Did High-Concentration THC Change Human Blood-Cell Epigenetics? What a 16-Donor Laboratory Study Shows

Short answer: Yes, under a narrowly defined cell-culture protocol. In a 2025 primary experiment, researchers isolated peripheral blood mononuclear cells (PBMCs) from healthy male blood donors and repeatedly incubated those cells with laboratory-added delta-9-tetrahydrocannabinol (THC). At a nominal culture concentration of 200 ng/mL, cells from 16 donors showed transient decreases in DNA methyltransferase (DNMT) expression and activity, increases in ten-eleven translocation enzyme (TET) expression and activity, and altered genome-wide DNA methylation and hydroxymethylation patterns. An eight-donor experiment using 50 ng/mL did not show statistically significant DNMT/TET or cannabinoid-receptor changes, although selected cytokine messenger RNAs increased at both concentrations. These findings do not establish that 200 ng/mL was a measured concentration in a living person’s blood, that a cannabis user sustained permanent DNA injury, that cannabis use disorder was diagnosed, or that anyone was impaired while driving.

The original primary study was Pulk, Somelar-Duracz, Rooden, Anier and Kalda, Concentration-dependent effect of delta-9-tetrahydrocannabinol on epigenetic DNA modifiers in human peripheral blood mononuclear cells, Translational Psychiatry (2025), 15:198, doi:10.1038/s41398-025-03419-y. The distinction between an experimental culture concentration and a forensic blood result is central to interpreting this experiment.

What question did the researchers actually answer?

They asked whether purified THC, placed directly into a controlled cell culture, changes enzymes involved in DNA methylation and demethylation, cannabinoid-receptor transcripts, selected inflammation-related transcripts, and genomic patterns of modified cytosines. They also tested whether a second exposure after a period without THC in the culture changed these laboratory responses. Those are mechanistic questions about cultured immune cells. They are not clinical tests of intoxication, diagnosis of cannabis use disorder, neurocognitive impairment, or the origin of a particular injury.

The distinction matters in forensic testimony because biological markers can sound more definitive than they are. Finding a statistically significant cellular association establishes that an assay detected a group-level response under its experimental conditions. It does not establish an individual person’s route of administration, dose, blood THC concentration, recent-use interval, driving performance, mental state or cause of an incident. Each such conclusion would require additional measurements and an appropriately validated study design.

Were 16 people given THC?

No. Fresh whole blood was collected from 16 healthy male donors, ages 20–40. Investigators then separated the PBMC fraction using a Ficoll density gradient, cryopreserved the isolated cells, subsequently thawed them and cultured them in RPMI medium. THC was administered to the cells in a laboratory dish, not to the donors. The donors’ original whole-blood samples provided cells; the paper did not report measured THC concentrations in their circulating blood following cannabis consumption. Participants confirmed no narcotic use during the preceding 12 months, according to the paper’s donor description.

The laboratory compared THC-exposed PBMCs with cells receiving an ethanol-containing vehicle control. The stated THC stock was dronabinol, prepared in an ethanol vehicle. A vehicle-control condition therefore mattered: a response attributed to THC had to be evaluated relative to the corresponding solvent handling. The investigators studied cells, nuclear extracts, RNA and DNA, not fresh post-dose forensic blood specimens from tested drivers.

How were the 50 and 200 ng/mL concentrations used?

In the repeated-exposure protocol, the investigators incubated cultures for one hour on each of five consecutive days. At the lower nominal THC concentration, 50 ng/mL, the study reports eight donors. At the higher concentration, 200 ng/mL, it reports 16 donors. Cultures were sampled 24 hours after each incubation. Thus five consecutive time points are repeated laboratory observations, not five distinct cohorts of 16 newly recruited people.

The authors chose these nominal values with reference to previously published serum concentrations following cannabis smoking. That rationale does not turn them into directly assayed serum or plasma values in the present experiment. Nor does it establish a one-to-one relationship between extracellular THC added to culture medium and unbound intracellular THC, circulating whole-blood THC, or cerebral receptor exposure. Serum, plasma and whole blood are different specimen matrices with different preparation methods and partitioning behavior. There is no validated forensic conversion from this culture dosing to an individual’s blood THC or impairment level.

What was the laboratory “abstinence and challenge” experiment?

