FormalPara Key Points

Evidence from preclinical and clinical studies indicates that cannabidiol (CBD) has antiseizure properties.

A highly purified CBD formulation (GW CBD, registered as Epidiolex®, 100 mg/ml CBD solution in sesame oil) derived from Cannabis sativa has received FDA-approval for the treatment of seizures associated with Dravet syndrome and Lennox Gastaut syndrome in patients 2 years of age and older, based on results from adjunctive-therapy placebo-controlled trials.

Preliminary data from another recently completed placebo-controlled trial indicate that GW CBD is also effective in the adjunctive therapy of seizures associated with tuberous sclerosis complex.

Interactions with concomitant antiepileptic drugs, particularly an increase in plasma levels of norclobazam (N-desmethyl-clobazam), may contribute to clinical effects in patients treated with CBD.

1 Introduction

Use of Cannabis products to treat seizure disorders was first reported in stone-inscribed cuneiform Mesopotamian texts describing medical practices in Akkadian and Sumerian cultures in the second millennium BCE [1]. Cannabis preparations were part of the therapeutic armamentarium against epilepsy used by Islamic physicians in the 11th century CE [2]. Interestingly, however, there is no evidence of these products being recommended for seizure disorders in Europe before 1843, when William O’Shaughnessy, an Irish physician serving in the Bengal Army, reported in a British journal the successful control of recurrent convulsive seizures in a 40-day-old girl treated with an extract of Cannabis indica [3]. In the following years, Cannabis-based products became recognized as a valuable therapeutic option for bromide-resistant seizures in Britain [4, 5] and North America [6], but their medical use declined in the first decades of the 20th century after cultivation of the plant was made illegal in many countries.

Interest in Cannabis products as antiseizure medications was rekindled in the second half of the 20th century following characterization of the many active constituents present in the plant and the subsequent discovery that some of these compounds display anticonvulsant activity in experimental models of seizures [7,8,9,10,11]. Based on these data, cannabidiol (CBD) emerged as an attractive candidate because of its comparatively favorable therapeutic index and lack of the undesirable psychoactive properties associated with tetrahydrocannabidiol (THC). A few small clinical trials of CBD in patients with epilepsy were conducted in the 1980 s with conflicting results [1], but it was not until the last decade that the use of CBD-enriched Cannabis preparations for the treatment of seizures became a truly hot topic [12]. A highly influential event in this context was the CNN broadcasting in 2013 of the story of Charlotte Figi, a young girl with Dravet syndrome whose convulsive seizures decreased from about 50 per day to just two to three per month after receiving a Cannabis extract (Charlotte’s Web) with a high CBD/THC concentration ratio [13].

Over the last several years, GW Pharma Research (Cambridge, UK) has been actively engaged in developing a liquid pharmaceutical formulation of highly purified CBD (Epidiolex®, 100 mg/ml CBD solution in sesame oil) derived from Cannabis sativa as a potential antiseizure medication. This formulation, henceforth referred to as GW CBD, has undergone well-controlled randomized, placebo-controlled, adjunctive-therapy trials, which led to its approval by the US Food and Drug Administration (FDA) on 25 June 2018 for the treatment of seizures associated with Dravet syndrome (DS) and Lennox-Gastaut syndrome (LGS) in individuals 2 years of age and older [14]. More recently, GW Pharma announced the successful completion of another adjunctive-therapy placebo-controlled trial that reportedly demonstrated that GW CBD is also effective in the adjunctive treatment of seizures associated with tuberous sclerosis complex [15].

The present article provides an overview of the pharmacological properties of CBD in models of seizures and epilepsy, its pharmacokinetic and drug interaction profile, and its efficacy and safety in the treatment of seizure disorders. Sources of information for this review were identified by conducting a literature search on Pubmed using the search terms “cannabidiol and seizures and/or epilepsy” and “cannabidiol and mechanism of action and/or pharmacokinetics and/or interaction.” Lists of references of retrieved relevant articles and authors’ personal files were also consulted.

2 Pharmacological Profile

2.1 Pharmacological Activity in Experimental Models of Seizures and Epilepsy

CBD exerts anticonvulsant activity in a wide range of seizure models in rodents. These include audiogenic seizures in rats [8], seizures induced by maximal electroshock [8, 11, 16,17,18] and pentylenetetrazole [16, 18,19,20,21] in rats and mice, and seizures induced by cocaine, 3-mercaptopropionic acid, picrotoxin, isoniazid, and bicuculline (but not strychnine) in mice [16, 22]. CBD is also active in rodent models considered to be predictive of efficacy in focal epilepsy, such as the pilocarpine model of temporal lobe seizures [23, 24], the 6-Hz model of psychomotor seizures [18], focal seizures induced by application of penicillin [24] and cobalt [25], seizures induced by corneal kindling [17, 18], and kindled seizures in rats chronically implanted with cortical and limbic electrodes [26]. In the latter model, CBD increases the afterdischarge threshold, and reduces the afterdischarge amplitude and duration [26]. At doses up to 300 mg/kg intraperitonally (i.p.), however, CBD does not exert seizure-protective activity in the lamotrigine-resistant amygdala kindled rat model of pharmacoresistant epilepsy [17]. In most models, CBD protection against seizures occurs at doses lower than those causing neurotoxic symptoms [18]. No tolerance to the anticonvulsant effects of CBD has been reported after repeated administration in various seizure models [27].

CBD is also active in in vitro models of seizure activity. In particular, CBD has been found to decrease epileptiform local field potential burst amplitude and burst duration in Mg2+-free medium and in the 4-aminopyridine models of epileptiform activity in hippocampal brain slices [19]. The suggestion has been made that, in addition to antiseizure effects, CBD might also have neuroprotective actions [18, 23, 28]. In a rat model of status epilepticus induced by intrahippocampal injections of pilocarpine, CBD attenuated post-status hippocampal neuronal degeneration and other histological changes, but it is unclear whether these effects were secondary to decreased seizure severity or to an independent protective action of the compound [23]. In another study conducted in rats with spontaneously recurring seizures following low-intensity status epilepticus, CBD attenuated the time-dependent increase in seizure burden and motor co-morbidities, and protected against associated cognitive deficits [18]. Likewise, in a mouse model of DS, CBD treatment attenuated deficits in social behavior in addition to suppressing seizures [28]. In other experiments conducted in a mouse model of DS, CBD prolonged the survival of SCN1α heterozygotes and null animals, and improved weight gain and delayed the worsening of welfare scores and behavioral scores in SCN1α null animals [29]. In many of these studies, it is difficult to disentangle effects resulting from seizure suppression from potential independent effects of CBD on co-morbidities.

