We evaluated the impact of flumazenil administration as an adjunctive treatment for dogs with symptomatic marijuana toxicosis. Symptoms were assessed using an established scoring scheme (MGCS) and a novel score (CMSS). Flumazenil administration was associated with improvements in both scores within 30-min, consistent with reduced severity of neurologic signs. Quantifiable THC concentrations were identified in all but one dog. Dogs with lower CMSS tended to have significantly higher serum THC concentrations. This novel score therefore shows promise as a novel indicator of clinical severity in affected dogs.

Marijuana toxicosis in dogs is common, usually causing significant but self-limiting morbidity in dogs. Fatalities have been reported however and more specific interventions might be necessary to hasten clinical recovery (1). In veterinary medicine, flumazenil is primarily used to reverse paradoxical excitation and prolonged recovery times following benzodiazepine administration (23). Flumazenil is used in human medicine to reverse the effects of several toxicants, including marijuana (14, 16–19). The aim of this study, therefore, was to assess whether flumazenil administration was associated with improvement in clinical signs in dogs with symptomatic THC toxicosis. In our study, 13/17 and 10/17 affected dogs had improved ($\ge$1 change in score from baseline) CMSS and MGCS scores, respectively, within 30-min of flumazenil administration, consistent with less severe neurologic signs (Supplementary Tables S2, S3). MGCS may be a less sensitive scoring system compared to the novel CMSS. The frequency of ataxia in this cohort was 70% and the MGCS does not assess this parameter unlike the CMSS. We did not observe statistical improvement in either score within 5 or 15-min of flumazenil administration. Interestingly, the peak effect of flumazenil occurs 6–10 min after administration in people (15), with an elimination half-life of 40–60 min. In dogs given tiletamine-zolazepam, there is a dose-dependent improvement in the ability to stand and walk 10 min after flumazenil administration (23). It is possible that an additional or higher dose of flumazenil would have achieved a faster clinical response. It is also plausible that the weak GABA_A antagonist effect of flumazenil on THC is delayed in comparison to the stronger competitive antagonistic effect seen with flumazenil and benzodiazepines. It is possible that the dogs included in our study might have improved on their own despite the use of flumazenil. However, given the biological half-life of THC is approximately 30 h, spontaneous clinically overt improvements within 30 min are less plausible (6, 9).

Cranial nerve deficits were present in 58% of affected dogs in our study, which appears higher than previous reports in the veterinary literature (4, 8). Delayed PLR in one or both eyes was the most common cranial nerve deficit noted in this cohort. Binagia et al. recently published a large retrospective study of patients with canine marijuana toxicosis, reporting delayed or absent PLR in 1.3% of cases (4). The reason for these differences is not known, although interobserver variation between clinicians performing cranial nerve exams is plausible. The PLR was assessed as normal by the end of the 30-min monitoring period in 6/10 patients suggesting there was an observable change in speed in the PLR response. In humans there appears to be a correlation between higher THC concentrations and more delayed PLR reflexes. The average serum THC concentration in this cohort of dogs was higher than previously reported, which could explain the increased frequency of PLR abnormalities (22).

One dog in this pilot study was diagnosed with both marijuana toxicosis and meningitis of unknown etiology. In humans, there is a possible association between THC and the acute onset of necrotizing encephalitis without evidence of infectious etiology (24). In dogs with steroid-responsive meningitis-arteritis there are increased circulating endocannabinoids, and THC can directly increase the concentration of plasma endocannabinoids (10, 25). While there is literature showing the anti-inflammatory properties of endocannabinoids, both in vivo and in vitro work has shown that AG-2 can play a pro-inflammatory role via the production of reactive oxygen species, promoting white blood cell migration and stimulating pro-inflammatory cytokine release (10). This suggests that the immunomodulatory action of THC could contribute to a pro-inflammatory state that led to the onset of meningoencephalitis in this patient.

There are sparse reports in the human literature stating reversal with flumazenil can cause seizures and cardiac arrhythmias. Reports are often associated with high flumazenil dosing, history of chronic benzodiazepine use, or history of multiple drug co-ingestion (15). In a large observational study in humans, flumazenil administration did not result in seizures or arrhythmias in those patients despite seizure history and multiple drug co-ingestion (18). Other minor side effects of flumazenil in humans are vomiting, salivation, and anxiety (18). In a study where beagles are given flumazenil to reverse tiletamine-zolazepam, seizures, hypersalivation, vocalization, and opisthotonos were reported at doses 8–16 times the dose used in this pilot study (0.01 mg/kg) (23). One dog in this pilot study vomited during the monitoring period but had a history of vomiting prior to presentation, which could be related to prior marijuana ingestion or flumazenil administration. One dog enrolled had a history of well-controlled seizures on phenobarbital and CBD and was seizure-free for more than a year, and no seizure activity was noted during monitoring or in the 24 h after discharge after follow-up with the owner. However, given the small cohort in this study it cannot be determined whether there is an increased risk of seizures when giving flumazenil at standard doses to patients with a history of seizures. It also is prudent to mention that if a dog does have a seizure after flumazenil administration, benzodiazepines may be less effective in terminating the seizure.

In our study, dogs with higher serum THC concentrations had lower CMSS at baseline. This could indicate that these patients were more severely affected; however, we collected serum on presentation and dogs were presented at different time points post-exposure. For three of these dogs, serum was stored in a −20°C freezer until a −80°C freezer was later purchased by the hospital. Resin-based marijuana products with varying levels of THC concentrations showed no statistical difference in THC concentrations over a four-year time period when stored at −20°C (26). Therefore, differences in THC stability between the samples stored in −20°C and −80°C freezer are unlikely.

One patient had a serum THC concentration > 1,500 ng/mL, the highest in this cohort, and concurrently had the lowest CMSS score (11/22). There were only 3 patients with CMSS scores of 12 or less. Those patients, while all having a significant change from baseline CMSS (+2 or greater) across the 30 min, 2/3 had dramatic changes in CMSS from baseline (+5 and + 6). Two of the dogs’ gait were unable to be assessed to due severe impairment (gait contributed 0 to their score) which impacted their overall baseline score. All patients had detectable concentrations of THC in serum except for one patient whose score was unchanged throughout the monitoring period. This patient’s serum was stored at −80°C immediately after collection until shipment for analysis.

Limitations of this study include lack of a control group, small sample size, different scoring assessors, different storage conditions, and the paucity of moderate and severe marijuana toxicosis patients. There was a significant change in CMSS and MGCS after 30 min following flumazenil administration and for some patients this resulted in either mild clinical improvement or no change from baseline. While it appears that flumazenil is well-tolerated in this small cohort of patients, it should still be used with caution depending on the patient factors and exposure scenario.

In conclusion, flumazenil might be considered as an adjunctive treatment for marijuana toxicosis based on our observations of improved neurological symptoms within 30 min of administration in affected dogs. The CMSS scoring system appeared to identify dogs with higher serum THC concentrations and more severe symptoms. Finally, we report on a novel case of concurrent marijuana toxicosis and meningitis of unknown etiology in one dog.