Kava for the treatment of generalised anxiety disorder (K-GAD): study protocol for a randomised controlled trial

A Cochrane review and meta-analysis of seven randomised clinical trials (RCTs) using kava mono-preparations (60–280 mg kavalactones) for the treatment of generalised anxiety symptoms found a significant reduction of anxiety on the Hamilton Anxiety Rating Scale (HAM-A) for kava compared with placebo (P = 0.01) [34]. Similarly, a more recent pooled analysis of six studies using kava versus placebo in the treatment of anxiety symptoms found a significant effect in favour of kava on the HAM-A, with an effect size (Cohen’s d) of 1.1 [35]. Our previous three-week double-blind, placebo-controlled, crossover RCT (n = 60), using a water-soluble rootstock extract of a noble kava cultivar in chronic generalised anxiety [36], involved a preparation standardised to a dose of 250 mg kavalactones. Following a placebo run-in phase there was a reduction of −9.9 points on the HAM-A in the first kava phase versus only a −0.8 point in the first placebo phase, and a reduction in the second crossover phase from kava of −10.3 points compared to a rise of +3.3 points for placebo (Cohen’s d = 2.24, P < 0.0001).

In a subsequent six-week parallel double-blind RCT [37] involving 75 participants with diagnosed GAD (58 randomised to 120 mg daily kavalactones titrated to 240 mg for non-response), a group × time interaction was found (P = 0.046) for a significant reduction in HAM-A scores in favour of kava over placebo. Further, kava significantly reduced participant anxiety by −4.2, representing a moderate effect size (Cohen’s d = 0.63). For participants with moderate-to-severe level anxiety (as assessed on the MINI Plus diagnostic interview), the treatment effect was more pronounced (P = 0.020), with a larger effect size (d = 0.80). The effects were still significant after controlling for baseline Montgomery–Åsberg Depression Rating Scale (MADRS) depression (P = 0.01), baseline Beck Anxiety Inventory (BAI) anxiety (P = 0.05), thyroid function (P = 0.02), and weekly caffeine use (P = 0.03). Further sub-analysis of participants with pure GAD and no other DSM-IV diagnosed co-morbid anxiety disorder (panic disorder, social phobia, post-traumatic stress disorder, obsessive-compulsive disorder) revealed a significant group × time interaction (P = 0.020; d = 1.28), with a reduction of −8.5 points for kava on the HAM-A compared to −2.3 points for placebo. As part of this trial we also examined five GABA transporter polymorphisms as potential pharmacogenetic markers of kava response. We observed a significant monotonic trend for two polymorphisms in which the number of rs2601126 T-alleles (P = 0.021) or of rs2697153 A-alleles (P = 0.046) was associated with significant reductions in HAM-A scores within the kava group.

The design of the study is a phase III, multi-site, two-arm, 18-week, randomised, double-blind, placebo-controlled trial using a standardised pharmaceutical-grade, water-soluble extract of kava (240 mg of kavalactones per day) or matching placebo (inert plant-based fibre) in 210 currently anxious participants with GAD. The trial sites are at the Centre for Human Psychopharmacology (Swinburne University of Technology), Hawthorn, Melbourne, Australia and the Academic Discipline of Psychiatry (The University of Queensland) based at the Royal Brisbane and Women’s Hospital, Herston, Brisbane, Australia. Recruitment is scheduled to commence September 2015 and through to June 2017 (Fig. 2).

Participants will be required to attend seven visits at the study sites at week 0 (baseline), 2, 4, 8, 12, 16, and 18. At the baseline visit, participants will complete informed consent forms, screening assessments, and mood and anxiety questionnaires (self-reporting and interview-based). Brain imaging component visits (Melbourne site only) are to be scheduled within one week of baseline and week 8 visits.

The study employs an additional two-week single-blind placebo-controlled post-study observation period (that is, for a total trial length of 18 weeks - refer Fig. 3).

