Alzheimer’s disease (AD) is the most common cause of dementia, exerting a substantial global burden. Although the pathological roles of amyloid-beta (Aβ) aggregation and tau have been increasingly clarified, translating these insights into effective treatments remains a major challenge (1). Mild cognitive impairment (MCI) is a transitional stage between normal aging and dementia, and individuals with amyloid PET–positive MCI are at particularly high risk of progressing to AD. Previous studies have characterized amyloid PET–positive MCI as prodromal AD (2, 3), with subtle impairments in memory and executive function that preserve functional independence. The annual conversion rate to AD among biomarker-confirmed MCI patients is estimated at 5–30% (4, 5). However, no disease-modifying treatments are currently approved for MCI. In South Korea and many other countries, standard anti-dementia drugs like donepezil, rivastigmine, galantamine, and memantine are not reimbursed for MCI (6), leading to widespread off-label use of cognitive supplements such as omega-3 fatty acids, choline precursors, and Ginkgo biloba.

Ginkgo biloba leaves have been traditionally utilized in East Asian ethnomedicine, particularly within Chinese and Korean medical traditions, as an oral decoction or herbal remedy to enhance memory, mitigate age-related cognitive decline, and alleviate circulation-related symptoms such as dizziness and tinnitus (7). Currently, standardized Ginkgo biloba extracts are widely employed both as prescription phytomedicine in Europe and as a non-prescription cognitive supplement globally. Despite their broad usage, the efficacy of Ginkgo biloba extracts as monotherapy in biomarker-confirmed mild cognitive impairment (MCI) remains uncertain.

At our institution, Ginkgo biloba extract has been occasionally prescribed depending on clinical context, and our accumulated clinical experience prompted a more systematic investigation. We employed Ginexin-F^®, a standardized extract approved by the Korean Ministry of Food and Drug Safety (MFDS-199702183). Its composition is equivalent to EGb 761, a widely studied formulation in clinical trials. Ginkgo’s neuroprotective effects have been attributed to antioxidative, anti-inflammatory, and vasodilatory properties, as well as modulation of neurotransmission and inhibition of Aβ aggregation (7–9). Although many studies have examined Ginkgo’s effects in MCI, results remain mixed. A placebo-controlled trial showed cognitive benefits with good adherence (10), whereas the large GEM trial in cognitively normal elderly individuals failed to show preventive effects, likely due to low incidence of dementia and poor compliance (11). The GUIDAGE trial in patients with subjective memory complaints also yielded inconclusive results (12, 13). Early meta-analyses indicated modest cognitive improvements (14), but a Cochrane review in 2007 questioned their robustness (15). More recent randomized trials and meta-analyses suggest EGb 761 may offer symptomatic benefit in MCI. One 12-month trial in China reported reduced dementia incidence in amnestic MCI patients treated with Ginkgo (16), while others noted improvements in memory and neuropsychiatric symptoms (17, 18). Still, a recent systematic review concluded that definitive evidence for Ginkgo in MCI remains insufficient (19).

This inconsistency likely reflects the heterogeneity of MCI itself. While some patients harbor prodromal AD pathology, others may have non-AD causes of cognitive symptoms. This variability complicates trials, as AD-targeting therapies may be tested on non-AD cases. Furthermore, the slow progression of early-stage AD challenges the sensitivity of conventional outcome measures. To address these issues, we integrated biomarker-based strategies for diagnosis and treatment monitoring. Our group previously reported a retrospective study combining donepezil and Ginkgo in amyloid PET–positive AD patients, using plasma MDS-Oaβ (Multimer Detection System–Oligomeric Aβ) as a biomarker (20). This blood-based assay measures the oligomerization tendency of Aβ and has shown associations with cognitive scores, disease progression, and CSF tau levels (21). In that study, the combination group demonstrated improved cognition and greater reductions in MDS-Oaβ. However, all patients received donepezil, limiting evaluation of Ginkgo’s independent effect. Because anti-dementia drugs are not reimbursed for MCI in Korea, Ginkgo biloba—being accessible and affordable—is often used as a practical alternative due to its accessibility and affordability. This creates a real-world setting to examine its effects without confounding treatments.

In this study, we focused on amyloid PET–positive MCI patients to enhance diagnostic specificity. We evaluated cognitive and functional changes as well as plasma MDS-Oaβ over 12 months. Unlike static biomarkers such as Aβ42 or total Aβ, MDS-Oaβ reflects the dynamic propensity for toxic oligomer formation, a key pathogenic step in AD (21). This study was designed to assess whether Ginkgo biloba monotherapy could provide both symptomatic and biomarker-level benefits in patients with amyloid-confirmed MCI. To this end, we analyzed real-world registry data and applied validated biomarker assays to evaluate its potential as a practical early intervention.