2.2 Mechanisms of Action

Several constituents of the Cannabis plant exert their actions, at least in part, by binding to cannabinoid (CB) receptors in the central nervous system (CNS) [30]. There are, however, major differences in the pharmacological actions of individual constituents of the plant. For example, studies in experimental models have shown that THC possesses prominent psychoactive effects, variable anti- or proconvulsant effects (with proconvulsant actions being especially seen after exposure to very high doses in rodents), and analgesic, cognitive, muscle relaxant, anti-inflammatory, appetite stimulant, and antiemetic actions [27, 31]. CBD, on the other hand, is mostly devoid of adverse psychoactive effects and displays anticonvulsant, analgesic, anti-anxiety, antiemetic, immune-modulating, anti-inflammatory, neuroprotectant, and anti-tumorigenic properties in experimental models [27, 31]. Although the mechanisms underlying these actions are complex, the finding that endogenous CB receptor ligands (endocannabinoids) play a role in the control in synaptic transmission and in the regulation of neuronal firing [27, 31,32,33,34,35,36] led initially to the hypothesis that antiseizure effects may result from an interaction with endocannabinoid signaling [37]. Consistent with this hypothesis, there is evidence that endocannabinoid systems are altered in several animal models of seizures and epilepsy [27, 31, 36, 38], and probably also in some forms of human epilepsy [39,40,41,42]. Many findings, however, argue against the hypothesis of the antiseizure effects of CBD being mediated by a direct action on CB receptors [31, 43]. In particular, it has been pointed out that CBD has a very low affinity for both cannabinoid type 1 (CB1) and type 2 (CB2) receptors, and it is unlikely to interact functionally with these receptors at concentrations achievable at therapeutic doses [31, 44]. Moreover, anticonvulsant effects in experimental models are generally mediated by activation of CB1 receptors [32], and CBD acts as an allosteric negative modulator, not as an agonist, at these receptors [45,46,47].

Although the precise mechanisms responsible for the antiseizure effects of CBD remain unclear, available data indicate that multiple actions are probably at play. These are probably unrelated to direct CB receptor binding [48,49,50], even though indirect effects mediated by inhibition of the breakdown of the endocannabinoid anandamide cannot be excluded [51]. Among the many actions described in experimental models and systems [31, 43, 48,49,50, 52,53,54,55,56,57,58,59,60,61] (Table 1), three mechanisms have emerged in particular for their potential role in mediating antiseizure effects. These include antagonism of G protein-coupled receptor 55 (GPR55) [28, 44, 62], desensitization of transient receptor potential of vanilloid type 1 (TRPV1) channels [63, 64], and inhibition of adenosine reuptake [44, 49, 65]. Specifically, CBD has been found to inhibit through GPR55 antagonism intracellular calcium release and neuronal hyperexcitability in epileptic tissue [28, 44]. Being a TRPV1 agonist, CBD also desensitizes TRPV1 channels leading to decreased extracellular calcium influx and consequently reduced neuronal hyperactivity. Consistent with these effects, the anticonvulsant effects of CBD are attenuated in GPR55 knockout animals, and in TRPV1 knockout animals [44]. CBD also blocks the equilibrative nucleoside transporter ENT1 and by this mechanism reduces adenosine uptake and increases extracellular adenosine concentration, thereby depressing neuronal excitability [44]. Other actions possibly contributing to antiseizure effects include blockade of voltage-gated sodium channels (see [66] and [67] for review), interactions with voltage-gated potassium channels, 5-HT1a receptors, and α3 and α1 glycine receptors, blockade of T-type calcium channels, modulation of voltage-dependent anion selective channel protein (VDAC1), and modulation of tumor necrosis factor alpha release [43, 51, 53, 59].

Table 1 A list of targets and actions reported for cannabidiol based on results of studies in different experimental models and systems.

3 Pharmacokinetic Properties

The pharmacokinetics of CBD show extensive variability in relation to route of administration (e.g., intravenous, oral, sublingual, oromucosal spray, inhalation, oral, transdermal), type of product administered, concomitant intake of food, drug-drug interactions, and other factors [29, 34, 68].

The present article focuses on oral administration, and particularly on data for GW CBD, the only formulation approved to date by the FDA for the treatment of seizure disorders.

3.1 Gastrointestinal Absorption

In studies using a variety of CBD products, CBD has been found to be absorbed relatively rapidly, with peak plasma concentrations occurring between 0.5–6 h after oral intake [68]. After single doses of GW CBD given in the fasting state to healthy adults, peak plasma CBD concentrations occur at 3–5 h (median 5 h) and increase less than proportionally with increasing dose within the 1500–6000-mg dose range [69]. During multiple administrations at doses of 750 mg twice daily (b.i.d.) and 1500 mg b.i.d, steady-state trough CBD concentrations were reached after about 2 days, and peak concentrations (geometric means) in the morning of day 7 were 330 ng/ml and 541 ng/ml, respectively. Interestingly, on the first day of treatment, CBD concentrations were considerably higher after the evening dose than after the morning dose, probably due to differences in the prandial state between morning (following overnight fasting) and evening administration.

No detailed studies are available on the absolute oral bioavailability of CBD. However, bioavailability is estimated to be low (around 6%) [29], due to limited solubility of CBD in the gastrointestinal fluids and a prominent first-pass effect [70]. In the phase I pharmacokinetic studies of GW CBD in healthy subjects, areas under the plasma concentration versus time curve (AUC) increased less than proportionally with increasing single doses, from 1618 ng × h/ml after a 1500-mg dose to 3900 ng × h/ml after a 6000-mg dose [69]. In the same study, disproportionality was not evident during multiple dosing, with the AUC at 750 mg b.i.d. being nearly double that recorded at 1500 mg b.i.d. in healthy adults. In studies conducted in chronically dosed patients, however, the increase in CBD exposure was less than dose-proportional over the range of 520 mg/kg/day [14].

Food intake can have a profound effect on the oral bioavailability of CBD. In healthy subjects, administration of GW CBD (1500 mg) together with a high-fat/high-calorie meal resulted in an approximately fivefold increase in peak plasma CBD concentration, and an approximately fourfold increase in AUC compared with the fasting state (Fig. 1) [69]. Variability in plasma CBD concentration is also lower after intake in the fed state compared with the fasted state in healthy volunteers. Based on these observations, it has been suggested that treatment outcomes may be improved by taking CBD consistently with food [69].

Fig. 1
figure 1

Reproduced from Taylor et al. (2018) [69], with permission

Geometric mean and individual (expressed as dose-normalized) peak plasma concentrations (Cmax, a) and area under the concentration versus time curve (AUC) 0–∞ values (b) of cannabidiol (CBD) following intake of a single 1500 mg dose of GW CBD in the fasting state and after a high-fat breakfast in healthy subjects. Geometric means are shown as red triangles.

The increase in the oral bioavalability of CBD when taken concurrently with food has been confirmed recently in a study conducted in patients with epilepsy [71]. Patients were given CBD in a soft gelatin capsule formulation containing a highly purified CBD extract from Cannabis sativa in a coconut oil vehicle (Vireo Health, Vancouver, Canada) under fasting (no breakfast) and fed (high-fat 840–860 calorie) conditions. In the eight patients who completed the study, mean peak CBD concentrations were 14 times higher, and AUC values four times higher, when CBD was taken in the fed state compared with the fasting state.

