This study is a randomized, double-blind, placebo-controlled clinical trial, registered with the Chinese Clinical Trials Registry (ChiCTR2400084102). The trial was conducted from February 2019 to December 2020 at the PLA General Hospital, Beijing, China, and was in accordance with the Declaration of Helsinki standards and guidelines.

The study product is a health supplement developed by the Nutrilite Health Institute of Amway, containing a combination of extracts from Cistanche tubulosa (Schenk) Wight ex Hook.f. [Orobanchaceae] and Ginkgo biloba L. [Ginkgoaceae]. Each CG tablet contained 150 mg of Cistanche extract (standardized to ≥28.0% echinacoside, yielding ≥42 mg) and 60 mg of Ginkgo biloba extract (standardized to ≥25.0% total flavonol glycosides and ≥6.0% terpene lactones, yielding ≥15 mg and ≥3.6 mg, respectively). Functionally relevant metabolites comprised 2.85 g total flavonol glycosides and 7.55 g echinacoside per 100 g of tablets (Patent ID US9737582B2, US20150320818A1). More information and determination methods were detailed in Supplementary Table S1.

The inclusion criteria were as follows: (1) healthy subjects aged between 30 and 65 years, good overall health with no significant disorders of the brain, heart, liver, kidneys, lungs, or hematological system; (2) no history of long-term medication use; (3) no prior participation in similar tests (e.g., memory quotient or IQ tests); and (4) no use of drugs or health supplements related to memory improvement within the past year. All participants provided written informed consent. The exclusion criteria were: (1) pregnant or lactating women; (2) individuals allergic to health supplement ingredients; (3) patients with severe systemic diseases; (4) recent use of substances that could influence the study’s outcomes; (5) non-compliance with study protocol; or (6) incomplete baseline data that could impact the efficacy or safety evaluation. Subjects were balanced based on education level, age, and gender, and were subsequently randomized into either the CG intervention group or the placebo control group.

In this double-blind trial, participants were randomized to receive either the CG tablets or sensorially indistinguishable placebo tablets. All participants were instructed to take two tablets daily (one tablet in the morning and one in the evening, approximately 12 h apart) for 30 days, adhering to the manufacturer’s recommended dosage. Placebo tablets contained glucose, microcrystalline cellulose, corn starch, caramel pigment, and inert excipients matched to CG tablets in appearance, taste, and texture. Participants consumed their assigned intervention for 30 days under strict double-blind conditions.

Memory function was assessed at baseline and on Day 30 using an online clinical memory scale developed by the Psychology Institute of the Chinese Academy of Sciences. This scale aligns with the 2003 “Technical Standards for Testing and Assessment of Health Food” and is specifically designed to evaluate memory enhancement interventions (Inspection, 2003). The assessment was conducted at baseline and post-trial. The scale included the following subtests: (1) directional memory (DM): Participants were required to recall 12 target words from a list of 24, which included distractor items. This primarily assessed working memory and episodic memory under conditions of selective attention and interference control. Participants must actively maintain target words while inhibiting distractors, reflecting the ability to encode and retrieve specific items within a context. (2) associative learning (AL): Involved learning six pairs of words (three with logical connections and three without), tested in various sequences over three trials. This evaluated associative memory, a core component of episodic memory formation. It specifically tests the ability to bind different pieces of information (word pairs), which is fundamental for learning new facts and personal experiences. The inclusion of both logically related and unrelated pairs examined the efficiency of strategic versus rote associative encoding. (3) graphic free recall (GFR): Participants recalled 15 sequentially presented images of common daily items. This measured visual episodic memory and free recall ability. Recalling a series of common images tested the encoding, consolidation, and retrieval of visual information without cues, which was indicative of everyday memory for events and objects. (4) meaningless graphics recognition (MGR): Involved the recognition of five types of abstract graphic stimuli. This assessed visual recognition memory and pattern separation ability. Recognizing abstract graphics among distractors relied on the ability to form and retain distinct, non-verbal memory traces, tapping into perceptual representation memory systems. (5) portrait memory (PM): Participants memorized and later recalled details (name, occupation, preferences) associated with six portraits. This was a complex task that integrated episodic memory (binding multiple features—face, name, occupation, preference into a coherent episode) and semantic memory (recalling factual details). It closely simulated real-world social memory demands. The total score was calculated by summing the results from the five subtests, and the Memory Quotient (MQ) was derived from this total score (Luo et al., 2013; Zhang et al., 2017). All test administrators underwent training on the software system, and the assessments were conducted in a controlled, uninterrupted environment.

