To the best of our knowledge, these results are the first on the effect of demographic and clinical variables on CBD plasma concentrations in “real” children and adults with LGS and DS. In our cohort, plasma concentrations of CBD were linearly related to matched daily dose, expressed as mg/kg/day, over a range of 5–23 mg/kg/day. This result is in keeping with findings obtained in pediatric patients by Devinsky et al. (2018a), over a 5–20 mg/kg/day dose range, and Wheless et al. (2019), over 5–40 mg/kg/day doses, and in healthy adult volunteers after multiple doses (750–1,500 mg/day) (Taylor et al., 2018). Plasma concentration-dose linearity is an important drug characteristic in clinical practice, as it may help physicians in patients’ dose adjustments. However, it was characterized by a large, up to 10-fold, intersubject variation in drug trough plasma concentrations at a given dosage. High intersubject variability in CBD bioavailability has been reported in patients (Devinsky et al., 2018a; Wheless et al., 2019) and in healthy subjects (Taylor et al., 2018; Crockett et al., 2020), partly ascribed to CBD incomplete oral absorption and large pre-systemic elimination (Perucca and Bialer, 2020). Moreover, food intake, especially high fat/high caloric meals, has a marked effect on CBD exposure, increasing drug bioavailability up to 4-5-fold (Taylor et al., 2018; Crockett et al., 2020). All our patients took their morning CBD dose in a fed state, but breakfast type was not standardized.

Notably, up to 8-fold fluctuations in intrasubject CBD plasma concentrations were observed in most patients’ samples between the morning trough and 2.5 h post-dosing, in line with previous evidence (Devinsky et al., 2018a). Reported times to peak of plasma CBD oral formulations are highly variable, mostly in the range of 1–4 h (Millar et al., 2018). We established the time of post-dosing blood sampling based on the clinical trial of Devinsky et al. (2018a), using the same CBD oral solution.

A novel finding was the significant effect of age on median trough CBD C/D ratio, which was higher in subjects aged 18 and over than in those under 18. Cannabidiol undergoes both an extensive first-pass effect and metabolism in the liver (Franco and Perucca, 2019; Perucca and Bialer, 2020), and age-mediated reduction in both these processes may partly explain this observation (Van den Elsen et al., 2014). The only data reported so far on the potential influence of age on CBD pharmacokinetics were confined within a cohort of pediatric patients (Wheless et al., 2019). At any given dosage, plasma CBD concentrations were lower in infants (aged 1 to <2 years) compared to children (2 to <12 years) and adolescents (12 ≤17 years).

Sex did not affect median trough C/D ratio of CBD in our patients. This observation should be considered cautiously as potentially influenced by the older age of the women group. Of note, no study has explored so far the potential differences between males and females in cannabinoids pharmacokinetics (Millar et al., 2018), which might contribute to observed sex-dependent differences in some of their effects (Fattore and Fratta, 2010). From a theoretical point of view, it has been hypothesized that the larger percentage of body fat in women might result in an increased volume of distribution of lipophilic compounds such as CBD, with a higher proportion of drug concentration sequestered in fat tissue and reduced drug plasma concentrations (Fattore and Fratta, 2010). This should also be linked to the peculiar half-life of CBD being initially shorter and then longer according to the possibility of compartmentalization of the drug in some not defined deep compartments (e.g., adipose tissue) (Lattanzi et al., 2020a).

No significant differences emerged in CBD C/D ratio from patients’ samples grouped based on metabolism inducing or inhibiting properties of concomitant ASMs. The interpretation of these results is limited by the small sample size per ASM cotherapy groups coupled with high within-group intersubject variability in CBD C/Ds, especially in patients taking strong enzyme inducers. Moreover, the influence of confounders, such as older age of the subgroup on inducers, cannot be ruled out. CBD is metabolized by the cytochrome P450 isoenzyme CYP2C19 to the active metabolite 7-hydroxy-CBD and further to inactive metabolites through CYP3A4 and uridine 5′-diphospho-glucuronosyltransferases (UGTs) (Landmark and Brandl, 2020). Enzyme-inducing ASMs, especially CBZ and PHT, would be expected to reduce CBD C/D ratio (Franco and Perucca, 2019), but no formal study has explored so far this potential interaction. Data on the effect of concomitant ASMs on CBD pharmacokinetics are scanty. From a phase I, open-label healthy volunteer trial (Morrison et al., 2019), concomitant intake of metabolism inhibitors such as STP (750 mg b.i.d., for 14 days) and VPA (500 mg b.i.d., for 14 days) had no significant effect on CBD bioavailability; 7-hydroxy-CBD exposure was decreased by 29% by STP, but the underlying mechanism is unknown. Clobazam (5 mg b.i.d., for 21 days) did not affect CBD exposure, while 7-hydroxy-CBD increased 1.5-fold, possibly by inhibition of UGTs (Morrison et al., 2019).

Dose-dependency for both efficacy and tolerability was not evidenced by our data. Attempts to find out a relationship between CBD plasma concentrations and both seizure control and AEs yielded no significant results. Plasma CBD values associated with therapeutic efficacy or AEs were overlapping. These findings might partly reflect high intersubject variability in CBD bioavailability, patients’ different clinical characteristics, and the heterogeneous contribution of different types and doses of concomitant ASMs.

The observed 54% responder rate was in line with the 38–52% previously reported in open-label studies (Devinsky et al., 2016; Thiele et al., 2019), a EAPs (Szaflarski et al., 2018; Laux et al., 2019) and randomized controlled trials (Devinsky et al., 2018b) involving patients with pharmacoresistant epilepsies treated with the same oral solution of purified CBD. Of note, distribution of CLB cotherapy did not differ between responders and nonresponders, in line with the evidence coming from randomized controlled studies that CBD has antiseizure activities irrespective of CLB coadministration (Bialer and Perucca, 2020; Lattanzi et al., 2020b).

The AE rate of 43% was lower than 79–94% reported in the abovementioned studies; the most common registered AEs, appetite loss, diarrhea, and somnolence were in line with the literature.

All these observations should be taken with caution due to the limited number of patients, the uncontrolled design of EAP protocol, and possible intersite variability in reporting methods, among others. Further studies in larger cohorts of patients are needed to confirm these findings.