Can Oral THC Peak Around Four Hours Yet Persist Before the Next Dose? A 24-Vervet Study

Short answer: In a controlled 2026 experiment involving 24 adolescent vervet monkeys, measured cannabinoid concentrations rose after oral Δ9-tetrahydrocannabinol (THC), with the strongest reported post-dose THC signal around four hours on both the first and eighth study days. THC also remained measurably higher before dosing on day eight than before the first dose. These are controlled animal concentration–time observations—not a four-hour rule for humans, an automatic estimate of the time of cannabis consumption, or evidence that a person was impaired.

The distinction matters when a toxicology report is used to reconstruct an alleged exposure. An administered oral dose is not an assayed concentration; a blood-derived chromatographic result is not a performance test; and an average or group-median curve does not disclose the specific absorption, metabolism, tolerance, or functional ability of an individual. The primary experiment also contains a laboratory-matrix reporting issue that must be resolved before translating its figures into a legal blood-test comparison.

The original primary research is Huppé-Gourgues, Spridzans and colleagues, “Pharmacokinetics of oral Δ9-tetrahydrocannabinol (THC) administration in vervet monkeys,” Journal of Cannabis Research (2026), doi:10.1186/s42238-026-00422-y. The findings below come from its methods, measured results, plotted data and discussion—not from generalized assumptions about edible cannabis.

What forensic question can this experiment actually answer?

It can answer a bounded pharmacokinetic question: under the investigators’ particular dosing, feeding, sedation, specimen-collection and analytical conditions, how did THC and selected metabolites change after oral administration and across eight consecutive days? The authors directly quantified cannabinoids in collected blood-derived material using liquid chromatography–tandem mass spectrometry. They tested THC, its hydroxylated metabolite 11-hydroxy-THC, its carboxylated metabolite 11-nor-9-carboxy-THC, and cannabidiol (CBD). These are chemically different analytes whose concentration curves need not follow the same pattern.

It cannot directly answer when an unknown person last consumed an edible, how much that person consumed, whether the person was driving safely, or whether a positive result caused an injury. No human participants drove, performed standardized sobriety tests, or supplied forensic roadside specimens in this experiment. Even in monkeys, the work was designed to characterize drug disposition rather than a measured dose–impairment relationship. Establishing an impairment threshold would require an entirely different kind of study.

Who was studied, and what was the dosing schedule?

Twenty-four approximately four-year-old adolescent vervet monkeys were randomly assigned to four oral THC dose groups: 0.5, 1.0, 2.0 and 3.0 milligrams per kilogram of body weight. Each group contained six animals—four males and two females. The animals weighed about 3 to 4.5 kilograms and had not previously been exposed to cannabinoids. Their small, controlled cohorts are helpful for observing drug concentration changes; they do not constitute a human clinical population or four independently validated human dosing standards.

The investigators gave the assigned THC doses on days one and eight under sedation and maintained the corresponding regimen on days two through seven by placing THC solution in banana. The THC reference formulation was more than 99.5% THC and contained less than 0.05% CBD. Day-one and day-eight concentration profiles are therefore not two independent groups of animals: they represent repeat sampling of the same four assigned groups at different phases of exposure. A study day, assay injection or serial blood specimen is not another independently exposed participant.

The reported milligrams-per-kilogram figure is an administered amount. It is not a blood concentration, a measured absorbed dose, an estimate of edible bioavailability, or a validated dosing equivalence between a vervet monkey and a human. Such equivalence would demand additional biological and formulation evidence.

When was blood collected, and why is the schedule important?

Samples were collected at nominal post-dose times of zero, 0.25, 0.5, one, two, four, eight and 24 hours on the intensively sampled days. The zero-hour sample establishes a point before that day’s administration; a day-eight zero-hour result is not necessarily a drug-free baseline because animals had received THC on intervening days. Blood was collected from the femoral site and preserved with EDTA. The researchers froze the specimens and shipped them for analytical testing.

Those collection times define the experiment’s temporal resolution. If the highest sampled group concentration occurred at four hours, that result supports a four-hour sampling peak under this protocol. It does not prove that every animal reached its true physiological maximum precisely at four hours. A maximum between the two-hour and four-hour samples, or shortly after four hours, would not be observed with finer precision under this schedule. Nor does the curve become a universally applicable clock for an individual whose dosing time, food intake, formulation and metabolism are unknown.

Which matrix was measured: whole blood or plasma?