All eligible participants will be randomly allocated to a treatment arm, and corresponding treatment will be provided following clearance of blood tests taken at the baseline visit. All subsequent visits will follow the same outline of the baseline session excluding consent forms and screening assessments and including a safety assessment. In addition, participants will be required to provide a blood sample within several days of the baseline, and at week 2, 8, and 16 visits. For female participants not taking hormone-modulating contraception who are otherwise eligible to partake in the brain imaging components, the baseline visit is to be scheduled on days 1–5 of the menstrual cycle.

This study has ethical clearance (Alfred Hospital Research Ethics Committee 137/14, University of Queensland Research Ethics Committee 2014000876, Swinburne University Human Research Ethics Committee 2014/204), and is registered on ClinicalTrials.gov (protocol number: NCT02219880).

Participants will be recruited through various advertisement sources:

The primary outcome is symptoms of GAD measured by the Structured Interview Guide for the Hamilton Anxiety Rating Scale (SIGH-A), at baseline and subsequent visits up to 18 weeks following randomization. The SIGH-A is a widely used measure of anxious symptomatology in GAD research. It includes measures of persistent worry, anxiety, mood, and somatic symptoms. The SIGH-A has been used in previous studies of pharmacological and psychological treatments for people with GAD. It is reliable (intraclass correlation coefficients for inter-rater reliability of 0.98 and test-retest reliability of 0.88), has high convergent validity with structured clinical ratings of symptoms of GAD (for example, MINI PLUS 6.0 [46]), and is sensitive to change [47, 48].

Inclusion criteria*

Exclusion criteria

Participants are assigned study identification numbers sequentially, and treatments are randomised via identification numbers using permutated 3×2 block randomisation, for example, ABAAABABBBA, based on these numbers. Randomisation is performed by staff unrelated to the study. Treatment is allocated sequentially to enrolled participants by trial clinicians who are blinded to treatment coding (with no identifying details on the bottle of which group [A or B] each participant is assigned to, thus providing double-blinding). Participants will take two kava tablets twice per day (total of 240 mg of kavalactones per day) or matching placebo for 16 weeks.

Analysis of data will be conducted with blinding to group allocations. The primary efficacy analysis will assess average treatment group differences for the primary outcome measure (SIGH-A) over the entire study period using a likelihood based mixed-effects model, repeated measures approach (MMRM). Results from the analysis of dichotomous data (for example, demographics and genetic data) will be presented as proportions (for example, relative risks), with 95 % confidence interval, and Fisher’s exact test P value where appropriate. Non-parametric statistics will be used when assumptions for parametric methods are violated. Cohen’s d effect sizes will be calculated. All tests of treatment effects will be conducted using a two-sided alpha level of 0.05 and 95 % confidence intervals. Data will be analysed using Statistical Package for Social Sciences software (SPSS) [55].

The study will recruit a sample size of 210 participants (105 participants per arm). The study is powered to detect a small-to-moderate difference between active treatment and placebo on efficacy outcomes. Data will be analysed using data with at least one post-baseline, and all data including baseline measurement, in an intention-to-treat analysis. As such, a conservative effect size F of 0.15 on the primary outcome measure (SIGH-A) for a two-tailed analysis (with alpha = 0.05), and the study powered at 80 % (Z beta = 0.80), with a correlation among repeated measures (analysis of variance (ANOVA) model) over six time points, will require 206 participants (critical F of 3.88). A sample size of 206 (rounded up to 210) will be sufficiently powered to provide a statistical difference between active treatment and placebo groups (using the intention-to-treat analysis) with the data.

The sample size of the imaging component of this study is 40 per treatment arm (n = 80). Following 12-week supplement administration, a recent MRS study with a sample size of 24 (12 in each arm) detected significant changes in metabolites by 39 % [60]. Our sample size of 40 participants in each arm will provide higher statistical power compared to previous studies, whilst still maintaining sufficient power with a 10 % participant attrition rate, as well as attrition due to unreliable metabolite estimates in one or both time points.

