The efficacy and safety of cannabidiol as adjunct treatment for drug-resistant idiopathic epilepsy in 51 dogs: A double-blinded crossover study

2.1 Animals

The study was approved by the Colorado State University (CSU) Institutional Animal Care and Use Committee (Protocol #16-6790A). A target of 60 dogs was chosen to find an estimated difference of 50% in the episodes between the 2 groups. The calculation considered a statistical power of 80% and confidence of 95%. Dogs were prospectively enrolled from January 2018 to September 2020 after owner consent and fulfillment of enrollment criteria. Dogs were included if they met the tier II confidence level for diagnosis of IE, as defined by the International Veterinary Epilepsy Task Force Consensus Statement.²⁶ Specifically, criteria included pretrial owner documentation of ≥2 seizures per month for at least 12 weeks while receiving at least 1 conventional ASD (phenobarbital, potassium bromide, levetiracetam, and zonisamide). Dogs receiving phenobarbital or potassium bromide were required to have blood or serum concentrations within the reported therapeutic range (20-40 μg/mL for phenobarbital and 0.88-3.0 mg/mL for potassium bromide).^27-29 Those receiving levetiracetam or zonisamide were required to be within the accepted dose range for these medications (≥20 mg/kg PO q8h immediate-release levetiracetam; ≥30 mg/kg PO q12h extended-release levetiracetam; and ≥5 mg/kg PO q12h zonisamide). Alanine transaminase (ALT) and alkaline phosphatase (ALP) were considered abnormal if activities were >2.5× the high end of the reference range. If either ALT or ALP was abnormal, postprandial bile acid concentration was required to be normal (<30 μmol/L) for inclusion. Activity of ALP >2× the high end of the reference range in dogs receiving either zonisamide or phenobarbital was still acceptable if ALT activity or postprandial bile acid concentrations were normal because those medications are known to cause a clinically unimportant increases in ALP activity.^{30, 31} Infectious disease testing was performed if cerebrospinal fluid (CSF) analysis was abnormal and had to be negative for Bartonella spp., Ehrlichia spp., Anaplasma spp., Neorickettsia spp., Dirofilaria immitis, Wolbachia spp., Rickettsia spp., Toxoplasma gondii, Neospora caninum, Borrelia burgdorferi, and canine distemper virus. Other exclusion criteria included anticipation of owner noncompliance and dogs with documented comorbidities associated with a poor prognosis.

2.2 Study design

The study was conducted as a randomized, double-blinded, placebo-controlled crossover clinical trial comparing the effect of CBD versus placebo on seizure frequency in dogs taking concurrent ASD medication. A computer-based random number generator (Microsoft Excel, Microsoft Corporation, Redmond, Washington) was used to assign a treatment order to each dog. The owners and primary researchers were blinded to the treatment order until the end of the trial. Once a treatment order was assigned, owners were given either CBD-infused hemp seed oil with chicken flavoring (Applied Basic Science Corporation, Castle Rock, Colorado) or the placebo (hemp seed oil with chicken flavoring) to be administered PO at the same volume and timing throughout the trial. Dogs were given the CBD or placebo for 12 weeks and then, after a 4-week washout period, given the opposite oil. The first 13 dogs enrolled in the trial were given a dosage of 5 mg/kg/day (2.5 mg/kg q12h) of each oil. However, because of lack of improvement during either treatment phase, we elected to increase the dosage to 9 mg/kg/day (4.5 mg/kg q12h) for the next cohort of 48 dogs. During the baseline period (12 weeks before enrollment) and the study period, no alterations could be made to the ASD protocol without withdrawing from the study. However, 3 to 5 days of “rescue dosing” with levetiracetam, gabapentin, clorazepate, or a combination of these medications were permitted for cluster seizure or status epilepticus management before and during the trial.

At enrollment, baseline seizure frequency was recorded, and blood tests were obtained. These included ASD serum concentrations, CBD plasma concentrations, CBC, serum biochemistry, and postprandial bile acid concentrations. Each week, owners recorded behavioral and temperament variables using the previously validated Canine Behavioral Assessment & Research Questionairre (CBARQ) evaluation.³² Owners recorded observed seizure activity in their pets using seizure logs, noting the frequency and type of seizures (focal or generalized). Re-evaluation appointments were performed once monthly, at which time blood was collected for CBD plasma measurements, CBC, and serum biochemistry monitoring. Blood sampling was not performed at fixed time points postdose administration and instead was performed at appointment times more convenient for the owners. Owners also were told not to change their pet's diet for the duration of the trial.

