Inclusion criteria
Sixteen adults were recruited from students, staff, and their families at Western Sydney University through advertisements on campus. All provided informed consent and were advised that they were free to withdraw at any time without penalty. Participants were excluded if they were pregnant, obese (to reduce the chance of recruiting participants with associated chronic diseases), had a history of allergies, serious gastrointestinal disorders, asthma or serious pulmonary disorders, diabetes, or any significant abnormality on laboratory tests including full blood count, liver, and renal function. Any participants taking anti-coagulant or cognitive enhancing medication/substances, including over the counter Ginkgo biloba or Panax ginseng capsules, were also excluded. All participants were non-smokers and showed no signs of neurological impairment (mini mental state exam; MMSE > 28).
Treatments
SLT treatment: SLT formula is standardised by the bioactive components of the three herbal extracts from Panax ginseng, Ginkgo biloba, and Crocus sativa. A range of in vivo pharmacological studies have been undertaken to test the bioactivity and determine the optimal ratio of the three individual herbal extracts. Each SLT capsule contains a 60 mg standardised mixture of 27.27 mg ginsenosides from Panax ginseng, 27.27 mg total ginkgo flavone-glycosides from Ginkgo biloba, and 5.46 mg crocins from Crocus sativa. The SLT preparations were manufactured in a Good Manufacturing Practice certified facility.
Placebo: Two capsules per day, containing an inert substance matched for the colour, taste, and smell of SLT.
Design
This pilot study employed a randomised, placebo controlled, double-blind crossover design. Participants were required to orally ingest two capsules of either SLT or a placebo every day for one week. This was followed by a one week washout period and the subsequent counterbalanced treatment week. This washout period length was determined by preclinical pharmacokinetic studies [33]. The active herbal formulation and placebo were prepared in an extract capsule form in accordance with Australian Goods Manufacturing Practices. Neurocognitive and cardiovascular function were tested before and after each of the interventions. Primary outcome measures included neurocognitive function test scores, and ERP component amplitudes. Secondary outcome measures were EEG spectral band amplitudes, aortic and peripheral pulse pressure, and resting HR. The procedure was approved by Western Sydney University's Human Research Ethics Committee (Ethics Approval: H8253), and the trial was registered with the Australian New Zealand Clinical Trials Registry (Trial Id: ACTRN12610000947000).
Materials and apparatus
Cognitive test battery
Cognitive function was tested using the Computerised Mental Performance Assessment System (Compass) neuropsychological test battery, which measures a range of cognitive abilities including attention, episodic memory, and working memory. These cognitive domains were assessed using 12 Compass tasks: Immediate Word Recall, Delayed Word Recall, Word Recognition, Simple Reaction Time, Choice Reaction Time, Numeric Working Memory, Alphabetic Working Memory, Corsi Block Span, N-Back Working Memory, Picture Recognition, Face Recognition, Serial Subtraction, and Rapid Visual Information Processing (RVIP). Parallel versions of each task were used for the repeated testing sessions.
Mental fatigue and Mood were self-reported on a Visual Analogue Scale (VAS) using a 100 mm horizontal line displayed on a computer monitor, with each endpoint marking the two extremes of mental fatigue or mood; participants responded using a computer mouse cursor. Three dimensions of mood were assessed (alertness, calmness, and contentment) with 16 separate Bond-Lader VAS items [39].
Electroencephalograph (EEG)
EEG data were recorded continuously from 18 scalp sites (F1, F3, Fz, F2, F4, C1, C3, Cz, C2, C4, P1, P3, Pz, P2, P4, O1, Oz, O2) with a 64 channel electrode cap using sintered Ag/AgCl electrodes. The nose was used as a reference and the cap was grounded by an electrode located midway between AF3 and AF4. Data were acquired 0–200 Hz using a Neuroscan Synamps 2 digital signal-processing system and Neuroscan 4.3.1 Acquire software. The display and stimulus markers were controlled by a separate stimulus computer (Dell Optiplex 760 with a 22 " LG Flatron W2253TQ screen) using Compumedics Stim2 (4.0.09302005) software.