This is an unusually important caution in the original report. Its methods describe collection of EDTA-preserved femoral blood, freezing, and analytical preparation; the results and figure legend sometimes label concentrations plasma or plasma-equivalent. The published methods do not clearly document a centrifugation or plasma-separation step that would reconcile those descriptions. In addition, the authors report preparing external calibration and quality-control materials in a rat blood matrix, rather than specifying matched vervet plasma standards.

Accordingly, the safest evidence-bound description is “reported blood-derived cannabinoid concentrations,” while retaining the authors’ own plasma-related labels when discussing their graphs. It would be unjustified to assume that a plotted ng/mL value is definitively a validated human whole-blood concentration or a directly interchangeable human plasma value. Matrix partitioning, sample preparation, calibration, and method validation are distinct matters. A litigation expert would need the laboratory worksheet, collection and processing records, validation package, and raw chromatograms to resolve this detail before numerical cross-matrix comparisons.

For a practical guide to these distinctions, see biological matrices in forensic toxicology. Neither the word “blood” nor the unit ng/mL, standing alone, establishes which biological compartment was quantitatively validated.

How did the investigators quantify THC and metabolites?

The researchers used high-performance liquid chromatography coupled to tandem mass spectrometry (HPLC–MS/MS), including a C18 chromatographic column, positive electrospray ionization, analyte-specific transitions and deuterated internal standards. The listed compounds were Δ9-THC, 11-OH-THC, THC-COOH and CBD. The reported calibration standards extended from 0.06 to 1,000 ng/mL and used weighted linear regression. Reported batch quality-control accuracy averaged approximately 95%, which is not itself proof that every compound and every specimen met all forensic accreditation criteria.

The authors stated an empirical detection limit of 0.18 ng/mL, with compound-specific signal-to-noise ratios provided in their methods. Values near that detection limit were treated qualitatively rather than used as precise quantitative inputs for pharmacokinetic modeling. A limit of detection is not automatically a validated limit of quantification. Results near a detection boundary cannot safely be interpreted with the same precision as those well within a validated reporting range.

This study also used an outlier rule: observations exceeding three standard deviations from a mean could be excluded, and some outcomes were not reported where exclusions undermined statistical analysis. That choice matters when reading small groups of six. A chromatographic identification result, a numerical concentration result, an adjusted group-model estimate and a participant-specific threshold decision are four different assertions. See the site’s discussion of confirmatory drug testing and analytical specificity.

What actually happened to THC concentrations after the first dose?

On day one, measured THC showed both a dose-group effect and a sampling-time effect. The reported mixed-effects dose test was F(3,19)=6.98, p=0.002; the time test was F(7,119)=14.18, p<0.001. The authors did not report a corresponding significant dose-by-time interaction for THC that day. Post hoc estimates identified clear increases in the 2.0 mg/kg group (β=0.452, p=0.007) and the 3.0 mg/kg group (β=0.517, p=0.003). These β values arise from the authors’ log-transformed statistical model, not from a new ng/mL concentration measurement in a particular animal.

Across doses, the four-hour sampling point was associated with the strongest reported post-dose THC effect (β=0.757, p<0.001). The authors described a decline toward baseline by 24 hours (β=−0.335, p=0.007). Their discussion reports group-median four-hour values of approximately 1.11 ng/mL at 0.5 mg/kg and 2.43 ng/mL at 3.0 mg/kg. Those numbers characterize these groups and the reported matrix convention; they are not person-specific predictions and should not be used to infer an unknown subject’s dose.

An oral route can introduce delayed absorption and variability, but this specific experiment does not isolate the gastrointestinal contribution from the effects of the formulation, overnight fasting or procedural ketamine. A delayed observed maximum is a finding to explain, not a standalone timestamp for legal reconstruction.

Did 11-OH-THC, THC-COOH and CBD behave the same way?

No. The first-day 11-OH-THC analysis showed a time effect, F(5,67)=4.01, p=0.003, but did not show a main dose-group effect, F(3,18)=0.21, p=0.887. The authors identified a significant four-hour 11-OH-THC elevation (β=0.713, p=0.001). This active hydroxylated metabolite is chemically and pharmacokinetically distinct from parent THC. Its concentration cannot simply be added to a THC reading and declared a measured dose or individual impairment score.

The paper did not demonstrate a significant four-hour THC-COOH increase over baseline on day one; its reported comparison was F(3,20)=0.58, p=0.635. That negative finding does not demonstrate zero metabolism, absolute absence of the metabolite, or an analytically universal period in which THC-COOH cannot appear. It states a lack of statistical demonstration under the particular samples and model reported. A delayed metabolite can be influenced by observation windows, censoring near the detection limit and inter-individual variation.

