Do CBD–THC Nanoemulsions Delay Seizures Without Restoring Memory? Two PTZ Mouse Models

Short answer: In a 2025 experiment using male Swiss mice, oral nanoemulsions containing cannabidiol (CBD), delta-9-tetrahydrocannabinol (THC), or a 1:1 combination improved selected seizure outcomes after chemically induced challenges. In a repeated-challenge experiment, CBD at 10 mg/kg and CBD–THC combinations at 1.5 or 3 mg/kg delayed severe seizure development, but the 6 mg/kg combination did not prevent full kindling. None of the cannabinoid regimens demonstrated a significant improvement over the vehicle group on the study’s novel-object-recognition or Y-maze measures. The researchers also reported changes in brain oxidative markers and hippocampal protein staining. They did not measure cannabinoid concentrations in blood, establish human therapeutic effectiveness, or determine an individual’s impairment.

This distinction matters whenever a claim about seizure protection, brain chemistry, or cannabinoid formulation is offered as though it were a toxicological measurement of a person. The original peer-reviewed study is de Aquino and colleagues, Nanoemulsions of Cannabidiol, Δ9-Tetrahydrocannabinol, and Their Combination Similarly Exerted Anticonvulsant and Antioxidant Effects in Mice Treated with Pentylenetetrazole (2025), Pharmaceuticals 18:782, DOI: 10.3390/ph18060782. The findings below are limited to that experiment, not drawn from a pooled clinical review.

What question did the primary experiment actually test?

The investigators asked whether separately formulated CBD and THC nanoemulsions, and a volume-mixed preparation containing both, could change seizure behavior in two pentylenetetrazole (PTZ) paradigms. PTZ is a laboratory convulsant administered to induce seizures; the mouse models were not observations of spontaneous human epilepsy. One arm used a single high-dose PTZ challenge and tracked immediate seizure and survival outcomes. The other used repeated lower-dose PTZ exposure to assess progression toward increasingly severe motor seizures, followed by memory tasks and hippocampal immunohistochemistry.

That division is essential. A longer latency to the first convulsion under one acute challenge is not the same endpoint as delayed kindling over three weeks. Neither is equivalent to protection of memory. These outcomes differ in what was administered, the timescale, how the endpoint was observed, and the biological material analyzed afterward. The study’s title reports anticonvulsant and antioxidant effects, but interpretation also requires its plainly reported unsuccessful memory results and high-combination-dose kindling result.

Who was studied, and what does “n = 15” mean?

The subjects were male Swiss mice weighing approximately 25–33 g. For each tested treatment arm in the acute experiment, the authors report 15 animals per group. Separate additional mice were used for the 21-day kindling protocol, again reported as 15 per treatment group. Fifteen therefore describes the independently assigned animals in one group of one experiment—not 15 human patients, 15 separate blood samples from one person, or an independently verified clinical trial.

Some downstream outcomes used fewer physical samples. The tissue biochemical analysis describes six to eight animals for certain determinations, while post-kindling hippocampal immunohistochemistry used four animals per group. Serial tests, microscopic regions, staining fields, laboratory determinations, and anatomical areas cannot be counted as additional independent mice. That matters when assessing whether visually striking differences have robust biological replication, particularly for the small immunohistochemistry subset.

How were the CBD and THC products formulated?

The authors prepared oil-in-water nanoemulsions using licuri oil, Pluronic F127 surfactant, and separately added cannabinoids. They mixed the CBD and THC preparations in a stated 1:1 volume ratio for combination testing. Dynamic light scattering characterized the individual preparations: THC droplets averaged 237.5 ± 1.46 nm in diameter, with polydispersity index 0.09 ± 0.03 and zeta potential −33.9 ± 2.14 mV. CBD droplets averaged 247.2 ± 1.15 nm, with polydispersity index 0.08 ± 0.02 and zeta potential −32.4 ± 1.37 mV.