Dogs that could not attend monthly appointments at CSU had blood samples collected by the primary veterinarian or a veterinary neurologist and shipped to CSU. At 0, 12, and 28-week time points all owners were required to bring their dogs to CSU for a neurological examination and to obtain additional blood samples for postprandial bile acids and ASD serum concentrations.

2.3 CBD oil and placebo information

The CBD was extracted from a crop that was registered as industrial hemp with the Colorado Department of Agriculture (#77046). The classification as industrial hemp certified that the plant contained a tetrahydrocannabinol concentration of ≤0.3%. Each batch of product was analyzed by at least 1 third-party laboratory for purity and cannabinoid concentrations (ProVerde Laboratories, Inc., Milford, Massachusetts; SC Labs [formerly Botanacor], Denver, Colorado; Aurum Labs, Durango, Colorado; (Table S1). The CBD oil was considered a full-spectrum product, containing approximately 100 mg/mL of CBD with trace amounts of other cannabinoids, varying slightly with different batches. The other ingredients were cold-pressed hemp seed oil and chicken flavoring. The placebo oil contained only the cold-pressed hemp seed oil and chicken flavoring. The CBD oil and placebo were masked in scent and appearance.

2.4 Laboratory analyses

Plasma concentrations of CBD were measured in canine plasma using a previously validated liquid chromatography-mass spectrometry technique³³ at the Pharmacology Laboratory of the Drug Development & Discovery Shared Resource (University of Colorado Cancer Center). Phenobarbital and potassium bromide serum concentrations were measured at the CSU Veterinary Diagnostic Laboratory. Phenobarbital was measured according to the package insert for an Immulite 2000 analyzer (Siemens Medical Solutions USA Inc., Malvern, Pennsylvania). Levetiracetam and zonisamide serum concentrations were measured at Auburn University's Veterinary Diagnostic Laboratory using previously published methods.^{34, 35}

2.5 Statistical analysis

A seizure day was defined as any 24-hour period during which a dog had at least 1 seizure, whereas total seizures were the total number of individual seizures reported for each dog. Percentage data were evaluated for normality using the UNIVARIATE procedure from SAS; it was not met for any of the endpoints, and thus, cubic root was used to approximate normal model assumptions. All endpoints were back-transformed for reporting in the original scale. The fixed effects included time (month), treatment, and their interaction term, which was not significant and, thus, removed from the model. Baseline absolute values for each percentage change endpoint were included as a covariate, and dog nested within treatment sequence was added as a random effect in the model. Pairwise treatment comparisons for all parametric data were conducted using the Tukey-Kramer test to protect against type I error.

Individual responsiveness to treatment was considered as an average monthly decrease of ≥50% in seizure numbers and seizure days. For that analysis, a 1-tailed t test was conducted comparing each treatment to a null hypothesis of 50% (SAS v 9.4). The number of AEs while on each treatment reported by the owner was analyzed as categorical data using Fisher's exact test wherein these were assigned a “yes” for each dog that had at least a single episode of each AE and a “no” if none was reported.

The ALP and ALT data were evaluated for normality using the Shapiro-Wilk test. If not normally distributed, the data were converted into logarithmic scale and GLIMMIX was constructed to evaluate the treatment and time point interaction effects. Tukey-adjusted differences and 95% confidence limits (CIs) were reported. A P value of .05 was used to determine significance. The program SAS v9.4 (SAS Institute Inc., Cary, North Carolina) was used for all statistical analyses.

3.1 Enrollment

A total of 502 dogs were screened for eligibility for the study, and 61 met the inclusion criteria and were enrolled. Thirteen dogs initially were enrolled at 5 mg/kg/day, and 12 dogs completed this portion of the trial (which included crossover with placebo). The dogs receiving the 5 mg/kg/day CBD dose showed no evidence of a treatment effect (P = .36) on the percentage change of total seizures and seizure days. Consequently, the principal investigator decided to increase the CBD dosage of the remaining dogs to 9 mg/kg/day, based on epilepsy data in humans showing higher dosages (10-20 mg/kg) being effective for Lennox Gastaut syndrome.^{19, 20} Forty-eight dogs were enrolled at the 9 mg/kg/day CBD dosage, and 39 dogs completed the trial. Reasons for withdrawal from the 5 and 9 mg/kg/day groups included euthanasia because of worsening seizures or status epilepticus (5/61), acute kidney failure (1/61), intestinal perforation (1/61), accidental death (1/61), change in ASD management (1/61), and lost to follow-up (1/61). All dogs that completed both arms of the trial were included in the analysis.

