This systematic review demonstrates potential efficacy of cannabinoids in improving motor functions in toxin-based rodent models of PD measured by both rotarod and pole behavioural tests. The meta-analyses provide evidence justifying the conduct of clinical trials investigating cannabis or cannabis-based treatment of PD in humans, particularly for studies intending to show an improvement of motor function as their primary focus. However, this does not guarantee successful translation clinically. Included studies have high or unclear risk of bias, which might have affected the outcomes. To investigate whether cannabis has any potential benefit for patients with PD, a properly designed RCT is needed.

A meta-analysis of rotarod results from nine studies demonstrated an estimated treatment effect of 31.63 s (95% CI 10.98 to 52.27 s; p=0.003) increased retention time on the rotating rod in favour of the cannabinoid-treated group. For studies that used only toxin-based rodent models, there was increased retention time on the rotating rod of 35.34 s (95% CI 11.67 to 59.01; p=0.003) and for studies that specifically used MPTP toxin PD rodents, 26.30 s (95% CI 1.33 to 51.27; p=0.039). A reduction in descending time from the pole was also demonstrated in the meta-analysis with an estimated treatment effect of −1.51 s (95% CI −2.85 to –0.16; p=0.028). For studies that only used toxin-based rodents, an estimated treatment effect of −2.03 s in descending time (95% CI −3.20 to –0.87; p=0.001) was observed while for studies that used MPTP toxin only had an estimated treatment effect of −1.72 s in descending time (95% CI −2.97 to –0.47; p=0.007) in favour of cannabinoids.

Rotarod and pole tests are common and reliable objective behaviour assays used in PD animal experiments.13 14 Both measurements are used to quantify progress or improvement in bradykinesia and postural stability, common symptoms in PD.61 The rotarod test requires animals to stay on the rod that is rotating faster than normal walking speed, and to have sufficient dopamine and reticulospinal tract function12 to execute this activity. Pole test involves forelimb grasping and manoeuvring to turn and climb down from the top of the pole, which requires intact basal ganglia and sufficient rubrospinal tract function.12 Although these two measurements are reliable and often used in PD animal experiments, there are identified limitations.

Both tests require animal models to be pretrained prior to the actual testing.62 Most of the studies in this meta-analysis described pretraining of their animal models. Additionally, as with any animal testing, there are always external factors that create stress to animals being tested, which may influence their overall performance and behaviour.63 64 For instance, mice are naturally social animals and follow social hierarchy. A male mouse may dominate other mice in the same cage by guarding food supply thereby limiting intake in other mice.65 66 Given the variability in animal testing environments, there are factors possibly impacting the results of behavioural examinations: the timing, dosage and total duration of interventions administered, the degree of dopamine loss in the disease course and when the behavioural assays were measured. Some mice were considered reverse cycled (nocturnal)67 but tested during daylight. Despite these limitations, there is no single behaviour assay in rodents that captures all the motor deficits of human PD. The rotarod and pole tests are the more common and reliable tests we have readily available for preclinical testing in PD.

The toxin rodent PD models are appropriate to use when investigating neuroprotective interventions such as the case with cannabinoids because of the high replicability of PD motor symptoms in this category of animal models.62 However, there are also known limitations with neurotoxic animal models such as absence of typical PD intraneural proteinaceous Lewy’s bodies, particularly in MPTP-treated or 6-OHDA-treated rodents.62 Toxin-based PD rodents are currently one of the most widely accepted experimental models that has been invaluable in better understanding PD mechanisms and in screening potential treatments. In fact, PD medications such as amantadine, a glutamate antagonist widely used for dyskinesia, were successfully translated to clinical use because of experiments conducted using MPTP-treated animal models. Deep brain stimulation for advanced PD is another treatment that was pioneered on toxin-based rodent models.62

There are also known differences of PD animal models even within the toxin-based category. Toxins used for PD animal models such as reserpine produce parkinsonian symptoms, however, their effects are temporary and do not lead to dopamine (DA) neuronal death.68 A better yet imperfect alternative is the 6-OHDA toxin which is more reliable in damaging DA neurons and producing parkinsonism. However, its effects are acute and does not produce complete composite symptoms of PD. This toxin is also often injected unilaterally rather than bilaterally into the substantia nigra, the nigrostriatal tract of the striatum. Bilaterally lesioned animal models often die from marked aphagia, adipsia and seizures.69 This outcome is similar with MPTP where effects are acute and insults to DA neuron are non-progressive. MPTP produces most but not all pathological hallmarks of PD.68 MPTP is a more favourable model than 6-OHDA because it is able to produce bilateral lesions, which is more relevant to PD since both hemispheres will have dopamine depletions consequently producing more PD-like symptoms.62 This bilateral lesion effect is achieved by systemic injections either i.p. or subcutaneously at higher doses in order to evoke desired PD symptoms while minimising toxic side effects of the intracranial injections. MPTP, when administered with probenecid, is more effective in producing similar pathological and clinical symptom in PD human as probenecid blocks MPTP clearance.62 This is, however, not to discount the significant roles of toxin-based animal models in the development of symptomatic therapies.69

We performed sensitivity analyses on different animal models used to assess if results changed. For rotarod analyses, one study used transgenic animal model, while the rest used toxin based. Of the eight studies that used toxin-based animal models, six studies used MPTP while two studies used 6-OHDA. We also tried excluding Chung et al31 due to cannabis introduction before and after toxin. We found that these sensitivity analyses did not significantly alter the overall results. This is similar to the pole test. Of the eight included studies, one study used transgenic mice, two studies used 6-OHDA toxin and the rest used MPTP toxin. The analyses did not significantly alter the results. For open field, of the six studies, two studies used transgenic animal models, two studies used MPTP and another two studies used 6-OHDA toxin. Due to the small number of studies included, we did not explore separating studies with these animal models. The aim of this review is to find if there is any effect of cannabinoids in PD animal models. Although there are significant differences between models, grouping all toxin-based animal models (with or without inclusion of the transgenic model) would make the most meaningful data for the review.