The paper also describes a high-concentration THC-challenge experiment. Cells received repeated THC at 200 ng/mL or vehicle for five days, then remained in culture for a further five days without the daily THC treatment, and were subsequently given an additional one-hour THC or vehicle challenge. The cells were examined 24 hours after that challenge. Four experimental histories were considered: vehicle followed by vehicle, vehicle followed by THC, repeated THC followed by vehicle, and repeated THC followed by THC.

This was abstinence in a cell-culture schedule, not a monitored withdrawal period in living cannabis consumers. The protocol cannot reveal how long THC was present in any donor’s blood, whether a person ceased cannabis use, how long cognitive effects persisted, or whether an individual’s tolerance changed. The design instead tests how past culture exposure history modifies the cells’ molecular response to a new dose. Calling the challenge a relapse model would go beyond the physical measurements.

Which assays measured the effects?

For gene expression, investigators extracted RNA, synthesized complementary DNA, and performed reverse-transcription quantitative PCR, with duplicate amplification and normalization to the B2M reference transcript. Separate nuclear-extract assays measured DNMT and TET enzyme activities. To examine DNA-level changes, the team used reduced-representation bisulfite and oxidative-bisulfite sequencing (RRBS and RRoxBS), which assess differential methylation and hydroxymethylation across selected genomic regions. The DNA sequencing component specifies n = 8, a subset rather than 16 independent genome-wide experiments at each condition.

These methods measure distinct things. A change in mRNA abundance does not automatically prove a corresponding change in protein concentration, catalytic activity, cytokine secretion or actual disease. A nuclear-extract enzyme assay is closer to function but remains a biochemical assay performed on processed cells. A differentially modified DNA region is a sequencing result, not a clinical diagnosis. The paper measured neither a validated THC intoxication biomarker in blood nor a clinical behavior outcome.

What changed at 50 ng/mL—and what did not?

The lower concentration provides an unusually important negative comparison. Across the five-day culture schedule, the study reported no statistically significant changes in measured DNMT1, DNMT3A, DNMT3B or TET1–3 messenger RNA, DNMT/TET enzyme activity, or CB1/CB2 receptor mRNA at 50 ng/mL in the eight-donor set. A nonsignificant result is not proof that a biological effect is impossible. The modest donor sample, measured time points and chosen assays limit how confidently one can exclude smaller, delayed or other effects.

There was also a positive finding at the same lower nominal concentration: after five daily incubations, IL-1β, IL-6 and IL-10 messenger RNA increased relative to control. The authors reported IL-1β and IL-6 at p < .0001 and IL-10 at p < .001 for their specified contrasts. TNF-α messenger RNA did not change significantly. Thus the study does not support a simplistic statement that “50 ng/mL did nothing.” Different cellular readouts had different responses.

What changed at 200 ng/mL?

In the 16-donor repeated-incubation experiment, messenger RNA for the three measured DNMT genes generally decreased, while the measured TET genes generally increased relative to their vehicle-control conditions. The most pronounced shifts were observed approximately 24 hours after the third daily THC incubation, with reported post hoc comparisons for multiple individual genes at p < .0001. DNMT enzyme activity declined while TET activity increased, again most prominently after the third incubation. CB1 and CB2 messenger RNA also increased around that stage.

But the response was not indefinitely progressive. By the end of the five-incubation schedule, several DNMT/TET and receptor measures had returned toward the control levels under the stated assay conditions. The figure shows meaningful intermediate-day changes that were not simply larger with each exposure. The authors discussed cellular adaptation as one possible explanation; the data do not establish a person’s clinical drug tolerance, permanent molecular damage or a lasting effect after ordinary cannabis use.

What happened after the additional THC challenge?

After the five-day culture interval and renewed THC challenge, the high-concentration group again showed decreased DNMT expression and activity and increased TET expression and activity compared with specified controls. The original analysis reports an omnibus DNMT3A F(3,60) = 35.94 and DNMT3B F(3,60) = 56.36, both p < .0001, with gene-specific tests for DNMT1 and the TETs also reported at p < .0001. DNMT activity after challenge was reported at p < .0001 and TET activity at p < .01 in designated comparisons. CB1 and CB2 challenge groups differed overall, with F(3,60) = 28.34 and 31.84, respectively.

