2C19 are members of the cytochrome P450 enzyme family. DNA variation can affect predicted enzyme activity, but a genotype-predicted activity category is not the same thing as an enzyme-activity measurement in the person’s liver at the time of exposure.

The investigators asked whether predicted normal, intermediate, rapid, or ultrarapid metabolizer categories corresponded to differences in measured cannabinoid concentrations or calculated exposure ratios after a controlled low-THC/high-CBD inhalation protocol. This is particularly relevant when a party argues that a laboratory THC concentration must have been unusually high or low because the person carries a particular genetic variant.

It is a much narrower question than whether genes explain all between-person variation in THC pharmacokinetics. Pulmonary uptake, timing, actual dose delivered by vaporization, distribution, and repeat exposure can vary independently of these two genotypes. The research also did not investigate the complete cannabinoid-metabolizing enzyme system.

Who was included, and why do the denominators matter?

The source trial had enrolled 60 adults across single- and repeated-consumption protocols, but only 27 had consented to additional use of their specimens and data for the pharmacogenetic analysis. Nineteen came from the single-consumption arm (11 men, 8 women), while eight came from the repeated-consumption arm (4 men, 4 women). These are separate subsets, not 27 participants who all underwent both protocols.

In the single-consumption source protocol, participants had received the study products in randomized crossover sessions separated by at least seven days. In the repeated-consumption source protocol, the assigned cannabis product was inhaled twice daily for ten days, with certain evening administrations and blood sampling supervised. The pharmacogenetic comparisons emphasized a product containing 14.6% CBD and 0.64% THC, identified in the source as Harley Quinn. It was administered as a 300-mg quantity of plant material for vaporization. The product’s labeled cannabinoid content is not proof of the quantity that reached systemic circulation.

This new genetic analysis was post hoc: investigators did not randomize people to a genotype, did not balance the genotype subgroups prospectively, and did not have a separately powered cohort of poor metabolizers. Repeated blood draws from one person also do not turn that person into several statistically independent participants.

What was measured in the blood, and what was only predicted?

Serial concentrations of THC, THC-OH, free THC-COOH, and CBD in whole blood were available from the original experimental case-report forms. The original source-study analytical method used gas chromatography–tandem mass spectrometry (GC–MS/MS). The pharmacogenetic report evaluates concentrations in ng/mL; its calculations of area under the concentration-time curve (AUC) have ng·min/mL units. Neither is a breath-alcohol reading or a forensic serum-to-whole-blood conversion.

Separately, the investigators extracted DNA from retained blood cellular material and identified selected CYP2C9 and CYP2C19 variants by TaqMan polymerase chain reaction genotyping. They then used allele-to-phenotype reference rules to predict metabolic categories. Those categories were genetic classifications; the study did not directly measure the corresponding enzyme’s activity in vivo.

The method measured the free, non-glucuronidated forms of the relevant THC metabolites for the reported cannabinoid comparisons. It did not separately quantify glucuronide conjugates or the key hydroxylated CBD metabolite, 7-OH-CBD. A claim about total conjugated THC-COOH or CBD metabolic clearance therefore cannot be established from these reported analyte concentrations alone.

How many participants fell into each genotype category?

Interpretive point: The totals in the second column refer to the same 27 people classified on two different genes. They must not be added together as independent groups. The smaller AUC denominator in the single-consumption group arose because three participants’ sampling stopped before the full intended interval. Apparent precision from numerous blood draws cannot overcome the very small genotype-group sizes.

What did the single-consumption study show?

Among the 19 single-consumption participants, no significant differences in pointwise THC, THC-OH, free THC-COOH, or CBD whole-blood concentrations were detected between the CYP2C9 normal- and intermediate-metabolizer groups at the reported sampling times (p > .05). The authors likewise did not identify statistically significant differences between CYP2C19 predicted-phenotype groups in those concentrations. Their tables of calculated exposure ratios and AUC comparisons were also nonsignificant.

The absence of statistical significance is not proof that genotypes have identical effects. The groups were small, within-group spread was substantial, and one CYP2C19 phenotype was represented by a single person. The study cannot exclude a clinically meaningful effect that it lacked power to detect, nor can it provide a generally valid interval correction for a future defendant’s result.

The exposure estimates and metabolite ratios

For CYP2C9, the authors reported mean THC AUC values of 255.4 ± 138.5 ng·min/mL for normal metabolizers (n = 9) and 288.4 ± 107.6 ng·min/mL for intermediate metabolizers (n = 7). For THC-OH the corresponding values were 56.1 ± 52.8 and 65.7 ± 53.9 ng·min/mL; for THC-COOH they were 513.0 ± 431.4 and 613.5 ± 303.4 ng·min/mL. These reported AUC group comparisons did not support a significant difference.