CBD was only a minor constituent of the administered THC material. On day one, the original analysis reported a CBD effect (F(3,17)=5.58, p=0.008), with a reported higher-dose finding for 3.0 mg/kg (β=0.192, p=0.01). This must not be presented as if participants received a separate controlled CBD treatment or as proof of CBD–THC pharmacodynamic interaction.

Did eight consecutive days change the THC profile?

After the short daily regimen, day-eight THC again varied with assigned dose (F(3,20)=3.61, p=0.031) and sampling time (F(7,138)=10.59, p<0.001). The paper again identifies the four-hour point as the principal sampled maximum and reports subsequent decline toward 24 hours (β=−0.722, p<0.001). These results support a repeatable group-level oral concentration–time pattern within this experiment, although they do not establish steady-state kinetics in each animal or long-term accumulation after months of use.

Day-eight 11-OH-THC findings also varied with dose and time, with the article reporting a dose-by-time effect (F(5,78)=2.63, p=0.003) and four-hour increases at the two highest doses. The two lower doses did not supply the same statistically supported metabolite comparison. For THC-COOH and CBD, the authors did not show a consistent statistically significant four-hour rise or time-dependent shift on day eight. These analytes should not be compressed into the inaccurate phrase “all cannabinoids accumulated.”

The distinction between a parent-drug curve and several metabolite curves is central when interpreting serial testing. Changing detectability or baseline in one analyte does not establish a universal change across all chemically related compounds.

What does a higher day-eight predose baseline mean?

At the nominal zero-hour sampling point, THC concentrations were higher on day eight than on day one in the authors’ model (β=0.419, p<0.001). The paper supplies two particularly useful examples: at 2.0 mg/kg, the group median predose value changed from 0.39 to 0.934 ng/mL; at 3.0 mg/kg, it changed from 0.275 to 0.936 ng/mL. That is approximately 2.4-fold and 3.4-fold, respectively, using the displayed medians. These are group comparisons, not estimated half-lives and not individual pharmacokinetic constants.

Residual THC before a new dose shows why a detected parent compound cannot always be attributed to an immediately preceding single administration without knowing exposure history. Yet this monkey result also cannot establish a human chronic-user detection window: the dose regimen was controlled, duration only eight days, animals were young and not previously exposed, and the analytical matrix label needs clarification. The underlying mechanisms and measurable magnitudes could differ substantially in human cases.

The source therefore supports the limited statement that a controlled daily dosing series produced higher predose THC on day eight in these animals. It does not validate an algorithm that converts an isolated concentration into the number of days, number of doses, or exact time since a human last used cannabis.

What are the two most useful sets of primary evidence?

Every number above belongs to the stated animal group, sampling design and reporting conventions. A coefficient or p-value is not itself a chromatographic concentration. A median is not the value for all six animals, and comparison across days is not proof of a new ingestion at any unknown time. When using a table in expert testimony, its denominators and analytical matrix should accompany the result.

Could ketamine sedation change how the curves should be understood?

Potentially. On sampled days, animals received ketamine during oral administration and as needed for safe blood collection, while doses on intervening days were offered in banana. The authors themselves identify possible effects of anesthesia on gastrointestinal transit and metabolic pathways, while acknowledging those mechanisms were not directly tested in this dosing experiment. Faster, slower or otherwise altered absorption cannot therefore be assigned definitively to oral THC alone. The experimental conditions differ from a person eating a commercial product while awake and moving normally.

This is not evidence that ketamine necessarily caused the four-hour peak. It is evidence that a competing procedural explanation was present and unresolved. Treating the curve as if the animals were unanesthetized human edible consumers would obscure a material part of the protocol. The overnight fast, edible vehicle, dose formulation and species also limit external validity. None was separately randomized to isolate its effect on the observed time profile.

Does the study establish bioavailability, a half-life, or a safe concentration?

No absolute oral bioavailability was determined. The study had no intravenous reference arm from which to calculate the absorbed fraction for the same animals, and it did not validate a patient-specific conversion between administered milligrams and measured ng/mL. A declining concentration over the observed 24 hours is not automatically a terminal elimination half-life; distribution, absorption, sampling boundaries, assay detection and ongoing exposure history can all complicate such calculations. The day-eight predose readings establish residual exposure in this controlled regimen, not a general duration of impairment.

The source also contains no objective driving or psychomotor criterion, no human legal threshold, and no controlled trial classifying cases as impaired versus unimpaired. It therefore cannot support a “safe” or “unsafe” cutoff for human THC concentration. Even a numerical result analytically measured in an appropriate human specimen would require independent temporal, clinical and behavioral context before reaching an impairment opinion. A nonhuman dose-to-concentration experiment is an even narrower kind of evidence.