Those statistics refer to laboratory treatment groups and the authors’ chosen comparison model. They are not clinical odds ratios for addiction, not estimates of the duration of intoxication, and not a validated method to determine whether a particular person had resumed drug use after abstinence. Results measured 24 hours after a laboratory challenge also cannot establish an individual driver’s performance at an accident scene.

What do the positive and negative outcomes show together?

A forensic reader should preserve the denominator attached to each result. Sixteen donors supplied the high-dose experimental biological material, eight supplied the lower-dose experiment, and the sequencing section identifies eight DNA samples. Duplicate PCR reactions, repeated days, multiple treatment conditions, thousands of sequence reads and hundreds of genomic regions do not enlarge the number of independent human donors.

What did DNA methylation and hydroxymethylation sequencing find?

Compared with vehicle controls, repeated 200 ng/mL exposure was associated with 737 differentially methylated regions (381 hypermethylated, 356 hypomethylated) and 468 differentially hydroxymethylated regions (310 hyperhydroxymethylated, 158 hypohydroxymethylated), after correction for the relevant multiple genomic tests. The methylation directions were comparatively balanced, but 66.24% of the altered hydroxymethylation regions were in the hyperhydroxymethylated category.

The challenge protocol, also relative to its control, yielded 693 differentially methylated regions (334 hypermethylated, 359 hypomethylated) and 442 differentially hydroxymethylated regions (174 hyperhydroxymethylated, 268 hypohydroxymethylated). Thus 60.63% of its altered hydroxymethylation regions were in the hypo category. The switch in proportional direction is a comparative genomic observation, not a stopwatch for recent cannabis use or proof of reversible clinical harm in a person.

Does a statistically enriched “addiction pathway” diagnose addiction?

No. The investigators performed Gene Ontology and KEGG pathway enrichment on genes associated with differentially modified promoter regions. The resulting terms included morphine addiction, alcoholism, synaptic signaling and other biological pathway labels. A pathway label is an annotation assigned to a collection of genes; it does not mean a donor exhibited addiction behavior or received a clinical diagnosis. Likewise, the observation of altered DNA modification around a gene does not alone demonstrate the gene’s protein product changed or that a clinical behavioral circuit was altered.

The authors validated selected gene-expression associations for GNAS, FOXO1, IL1R1 and CD93, but also noted that several changes in DNMT, TET and cannabinoid-receptor transcription did not map neatly onto corresponding promoter DNA-modification findings. That internal lack of perfect correspondence deserves attention: biological regulation is more complex than a direct one-marker, one-disease pathway. The genomic analysis identifies candidate mechanisms for future testing, not an evidentiary substitute for personal history, direct blood testing or clinical evaluation.

What about the cytokine findings?

IL-1β and IL-6 are generally discussed as proinflammatory signaling molecules; IL-10 is often discussed in an anti-inflammatory context. Increased messenger RNA for all three after repeated laboratory THC incubations does not establish net systemic inflammation, clinical immune suppression or a particular disease. The investigators did not demonstrate corresponding circulating cytokine protein concentrations in exposed humans, clinically diagnosed infections or psychiatric outcomes. And the nonsignificant TNF-α gene result limits any claim that every cytokine followed the same direction.

After the high-concentration challenge, certain cytokine-gene responses increased further in the analyzed cultures, while TNF-α still did not show a significant effect. The investigators also reported that the promoter-level methylation/hydroxymethylation analysis did not detect matching modifications in the relevant cytokine genes, suggesting regulation may involve other processes. The proper forensic conclusion is specific: the RNA readouts changed under defined cell-culture conditions; the study did not validate a real-world inflammatory biomarker of cannabis intoxication or functional impairment.

Is this the same as the earlier cannabis-smoking blood-lymphocyte study?

No. A previously published smoking study examined messenger RNA after actual acute inhalation in cannabis users. Here, Pulk and colleagues removed immune cells from donors before exposing those cells to purified THC in culture for five days and applying a subsequent culture challenge. The earlier experiment involved real human use and specified post-use sampling intervals; this one isolates the effect of nominal THC incubation from many features of smoking, including combustion products, other cannabinoids, metabolism, stress and behavior. The present authors explicitly acknowledged that their in-vitro results did not reproduce every finding from investigations of cannabis users.

The difference is substantive rather than cosmetic. These studies are relevant to different scientific questions. One cannot combine their donor numbers as though they were a single cohort, treat the cell-culture dose as an observed post-smoking whole-blood concentration, or substitute one experiment’s post-exposure time for the other’s. This matters in litigation when expert testimony uses broad phrases such as “THC changes blood gene expression” without specifying the actual exposure route, biological material, and independent sample size.