Documented comorbidities at enrollment included allergies (5/61), hypoadrenocorticism (1/61), cranial cruciate ligament disease (1/61), suspected renal dysplasia (1/61), hypothyroidism (1/61), and low-grade heart murmur (2/61).

At the time of enrollment, 1 dog had a bromide concentration that was 16% below previously reported therapeutic margins. This dog also was receiving zonisamide and levetiracetam. The bromide concentrations were measured and found to be within the therapeutic range at the 12th- and 28th-week ASD concentration checks.

During enrollment, 3 dogs were considered to have increased nucleated cell counts on CSF analysis (118/μL, 45/μL, and 10/μL). The 1st had a small amount of suspected iatrogenic blood contamination (4297 RBC/μL) and was re-evaluated 1 week later and found to have a nucleated cell count of 0/μL. The 2nd had a large amount of suspected blood contamination (>200 000 RBC/μL). Given normal magnetic resonance imaging (MRI) findings and lack of interictal neurological deficits, the increased nucleated cell count was attributed to the large amount of blood in the sample, and CSF was not re-evaluated. The 3rd dog also had a large amount of suspected iatrogenic blood contamination (>18 000 RBC/μL). Again, given the normal MRI findings and lack of interictal neurological deficits, the increased nucleated cell count was attributed to blood contamination, and CSF was not re-evaluated. Two dogs had albuminocytologic dissociation with protein concentrations of 32 and 42 mg/dL. On MRI, 1 dog had cerebral microbleeding with normal CSF analysis and was negative for Anaplasma phagocytophilum, D. immitis, B. burgdorferi, and Ehrlichia ewingii on 4Dx testing. The microbleeding was not considered to be the source of seizures for this patient. All 6 dogs described were included in the trial.

During the trial, 1 dog in the 9 mg/kg/day CBD phase of the trial had its potassium bromide dosage decreased 8 weeks into the trial because of concern for possible bromism. This dog began the study with a bromide concentration of 2.2 mg/mL, and also was receiving phenobarbital and zonisamide during the trial. At approximately 8 weeks into the trial, the dog was hospitalized for lethargy and weakness and its bromide concentration at this time was 2.3 mg/mL. Without other findings to explain this change, it was elected to decrease the potassium bromide dosage by 25%. This dog was receiving placebo at the time and received CBD in the 2nd half of the trial. Three crossed over from the 1st trial oil to the 2nd at 6, 8, and 9 weeks without a washout period because of a large number of seizures at home. Of these 3 dogs, 2 were receiving CBD and 1 was receiving placebo during the 1st phase. Another dog was started on the 2nd trial oil after a 3-week washout. It was receiving CBD in the 1st phase and did not have detectable CBD in the plasma at the start of the placebo oil.

3.2 Animal demographics

Given the lack of significant effect at the 5 mg/kg/day dosage and small sample size, that data will not be reported here but can be made available upon request.

Of the 39 dogs in the 9 mg/kg/day dose group, the most represented breed was mixed breed (15/39), followed by golden retriever (6/39), French bulldog (3/39), Labrador retriever (2/39), Border collie (2/39), and Neapolitan mastiff (2/39), with 1 each of Australian shepherd, Australian cattle dog, Boston terrier, Gordon setter, Irish setter, Jack Russell terrier, Nova Scotia duck tolling retriever, Rottweiler, and Yorkshire terrier. The most represented sex was male castrated (24/39), followed by female spayed (13/39), and male intact (2/39). No female intact dogs were included in the study. The mean age at the start of the study was 3.5 ± 2.0 years. The mean body weight was 28.7 ± 13.0 kg, with a range of 2.9 to 60.4 kg.

3.3 Animal seizure characteristics and ASD medications: 9 mg/kg/day dose

During the pretrial period, 28/39 dogs were reported to have only generalized seizures, 8/39 dogs experienced both generalized and focal seizures, and 3/39 dogs experienced focal seizures alone. The most common ASD combination was phenobarbital and levetiracetam (26/39) and most dogs were receiving ≥2 ASDs (36/39). Three dogs were receiving monotherapy with either phenobarbital (n = 2) or levetiracetam (n = 1).

3.4 Percentage change in total seizures and seizure days from baseline: 5 and 9 mg/kg/day doses

No significant changes (P = .36) to total seizures or seizure days were observed in dogs receiving 5 mg/kg/day of CBD.

For dogs receiving 9 mg/kg/day, a post hoc power analysis assessment using the power procedure from SAS (v 9.4) for crossover designs, considering a 30% intra-dog correlation, indicated that percentage changes of total seizures had a power of 71.3%, whereas that of seizure days was 93.0%.