3.2 Distribution

The apparent volume of distribution (Vz/F) of CBD in healthy adults has been estimated to be about 21,000 L after a dose of 1500 mg given in the fasting state, and increases with increasing doses up to approximately 43,000 L after intake of a 6000-mg dose [69]. These values represent gross overestimates of the actual volume of distribution, because calculations were based on the assumption of complete oral bioavailability.

In in vitro studies, CBD and its metabolites have been found to be highly bound (> 94%) to plasma proteins [14].

3.3 Elimination

CBD is eliminated through a multiphasic process, with a terminal half-life of about 60 h having been estimated after withdrawal of multiple-dose treatment. Half-life values assessed during the early part of the elimination phase are much shorter, as reflected by an effective half-life in the range of 10–17 h [69]. The effective half-life provides a better estimate of fluctuations in plasma concentrations during the dosing interval, as well as of the time needed to reach steady state (about four effective half-lives).

CBD undergoes extensive metabolism in the liver and the gut by the cytochrome P450 enzymes (CYP) CYP2C19 and CYP3A4, and by the uridine 5′-diphospho-glucuronosyltransferase (UGT) enzymes UGT1A7, UGT1A9, and UGT2B7 [72,73,74]. CYP2C19 is the main enzyme responsible for the conversion of CBD to 7-hydroxy-CBD, which is further metabolized to 7-carboxy-CBD by CYP3A4 [75]. The primary metabolite 7-hydroxy-CB retains pharmacological activity in the maximal electroshock seizure threshold test in the mouse, whereas 7-carboxy-CBD is pharmacologically inactive in this model [76]. After multiple dosing in healthy adults (1500 mg b.i.d.), the AUC of 7-hydroxy-CBD is 62% lower than the AUC of the parent drug. Conversely, 7-carboxy-CBD accumulates in plasma at comparatively high concentrations, with an AUC more than 40-fold greater than that of CBD [69]. A third metabolite, 6-hydroxy-CBD, is found in plasma at concentations much lower than those of the 7-hydroxy-metabolite.

The renal clearance of CBD is low, but significant amounts of unchanged drug are excreted in the faeces [14].

3.4 Pharmacokinetics in Children

The pharmacokinetics of GW CBD have been evaluated in a safety study in children with DS aged 4–11 years, who were randomized to different doses (n = 7–10 per group) [77]. CBD was administered twice daily (relationship to meal times not reported) in addition to background antiepileptic drugs (AEDs), which included most commonly clobazam and valproate. Pharmacokinetic assessments were based on sparse concentration data obtained on day 22, at the end of the maintenance period. Plasma CBD concentrations increased in an approximately dose-proportional manner across the three investigated dose groups (5, 10, and 20 mg/kg/day). Variability in CBD exposure among individuals was considerable, with coefficient of variation in AUC being in the order of 20–121%. 7-carboxy-CBD was the most abundant metabolite in plasma, with concentrations 13- to 17-fold higher than those of CBD. AUC values for 6-hydroxy-CBD were < 10% those of CBD, and those of 7-hydroxy-CBD (not reported quantitatively) were also lower than those of CBD.

A separate parallel-group safety study evaluated the single- and multiple-dose pharmacokinetics of a pharmaceutical-grade synthetic formulation of CBD (INSYS Manufacturing LLC, Chandler, AZ, USA) in 61 patients aged 1–17 years (mean age 7.6 years) [78]. Patients were divided into three cohorts and assigned to receive one of three doses of CBD oral solution (10, 20, or 40 mg/kg/day, in two divided daily administrations) for 10 days (4-day titration and 6 days’ maintenance). Each cohort included similar numbers of infants, children, and adolescents. Apart from pre-set volumes of water/clear liquid consumed immediately after CBD dosing, intake of liquid or solid food was not permitted for 1 h (age < 2 years) or 2 h (age ≥ 2 years) before and after drug administration on sampling days. Steady-state CBD levels were achieved within 2–6 days of dosing, and showed considerable interindividual variability. AUC values at steady state increased proportionally with increasing doses. Half-life values (geometric means) across dose groups ranged from 19.5 to 29.6 h. At any given dose, plasma CBD concentrations were usually lower in infants than in children and adolescents.

Due to differences in dosing and sampling protocols and to potential confounding effects of concomitant medications and food intake, comparisons of pharmacokinetic data in children and adults based on available studies are difficult to interpret. Further studies are needed to assess the influence of age on CBD pharmacokinetics.

3.5 Influence of Hepatic Disease

The pharmacokinetics of a single dose of GW CBD were evaluated in patients with hepatic disease. AUC values of CBD and its metabolites in patients with mild hepatic impairment (Child-Pugh A) were only slightly increased compared with those observed in healthy control subjects [79]. In patients with moderate (Child-Pugh B) or severe (Child-Pugh C) hepatic impairment, CBD AUC values were approximately 2.5- to 5.2-fold higher than those found in controls [14, 79]. Patients with moderate to severe hepatic impairment also showed elevated plasma concentrations of the metabolites 6-hydroxy-CBD and 7-hydroxy-CBD, though to a lesser extent than the parent drug.

4 Drug–Drug Interactions

4.1 Effect of Other Drugs on the Pharmacokinetics of CBD

In studies conducted in healthy volunteers, concomitant administration of the enzyme inducer rifampicin (600 mg/day) reduced the peak plasma concentrations of CBD and the AUC of CBD by about 50% and 60%, respectively [80]. Conversely, co-administration of the CYP3A4 inhibitor ketoconazole (400 mg/day) was associated with a twofold increase in both peak plasma CBD concentrations and CBD AUC [80]. A smaller increase in peak CBD concentrations (about 40%) without significant change in AUC was reported during concomitant treatment with the CYP2C9 inhibitor fluconazole [81]. The CYP2C19 inhibitor omeprazole (40 mg/day) had no significant influence on CBD pharmacokinetics [80]. It should be noted that these studies were conducted with an oromucosal spray formulation of nabiximols (a mixture of CBD and THC in a 1:1 ratio), and the magnitude of interactions involving induction or inhibition of CBD metabolism might be amplified with the use of oral formulations, because oral bioavailability is highly vulnerable to changes in extent of first-pass metabolism.

Available information on the influence of concomitant AEDs on CBD pharmacokinetics is limited. In a study conducted in healthy volunteers, co-administration of clobazam (5 mg b.i.d) was associated with a 50% increase in the AUC of the active CBD metabolite 7-hydroxy-CBD, possibly mediated by inhibition of UGTs [75]. In the same study, clobazam also caused a modest (about 30%) increase in the AUCs of unchanged CBD and its inactive 7-carboxy metabolite. In a pediatric study of a synthetic CBD formulation, CBD exposure on day 10 after dosing at 40 mg/kg/day was 2.5-fold higher (geometric means ratio) in patients receiving concomitant clobazam therapy compared with patients not receiving clobazam, but at lower CBD doses (10 and 20 mg/kg/day) CBD exposure was similar in patients on and off clobazam and an influence of confounders could not be excluded [78]. In healthy volunteers, multiple doses of stiripentol (750 mg b.i.d.) were found not to affect the exposure to concomitantly administered CBD, but there were minor decreases in the exposure to 7-hydroxy-CBD (− 27%) and 7-carboxy-CBD (− 13%) [75]. In another study in healthy volunteers, valproate given at a dose of 500 mg b.i.d. had no significant effects on the AUC of CBD, 7-hydroxy-CBD, and 7-carboxy-CBD [75]. Enzyme-inducing AEDs, particularly carbamazepine and phenytoin, would be expected to reduce plasma CBD concentrations, but these potential interactions do not appear to have been formally investigated [14].