Calculated average ratios of THC-COOH AUC / THC AUC were 1.76 ± 1.14 (NM) and 2.04 ± 0.98 (IM), also nonsignificant. The ratio is a calculation from serial concentrations, not a compound separately assayed in a second specimen. It does not independently establish the hour of inhalation or a fixed genotype-specific metabolic rate.

What changed after repeated high-CBD cannabis use?

The repeated-consumption arm is the most tempting source of overstatement. It contained only eight people, divided into four CYP2C9 predicted normal and four intermediate metabolizers. At specific times the authors observed statistically significant between-group results: THC at 60 and 180 minutes and CBD at 10, 20, and 60 minutes after the observed inhalation. The study’s Figure 2 marks those selected comparisons.

But those individual time-point findings did not translate into statistically significant differences in the reported THC-to-metabolite AUC ratio comparisons. The repeated-exposure groups were too small to support an individual-level classifier. The authors themselves warn that the significant group effects in this four-versus-four comparison could be overestimated and that the source experiment did not measure CBD metabolites necessary to distinguish alternative CBD biotransformation explanations.

For example, mean THC AUC was 365.9 ± 130.8 versus 173.9 ± 50.9 ng·min/mL (NM versus IM, each n = 4), but the genotype-focused paper did not demonstrate a reliable person-specific THC adjustment or report a significant difference in the associated metabolite exposure ratios. Repeated testing of several analytes at several time points also creates an interpretive multiple-comparison concern; it is not a substitute for independent replication.

Evidence table: significant findings and important null results

The distinction between a nonsignificant statistical test and an untested comparison matters. The repeat-exposure CYP2C19 analysis cannot properly be summarized as having “proved no effect”; the authors declined formal statistical testing.

Does the analysis establish how CYP2C9 or CYP2C19 will behave in a defendant?

No. Genotyping can identify a person’s sequence at the chosen variants. The predicted enzyme phenotype follows from a reference classification, which was applied correctly for the study’s exploratory grouping, but it does not measure a person’s actual hepatic enzyme activity, inhaled dose, receptor response, or level of intoxication. In this cohort no CYP2C9 poor metabolizers were observed. Stronger claims about individuals with markedly impaired enzyme activity would go beyond the tested participants.

The experiment evaluated a CBD-dominant cannabis variety under a controlled inhalation protocol. It does not furnish validated adjustment equations for high-THC concentrates, edible products, other delivery systems, combinations with medications, or a person whose genetic and exposure details are unknown. Drug timing and repeated dosing could matter, but this pilot study cannot quantify their relative contribution for any individual case.

Why whole blood is not serum, a metabolite ratio, or a roadside behavior test

The laboratory’s whole-blood GC–MS/MS results are direct measurements of particular cannabinoids in a defined physical matrix. The DNA test is a separate molecular examination of collected blood material. The phenotype category is a prediction derived from DNA; AUC and parent/metabolite ratios are calculations across observations. None is interchangeable with the other.

For a more general explanation of those specimen boundaries, see biological matrices in forensic toxicology. A measured blood THC value cannot simply be equated to a serum value from another study. Likewise, a calculated AUC ratio does not tell a court what an eyewitness observed, whether a driver crossed a lane marker, or whether a traffic collision was caused by a drug.

The source analysis did not assign individual impairment classifications from CYP2C9 or CYP2C19 genotype. It did not validate sensitivity, specificity, positive predictive value, or a legally defensible threshold for those categories. For principles of analyte identification and method-specific confirmation, see confirmatory drug testing.

Why the reported blood THC-COOH is specifically the free fraction

THC-COOH can occur in free and conjugated forms. The source method measured free THC-COOH, alongside THC, THC-OH, and CBD. The investigators did not apply a separate validated quantitation scheme for THC-COOH glucuronide or other phase-II conjugates. A statement about total THC-COOH from this dataset would conflate measured and unmeasured fractions.

Similarly, although the paper discusses CYP2C19’s involvement in CBD metabolism, it did not assay a full CBD metabolite profile. Thus, a small null result for the measured parent CBD concentration cannot establish that CYP2C19 plays no role in CBD metabolism. These are limits of the measured analyte panel, not reasons to ignore the reported observations.

What are the main design limitations?