There are also differences with how human and animal models such as rodents metabolises environmental toxins, as well as differences in capacity in blood–brain barriers,62 hence interpretations from animal model experiments are to be taken with caution.

Most of the toxin-based studies in this review used male mice model. It is known that higher doses of MPTP can kill female mice.70 Traditional animal studies in PD use male mice. For studies that used both male and female mice, they commonly fail to provide comparison of outcomes between sex. This is a known limitation in translating positive animal studies to humans.71 Development of PD is twice higher for male than female. However, the latter has faster disease progression and higher mortality rate.71

Despite all the identified limitations, toxin-based animal models in PD are indispensable in understanding and finding therapeutic treatment. This review is not to investigate and present the histoclinical pathology of PD. This review aims to find any treatment effect of cannabis-based treatment in PD symptoms. Positive outcomes for both rotarod and pole tests did not significantly change even after analysing studies that only used MPTP-based animal models. A systematic review of cannabis derived phytocannabinoids (CDCs) in PD animal models demonstrated neuroprotective effects evidenced by increased dopamine and dopaminergic neurons levels.72 Cannabis was able to reduce losses of dopaminergic neurons73 74 and increase TH-positive neurons.74–76 The authors account these results from CDC’s ability to combat oxidative stress, reduce neuroinflammation and their antiapoptotic effects. These results may provide some support to why this review detected significant motor improvement of PD animal models.

We also noticed the variations on how the different outcomes were measured between studies. For rotarod, most of the studies evaluated rotarod using accelerating rotating speed between 4 and 40 rpm over 5 min20 22 30 41 50 53 while one study evaluated rotarod using accelerating speed up to 20 rpm over 20 min,31 4–20 rpm over 3 min51 and another study 10–60 rpm over 5 min,47 taking the average of three trials for most studies. For pole test, the height of the pole ranges from 50 cm to 60 cm high, with diameter ranging between 0.8 cm and 1 cm between studies. Most of the studies took the average time to turn head and descend using the average of three trials, while one study measured time to descend to the floor. It is difficult to determine how these measurement variances affected the overall outcomes. However, because both tests have the same intent to measure motor performances and data were presented with similar measurement units, we decided to proceed to meta-analyses.

This improvement of motor function, however, was not replicated in the open field test with an SMD of 0.36 (95% CI −0.58 to 1.29; p=0.453). This could be due to the high heterogeneity between studies which included genetic-based and toxin-based PD models and the use of a variety of cannabinoids (two CBR agonists, three MAGL/FAAH inhibitor or endocannabinoid modulator) or more significantly, had variations in how the open field tests were undertaken in each included study. The duration of testing ranged from 5 min,20 23 53 10 min,34 59 up to 15 min54 and the area of the test arena also differed between studies: 1 m × 1 m,59 45 cm × 45 cm20 23 and unspecified space measurements.34 53 54 There are significant variabilities in published open field protocols in the literature,65 66 and there is a lack of reproducibility in studies when the same open field is carried out in different laboratories63 64 thereby making open field results challenging to compare and interpret.66

In some studies, multiple different doses of cannabinoid were used. For this meta-analysis, to ensure independence of study results, it was predetermined that only the highest dose was included in meta-analyses. This was based on an assumption of a dose-response effect if there was a detectable effect. It is possible that such an assumption is in error, though results from Shi et al and Chung et al, in which multiple doses were used, suggest that the highest dose did produce the largest effect (although lesser doses may have also produced an equivalent effect).

Another limitation of this study is the potential effects of publication bias and unclear and high risk of biases. Most studies did not specify details about baseline symptoms, randomisation or blinding processes. These are known limitations and are inherently problematic when conducting systematic reviews of preclinical studies. Included studies in this review also have variability in the toxin dosing, delivery method as well as in the treatment schedule and when it was commenced after the toxin (Table S9) on https://doi.org/10.6084/m9.figshare.19695004.v3).18 This would have likely influenced the extent of the disease observed in the animals and the subsequent behaviours. All these mentioned factors may have influenced the results of our meta-analyses. However, one potential benefit of the heterogeneity in the studies is that it might capture the different facets and diverse range of PD symptoms throughout disease progression. PD animal models are diverse and no one standard is used due to the difficulty in replicating the full myriad of PD symptoms in animal models.77

The results of this review do not guarantee successful translation clinically. In fact, the translation of preclinical results clinically has been elusive. Systematic reviews of medicinal cannabis in patients with PD showed subjective alleviation of motor and non-motor symptoms, however, the evidence is weak.8 9 More robust and symptom-specific RCTs are more required to further elucidate any cannabis effect. This review suggests that a study of cannabis on motor functions may provide the best clinical benefit for patients.

Overall, this systematic review and meta-analysis provides evidence of the benefit of cannabinoid treatment in PD animal models, which warrants further investigations. This review supports clinical trial of cannabis or cannabis-based treatments in humans with PD.