A summary of drug interactions that have been reported to affect the plasma concentration of CBD and its active 7-hydroxy-metabolite is provided in Table 2.

Table 2 Summary of reported drug–drug interactions resulting in increased or decreased plasma concentrations of cannabidiol (CBD) and its metabolites

4.2 Effect of CBD on the Pharmacokinetics of Other Drugs

In vitro, CBD has been found to inhibit at clinically relevant concentrations the activity of CYP2C8, CYP2C9, CYP2C19 [14, 74, 82,83,84], and, possibly, CYP2D6 [85]. Although CBD may inhibit CYP3A4 in vitro [86], a phase I study in healthy volunteers showed that CBD given at a dose of 750 b.i.d. had no effect on the plasma concentration profile of midazolam, a sensitive probe of CYP3A4 activity [87]. CBD may also induce or inhibit CYP1A2 and CYP2B6, and inhibit UGT1A9 and UGT2B7. Other UGT isoforms (UGT1A1, UGT1A3, UGT1A4, UGT1A6, and UGT2B17) are not inhibited by CBD in vitro [14]. At clinically relevant concentrations, the metabolite 7-carboxy-CBD can inhibit the drug transporter systems BCRP (breast cancer resistance protein) and BSEP (bile salt export pump) [14]. US prescribing information for GW CBD provides a list of potential interactions that could be predicted based on these in vitro effects [14].

The potential influence of CBD on the pharmacokinetics of concomitantly administered AEDs has been investigated in a number of studies. The best documented interaction reported to date consists of a clinically relevant increase in the concentration of norclobazam (N-desmethyl-clobazam), the active metabolite of clobazam [75, 78, 88,89,90,91,92]. The average increase in norclobazam levels has been estimated to be about fivefold in a study of 13 mostly pediatric patients receiving CBD doses of 20–25 mg/kg [92], and about 3.4-fold in a formal drug interaction study in 13 healthy adults exposed to CBD 750 mg b.i.d. [75]. Plasma levels of the parent drug clobazam increase to a much lesser extent (20–60% on average) [75, 92]. The interaction is probably mediated by inhibition of CY2C19, the primary enzyme responsible for norclobazam clearance. Clobazam-treated patients may experience benzodiazepine-like adverse effects, particularly somnolence, when CBD is added onto their treatment regimen, and may require a reduction in their clobazam dose [14, 92]. Interestingly, a recent study conducted in children with DS provided suggestive evidence that no increase in norclobazam levels occurs when CBD is administered to patients co-medicated with clobazam and stiripentol [77]. Specifically, the increase in norclobazam levels after adding CBD was almost threefold in 13 children treated with clobazam without stiripentol, and completely absent in four children who received clobazam and stiripentol in combination. Like CBD, stiripentol is a potent inhibitor of CYP2C19 [93], and presumably inhibition of the enzyme was already maximal in those patients. Further studies are required to confirm this finding.

In a study in healthy volunteers, administration of CBD at a dose of 750 mg b.i.d was found to increase the AUC of concomitantly administered stiripentol by 55% (90% confidence interval (CI) 42–69) [75]. As stiripentol is partly metabolized by CYP2C19 [94], the interaction could be related to inhibition of CYP2C19 by CBD [75].

In a recent report, addition of CBD at a dose of 10–40 mg/kg/day in five patients stabilized on brivaracetam was found to result in increased trough plasma brivaracetam concentrations by 95–280% [95]. This possible interaction was tentatively ascribed to inhibition of brivaracetam metabolism by CYP2C19 and possibly other enzymes.

The majority of studies conducted to date failed to show clinically significant effects of CBD on the plasma concentrations of other commonly co-administered AEDs [77, 90]. In a phase II study in ten adults with epilepsy, addition of CBD (10 mg/kg b.i.d.) was associated with a minor reduction (< 16%) of plasma valproic acid concentrations at steady-state [96], but another study in healthy volunteers failed to identify any effect of CBD (750 mg b.i.d.) on the AUC of concomitantly administered valproic acid [75]. The binding of valproic acid to plasma proteins in vitro is also unaffected by CBD [96]. One isolated report suggested that CBD may increase the plasma levels of topiramate, rufinamide, zonisamide, and eslicarbazepine, but these data were difficult to interpret because of the retrospective nature of the observations, small number of patients, changes in dose of co-medications, and uncertainties about standardization of sampling times [91].

The influence of CBD on the pharmacokinetics of drugs used for the treatment of non-epilepsy conditions has been little investigated. Lack of interaction with the CYP3A4 substrate midazolam suggests that CBD is unlikely to affect the pharmacokinetics of drugs that are primarily metabolized by CYP3A4 [87]. However, potential interactions related to the inhibitory effect of CBD on other drug-metabolizing enzymes should be considered [14]. In a recent report, administration of CBD to a 44-year-old man with drug-refractory epilepsy on stable warfarin therapy was found to cause a prominent increase in anticoagulant effect, which ultimately required a 30% reduction in warfarin dose [97]. The interaction was probably due to inhibition of CYP2C9, which plays a major role in S-warfarin metabolism. In a study in vitro, CBD has also been found to inhibit the glucuronidation of ethanol, presumably by its effects on UGT1A9 and UGT2B7 [98].

Table 3 summarizes drugs that have plasma concentrations that have been reported to be modified by concurrent administration of CBD.

Table 3 Effect of cannabidiol on the pharmacokinetics of other drugs

4.3 Pharmacodynamic Drug Interactions

There is evidence from preclinical and clinical studies that CBD may antagonize some of the actions of THC at CB1 receptor sites [99,100,101,102], although the interaction may differ depending on whether testing is done under acute or chronic conditions [103]. In animal experiments, CBD and THC may even display some superadditive effects after prolonged exposure [103].

In clinical studies conducted with GW CBD, dose-related elevations of liver enzymes during CBD administration were far more common among patients co-medicated with valproic acid, particularly when combined with clobazam, than in patients comedicated with other AEDs (see section 6 for details) [104, 105]. The mechanism underlying this apparent interaction is unclear. Plasma levels of valproic acid and 4-en-valproic acid, a metabolite potentially associated with valproic acid-induced liver toxicity, are not increased by CBD co-administration [96]. A pharmacodynamic interaction at the level of hepatocytes cannot be excluded, possibly related to CBD-related impairment of mitochondrial function [106, 107].