Resting-state functional MRI (fMRI) technology leverages the spontaneous fluctuations of blood oxygenation level-dependent (BOLD) signals to obtain functional brain imaging in a non-invasive and relatively fast manner. This technique measures hemodynamic changes induced by neuronal activity across the whole brain at multiple time points (Chu et al., 2021). Whole-brain MRI was conducted at the PLA General Hospital using a 3.0T Discovery MR 750 scanner (GE Healthcare, Milwaukee, WI, United States). Participants willing to undergo MRI testing were randomly selected, and both 3D-T1 high-resolution structural images and fMRI scans were performed at baseline and on Day 30.

The MRI data were processed using Statistical Parametric Mapping software (SPM12, http://www.fil.ion.ucl.ac.uk/spm) and the toolkit DPARSF (http://rfmri.org/DPARSF). Slice timing corrections were applied to all datasets. A 2x2 mixed analysis of variance (ANOVA) model, referred to as flexible ANOVA in SPM, was established to analyze the data. This model included both within-group and between-group factors. An F-test was used to determine the significance of main effects (group and time) and interaction effects. Post-hoc T-tests were then performed to further examine significant differences within and between groups. Areas with organic changes in gray matter volume were selected as seed regions, and known cognition-related regions were further selected as target region of interest (ROI). Further, the association of changes of gray matter volume or fractional amplitude of low-frequency fluctuations (fALFF) and clinical memory scale were tested to determine the potential causal association.

Descriptive statistics were reported as mean ± standard deviations (SD) for continuous variables and percentage for categorical variables. All primary analyses were based on the intention-to-treat (ITT) principle. Baseline categorical variables were analyzed using the Chi-square test or Fisher’s exact test, while continuous variables were assessed using the independent sample t-test. Changes within groups on the clinical memory scale were evaluated using the paired t-test and changes between groups were tested using independent t-test. In addition, to explore the influence of demographic factors on CG tablet efficacy, subgroup analyses were performed, stratified by age (30–49 vs. 50–65 years) and gender. For the MRI population, within-group changes and between-group differences were assessed using the Mann-Whitney U test due to the limited sample size. Associations between ROI and clinical memory scale score were tested using linear regression. A significance level of 0.05 was applied for all tests unless otherwise specified. Statistical analyses were conducted using R software (Version 4.2.2).

The study initially enrolled 117 participants (ITT population), with 59 randomized to the CG group and 58 to the placebo group (Figure 1). Demographic characteristics and baseline neuropsychological assessments are summarized in Table 1. The cohort had a mean age of 48.2 ± 6.8 years with 28.2% male representation. Two participants of the placebo group withdrew from the study due to personal reasons. A neuroimaging subsample (n = 15; CG = 9, placebo = 6) underwent serial MRI assessments at baseline and Day 30, with age ranging from 31 to 50 years. Importantly, baseline comparisons demonstrated no significant intergroup differences in age, gender distribution, educational attainment, or memory performance metrics across both the full cohort and MRI subsample (all p > 0.05), indicating successful randomization.

As shown in Table 2, the ITT analysis showed that participants in the CG group exhibited significant within-group improvements on DM, AL, GFR, and PM domains by Day 30 (all p < 0.01). Between-group comparisons revealed significantly greater improvement in the CG group compared to the placebo group for DM (mean change of 3.05 in the CG group vs. −1.21 in the placebo group, p = 0.0035), AL (1.86 vs. 0.16, p = 0.0015), and GFR (3.76 vs. −1.24, p = 0.0144). MGR scores decreased significantly within both groups post-intervention. However, the CG group showed a significantly smaller decline compared to the placebo group (−4.34 vs. −9.57, p = 0.0007). As shown in Figure 2, the CG group showed significant increase in the total score of clinical memory scale and MQ, and was significantly higher than those in the placebo group after intervention (P < 0.01).

