This was a single-center, randomized (2:1), placebo-controlled, double-blind trial. Patients were recruited nationally by the principal investigator (VH). During enrollment written informed consent was provided, and an application for a cannabis treatment license was arranged (issuance took an average of seven weeks). Over 16 weeks of the treatment period, participants were followed every two weeks, with the option to terminate their participation. The trial took place in a tertiary medical center in northern Israel from December 2017 to September 2019.
Study procedures were conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonisation Consolidated Guidelines on Good Clinical Practice and followed the CONSORT reporting guideline.(47).
At screening, participants were evaluated for eligibility criteria, which included an age of 60 years or older, major neurocognitive disorder according to the DSM-5 criteria (all types of dementia), Mini–Mental State Examination (MMSE)(48) score of <26 for cognitive impairment measurement, clinically relevant neuropsychiatric behaviors defined as Neuropsychiatric Inventory–Nursing Home Version (NPI-NH)(49-51) sub-score of agitation ≥3, a stable medication regimen for at least two weeks prior to baseline visit, and residing in either an institutionalized setting or in a non-institutionalized setting subject to 24-hour supervision. Exclusion criteria included severe heart disease, epilepsy, anxiety disorder; psychotic condition in the present or in the past (not related to dementia); family history of schizophrenia; current substance use disorder, recent cannabis experience, or scheduled surgery during the trial.
Eligible participants were randomly assigned by a computerized random-number generator system in a 2:1 ratio to receive either Avidekel oil or a placebo. The 2:1 ratio was intended to encourage interest in caregivers to participate in the trial. The randomized list of patients was set before the trial was initiated; based on it, the investigational product and placebo were prepared. Patients, families, and the medical team were masked to the individual patients’ treatment assignment. To ensure masking was maintained, Avidekel and placebo oils were manufactured to have identical appearances and smell.
Investigational product (IP)
The investigational product or the placebo were added to the routine medication regiment (Table 1). Subjects received the IP or the placebo as drops applied under the tongue three times a day. Participants in the investigational group received Avidekel (made in Israel by Tikun-Olam Cannbit Pharmaceuticals), an ethanol extraction of high-CBD (~15%), low-THC (~0.5%) cannabis chemovar dissolved in olive oil. The IP contained 30% CBD, 1% THC, 1% Cannabichromene (CBC), 0.5% Cannabigerol (CBG), and 0.5% Cannabidivarin (CBDV). One drop ~0.04 ml contained 11.8 mg CBD and 0.5 mg THC. Patients in the control group received a placebo containing olive oil and chlorophyll. We selected this specific chemovar aiming mainly to minimize side-effects. This was based on earlier clinical experience with 39 patients with dementia (unpublished data), and with pediatric autism spectrum disorder patients suffering from behavioral disorders.(52) Patients receiving this product showed improvement in agitation.
The initial dose was one oil drop at morning, noon, and evening, for two days; then two drops, three times a day, for two more days, and so forth. The dose was titrated gradually depending on the tolerance of each patient, to a maximum dose of 21 drops per administration or until an adverse reaction occurred; the patients were then tapered down one level to a pre-adverse reaction dose. The time for each patient to “find” the therapeutic dose, a balance between maximum reduction in agitation and minimum side-effects, lasted up to six weeks. After the titration phase, patients entered a ten-week treatment phase of fixed-dose (Table 2).
For safety evaluation, serious adverse events (SAEs; defined as: death, life-threatening events, hospitalization, debilitation, or immobility), and all adverse events (AEs), with a severity score on a Likert scale of 1 to 10, were collected in all trial visits. In this population, with many comorbidities and medications, the symptom list of main AEs was also evaluated at baseline and documented as a non-IP-related AE report. An AE was defined as any unfavorable symptom, sign, syndrome, or disease that occurred during the study, having been absent at baseline, or if present at baseline, appeared to worsen. Clinical data included vital signs and physical examination in all trial visits, as well as blood chemistry and hematology every other visit.
The primary efficacy endpoint was the proportion of subjects achieving a 4-point decrease at week 16 compared to baseline in the Cohen-Mansfield Agitation Inventory (CMAI)(53-57), with higher scores indicating greater severity. We evaluated a decrease in CMAI score to demonstrate a better outcome compared to a similar randomized controlled trial using oral THC (in which a 2.3 points reduction in the active group was deemed not significant),(58) and to set the bar above the placebo effect (two points decrease in the placebo group).(59)
Secondary outcomes included: mean change in the CMAI, the time necessary to achieve a 4-point reduction in CMAI, mean change in NPI-NH agitation/aggression sub-score. In the NPI-NH, the higher the score, the more severe and frequent the behavioral disturbances.
At each visit, a geriatrician and a trained occupational therapist examined the patients. All study questionnaires were administered and completed by the trained staff and answered by the patient’s main caregiver on every visit. The following questionnaires were also administered: the Geriatric Depression Scale (GDS), the Pain Assessment in Advanced Dementia Scale (PAINAD), the Clinical Global Impression for Agitation and Aggression (CGI-S-A/A), and the MMSE.
Sample size was calculated for a power of 80% and for two-sided α level of 0.05 to detect difference in proportion of successful reduction in CMAI score between investigational group versus control, at week 16. Success was defined as at least a 4-point reduction. For an expected difference of 35% in the proportion of success between the groups, an unbalanced sample of 42 and 22 was selected for the investigational and placebo group, respectively. Thus, 64 patients were randomly assigned to the investigational or control group (4 patients withdraw immediately after randomization, therefore 60 patients who started treatment were analyzed).
The efficacy analyses were performed according to the intention-to-treat (ITT) principle, to provide unbiased comparisons among groups. The ITT analysis was done in all patients randomized and receiving treatment, with multiple imputation of missing data of those patients who did not complete the trial (to get more accurate results than provided by a single imputation method, the number of imputations was 5). We further performed a per protocol (PP) analysis (for 52 patients) as a sensitivity analysis, where only patients who completed the trial according to protocol were counted towards results.
Baseline characteristics between groups are presented as means and standard deviations for continuous variables and as frequencies and percentages for categorical variables. Chi-square tests and independent t-tests were performed to compare groups for categorical and continuous baseline variables, respectively.
The primary outcome was analyzed with the chi-square test including Yates' corrected chi-square (continuity correction). The baseline CMAI distribution was tested for normality using the Kolmogorov-Smirnoff test, and the Mauchly's Test of Sphericity tested whether the variances of the differences were equal.
The GLM (general linear models) Repeated Measures procedure was used to provide an analysis of variance for repeated CMAI measurements for nine visits on each subject. The analyses involve one within factor (time) and one between factor (groups). Changes over time and differences within groups were calculated (time*group), including contrasts tests to test differences among factor levels (1 factor, 9 levels), with a total significant level of 5%. The contrast was compared by method: difference; each level was compared to baseline. Analyses were performed on two sets of full data (without missing data), the ITT set (n=60) and the PP set (n=52).
Kaplan Meier survival analysis was performed to compute the time to achieve a CMAI ≥4-point reduction (success) for each group and tested the group difference using the log rank chi-square test. Comparison of CMAI mean score between the two groups was analyzed by the independent t-test.
Comparison between groups in NPI-NH frequencies of all sub-categories (as dichotomous variables: yes/no) were analyzed by the Fisher’s exact test for baseline and end of study. NPI-NH factors scores, total NPI-NH, and all other variables were tested by independent t-test. Frequency of AEs and medications consumption between the two groups was compared by using the Fisher's exact test.
Data were analyzed with IBM SPSS statistics software version 27.0 (SPSS Inc. Headquarters, Illinois, USA). Significance levels were set at 0.05.