Those are physical properties of the prepared emulsion droplets. The nanometer diameter is neither a blood cannabinoid level nor evidence of an absorbed human dose. The study did not directly compare systemic cannabinoid exposure with ordinary non-nanoemulsified products in a pharmacokinetic crossover. Consequently, a plausible formulation rationale must not be rewritten as demonstrated human bioavailability, more reliable intoxication, or a validated conversion from oral milligrams to blood nanograms per milliliter.

How was the acute seizure experiment conducted?

The acute experiment administered CBD or THC individually by oral gavage at 1, 3, 6, or 10 mg/kg, or a 1:1 CBD–THC combination at the study’s nominal 1.5, 3, or 6 mg/kg combination dose. After 60 minutes, mice received PTZ 80 mg/kg intraperitoneally. Observers recorded the time until the first generalized seizure, time until death, and survival information after the challenge. These are observed mouse outcomes timed in seconds or assessed as a survival proportion—not plasma drug concentrations or a clinical seizure-frequency diary.

The authors describe the mixed preparation in terms of a 1:1 volume ratio and a nominal combination dose. Without an independent component-specific administered-dose table, it would overstate the record to identify each component’s mg/kg in every mixed-dose arm. The route distinction also matters: cannabinoids were given orally, but PTZ was injected into the peritoneal cavity. This is not a scenario in which a person used a commercial cannabis product and subsequently provided a forensic blood or urine specimen.

Which treatments delayed the first acute generalized seizure?

Figure 2 reports that individual CBD and THC at the lowest 1 mg/kg dose did not significantly prolong first-seizure latency relative to vehicle. Several intermediate and higher individual-dose groups did, and the tested CBD–THC combinations were marked significantly different from vehicle. The figure’s stars provide dose-specific comparisons, while the accompanying text groups intermediate doses together more broadly. That discrepancy cautions against turning the narrative’s general phrase into an assertion of a significant effect at every dose and for every individual compound.

In particular, the plotted CBD 6 mg/kg arm does not bear the same vehicle-comparison significance marking as the CBD 3 and 10 mg/kg arms. THC treatment at 3, 6, and 10 mg/kg and all three combination bars carried favorable significance markings in the figure. The study used one-way analysis of variance and Tukey post-hoc comparisons for these plotted acute means. An effect demonstrated against vehicle is not, by itself, proof that a specific CBD–THC combination is synergistic rather than additive or merely more effective than a weaker, differently dosed comparator.

Did the formulations also alter time to death and survival?

Figure 3 reports longer latency to death for selected treatments after high-dose PTZ, particularly combinations at 3 and 6 mg/kg. The paper reports 20% survival in each of those two combination groups. Given the reported starting size of 15 mice per group, 20% corresponds to three surviving animals within each group; it is not a precision estimate of any future mouse’s survival probability. The study observed survivors for its stated post-challenge period, rather than establishing long-term neurological recovery or recurrence prevention.

The distinction between first seizure and death is not cosmetic. A treatment may shift the timing of a convulsion without rescuing all subjects from a severe intoxication-model challenge. Conversely, a better short-term survival proportion does not identify the mechanism. The experiment was not designed to establish an effective human epilepsy dose, safety margin, treatment compliance marker, or legal intoxication level for either cannabinoid.

Evidence table 1: acute experiment and formulation measurements

What did the repeated PTZ-kindling experiment add?

The second experiment tested progression of seizures over time rather than a single overwhelming convulsant exposure. Additional mice received PTZ 30 mg/kg intraperitoneally every other day for 21 days. One hour beforehand they received vehicle, CBD 10 mg/kg, the CBD–THC combination at nominal 1.5, 3, or 6 mg/kg, or valproic acid 50 mg/kg. A separate group received the 6 mg/kg mixed preparation without PTZ. The comparator valproate was included in this repeated protocol, not the separate acute oxidative-stress experiment.

The authors scored motor-seizure severity on a five-stage modified Racine scale. The scale tracks escalating visible behaviors, culminating in severe generalized motor phenomena and loss of righting. A delayed transition to a severe stage is a relevant preclinical finding, but it remains distinct from electrically documented human seizure types, number of spontaneous epilepsy attacks, and prevention of cognitive consequences. Repeated observations on the same mouse should remain connected to that mouse and should not be treated as independently recruited subjects at each time point.