Dogs receiving CBD at the 9 mg/kg/day dose had a small increase in percentage change (3.31%) of total seizures from baseline, which was 10 times lower in magnitude compared with placebo (30.72%) and was significantly different (P = .04; Figure 1). Conversely, the percentage change of seizure days while on CBD decreased by 24.1% (P = .002), whereas dogs on placebo had a 5.81% increase (Figure 2). When analyzing the effect of treatment on the percentage change of seizure type, generalized seizure percent change was not different between CBD and placebo groups (P = .12). Model fitting of focal seizures was not feasible because of the large number of zero values. No significant differences related to the sequence of placebo and CBD oil for total seizures (P = .50) and seizure days (P = .94) were found.

There were 9 responders (percentage change ≥ −50%) in the CBD group and 8 in the placebo group relating to total seizures (P = .78), and 13 versus 8 responders relating to seizure days for the CBD and placebo groups, respectively (P = .20). These differences were not significant.

3.5 Interaction between ASDs and CBD oil: 9 mg/kg/day dose

The interactions between phenobarbital, potassium bromide, zonisamide, and levetiracetam and treatment had no effect (P > .07) on the percentage change of seizure days or total seizures.

No evidence was found for differences between CBD administration and placebo for percentage change from baseline in ASD concentrations for any of the measured ASDs (P > .15). Large variation was found in all measured ASDs, which ranged from −42% to +144% for phenobarbital, −28% to +145% for potassium bromide, −45% to +174% for zonisamide, and −88% to +1311% for levetiracetam. Although not significant, the average percentage change of phenobarbital while receiving CBD was +11% and during the placebo phase was −1%.

3.6 Bile acid monitoring and change in liver enzyme activities: 9 mg/kg/day dose

When comparing liver enzyme activities between months, a significant difference was found when comparing CBD and placebo treatment, and CBD-treated dogs consistently had higher ALP activities (P ≤ .001). Also, a significant difference in ALP activity was found when comparing CBD treatment during all 3 months and baseline (P ≤ .001). No difference was found between placebo and baseline for any month (P ≥ .88; Figure 3). A significant difference was found in mean serum ALT activity between CBD and placebo groups (P = .003; Figure 4). The mean baseline ALP activity was 278 (9-4112) IU/L. The mean ALP activity after the 3rd month of CBD treatment was 1585 (16-9642) IU/L. Of 30 dogs with ALP activity ≤2× the high end of the normal range at baseline, 24 (80%) had activities above twice the high end of the normal reference range when receiving CBD (95% CI, 368.7-849.9). Two of these dogs also had a single-month result >2× the high end of the reference range during placebo administration. The mean baseline ALT activity was 64.4 (9-483) IU/L. The mean ALT activity after the 3rd month of CBD treatment was 114.5 (7-646) IU/L. In 5 dogs (12.8%), ALT activity increased >2× the high end of the reference range during CBD administration (95% CI, 52.1-75.4). One dog had ALT activity >2× the high end of reference range only during the placebo phase. Two dogs had ALT activity at ≥2× high end of the reference range during both the placebo and treatment phases. Of these 2 dogs with increased ALT activity during both phases, both had baseline ALT activity above the high end of the reference range with 1 above twice the high end of the reference range.

Postprandial bile acid concentrations were increased in 2 dogs at the end of week 28. One had received CBD in the 1st phase (52 μmol/L) and the other had received CBD in the 2nd phase (81 μmol/L). All other bile acid concentrations at weeks 12 and 28 were normal.

3.7 Adverse events

Owner-reported AEs during 9.0 mg/kg/day dosing included decreased appetite, vomiting, soft feces or diarrhea, foraging, anxiety, increased appetite, increased activity, decreased activity, ataxia, and aggression (Table 1). Of these, only decreased appetite (P = .01) and vomiting (P = .05) were more frequent when dogs received the CBD treatment compared with placebo (Table 1). No differences were found between CBARQ scores for trainability, aggression, fear and anxiety, separation-related anxiety, excitability, attachment/attention seeking, or other behavioral problems while dogs were on CBD compared with placebo in both dosing groups.

3.8 Plasma CBD concentration

The mean plasma CBD concentration at the 5.0 mg/kg/day dosage was 173.6 ± 101.5 ng/mL. Plasma CBD concentrations in the 9 mg/kg/day treatment group ranged from 226.7 to 2833.7 ng/mL with a mean ± SD of 1207.4 ± 571.3 ng/mL.