Small, selected population: The genetic study included only 27 consented people out of an original 60. The original exclusion criteria and selection for specimen reuse may make it difficult to generalize beyond this sample. The repeated-consumption CYP2C9 comparison had four people per category.

Retrospective grouping: The participant’s genotype was not randomized or manipulated. Apparent group differences therefore cannot isolate the causal contribution of enzyme variants from inhalation behavior, dosing, and other unmeasured influences.

Unrepresented or tiny categories: There were no CYP2C9 poor metabolizers. There was only one CYP2C19 ultrarapid metabolizer overall, and only two participants each in some repeated-exposure CYP2C19 strata.

Limited cannabinoid panel: The measured results were whole-blood THC, THC-OH, free THC-COOH, and CBD. They did not include the full conjugated cannabinoid panel or hydroxylated CBD metabolites needed to test every proposed mechanism.

Multiple exploratory comparisons: Several analytes and sampling times were examined. A few significant pointwise results, particularly with n = 4 in each of two groups, are not the same as a robust, externally validated prediction rule.

Pharmacokinetics, not individual effects: The paper’s new outcome was a genotype–concentration association. It was not a randomized experiment on criminal responsibility, actual road safety, individual impairment, or causation. The reported findings require evaluation within the boundaries of this protocol.

How should a forensic reviewer evaluate a genotype-based claim?

A scientifically useful review would first separate the measured facts from the proposed explanatory model. Which biological specimen was actually tested, and did the method measure free or total analytes? Were THC, THC-OH, THC-COOH, and CBD reported from one contemporaneous specimen, or are values being combined across sources and times? Was the claimed genotype actually tested for the person, or merely proposed as a possible explanation?

Next, the reviewer would ask whether the person’s sequence was translated into a validated enzyme phenotype, whether the source population contained that phenotype, and whether the source tested the same route, exposure pattern, and analyte. Even with genotyping, a phenotype category does not by itself specify the time course of a case-related concentration.

Finally, an extrapolation to impairment or causal involvement would require independent event-specific information and an appropriate validated interpretive framework. For the broader reasons that absorption, distribution, metabolism and elimination cannot generally be collapsed into one case-specific number, see pharmacokinetics and forensic interpretation.

Frequently asked questions

Can CYP2C9 testing tell a court why someone’s THC concentration was high?

Not from this study alone. The researchers tested whether categories predicted from variants were associated with whole-blood cannabinoid observations. The sample was small, single-consumption comparisons were nonsignificant, and repeated-consumption pointwise signals did not produce a validated individual concentration model.

Did this study find absolutely no genetic associations?

No. In the eight-person repeated-use arm, several isolated CYP2C9 comparisons of THC and CBD concentrations at specified times met p < .05. That exploratory result must be reported alongside the four-versus-four denominator, the nonsignificant AUC ratios, and the lack of individual predictive validation.

Did CYP2C19 have no effect on CBD metabolism?

The single-exposure measured-parent-CBD comparison was not significant, while repeat-exposure CYP2C19 statistics were not performed. The investigators did not measure 7-OH-CBD or a full CBD metabolic profile; therefore the stronger conclusion about all CBD metabolism is unsupported.

Were the THC levels measured in whole blood or serum?

The genotype analysis reused GC–MS/MS cannabinoid concentrations reported for whole blood. The DNA was recovered from blood cellular material. The analysis did not establish or apply a serum-to-whole-blood conversion.

Does a THC-COOH-to-THC ratio directly measure when cannabis was consumed?

No. A ratio is computed from measured concentrations or exposure estimates. Its interpretation depends on specimen type, dose, route, repeated exposure, timing, and metabolism. This study did not validate a genotype-specific clock for an individual case.

How many participants received cannabis repeatedly?

Eight of the 27 in the genetic analysis were drawn from the repeated-use protocol. They were not eight people in each genetic group; CYP2C9 comparisons were four versus four, and the CYP2C19 categories were four, two, and two.

Can these data establish that a person was impaired when driving?

No. Genotype and blood cannabinoid concentrations are different types of evidence from observed psychomotor performance. This pharmacogenetic sub-analysis did not validate an impairment decision rule or identify crash causation in particular people.

Primary scientific source

Schulte J, Potzel L, Frei P, et al. Assessing the influence of CYP2C9 and CYP2C19 genotypes on the metabolism of CBD-cannabis after controlled single and repetitive consumption. International Journal of Legal Medicine. 2026;140:1299–1311. doi:10.1007/s00414-025-03708-7. The statements above are confined to this original study’s measured analytes, defined sample, reported statistical comparisons, and documented limitations.