Why does the ineffective 6 mg/kg combination matter?

The authors report that CBD 10 mg/kg, the CBD–THC combination at 1.5 or 3 mg/kg, and valproic acid 50 mg/kg significantly slowed progression toward severe kindling. By contrast, the 6 mg/kg combination did not prevent full kindling. Figure 6 uses nonparametric group comparisons, reported as medians, and marks significance versus vehicle for the effective arms. The result is not a straight line in which a larger nominal mixed-cannabinoid dose guarantees a better outcome.

There are several possible interpretations of the nonmonotonic pattern, but this experiment does not isolate a unique explanation. Differences in exposure, cannabinoid interaction, dose-dependent pharmacology, measurement variability, and model properties are all possibilities that would require further experiments. It would be scientifically incorrect to claim that the ineffective arm proves the 6 mg/kg formulation caused seizures, or that the effective lower arms prove a safe therapeutic window for humans. The defensible conclusion is narrower: across the doses tested, the mixed preparation did not have a uniformly increasing protective effect.

Did fewer severe seizures translate into improved recognition memory?

No significant cannabinoid-associated rescue was demonstrated. After the repeated PTZ protocol, mice completed a novel object recognition test (ORT). The discrimination index measures relative exploration of a new object compared with a familiar object. CBD 10 mg/kg and each tested CBD–THC combination did not show a statistically significant improvement over the vehicle group on this endpoint. Valproic acid at 50 mg/kg did show a significant difference against vehicle at the authors’ p < 0.05 threshold.

Importantly, lack of statistical significance is not proof that CBD or THC produces exactly no memory effect under every condition. The study itself notes that comparisons between cannabinoid arms and both vehicle and valproate do not settle an equivalence question. The no-PTZ mixed-dose arm helps provide context, but the design does not provide comprehensive baseline and sham-kindled comparators that would permit exact calculation of how much impairment was induced or reversed for each individual mouse. These boundaries belong in any faithful assessment of the ORT finding.

What did the Y-maze show about working memory?

The Y-maze assessed spontaneous arm alternation over an eight-minute observation, a rodent laboratory proxy for working-memory processes. Here too, none of the cannabinoid pretreatments produced a significant protective improvement against vehicle after the repeated PTZ challenges. The valproate comparator produced a favorable group difference reported at p < 0.05. The authors therefore did not find the cognitive improvement that might have been predicted merely from observing benefits on seizure progression.

Recognizing the assay’s scope is essential. An ORT discrimination index or maze alternation rate is not a human driving reaction-time test, neuropsychological battery, memory truthfulness test, or assessment of whether a witness accurately remembers an event. The outcome is a mouse’s behavior in a tightly defined apparatus. It cannot be assigned a forensic threshold in whole-blood THC concentrations, because whole-blood THC concentrations were never determined and the paper does not supply a validated bridge between the experimental behavior and an individual legal event.

Evidence table 2: repeated PTZ, cognition and tissue measurements

What were the acute brain oxidative-stress results?

After the acute PTZ challenge, the investigators examined the prefrontal cortex, hippocampus and striatum. A Griess-based procedure assessed nitrate/nitrite, frequently discussed as a downstream marker associated with nitric-oxide metabolism. A thiobarbituric acid-reactive-substances method assessed malondialdehyde-related lipid-peroxidation signal. Compared with PTZ-challenged vehicle animals, CBD 10 mg/kg and the tested mixed 1.5, 3, and 6 mg/kg regimens were reported to attenuate the increases in both measures across the sampled brain regions (Figures 4 and 5; source-reported p < 0.05).

These are biochemical assays of dissected tissue, not a clinical blood test and not a measure of the amount of CBD or THC inside a particular neuron. Nitrate/nitrite and TBARS-related MDA are commonly used indirect indices of biochemical stress; they do not identify an exclusive cause for each signal, nor establish that an individual animal’s long-term brain function was restored. The original paper’s methods and figure captions are not fully consistent in the displayed mass-normalized unit for these readouts, so interpreting the direction and reported comparisons is safer than copying an unsupported precise absolute unit into a forensic concentration table.

