Can Blood-Lymphocyte CB2 or DNMT mRNA Prove Recent Cannabis Impairment? A Placebo-Controlled Smoking Study

Short answer: No. A small randomized, double-blind human study found differences in how selected blood-lymphocyte messenger RNA (mRNA) measurements changed after smoking cannabis cigarettes containing different percentages of THC. The strongest overall three-group finding concerned cannabinoid receptor 2 (CB2) mRNA. But these measurements were not validated as tests of when a person consumed cannabis, the amount taken, or whether that person was impaired while driving. The study also found no consistent relationship between its lymphocyte mRNA changes and measured whole-blood THC concentration changes. These are important limits for anyone considering gene-expression evidence in a forensic case.

The original source is Smith and colleagues’ 2022 randomized-smoking sub-study in Frontiers in Psychiatry. It investigated peripheral immune-cell transcription and other biological measures, not a validated diagnostic test for cannabis-related driving impairment or a method of reconstructing an individual’s cannabis dose.

What forensic question did the investigators actually study?

The central laboratory question was whether controlled assignment to cigarettes labeled with different THC percentages produced measurable changes in selected mRNAs in lymphocytes isolated from blood. Candidate transcripts included cannabinoid receptor genes, DNA-methyltransferase genes, genes involved in DNA demethylation, and inflammation-related targets. The research tested differences between experimental groups at defined collection times, not whether a single unknown roadside blood specimen could correctly identify a drug dose or establish an impaired driver.

There were related measurements of whole-blood THC and THC-COOH, participants’ ratings of feeling “high,” and simulator-derived driving performance. These measurements address different evidentiary propositions. An assay of peripheral-cell gene expression is not itself an assay of THC concentration, a direct measure of neuronal activity, a functional driving test, or evidence that a particular collision occurred because of a drug.

How many people actually contributed to the mRNA comparison?

Twenty-four cannabis-using adults entered this laboratory sub-study from a larger randomized smoking experiment. The assigned arms were a placebo cigarette labeled 0.02% THC (11 participants), a 5.9% THC cigarette (five initially), and a 13.4% THC cigarette (eight). One member of the 5.9% arm lacked detectable THC after smoking and was excluded from the authors’ analysis. Therefore the analyzed group counts were 11, four, and eight—23 people total. The 191-participant parent driving study must not be substituted for this much smaller molecular-analysis denominator.

All participants reported prior cannabis use and were asked to abstain for two days before the experimental visit. Cigarettes were smoked ad libitum, with instructions approximating ordinary smoking to reach a desired high. The labeled THC percentages therefore described the cannabis material, not identical absorbed THC doses. A higher labeled percentage cannot be assumed to have produced a proportionally higher systemic dose.

The study observed material pre-smoking differences. The group assigned the 13.4% cigarette was older on average than the other groups, and the 5.9% group had the highest baseline CB2 mRNA value. Those differences matter when assessing subsequent between-arm contrasts, particularly with only four analyzed participants in one active group.

What specimens and assays were used?

The investigators collected venous blood at baseline and specified intervals after smoking. Separate assays served separate purposes. Whole-blood THC and THC-COOH were measured by liquid chromatography–mass spectrometry under previously described validated methods. For gene-expression work, a lymphocyte layer was separated from blood, RNA was extracted, converted to complementary DNA, and analyzed by quantitative polymerase chain reaction (qPCR) with TaqMan probes. The authors normalized the target transcript results to beta-actin and used relative expression calculations.

These qPCR measurements describe relative mRNA abundance in isolated peripheral blood cells. They do not directly count CB2 receptor proteins on cells, measure DNA methylation at a particular genomic locus, or measure THC in brain tissue. The mathematical expression units and baseline-change scores should not be reported as ng/mL of a compound, nanograms of THC, or a probability of impairment.

The authors’ main mRNA comparisons occurred at about 55 minutes, 80 minutes, and four hours after smoking. Whole-blood THC was also assessed at an early interval near ten minutes, when active-group concentrations tended to peak. Missing the early lymphocyte specimen at that peak means an mRNA value at 55 minutes is not simultaneous with the reported ten-minute peak blood THC.

Which CB2 mRNA results were statistically significant?

CB2 mRNA change from an individual’s own baseline differed across the three analyzed groups at all three reported post-smoking transcription time points. Importantly, the 5.9% group generally showed decreases while the 13.4% group showed increases, despite the study finding no statistically significant difference in whole-blood THC concentrations between the two active arms. This does not establish a monotonic dose-response biomarker.

Table interpretation: These are published group means and standard deviations of baseline-adjusted relative mRNA-expression scores, not cannabinoid concentrations or clinically validated cutoffs. There were 11 placebo participants, four in the 5.9% arm, and eight in the 13.4% arm. A statistically significant overall F test identifies a difference somewhere among the small groups; it does not classify individual people as impaired.