5 Clinical Efficacy in the Treatment of Seizure Disorders

5.1 Non-Randomized Studies

Although reports of antiseizure effects of Cannabis products go back to antiquity [1], it is only in the last 10 years that results of patient surveys and exploratory studies started to emerge [12, 108]. The vast majority of these studies included predominantly pediatric patients with multiple forms of drug-resistant epilepsies [108], with a few studies focusing specifically on DS [109], tuberous sclerosis complex (TSC) [110], Sturge Weber syndrome [111], febrile infection-related epilepsy [112], epilepsy of infancy with migrating focal seizures [113], CDKL5 deficiency disorder and Aicardi, Dup15q, and Doose syndrome [114, 115]. A 2019 systematic review of investigations that included a majority of pediatric patients identified a total of 19 evaluable non-randomized studies involving 1115 participants (range 5–214) [116]. Mean ages ranged across studies from 7 to 14 years, and treatment duration ranged from 10 days to 57 months. Estimates of seizure freedom varied across studies from 1 to 20% (pooled proportion 7%; 95% CI 4–11; duration 8 weeks to 33 months). The percentage of patients with at least 50% reduction in seizure frequency compared with pre-treatment ranged from 24 to 100% (duration 8 weeks to 50 months). These studies were found to have a high risk of bias, due to at least one of the following: lack of a control group, lack of blinding and subjective outcomes, self-selection or unclear selection of participants, inconsistency in interventions across participants, or lack of detail about the interventions [116]. Similar conclusions were reached in an earlier systematic review that analysed data from 30 non-randomized studies, including six open-label interventional studies, ten case studies, eight self-report surveys, five retrospective chart reviews, and one study of unclear design [108]. Because of these shortcomings, the results obtained in these studies do not allow meaningful conclusions to be drawn about the efficacy of the administered treatments. As an added source of interpretative difficulties, the products tested differed across studies and included CBD preparations of variable purity derived from plant extracts, synthetic CBD, combination products containing variable mixtures of CBD and THC, various forms of artisanal CBD including homemade Cannabis extracts, and Cannabis products obtained from illegal suppliers with unknown composition. Interestingly, a recent meta-analysis of observational studies found more reports of improvement (and less reports of adverse effects) among patients with pharmacoresistant epilepsy treated with CBD-rich extracts compared with patients treated with purified CBD, but the proportion of individuals with a 50% or greater reduction in seizure frequency did not differ between products [117, 118]. Patients treated with CBD-rich extracts reported a lower average dose (6.0 mg/kg/day) than those using purified CBD (25.3 mg/kg/day), an observation that the authors tentatively ascribed to possible synergistic effects of CBD with other phytocompounds (the so-called “entourage effect”). Because of the uncontrolled nature of these observations and the many possible biases, however, no meaningful conclusions can be drawn about the actual efficacy of the products compared. When discussing non-pharmaceutical grade products, it is important to emphasize that regulatory control of galenic CBD formulations, particularly those that can be purchased online or over the counter, is inadequate in most countries [12, 119]. As a result, the CBD content of many galenic products may differ markedly from that stated in the label, and some products may contain appreciable amounts of THC or harmful levels of pesticides and other contaminants [120,121,122,123,124,125,126]. Loss of stability during storage is also a concern with some products [127].

Among non-randomized studies conducted with a pharmaceutical-grade standardized formulation, the most informative are those that assessed outcomes in patients enrolled in physician-sponsored GW CBD expanded-access programmes [105, 128, 129]. The first of these studies reported on a cohort of 214 patients aged 1–30 years followed up for 12 weeks at 11 sites in the USA [105]. All patients had severe forms of childhood-onset epilepsy, and CBD was started at a dose of 2–5 mg/kg/day and titrated according to clinical response up to a maximum dose of 25 or 50 mg/kg/day depending on study site. An exploratory assessment of efficacy was conducted in a subgroup of 137 patients, which excluded those treated for less than 12 weeks (n = 52), those with no motor seizures (n = 21), and those below 1 year of age or affected by progressive metabolic diseases (n = 3). The median reduction in total seizure frequency in this subgroup was 35%, with the best outcome being recorded in patients with focal seizures (55% seizure reduction, n = 42) and atonic seizures (54% seizure reduction, n = 32). Nine patients (7%) were seizure-free during the last 4 weeks of follow-up. Interestingly, the proportion of individuals with a reduction in seizure frequency by 50% or more (compared with pretreatment) was almost twice as large in patients co-medicated with clobazam than in those not receiving clobazam (51% vs. 27%). Exploratory analysis in relation to type of epilepsy syndrome revealed a median reduction in seizure frequency of 43% in patients with DS (n = 32), and 36% in patients with LGS    (n=30). A follow-up study reported data for 607 patients (mean age 13 years, range 0.4–62 years) with drug-resistant epilepsies enrolled at 25 US sites [128]. CBD was added to pre-existing AEDs at a starting dose of 2–10 mg/kg/day and titrated to a maximum of 25–50 mg/kg/day. The median dose of CBD was 25 mg/kg/day and the median follow-up was 48 weeks (range 2–146 weeks), with 138 patients providing seizure outcome data for 96 weeks or longer. At the time of the analysis, 76% of patients remained on treatment. Compared with baseline, the frequency of convulsive seizures decreased by 51% at 12 weeks, and a similar reduction persisted through consecutive 12-week periods for up to 96 weeks. About 11% of patients were free from convulsive seizures during each 12-week period. Results from a cohort of 72 children and 60 adults evaluated at a single center were consistent with those reported for the larger cohort [129]. As discussed above, seizure outcome data from these studies are difficult to interpret, due to the lack of a control group and the potential influence of confounders.

A number of non-randomized studies have suggested that CBD may also have favorable effects on non-seizure outcomes such as mood, behavior, cognition, sleep quality, adverse AED effects, and/or overall quality of life [108, 116, 129,130,131]. It is unclear whether these reported benefits reflect an independent effect of CBD or the influence of confounders such as improved seizure control and changes in concomitant therapies. Evidence also exists, however, that in some patients CBD can be associated with adverse outcomes on measures of behavior [14] and sleep [116]. In the randomized studies discussed in the section below, no statistically significant effects of CBD have been found on Quality of Life in Childhood Epilepsy scale scores or Sleep Disruption rating scale scores [77, 104, 132].

5.2 Randomized Studies

The first randomized studies of CBD in drug-resistant epilepsy were conducted in the late 1970s by Brazilian investigators using a purified formulation developed by Raphael Mechoulam at the Hebrew University in Jerusalem [133, 134]. Further investigations followed, and by 1990 four small placebo-controlled trials had been completed using CBD doses in the range of 200–300 mg/day for treatment periods between 8 and 26 weeks [133,134,135,136]. Unfortunately, these trials had major methodological shortcomings, including inadequate sample size (the largest study included only 15 patients) and failure to provide critical details about the intervention and the characteristics of the patients [53]. Two systematic reviews published in 2014 evaluated these and other studies, and determined that the data did not allow any meaningful conclusions about the potential antiseizure effects of CBD [137, 138].