What do GFAP, GAT-1 and Kir4.1 staining establish?

In the hippocampus after the repeated PTZ protocol, researchers assessed immunoreactivity for GFAP (an astrocyte-associated structural protein), GAT-1 (a GABA transporter) and Kir4.1 (a potassium-channel protein). Their figures describe differences in staining across CA1, CA3 and the dentate gyrus, broadly involving lower GFAP/GAT-1-associated changes and increased Kir4.1 expression measures in selected cannabinoid groups compared with the kindled vehicle group. The patterns are compatible with altered glial responses in the experimental model.

However, the immunohistochemistry sample was four mice per group, not fifteen independent brains per staining marker. A staining-area or intensity change is an assay-specific protein-localization outcome. It is not direct evidence of restored network signaling, receptor occupancy, metabolite formation, CBD–THC synergy, or lower blood THC. Without interventions that selectively perturb GFAP, GAT-1, or Kir4.1 while tracking downstream seizure rescue, the measured association cannot by itself establish which pathway caused the observed behavioral changes.

Did the study measure THC or CBD in blood, plasma, urine or breath?

No. The administered dose was reported as milligrams per kilogram by oral gavage, while the directly assessed outcomes were behavioral events, formulation droplet properties, brain-tissue biochemical readouts, and immunohistochemistry. The paper did not present a validated panel of measured parent THC, CBD, 11-hydroxy-THC or THC-COOH concentrations in venous whole blood, separated plasma, serum, urine, oral fluid, exhaled breath, vitreous humor or hair from the experimental mice.

A nominal mg/kg oral dose is therefore not an empirical toxicology result in ng/mL blood. The article does not supply a pharmacokinetic calibration curve, paired blood-and-brain concentration time course, or conversion equation from dose to an individual specimen concentration. It also does not directly establish the concentration of THC in the brain at the time of seizure protection or memory testing. In a forensic setting, a quantitative concentration requires validated analysis of the actual relevant specimen and matrix; see the site’s guide to biological matrices.

Why are combination effects not the same as proven synergy?

The mixed preparation outperformed some low-dose single-cannabinoid comparators on the acute seizure figures, but that is insufficient to characterize the drug interaction formally as synergistic. A synergy analysis requires an appropriate definition of expected additivity, matched or factorial component doses, exposure information, and a comparison of the observed mixture effect to that expectation. The paper did not establish those conditions in a way that validates a general numerical CBD–THC interaction law.

There is also a distinction between delivery formulation and pharmacology. Small droplets can improve dispersibility, but no head-to-head in vivo exposure comparison measured systemic absorption from the nanoemulsion against a conventional oral vehicle. The lower-dose mixed kindling advantages and the 6 mg/kg failure underscore why product composition, nominal gavage dose, effective internal exposure and final biological outcome should remain separate variables rather than collapsed into the assertion that “the combination works better.”

What are the principal design and statistical uncertainties?

Several boundaries apply. Only male Swiss mice were used; the acute and repeated experiments employed different PTZ regimens and subject groups; group sizes narrowed for molecular outcomes; and the cognitive experiments did not include a full range of sham and exposure-matched controls. The acute graphs present mean values with standard errors, whereas kindling severity is presented through nonparametric ordinal comparisons. The sample sizes and measurement scales therefore cannot be pooled as though every endpoint came from a single uniform dataset.

The original figures sometimes provide significance symbols without a full numerical effect-size interval for each dose contrast. A significant p value identifies evidence against a specified null under the test used; it does not quantify clinical relevance or validate an individual prediction. Conversely, the cognitive null comparisons do not prove equality. The dose designation for the 1:1 combination describes the mixed preparation and should not be rewritten into separately verified per-component systemic doses. These are methodological limitations, not reasons to erase the measured effects that were actually observed.

How should the findings be used in a forensic-toxicology dispute?