While kava is a Therapeutic Goods Association (TGA) listed, over-the-counter nutraceutical product in Australia and has not been shown to cause any confirmed serious adverse reactions in this country, concerns over hepatotoxicity have led to withdrawal or restriction in some other countries. As discussed in the literature [61‐63], many reported cases have involved concomitant ingestion of other compounds with potential hepatotoxicity (such as other medications and/or alcohol). At the clinical level, a variety of case study data from patients with kava hepatotoxicity have been gathered, with probable causation reported in a few cases [63]. Previously reported hepatotoxicity issues identified with German kava products may also be due in part to the extraction (ethanol or acetone extraction) method; the use of non-water soluble chemical (ethanol or acetone) extraction techniques (the traditional solvent is water), and use of aerial parts of the plant and root and stem peelings, and poorly prepared, potentially contaminated raw material [63].

In response to safety concerns, the World Health Organization commissioned a report assessing the risk of kava products [64]. Recommendation 2.1.3 from the report suggested that products from water-based suspension preparations be preferentially used over acetonic and ethanolic extracts. The current study addresses these safety concerns by using standardised aqueous formulations of kava from the peeled rootstock of a noble cultivar (such cultivars are higher in kawain and dihydrokavain and lower in dihydromethysticin). Note that in June 2014, the German Federal Institute for Drugs and Medical Devices (Bundesinstitut für Arzneimittel und Medizinprodukte) overturned the previous 2002 ban on the use of kava products. As a European Union regulatory and guideline body, such a change has positive ramifications for the potential reinstatement of kava products based on ongoing empirical evidence from clinical trial research.

Study staff involved in all aspects of participant recruitment, assessment, and monitoring hold postgraduate psychology qualifications, undergo training including inter-rater reliability of interviewer scale measures, and are supervised by psychiatrists and medical staff. This study has comprehensive medical supervision. Regardless, the occurrence of an adverse event (AE) may come to the attention of study personnel during study visits and interviews of a study subject presenting for medical care, or upon review of subject data by a study monitor. Any medical condition that is present at the time that the subject is screened for inclusion will not be reported as an AE. However, if it deteriorates at any time during the study, it will be recorded as an AE. All AEs will be graded for severity and relationship to study product. Participants will also be required to complete the Systematic Assessment for Treatment of Emergent Effects (SAFTEE) [65] at each assessment session to assess for AEs. All AEs will be recorded in the adverse event log in the participant case report form (CRF), including the seriousness, severity, and relationship to study product, duration, and outcome. In all cases, researchers will maintain contact with participants who experience an AE until it has been resolved and symptoms disappear. They will also be asked to notify their general practitioner. If a worsening of a participant’s mental state occurs, the study psychiatrist and/or clinical psychologist will be consulted and appropriate steps taken.

Liver function will be assessed pre-treatment, and participants with any abnormal results will be excluded as a safety precaution. In people with healthy liver function, it is not expected that any elevation of serum liver enzyme levels will occur (based on our previous studies using similar standardised pharmaceutical-grade extract of kava). Nevertheless, liver function will be closely monitored via blood samples at regular intervals throughout the study (weeks 2, 8, and 16).

Participants will be required to take the tablets as instructed during the 18 weeks of the study. At each assessment session they will be asked to return all unused tablets, which will be counted for compliance rate (tablet counts will be employed by an independent researcher).

Source documents will be kept in the participant files, which will be stored in a locked cabinet in a locked room accessible only to the investigators. Data from study measures will be entered into a password-protected electronic database on a secure network drive and backed up onto a USB/external hard drive stored alongside the participant files in a locked cabinet. All electronic data will be verifiable against source documents. As per Good Clinical Practice guidelines, upon request of regulatory authorities, the principle investigator will make all requested trial-related records, including source documents, available for direct access. The study files and all source data will be retained for 15 years from the date of publication of study results, in accordance with university policy. Study protocol and processes have been developed in line with Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) guidelines [66].