To date, evidence on the antiseizure efficacy of CBD rests primarily on five adequately powered, placebo-controlled, randomized, double-blind, adjunctive-therapy trials of GW CBD reported in the last 3 years [15, 88, 104, 132, 139]. All trials for which full details have been published had a similar design, with a 4-week prospective baseline, a 2-week titration period, a 12-week maintenance period and a tapering period of up to 10 days [88, 104, 132]. After completion of the double-blind period, patients were offered the opportunity to enter a long-term open-label follow-up study.

Two of the trials were conducted in a total of 396 patients with LGS and explored the efficacy and safety of GW CBD at doses of 10 mg/kg/day (one trial) [88] and 20 mg/kg/day (two trials) [88, 132], divided into two daily administrations. In both trials, main eligibility criteria were age between 2 and 55 years, a diagnosis of LGS with an electroencephalogram showing slow (< 3 Hz) spike and wave complexes, a history of at least two generalized seizure types including drop seizures for at least 6 months, and being on treatment with one to four concomitant AEDs. Additionally, to access double-blind treatment, patients had to have a seizure frequency of at least two drop seizures per week during the prospective 4-week baseline. The main features of the patients enrolled and key trial results are summarized in Table 4. In both trials, the median percent seizure reduction from baseline in drop seizure frequency (primary endpoint) and the percentage of patients with at least 50% seizure reduction from baseline (key secondary endpoint) were significantly greater in the CBD groups than in the corresponding placebo groups. Most other secondary endpoints, including patient and caregiver global impression of change, also favored CBD over placebo. In the trial that explored two doses, median reduction in drop seizure frequency was 41.9% in the 20 mg/kg/day group, 37.2% in the 10 mg/kg/day group, and 17.2% in the placebo group (p = 0.005 for the 20 mg/kg/day group vs. placebo, and p = 0.02 for the 10 mg/kg/day group vs. placebo). Between 3.5 and 7% of patients randomized to CBD were free from drop seizures during the 12-week maintenance phase, whereas overall < 1% of patients randomized to placebo were seizure-free during the maintenance period (Table 4). Of the 368 patients who completed treatment, 366 accessed the open-label extension study. An interim analysis of open-label data after a median follow-up of 38 weeks at a modal CBD dose of 23 mg/kg/day was reported recently [140]. At the time of follow-up, 82% of patients remained on treatment. Median reduction from baseline in monthly drop seizure frequency ranged from 48% to 60% across 12-week periods through week 48. Although a number of confounders can influence outcome data in the extended follow-up studies, there was no indication of loss of response over time.

Table 4 Summary of study design, patient characteristics and main outcome data in the two randomized placebo-controlled adjunctive therapy trials of cannabidiol (CBD) in patients with Lennox Gastaut syndrome

Of the two randomized double-blind trials completed in patients with DS (Table 5), one has been published in full [104], while the other has been reported in abstract and poster form only [139]. Eligibility criteria for the fully reported trial were an established diagnosis of DS, receiving one or more AEDs, and having had at least four convulsive (tonic, clonic, tonic-clonic, or atonic) seizures during the 28-day baseline period. A total of 120 patients, aged 2.3–18.4 years, were allocated to receive CBD (titrated up to the target dose of 20 mg/kg/day, given in two divided administrations) or placebo at 23 centers in the USA and Europe. Fifty-two of 61 patients in the CBD group and 56 of 59 patients in the placebo group completed the study. The median percent change in convulsive seizure frequency from baseline (primary endpoint) was − 38.9% in the CBD group compared with − 13.3% in the placebo group (p = 0.001 for the adjusted median difference between groups). As for secondary endpoints, responder rates (proportion of patients with a reduction in seizure frequency by 50% or more) and median percent reduction in total seizure frequency also favored CBD over placebo, whereas changes in nonconvulsive seizure frequency did not differ between groups (Table 5). Scores in the Caregiver Global Impression of Change scale showed a significant improvement in the CBD group compared with placebo (p = 0.02). No significant differences between groups, however, were found for changes in Sleep Disruption score, Epworth Sleepiness Scale score, Quality of Life in Childhood Epilepsy score, Vineland-II score, and inpatient hospitalizations due to epilepsy. The second double-blind DS trial, reported in abstract form, was conducted in a total of 199 patients randomized to receive CBD 10 mg/kg/day, 20 mg/kg/day, or placebo. Both CBD  doses tested were found to be significantly superior to placebo for all three endpoints reported (percent reduction in convulsive seizure frequency, responder rate, and percent reduction in total seizures) (Table 5). Of the 278 patients who completed the two randomized DS trials, 264 enrolled in an open-label extension study. An analysis of the latter was conducted after a median treatment duration of 274 days (range 1–512 days) [141]. Treatment was discontinued in 75 patients (28.4%) overall, and in 17 (6.4%) due to adverse effects. Seizure frequency in patients remaining on treatment showed a consistent reduction compared with baseline, with no indication of a loss of response over time.

Table 5 Summary of study design, patient characteristics and main outcome data in the two randomized placebo-controlled adjunctive therapy trials of cannabidiol (CBD) in patients with Dravet syndrome

A fifth multicenter, randomized, double-blind, adjunctive-therapy placebo-controlled trial of GW CBD has been completed in patients with seizures associated with TSC, and results were announced on 6 May 2019 in a press release by GW Pharmaceuticals and its US subsidiary Greenwich Biosciences Inc. [15]. In this trial, 224 TSC patients aged 1–57 years (mean age, 14 years) were randomized to receive CBD 25 mg/kg/day (n = 75), 50 mg/kg/day (n = 73), or placebo (n = 76) added on to an average of three pre-existing AEDs. Most common concomitant AEDs were valproic acid (45% of patients), vigabatrin (33%), levetiracetam (29%), and clobazam (27%). The primary endpoint was the change in seizure frequency of countable seizures (focal motor aware or unaware seizures, focal to bilateral convulsive seizures, and generalized tonic–clonic, tonic, clonic, or atonic seizures) over the 16-week treatment period (4-week titration followed by 12-week maintenance at the target dose) compared to baseline for the 25 mg/kg/day CBD arm versus placebo. Median baseline seizure frequency was 57 per month. Seizure reductions from baseline were 48.6% for the 25 mg/kg/day arm, 47.5% for the 50 mg/kg/day arm, and 26.5% for the placebo arm (25 mg/kg/day vs. placebo, p = 0.0009 and 50 mg/kg/day vs. placebo, p = 0.0018).

Overall, the results of the pivotal trials summarized above provide Class 1 evidence that adjunctive therapy with GW CBD is effective in reducing the frequency of the most disabling seizure types associated with DS, LGS, and, based on a preliminary report released to date, also seizures associated with TSC. At least for the trials in DS and LGS, one concern in interpreting the findings relates to the observation that about one-half of the enrolled patients were on concomitant clobazam therapy. This raises the question as to whether, and to what extent, the improvement in seizure control associated with CBD treatment was mediated by the increase in serum levels of norclobazam (and a more modest increase in the levels of clobazam itself) rather than to an independent effect of CBD [12]. To address this question, Thiele et al. [142] analyzed data from the two LGS trials and compared seizure outcomes on and off clobazam co-medication. Among patients co-medicated with clobazam, responder rates (proportion of patients achieving at least 50% reduction in drop seizure frequency) were 46% in the CBD 10 mg/kg/day group compared with 19% in the placebo group. Among patients not receiving clobazam, responder rates were 33% in the CBD 10 mg/kg/day group compared with 8% in the placebo group. When patients assigned to CBD 20 mg/kg/day were assessed (a more meaningful comparison, because sample size could be enlarged by pooling data from two trials), responder rates in the CBD groups were 57% (vs. 24% on placebo) among patients on clobazam, and 35% (vs. 13% on placebo) among patients off clobazam. While no statistical analysis was reported and the placebo-corrected effect-size at the 20 mg/kg/day dose was numerically inferior in patients not receiving clobazam, these data are consistent with CBD having an antiseizure effect independent from its interaction with clobazam.