The pretreatment CB2 group values already differed, with placebo 248.77 ± 179.16, the 5.9% group 419.42 ± 262.17, and the 13.4% group 148.10 ± 73.42 (overall p = 0.050). Subtracting baseline can help describe changes but cannot make non-equivalent groups equivalent or cure an extremely small sample. The investigators also reported that controlling for age weakened the statistical strength of some molecular contrasts.

What did the DNMT findings show—and not show?

DNA methyltransferases participate in processes relevant to DNA methylation. This investigation measured mRNA for genes associated with those processes; it did not directly map genome-wide DNA methylation changes or establish an epigenetic diagnosis. At four hours, the 13.4% THC group had a higher mean DNMT3A mRNA change than the placebo group. The authors reported a significant exploratory high-versus-placebo pairwise comparison (p = 0.027), while the overall three-arm DNMT3A comparison was not statistically significant (F(2,20) = 3.075; p = 0.068).

This distinction is consequential. A pairwise contrast that passes a nominal p-value threshold does not retroactively convert a nonsignificant omnibus test into a positive three-group finding. Multiple transcript targets, multiple time points, and several post-hoc contrasts were examined in this small study; their interpretive uncertainty is greater than a single prespecified, independently validated biomarker test.

DNMT1 mRNA showed a numerically higher mean change in the 13.4% group at four hours, but its overall result was p = 0.103; the authors’ high-versus-placebo comparison was described as a trend at p = 0.056. These are not established significant overall group effects. A transient measured transcript difference should also not be transformed into an assertion that someone has experienced a permanent epigenetic alteration.

Which molecular findings were nonsignificant or inconsistent?

A scientifically faithful explanation must report negative results alongside positive ones. The CB1 mRNA group comparisons were nonsignificant at 55 minutes (p = 0.104), 80 minutes (p = 0.346), and four hours (p = 0.172). Other measured transcripts frequently showed numerical shifts without significant overall differences, including inflammatory and DNA-demethylation targets. The original report discussed an exploratory TET3 pairwise contrast at 55 minutes, but its overall test had p = 0.100.

The investigators additionally reported that combining the two active-cannabis groups did not yield significant mRNA differences versus placebo. That result strongly limits any claim that the study demonstrated a uniform “cannabis-positive lymphocyte expression signature.” It also illustrates why selecting one attractive nominally significant gene comparison from a large set can overstate what a small exploratory experiment established.

Did labeled cigarette potency predict actual whole-blood THC?

No statistically significant between-active-arm difference in measured whole-blood THC concentrations was found, even though the cigarettes were labeled 5.9% and 13.4% THC. The investigators proposed that participants may have adjusted how much smoke they inhaled to reach a desired effect. That is a plausible interpretation, not a directly measured identical delivered dose. Whole-blood THC rose after active smoking and generally peaked at the early ten-minute blood sampling point before decreasing.

The measured whole blood and the separated lymphocyte RNA preparation are analytically distinct. Serum or plasma concentrations, THC metabolites measured in urine, oral-fluid drug screens, and breath alcohol results cannot be substituted into the study’s whole-blood correlations without their own validated measurement and interpretation methods. For the broader specimen distinction, see the published discussion of biological matrices in toxicology.

Baseline whole-blood THC-COOH reflects prior cannabis exposure, not necessarily the acute smoked dose that day. A detected parent THC concentration can provide evidence relevant to recent exposure under a validated test, but it is still not a direct measurement of impaired psychomotor performance. The paper did not construct an individual-impairment classification threshold from any of these chemical values.

Were mRNA changes correlated with blood THC concentrations?

The authors did not find consistent correlations between changes in whole-blood THC and the measured mRNA changes, or persuasive relationships between baseline whole-blood THC-COOH and later mRNA shifts. This is central to the biomarker question. If a proposed forensic method cannot consistently track even the measured exposure variable in its source experiment, it cannot be presumed capable of reconstructing a particular person’s THC dose, use time, or impairment.

There were isolated significant correlations among different biological or behavioral variables. For example, a CB2 change and an inflammation-related mRNA measure had a reported association at one early time point. Such correlations are descriptive in a small, repeatedly sampled panel; they do not demonstrate that the molecular change caused a driving decrement or identify the direction of a mechanism. Requiring corroboration is especially important where numerous correlations were examined.

How do the subjective “high” ratings compare with the gene-expression results?

Participants rated the subjective feeling of being “high” on a 0–100 scale. In contrast to the inconsistent link between THC concentration changes and mRNA measurements, the authors reported a relationship between the rise in measured whole-blood THC near ten minutes and cumulative subjective-high ratings (r = 0.641; p = 0.001). That association does not establish that the participant was unable to perform any particular activity. It describes a relationship between a measured drug change and a self-reported experience in this experimental sample.

Subjective drug effect and impairment are not synonyms. Someone may report a strong subjective experience without a known quantified driving deficit; another person may show altered task performance without accurately perceiving it. Neither a molecular profile nor an individual’s remembered sense of intoxication is automatically a validated substitute for relevant observed behavioral evidence.

Did this study measure actual driving performance?