Can the cell-culture concentration be used to back-calculate a driver’s dose or impairment?

No validated method is provided. A nominal culture-medium concentration is an experimental input. A legal venous whole-blood THC concentration is an analytical output obtained from a specific specimen using a laboratory method. A separated plasma or serum concentration is yet another matrix. Oral fluid, urine metabolites, hair and exhaled breath have different analytical and temporal meanings. The Pulk study neither measured a driver blood level nor tested a conversion among these compartments.

The study also did not ask participants to drive, track lanes, perform divided-attention tasks or receive clinical intoxication ratings. DNMT/TET changes are cellular responses and cannot be labeled a concentration-specific impairment test. Even if the general mechanism proves reproducible, it would require independent validation across people, dose ranges, sampling times, confounders and outcome measures before anyone could discuss an individual forensic sensitivity, specificity, cutoff or causation. Treating a culture condition of 200 ng/mL as a forensic impairment threshold would be scientifically unsupported.

How should a forensic report or testimony describe this evidence?

A defensible description should name the original physical material—PBMCs separated from 16 healthy male donors—the nominal treatment—50 or 200 ng/mL THC in culture—the repeated protocol and the measured molecular endpoints. It should identify both significant and null outcomes and disclose that the genome-wide analysis specifies an eight-donor DNA subset. It should also distinguish relative mRNA expression, activity assays and altered sequencing regions from measured parent drug concentrations.

It would be appropriate to say the investigators generated mechanistic evidence that repeated high-concentration THC incubation can alter DNMT/TET-related molecular responses and selected genome-wide DNA-modification patterns in cultured human immune cells. It would not be appropriate to testify that this paper proves permanent genetic injury, diagnoses cannabis use disorder, establishes a reliable recent-use interval, predicts a specific person’s cognition, or proves a collision. Those are different questions involving unmeasured outcomes and potentially substantial person-to-person heterogeneity.

Frequently asked questions

1. Does 200 ng/mL mean the donors actually had that THC level in blood?

No. It was the concentration added to cultured PBMCs. The researchers collected donor whole blood to obtain cells but did not give THC to the donors or report THC measurements in their circulating whole blood after use.

2. Were there 16 people in each exposure group?

No. The study describes 16 independent healthy male donors overall, eight in the 50-ng/mL condition and 16 for higher-concentration work. Repeated cell-culture time points and assay duplicates are not additional human participants.

3. Did the lower THC concentration have no effect at all?

No. The 50-ng/mL series showed no significant DNMT/TET or CB1/CB2 responses on the reported endpoints, but selected cytokine messenger RNAs increased, while TNF-α was not significantly changed.

4. Are 737 changed DNA regions equivalent to 737 mutated genes?

No. They are statistically identified differential methylation regions, not a count of mutation events, affected patients or clinically damaged organs. Methylation and DNA-sequence mutation are different concepts.

5. Was the five-day abstinence period observed in actual cannabis users?

No. It was an incubation-free interval in cultured cells between controlled THC exposures. It cannot establish the duration of a person’s intoxication, withdrawal, recovery or relapse.

6. Did the study prove that THC causes cannabis use disorder?

No. It generated molecular evidence relevant to possible mechanisms, but it did not enroll a clinical cohort to observe diagnoses, dependence trajectories or drug-seeking behavior.

7. Can DNMT/TET results establish whether a driver was impaired?

No. The investigators collected no driving-performance data and validated no person-level classifier connecting the cell assays to a legally relevant state of impairment, let alone a particular accident.

Primary scientific reference and essential limitation

Pulk K, Somelar-Duracz K, Rooden M, Anier K, Kalda A. Concentration-dependent effect of delta-9-tetrahydrocannabinol on epigenetic DNA modifiers in human peripheral blood mononuclear cells. Translational Psychiatry. 2025;15:198. Original primary study: doi:10.1038/s41398-025-03419-y. The paper is an original controlled in-vitro donor-cell experiment—not a newly conducted human cannabis-administration trial. It provides mechanistic molecular observations; it does not establish an individual person’s blood THC concentration, impairment, neurological injury, addiction diagnosis or event-specific legal causation.