A comparison of seizure outcomes on and off clobazam has not been reported for patients included in the DS trials. The impact of the interaction between CBD and clobazam might be attenuated in these patients, many of whom were also receiving stiripentol. As discussed in sect. 4.2, there is suggestive evidence that no major changes in norclobazam levels occur when CBD is added to the combination of clobazam and stiripentol [77]. On the other hand, addition of CBD has been reported to increase plasma stiripentol levels by an average of 55% [75], an interaction that can also confound interpretation of seizure outcomes in these patients.

6 Adverse Effects

The first well-controlled study on the safety of CBD in patients with epilepsy was a dose-ranging double-blind, parallel-group study conducted in 2014–2015 at 12 US and UK sites [77]. A total of 34 children (aged 4–10 years) with DS were randomized to receive GW CBD 5, 10, or 20 mg/kg/day or placebo for 3 weeks, including 3- to 11-day titration. CBD was relatively well tolerated in all groups. A clear relation to dose was identified for decreased appetite, which was reported in five children (44%) in the 20 mg/kg/day group versus one of 25 children (4%) in the other groups combined, and for elevation in liver enzymes (ALT or AST) above three times the upper limit of normal (> 3 × ULN), which occurred in six children overall, four of whom were in the highest dose group. The study provided the rationale for testing doses of 10 and 20 mg/kg/day in the pivotal double-blind efficacy/safety trials in DS and LGS.

A summary of the main adverse events reported in the four fully published double-blind trials of GW CBD [77, 88, 104, 132] in patients with epilepsy is provided in Table 6. The most common events that differentiated CBD-treated patients from placebo-treated patients were somnolence/sedation, decreased appetite, increases in transaminases, diarrhea, behavioral changes (including irritability, agitation, aggression), skin rashes, fatigue, and sleep disturbances. Infections (particularly peneumonia and viral infections) tended to be more common in the CBD groups [14]. A clear relationship with dose was especially evident for decreased appetite, diarrhea and gastroenteritis, increases in transaminases, and skin rashes (Table 6). In a meta-analysis that included data from the same trials, adverse events significantly associated with CBD were somnolence, decreased appetite, diarrhea, and increased transaminases [143]. Interestingly, vomiting was reported in 6–16% of CBD-treated patients and in a similar percentage of placebo-treated patients [88, 104, 132], suggesting that this event is either unrelated to treatment or is related to the vehicle (sesame oil) rather than CBD itself. In these trials, adverse events led to discontinuation of treatment in 8–14% of patients in the CBD groups, compared with less than 2% of patients allocated to placebo. Increased transaminases was the most common adverse event leading to discontinuation [88, 132]. Adverse event reports from long-term studies have generally been consistent with those from short-term randomized trials [128, 129, 140, 141, 144].

Table 6 Adverse events reported in the four fully published double-blind placebo-controlled trials of cannabidiol in Lennox-Gastaut syndrome and Dravet syndrome [77, 88, 104, 132] and occurring in > 3% of patients in one of the cannabidiol groups.

In the placebo-controlled trial conducted in TSC patients, which evaluated higher doses (25 and 50 mg/kg/day) than those used in the other double-blind trials, adverse events occurring in ≥ 10% of CBD-treated patients with a frequency greater than on placebo included somnolence, decreased appetite, diarrhea, constipation, vomiting, transaminase elevations, pyrexia, seizures, cough, and infections. Diarrhea, vomiting, transaminase elevations, somnolence, and rash occurred at higher rates in the 50 mg/kg/day group than in the 25 mg/kg/day group. Patients on concomitant valproic acid receiving the higher CBD dose had a higher frequency of transaminase elevations. There were no cases of Hy’s law observed, and no deaths were recorded.

In the fully published double-blind trials, the frequency of somnolence and sedation, including lethargy, was 34% at 20 mg/kg/day compared with 27% at 10 mg/kg/day and 11% on placebo [14]. Somnolence and sedation were twice as common in patients co-medicated with clobazam than in those not co-medicated with clobazam (46% vs. 16%) [14], probably due to the associated increase in serum norclobazam levels [77, 88, 132]. In fact, during the double-blind phase of the trials some patients did require a reduction in clobazam dose: in one of the LGS trials, clobazam dose was reduced in 11 (27%) of 41 patients in the CBD group, versus four (9%) of 43 patients in the placebo group [132]. Somnolence was also one of the most frequently reported adverse effects in open-label trials, particularly in patients treated with clobazam [105, 128, 129], who may require a reduction in clobazam dose. In fact, in a recent long-term study, the majority of clobazam-treated patients had their clobazam dose reduced during CBD therapy [128].

Another adverse effect that appears to be partly mediated by a drug interaction is increased serum transaminases [77, 88, 132, 140]. In the fully published double-blind trials, ALT elevations > 3 × ULN occurred in 30% of patients co-medicated with both valproic acid and clobazam, in 21% of those co-medicated with valproic acid (without clobazam), in 4% of those co-medicated with clobazam (without valproic acid), and in 3% of patients co-medicated with neither drug [14]. Elevations in transaminases are strongly dose-related, with ALT elevations > 3 × ULN being reported overall in 17% of patients treated with CBD 20 mg/kg/day compared with 1% of those taking 10 mg/kg/day. Patients with elevated transaminases at pretreatment are also at greater risk of developing ALT increases > 3 × ULN [14]. In most cases, elevations in transaminases occur during the first 2 months of CBD therapy. In about two-thirds of cases, elevated transaminases resolve with discontinuation of CBD or a reduction in dose of CBD and/or concomitant valproic acid therapy. In about one-third of cases, elevations resolved during continuation of CBD treatment without dose reduction [14]. In one of the double-blind LGS trials, two individuals on CBD discontinued treatment reportedly because of hepatic failure. However, on review neither of these cases meet diagnostic criteria for hepatic failure or Hy’s law criteria for severe liver injury, because the events were without elevations in bilirubin [132]. Both patients recovered fully.