Electro-oculogram (EOG) was recorded using sintered Ag/AgCl electrodes placed 2 cm above and below the left eye for vertical movements, and on the outer canthus of each eye for horizontal movements. Impedance was less than 10 kΩ for cap, EOG, and reference electrodes. Scalp and EOG potentials were amplified with a gain of 2816 and digitised at 1000 Hz.
EEG was recorded during two resting conditions (5 min eyes open and 5 min eyes closed) followed by two oddball tasks (visual then auditory). For the visual oddball task, stimuli were white 37 × 43 mm letters (target: X; nontarget: O) presented for 500 ms on a black background. For the auditory task, acoustic stimuli were delivered binaurally through foam stereo eartips (ER3-15A) using the Stim Audio System (P/N 1105), and consisted of 1000 Hz (target) and 500 Hz (nontarget) 80 dB tones (500 ms duration, 10 ms rise/fall time). For each of these tasks, a randomised stimulus sequence consisting of 200 trials was presented (target p = .2, nontarget p = .8) with an ISI of 1.5 s in a single block that was approximately 5 min in duration. Targets were responded to with a button press using the Stim System Switch Response Pad (P/N 1141).
Cardiovascular measures
Aortic and peripheral measures of pulse pressure, and resting HR were assessed using an electronic sphygmomanometer and tonometer (AtCor Medical SphygmoCor CVMS v8.2); the mean across three consecutive measurements of each index was computed. To ensure reliability, only measurements equal to or greater than an Operator Index of 80 were used. The Operator Index is a measure of quality control derived from the mean and variability in pulse strength, diastolic variation, and systolic shape deviation. The score is out of a possible 100, and an Index over 80 is considered acceptable.
Procedure
Participant eligibility was confirmed during a brief telephone interview. Then, during a screening session, participants provided informed consent and completed the MMSE before a medically trained professional conducted a medical questionnaire, physical examination, and provided participants with a pathology request form so that blood samples could be taken and analysed at a Douglas Hanley Moir pathology laboratory. Blood sampling was only deemed necessary for participants aged 50–75; pathology examined were full blood count, and liver and kidney function. If no abnormalities were discovered participants completed a training session on the Compass test battery, and were randomly allocated to a treatment group via a computer randomisation package. The randomisation was conducted by the research program coordinator externally to the research team and the assignments were blinded to all investigators and assessors as well as to the trial participants. All data were collect at the National Institute of Complementary Medicine Clinical Laboratory, Western Sydney University during 2012–2013.
At the beginning of each of the one-week intervention cycles, participants attended a baseline session where they were asked a series of questions about any illnesses or adverse events, and whether they had started, stopped, or changed any medications. Then, whilst sitting comfortably, participants had their cardiovascular measurements taken, completed the Compass testing battery, were fitted with EEG recording apparatus, and had their EEG activity recorded. Each of these baseline sessions ran for approximately 2.5 h. Efforts were made to ensure that each timepoint for each participant was at the same time of day. Participants were dispensed eighteen capsules containing either the SLT formulation or placebo (depending on randomisation) and were instructed to take two capsules per day for one week (morning and night with food), and to return any unused capsules at the post-treatment session. Four extra capsules were provided in the event that participants were unable to attend the next scheduled appointment. Halfway through the treatment cycle, a brief telephone follow-up was conducted to ensure participant safety in regards to adverse reactions and illness, and to confirm the time and date of the next testing session. The post-treatment sessions were identical to the baseline, with the exception that participants returned any excess capsules and also completed a short survey to determine the successfulness of the blinding procedure.
After the initial intervention cycle, there was a seven day washout period to allow for the elimination of any residual herbal compound from the body. Following this, participants began a second week of treatment in which they received the counterbalanced treatment condition (SLT or placebo), and again were tested pre- and post-intervention cycle.