First identify the proffered proposition. If a party claims the study demonstrates an antiseizure effect in PTZ-challenged mice under the specified regimens, there is relevant original experimental evidence, including both acute effects and certain repeated-kindling contrasts. If the claim extends to a person taking a particular medicine, product potency, effective serum or whole-blood THC concentration, route of cannabis use, or alleged intoxication at the time of an incident, the experiment supplies no validated direct measurement or transfer function.

The same principle applies to negative and positive findings. Better seizure endpoints did not automatically yield statistically demonstrable restoration of ORT or Y-maze behavior. Changes in oxidative chemistry and hippocampal staining cannot be substituted for demonstrated normal cognitive performance. Conversely, absent measured human cognition does not mean no mechanism could ever be explored; it means that this particular primary source does not decide it. For evidentiary work, analytical identity and reporting limits are addressed separately in confirmatory drug testing, and specimen provenance in specimen collection.

Seven questions frequently asked about this experiment

1. Did the CBD–THC mixture completely prevent PTZ seizures?

No. In the acute model, selected treatment groups prolonged latency to seizure and death, while survival reached a reported 20% in two mixed-dose groups. In the repeated model, some regimens delayed more severe kindling, but the 6 mg/kg mixture did not prevent full kindling. None of those findings establishes complete seizure prevention.

2. Was 6 mg/kg always more effective than 3 mg/kg?

No. That would reverse a critical result. The 6 mg/kg mixture showed favorable acute-challenge measures, but in the 21-day repeated-kindling model it failed to demonstrate the protection reported for the 1.5 and 3 mg/kg combination regimens. Acute and repeated challenges cannot be treated as interchangeable dose-response experiments.

3. Does the study show that THC improves memory?

No. Neither CBD 10 mg/kg nor the tested CBD–THC combinations showed a significant benefit relative to vehicle on the post-kindling novel-object-recognition or Y-maze assessments. The valproate comparator did show favorable statistical comparisons. The null cannabinoid findings do not prove an identical lack of effect under every other design.

4. What does the reported 20% survival mean?

It is the authors’ reported survival proportion after a severe acute PTZ challenge in the two higher combination arms. With fifteen mice assigned per arm, it corresponds to three survivors in each arm. It is a small experimental group result, not an estimate of survival for an identified person exposed to cannabis or experiencing epilepsy.

5. Were THC, 11-hydroxy-THC or THC-COOH quantified in venous whole blood?

No. The study characterized prepared nanoemulsions and measured mouse responses and tissue biomarkers, but it did not report direct cannabinoid quantitation in whole blood, plasma, serum or a validated casework matrix. An oral dose in mg/kg cannot be converted into a case blood concentration from these data.

6. Do lower MDA and nitrate/nitrite signals prove a safer human drug?

No. These were brain-tissue oxidative and nitrosative surrogate measures collected under a laboratory convulsant challenge. They neither establish systemic clinical safety nor prove that the markers caused the anticonvulsant effect. Human safety, pharmacokinetics, and functional outcomes require independent evidence.

7. Can this mouse study establish impairment or causation in a DUI, injury or death case?

No. It included no tested drivers, no human behavior, no forensic person-specific drug concentration, no validated impairment cutoff, and no model relating a specimen value to a particular incident. The scientifically appropriate use is to describe what the mouse challenge did and did not show, not to make a person-specific causal finding from an unmeasured biological compartment.

Source and bottom line

The controlling primary report is de Aquino P.E.A. and colleagues (2025), Pharmaceuticals 18(6):782, https://doi.org/10.3390/ph18060782. Its complete Methods, original dose-specific figures, behavioral findings and histochemical results are needed for evaluating any extrapolation.

Bottom line: A laboratory formulation can shift seizure latency and selected kindling outcomes without demonstrating restored memory. The investigators documented tissue biochemical and immunostaining differences, not measured cannabinoid concentrations in human forensic specimens. The distinction between a mouse anticonvulsant effect, a molecular surrogate, a clinical treatment claim and an individual impairment or causation opinion is the central evidentiary lesson.