Participants completed simulator driving tasks as part of the parent experiment. In this small molecular sub-sample, the active-cannabis participants had worse Composite Drive Scores than placebo at selected post-smoking times. The score aggregated normalized simulator-performance measures; it was not an on-road traffic collision, a court finding of impairment, or a biological diagnosis.

Crucially, the study did not demonstrate a dependable relationship between the selected lymphocyte mRNA changes and changes in Composite Drive Score. One exploratory DNMT3A–simulator correlation in the eight-person high-THC group was reported at a particular time-point pairing; the authors themselves described its substantive meaning as unclear. It is not a validated personal prediction rule. The parent driving trial’s larger participant count must not be misrepresented as the denominator for these 23-person lymphocyte analyses.

The proper evidentiary chain separates observed task behavior, timing of specimen collection, analytical drug measurements, molecular markers, participant history, and possible alternative influences. A simulator finding might be relevant to general psychopharmacology without proving why a particular road incident occurred.

Why don’t these gene-expression results establish a cannabis impairment test?

A usable forensic classifier would need a defined intended population, standardized specimen collection and processing, analytical reproducibility, clinically or legally relevant cutoff, blinded external validation, and reported person-level error rates. Ideally it would be tested against a clearly specified functional outcome under realistic time intervals and with alternative causes evaluated. This study did not establish sensitivity, specificity, positive or negative predictive value, or a validated threshold for CB2, DNMT3A, or any other mRNA to infer recent use or impairment.

Its molecular measures also are biologically indirect. Lymphocyte transcription may be influenced by immune-cell composition, individual baseline biology, age, chronic cannabis history, sampling time, handling, other substances, and experimental conditions. Transcript quantity can change without a parallel change in protein abundance or measurable behavioral function. Because the experiment contains small groups and unequal baselines, a nominal molecular group difference cannot safely be converted into a personal legal conclusion.

For evidence-handling issues relevant to tissue and blood processing, see specimen collection and preanalytical integrity. For why the identity of a chemical analyte and assay matters, see confirmatory drug testing. Neither existing topic page converts this experimental gene-expression method into a validated impaired-driving assay.

What would a case-specific forensic review need?

A scientist assessing an asserted cannabis effect should identify the exact issue to be decided—recent exposure, specimen provenance, a measured THC concentration, observed coordination, subjective symptoms, or alleged causal impairment—and match each proposition to admissible, fit-for-purpose evidence. Relevant case records may include documented specimen type and collection time, chain of custody, validated analytical methodology and uncertainty, medication and drug history, contemporaneous observations, the actual task or driving behavior, and alternative explanations.

Research-stage blood-cell mRNA findings can inform potential biological mechanisms and future validation studies. Without the required individual validation and case facts, they cannot establish when an individual smoked cannabis, the grams inhaled, impaired functioning at a legally relevant moment, or the cause of an accident. That limit protects against overstating both statistically significant group findings and nonsignificant endpoints.

Frequently asked questions

Was the trial randomized and placebo-controlled?

Yes. The parent experimental assignment used blinded placebo and active smoked cannabis conditions. The lymphocyte sub-study analyzed only 23 participants—11 placebo, four in the 5.9% THC arm, and eight in the 13.4% arm—after one post-assignment exclusion.

Did the 13.4% cigarette produce twice the measured blood THC of the 5.9% cigarette?

No such proportional result was established. The participants smoked ad libitum and the researchers found no statistically significant difference in measured whole-blood THC concentrations between the two active arms.

Did the investigators directly measure DNA methylation?

No. They measured mRNA for genes including DNMT1 and DNMT3A, which relate to methylation processes. Relative transcript levels are not the same measurement as DNA methylation changes at particular genetic sites.

Was DNMT3A a statistically significant three-group finding?

Not at four hours under the study’s overall comparison: p = 0.068. The authors reported a nominal high-THC-versus-placebo pairwise p = 0.027, which must not be substituted for the nonsignificant overall test.

Does an increased CB2 mRNA result prove that THC was recently smoked?

No. The group differences occurred in a small exploratory laboratory sample with different baselines and no validated individual exposure classifier. A single person’s measured CB2 mRNA value cannot be interpreted as a confirmed use-time result from this study.

Can lymphocyte gene expression establish an individual’s driving impairment?

No. Selected group differences in peripheral blood-cell mRNA were not validated to predict individual simulator performance, on-road driving, or impairment at an event time.

Does a negative or nonsignificant molecular result prove that a person was unimpaired?

No. A null comparison in this small transcription study does not establish chemical absence or unimpaired behavior. Any impairment conclusion requires different, case-appropriate evidence.

Primary scientific source

Smith RC, Sershen H, Janowsky DS, et al. Changes in Expression of DNA-Methyltransferase and Cannabinoid Receptor mRNAs in Blood Lymphocytes After Acute Cannabis Smoking. Frontiers in Psychiatry. 2022;13:887700. DOI: 10.3389/fpsyt.2022.887700.

The article summarizes a completed original human experiment. It does not constitute a new biological test, a legal impairment threshold, or an individualized expert opinion about an incident.