EEG data extraction and quantification
Using Neuroscan Edit software (Compumedics, Version 4.3.1), the EEG data were EOG artefact [40] and DC-offset corrected, and low-pass filtered at 30 Hz (24 dB/Octave). Any additional artefacts exceeding ± 100 μV were excluded. Oddball task data were epoched from 200 ms pre- to 1400 ms post-stimulus and baseline corrected using the pre-stimulus period. Trials containing omission (misses) and commission (false alarms) errors were excluded. For each subject and each modality, averages were computed for each of the four treatment sessions and the two stimulus conditions. Averaged half-sampled data (epoched −100 to 600 ms: 350 datapoints) from 18 scalp locations were submitted to two separate temporal Principal Components Analyses (PCAs; one for each modality) using the ERP PCA toolkit (v. 2.23 [32]) in MATLAB (The Mathsworks; Version 8.0.0.783, R2012b). In each of the PCAs, factors for all conditions were quantified simultaneously (2304 observations: 16 participants × 2 stimulus conditions × 4 testing sessions × 18 electrodes). The PCAs used the unstandardised covariance matrix with Kaiser normalisation, and all 350 unrestricted factors underwent Varimax rotation [41]. PCA factors were identified as ERP components based on their latency, topography, and polarity of their conspicuous maximum loading. The peak amplitude of each identified component was output and entered into subsequent statistical analyses.
The 5 min of artefact-corrected data from each of the resting conditions were epoched to 2000 ms and then quantified in MATLAB and EEGLAB (Version 9.0.8.6b [42]). For each subject and each condition, spectral amplitudes were computed using discrete Fast Fourier Transforms with a 10 % Hanning window for each of the four treatment sessions. Frequency data from 0 to 30 Hz were extracted with a resolution of 0.5 Hz, before averaging across epochs. The spectral band amplitudes were calculated by summing the activity across the frequency bins for delta 0.5-3.5 Hz, theta 4.0-7.5 Hz, alpha 8.0-13.0 Hz, and beta 13.5-29.5 Hz.
Statistical analyses
As this was a pilot trial, no formal sample size calculation was made. Sample size was approximated based on a trial [65] which employed two of the three SLT constituents as a combined intervention using a similar cross-over design. A post-hoc sample size calculation showed that in order to detect a difference of d = 1.00 at 80 % power, two-tailed, α = .05, 17 participants were required, suggesting that the sample size used here, although small, was sufficient to detect a meaningful change in the primary outcome measure.
Cognitive and cardiovascular measures
Relevant outcome measures were recorded automatically by the Compass and SphygmoCor software. Separate repeated measures analyses of variance (ANOVAs) with Treatment (SLT vs. placebo) and Time (baseline vs. post-treatment) as within-subject factors was used to compare the effects of SLT with placebo for each of the Compass and cardiovascular measures. As this was a pilot study, results approaching significance (p < .10) are reported as they may potentially reflect the existence of real effects.
EEG and ERPs
Expected EEG band and ERP component topographies were defined by selecting a cluster of electrodes surrounding the site of maximal amplitude. Using a mean across a region defined by multiple sites, rather than a single electrode, reduces the impact of chance variance at a single site [43].
Separate repeated-measures MANOVAs then assessed each band and component’s amplitude for the effects of Condition (EEG: eyes open vs. eyes closed; ERPs: target vs. nontarget), Treatment (SLT vs. placebo) and Time (baseline vs. post-treatment). Cohen’s d effect sizes are reported. The violations of sphericity assumptions associated with repeated-measures analyses do not affect single degree of freedom contrasts, so Greenhouse-Geisser-type correction was not necessary [44]. All F-tests reported have (1, 15) degrees of freedom.
One-way tests were utilised for all analysed predictions. It should also be noted that, as this paper details results for a number of dependent measures, the frequency of Type I errors increases. However, this increase in frequency of Type I errors cannot be controlled by adjusting α-levels, because the probability of Type I error remains the same [45].