How should an expert use this paper in litigation?

The paper is useful for testing overconfident claims about a fixed oral THC peak, a universal metabolite clock, or the meaning of a residual predose cannabinoid result. It demonstrates that the parent and metabolites can have different significant and nonsignificant findings; that oral exposure was associated with a multi-hour profile; and that repeat dosing affected the observed baseline in this small controlled population. Those are legitimate scientific propositions, limited to the reported conditions.

In a disputed human case, the relevant records would include the actual specimen type and processing, collection time, chain of custody, quantitative chromatograms and calibrators, validated lower reporting limits, separately measured metabolites, prescribed or reported dosing history, other drugs, timing of observations and contemporaneous functional evidence. The distinction between a laboratory method and a case inference should be preserved. The related pharmacokinetics and interpretation guide explains why population curves cannot mechanically be applied to reconstruct a particular event.

A defensible expert opinion would state whether the method actually measured the reported compound in the reported human matrix, then separately identify which conclusions are supported by the individual record. It would not use the animal group median as a human blood conversion factor, call residual cannabinoid positivity proof of immediate new consumption, or equate a concentration with evidence of driving behavior.

What specific limitations should appear alongside any citation?

At minimum, the reader should recognize four dose cohorts with only six animals each; differences in sex that were not adequately powered for reliable sex-specific conclusions; oral maintenance dosing using banana between sampled days; ketamine sedation on intensive sampling days; no documented direct human comparison; outlier removal and an empirical detection boundary; and mixed language for the laboratory matrix. The statistical model assesses reported group and time effects after transformation; it does not validate an individualized forensic prediction equation.

The original authors also observe that THC-COOH was not significantly raised in the reported window. That cannot be transformed into a claim of absent THC metabolism, particularly where the source itself discusses timing and sensitivity. Similarly, the absence of overt abnormal behavior during veterinary monitoring was not a standardized negative performance test. These distinctions are not technical quibbles: they determine whether a citation supports an observation, a qualified inference or an assertion the experiment never investigated.

Frequently asked forensic questions

1. Does oral THC always reach its peak exactly four hours after ingestion?

No. The four-hour point was the strongest sampled group signal under this vervet dosing and collection schedule. It is not a validated clock for humans or even a precisely measured peak time for each monkey between blood draws.

2. Did the researchers actually measure human blood THC?

No. The 24 subjects were vervet monkeys. Their femoral EDTA specimens were analyzed using HPLC–MS/MS. The report’s inconsistent blood-versus-plasma terminology should be clarified before making numerical comparisons with human whole-blood or plasma data.

3. Does a higher day-eight predose THC result prove a new dose immediately before sampling?

No. The animals had received assigned daily THC on intervening days. The higher reported group predose baseline follows repeated known exposure, and the experiment does not identify the timing of an unknown individual dose from one result.

4. Was THC-COOH absent because its four-hour comparison was nonsignificant?

No. A nonsignificant reported test is not proof that the metabolite was never formed or detected. It means the study did not demonstrate the specified elevation in that analysis under its sampling and analytical constraints.

5. Can the 0.5–3 mg/kg dosing groups be converted directly to a human edible dose or blood concentration?

No. The groups received known animal doses; human absorption, formulation, species-specific metabolism and analytical matrices were not jointly validated by a conversion study. Administered dose and measured concentration answer different questions.

6. Does the chromatographic detection limit establish an impairment cutoff?

No. A detection limit concerns instrument sensitivity and reporting. It says nothing by itself about driving, cognition, motor performance, legal capacity or accident causation. The paper did not validate a functional impairment classification.

7. What is the strongest defensible use of this study in a forensic report?

It supports carefully qualified discussion of delayed sampled oral THC peaks, differing parent/metabolite behavior, and residual predose findings during repeated controlled exposure in monkeys. It should accompany explicit species, method, matrix and case-inference limitations rather than replace human case-specific evidence.

Primary-source conclusion

A rigorously bounded reading of this original 2026 vervet pharmacokinetic experiment is straightforward: THC rose after oral administration, its strongest reported sampled signal occurred around four hours on the two profiled days, and controlled repeat dosing produced higher predose THC on day eight. Metabolites did not all behave identically. The study is informative about this experimental animal regimen and analytically inappropriate as a stand-alone human consumption-time or impairment calculator. The controlling question in litigation remains what was actually measured, in which matrix, by which method, at what time, and what independent facts connect that result to the individual event.