After diving the participants into two groups based on age, both age below 50 years and above 50 years groups showed significant increase in DM, GFR, PM, total score of clinical memory scale, and MQ after CG intervention (Table 3). Although the CG group in the subgroup with age over 50 showed significantly better outcomes in terms of DM, AL, GFR, and PM compared to the placebo group, this significant result was not observed in the subgroup with age under 50. This might be due to the fact that in the subgroup of patients under the age of 50, the baseline values of DM, AL, GFR and PM in the CG group were slightly lower than those in the placebo group. Furthermore, the total scores and MQ values of both subgroups after the intervention showed that the CG group was significantly better than the placebo group.

After stratifying the participants by gender, females demonstrated a significant increase in DM, AL, GFR, PM, the total clinical memory score, and MQ after the CG intervention (Table 3). A similar trend was observed in males, although only the improvements in DM and GFR reached statistical significance, a finding potentially attributable to the smaller sample size of males within the CG group.

Analysis of the MRI population (n = 15) showed a significant within-group decline in MGR for the placebo group (p = 0.047), but no significant changes in other domains were detected, likely attributable to limited sample size. In the full cohort, global cognitive metrics including the total composite score and MQ showed significant between-group improvement favoring the CG group (p < 0.001). As depicted in Figure 2, these improvements in total score and MQ favoring the CG group were also evident in the neuroimaging subsample, although statistical significance was not reached due to the small sample size. However, significant higher total score and MQ in the CG group than the placebo group after intervention can be found.

Voxel-based morphometry revealed significant GMV reductions localized to three motor-cognitive hubs exclusively within the CG group (FWE-corrected cluster-level p < 0.05; Table 4; Figure 3): 1. Left precentral gyrus (Brodmann area 4, peak MNI: −40, −16, 22), extending to the adjacent middle frontal gyrus. 2. Right precentral gyrus (Brodmann area 4, peak MNI: 54, 4, 34), with partial overlap in the inferior frontal gyrus. 3. Left supplementary motor area (SMA, peak MNI: −8, 4, 58), involving the medial frontal cortex. No significant GMV changes were observed in the placebo group at the same statistical method, confirming the intervention-specific nature of these structural alterations.

Resting-state fMRI analysis of fALFF demonstrated significant CG-specific increases in fALFF within the left cerebellar lobule VI (peak MNI coordinates: 3, −72, −6; 12, −78, −12) - a key node in cognitive-motor networks (Stoodley and Schmahmann, 2009) (Figure 4). Post-hoc analysis confirmed robust fALFF increases in the CG group (e.g., peak T = 8.43, p < 0.001; peak T = 7.83, p < 0.001). In contrast, no significant fALFF changes were observed in the placebo group across cerebellar-cortical circuits.

Within the CG group, exploratory correlation analyses were conducted between changes in clinical memory scale scores and changes in neuroimaging metrics (Figure 5). 1. GMV Changes: A significant negative correlation was observed between the change in PM score and the change in GMV within the left SMA (R² = 0.58, p = 0.018). This indicates that greater improvement in PM was associated with greater reduction in left SMA GMV. Similarly, a negative correlation between the change in AL score and the change in GMV within the left precentral gyrus approached statistical significance (R² = 0.42, p = 0.061), suggesting a potential association where greater AL improvement tended to accompany greater left precentral GMV reduction. 2. fALFF Changes: Significant positive correlations were found between the change in fALFF within left cerebellar lobule VI and both the change in DM score (R² = 0.51, p = 0.031) and the change in total memory score (R² = 0.68, p = 0.007). These findings demonstrate that increased spontaneous neural activity in the cerebellum is robustly linked to enhanced memory performance following CG intervention.

In the present 30-day trial, the intervention was well-tolerated by all participants. No adverse events (AEs) were spontaneously reported by any participant in either the intervention or control group throughout the study period. No participants withdrew from the study due to intervention-related discomfort.