Decreased appetite is a dose-related adverse effect that seems to be related to CBD itself, rather than being mediated by a drug interaction. Together with diarrhea and other gastrointestimal adverse effects, it is likely to contribute to weight loss, which was reported as an adverse event in 5% of patients receiving CBD 20 mg/kg/day compared with 1% of patients on placebo (Table 6). The actual proportion of patients experiencing a decrease in body weight, however, is greater than estimated by adverse event reports. In the three fully published controlled trials in patients with DS and LGS, a decrease in weight ≥ 5% of the initial body weight occurred in 18% of patients receiving CBD 20 mg/kg/day, 9% of those receiving 10 mg/kg/day, and 8% of those receiving placebo [14]. The proportion of patients with weight loss could be greater during long-term treatment. In a group of 26 children followed up for a mean of 21 months, weight loss, which occurred in 30% of the cases, became evident only later during treatment [144].

In a drug interaction study in healthy volunteers, nine out of 45 subjects (20%) assigned to receive GW CBD 750 mg b.i.d without titration or with fast (3-day) titration developed an erythematous, maculopapular, or follicular rash, which was severe in two cases [75]. The study protocol was then amended to require a 10-day titration, and no further cases of rash were recorded in the 32 subjects treated according to the revised treatment schedule, which also included avoidance of dosing units prefilled up to 1 week before intake. In the four fully published randomized trials, all of which involved a 10-day titration, the frequency of skin rashes was 13% on CBD 20 mg/kg/day, 7% on CBD 10 mg/kg/day, and 3% on placebo (Table 6).

Consistently with results obtained in animal models, CBD has not been found to induce symptoms suggestive of abuse potential or physical dependence in human studies [14, 145]. When tested in a highly sensitive population of polydrug users, a 750-mg dose of CBD was found to have a low abuse potential [146]. High (1500 mg) and supratherapeutic (4500 mg) doses of CBD were associated with subjective effects suggestive of a potential for abuse, but these were significantly lower than those associated with alprazolam (2 mg) or dronabinol (10 mg and 30 mg) [146]. In the same study, CBD was found to have minimal impact on an alertness scale, and no observable effects on cognitive/psychomotor tests.

7 Dosing Recommendations and Precautions for Use

Because at present GW CBD (Epidiolex) is the only CBD-based medication approved by the FDA for the treatment of epilepsy, the recommendations below are largely based on the US prescribing information for this product. A licensing application for the same product is under evaluation by the European Medicines Agency (EMA).

CBD is currently indicated for the treatment of seizures associated with DS and LGS in patients 2 years of age and older. The suggested starting dose is 2.5 mg/kg twice daily (5 mg/kg/day), which can be increased after 1 week to the maintenance dosage of 5 mg/kg twice daily (10 mg/kg/day). Based on clinical response, further weekly dose increments of 2.5 mg/kg twice daily (5 mg/kg/day) can be made up to the maximum recommended dose of 10 mg/kg twice daily (20 mg/kg/day). If more rapid titration from 10 mg/kg/day to 20 mg/kg/day is considered indicated, the dose may be increased no more frequently than every other day. Lower doses and, possibly, slower titration rates are recommended for patients with moderate to severe hepatic impairment. Although prescribing information does not specify whether CBD should be administered with food or in the fasting state, the prominent effect of food, particularly a high-fat meal, on CBD bioavailability [69] suggests that standardization of intake in relation to meal times would be desirable.

Because in the absence of alternative explanations transaminase elevations > 3 × ULN in the presence of increased bilirubin levels are considered to be predictive of severe liver injury, it is recommended that serum transaminases and bilirubin be measured at pretreatment and after 1, 3, and 6 months of treatment, and periodically thereafter or as clinically indicated [14]. Monitoring of transaminases and bilirubin is also recommended 1 month after increasing the CBD dose, or after adding other medications known to adversely affect liver function. More frequent monitoring may be appropriate in patients on valproic acid, and those with elevated transaminases at pretreatment. Discontinuation of CBD treatment is recommended in patients who develop signs and symptoms suggestive of hepatic dysfunction, in those with transaminase elevations > 3 × ULN and bilirubin > 2 × ULN, and in those with sustained transaminase elevations > 5 × ULN. Consideration should also be given to feasibility of reducing the dose of concomitant drugs adversely affecting liver function, particularly valproic acid.

Following initiation of CBD treatment or an increase in CBD dose, patients should be closely monitored for the occurrence or worsening of somnolence, particularly when clobazam is a concomitant medication. A reduction in the dose of concomitantly administered clobazam may be considered.

8 Current Place in the Treatment of Epilepsy and Future Perspectives

Several CBD products, including many artesanal formulations, are currently marketed in many countries or accessible through the Internet. Because the quality of many such products can be substandard [12, 120], treatment recommendations can only be made for the only pharmaceutical-grade product for which extensive efficacy and safety data are available [77, 88, 104, 132]. Despite reports suggesting broad-spectrum efficacy against a wide variety of seizure types and syndromes [12, 110,111,112,113, 115, 116, 128, 129], to date adequate evidence of clinical benefit is limited to “convulsive” (tonic-clonic, tonic, clonic, and atonic) seizures in patients with DS, drop seizures in patients with LGS, and, based on a preliminary report, seizures associated with TSC. At present, utilization of CBD in the management of other seizure types, or any seizure type in patients with epilepsies other than DS and LGS, should be regarded as investigational.

In all controlled trials conducted to date, CBD has been administered as adjunctive therapy in patients with persisting seizures despite ongoing treatment with other AEDs. Although FDA-approved indications do not make any reference to adjunctive use, it would seem reasonable for the time being to restrict the prescription of CBD to patients who are receiving concomitant AEDs. Based on available evidence, CBD can be combined with any of the AEDs commonly used in the management of DS and LGS. However, as discussed earlier in this article, caution is required when CBD is added to valproic acid therapy because of the higher risk of transaminase elevations in these patients. Clobazam is also frequently used in the management of DS and LGS, and patients taking this drug are at higher risk of developing sedation and somnolence, which could be managed by reducing the dosage of clobazam. There is suggestive evidence that the increase in serum norclobazam levels after adding CBD does not occur in patients who also receive stiripentol as co-medication. This finding could have clinically important implications, and requires confirmation in an adequately sized cohort. Further studies are also required to clarify the contribution of increased serum norclobazam levels to the improvement in seizure control associated with intake of CBD in clinical trials [12].

Because CBD has been introduced only recently in the armamentarium against DS and LGS, its place in the treatment algorithm remains to be defined. Many prevalent patients with DS and LGS are likely to have been exposed unsuccessfully to several AEDs, and a trial of CBD in these individuals is generally justified. The question as to how early CBD should be prescribed in the course of the disease, however, becomes increasingly relevant as incident (recently diagnosed) cases accrue, and newer medications are introduced for the same indications [29]. Further studies, preferably including head-to-head comparisons with alternative treatment options, are required to determine the optimal use of CBD in currently approved indications. Its potential when used as monotherapy also remains to be assessed.

Apart from the need to evaluate whether the therapeutic value of CBD extends to other types of seizures and epilepsies beyond those assessed in randomized trials completed to date, well-designed studies are required to investigate potential predictors of therapeutic response, risk factors for adverse effects, drug-drug interactions, and long-term safety overall. Aspects that require particular attention include potential consequences of early life exposure on neurodevelopment [147] and any potential risks associated with use during pregnancy, including risks for fetal development [14].