This retrospective cohort study aimed to evaluate the efficacy of Ginkgo biloba monotherapy in patients with amyloid PET–positive MCI, using both clinical and biomarker endpoints. Our findings suggest that Ginkgo, when used alone without concomitant prescription anti-dementia medications, may offer measurable benefits in terms of cognitive stability, functional maintenance, and reduction in plasma MDS-Oaβ levels over a 12-month follow-up period. In multivariate linear regression with responder status as the dependent variable, only Ginkgo group was significantly associated with being a responder, no other covariates reached statistical significance (Table 4). Our results are consistent with findings from recent studies that explored the use of EGb 761 in MCI and early dementia. For example, Tian et al. demonstrated a reduced incidence of dementia over 52 weeks in amnestic MCI patients treated with Ginkgo (16). Similarly, García-Alberca et al. found that Ginkgo in combination with acetylcholinesterase inhibitors improved cognitive and neuropsychiatric outcomes (28). While most studies have examined Ginkgo as an adjunct therapy, our study uniquely evaluates it in isolation, supported by biomarker evidence.

These results build upon our earlier study, which showed that the addition of Ginkgo to donepezil in amyloid PET–positive Alzheimer’s disease (AD) patients led to better cognitive outcomes and a larger decrease in MDS-Oaβ (20). However, in that study, Ginkgo was not administered independently, and all patients received donepezil due to ethical standards. This limited our ability to isolate Ginkgo’s effect. In contrast, the current study focused on MCI patients—a population where anti-dementia drugs are not reimbursed in Korea—allowing us to observe the standalone effects of Ginkgo in a real-world setting.

The cognitive preservation observed in the Ginkgo group is notable, especially considering that approximately 5–30% of amyloid PET–positive MCI cases convert to AD annually (4, 5). While the non-Ginkgo group showed mild decline in K-MMSE and modest functional worsening, the Ginkgo group not only maintained but slightly improved in both cognitive and functional outcome measures. These changes, though modest in magnitude, are clinically relevant given the context of early AD intervention, where even stabilization is a meaningful therapeutic target. The improvement in MDS-Oaβ levels further supports the potential disease-modifying effect of Ginkgo. This plasma biomarker reflects the dynamic propensity of Aβ monomers to oligomerize, a process that is believed to play a critical pathogenic role in AD by disrupting synaptic function and triggering neuroinflammation. Traditional biomarkers like CSF Aβ42 and tau levels are static measures, whereas MDS-Oaβ offers insight into ongoing amyloidogenic activity. The observed reduction in MDS-Oaβ in the Ginkgo group may indicate a suppression of toxic oligomer formation.

Mechanistically, standardized extracts of Ginkgo biloba contain bioactive components including flavonol glycosides (e.g., quercetin, kaempferol, and isorhamnetin) and terpene lactones (ginkgolides and bilobalide), which provide distinct yet complementary protective effects against amyloid-beta (Aβ) pathology. Flavonol glycosides directly interact with Aβ peptides, disrupting the formation of β-sheet-rich oligomeric structures and inhibiting fibril nucleation, thus attenuating oligomer-induced synaptic toxicity. Additionally, their antioxidative activity mitigates oxidative stress-induced neuronal damage, a known accelerator of Aβ aggregation (8, 9). Ginkgolides, particularly ginkgolide B, have demonstrated potential inhibitory effects on β-secretase (BACE-1), the enzyme critically involved in amyloidogenic cleavage of amyloid precursor protein (APP), thereby reducing Aβ generation at an early stage. Bilobalide further contributes neuroprotective effects through mitochondrial stabilization and promotion of autophagy-mediated clearance of misfolded proteins (29, 30). Collectively, these pharmacological activities likely synergize to reduce Aβ oligomer formation and deposition, as reflected by decreased plasma MDS-Oaβ levels observed in our study. Importantly, these effects were achieved without the use of donepezil, rivastigmine, or memantine, highlighting the potential of Ginkgo as a monotherapy in the prodromal stage of AD.

The robustness of these findings is supported by two converging observations. First, our earlier add-on trial in amyloid-positive AD dementia showed that Ginkgo biloba produced a comparable magnitude of benefit when layered on top of donepezil, and that benefit tracked with a parallel reduction in plasma MDS-Oaβ levels (20). Second, in the present study, the non-Ginkgo group appeared to follow a clinical and biomarker trajectory similar to the natural course reported in longitudinal cohorts of amyloid-positive MCI: CDR-SB worsened over 12 months (4, 31), and MDS-Oaβ levels increased in a pattern consistent with previously reported biomarker trajectories in early-stage AD (27). The observation that the non-Ginkgo group showed clinical and biomarker trajectories similar to those reported in previous studies suggests that the likelihood of systematic bias or measurement artefact may be low and provides some support for the interpretation of stability observed in the Ginkgo group. Nonetheless, because our design is retrospective and unblinded, these signals should be viewed as hypothesis-generating rather than confirmatory.

The study also holds significance in the context of current therapeutic limitations. Recently approved anti-amyloid agents such as lecanemab and donanemab have shown promise in modifying disease progression. While direct comparisons must be approached with caution due to substantial methodological differences, nonetheless the cognitive and functional outcomes in the Ginkgo group appeared relatively stable, and this observation may be of interest in the context of findings from recent monoclonal antibody trials. For example, in the Clarity AD study of lecanemab, all participants exhibited some degree of cognitive decline over 12 months, with a mean increase of approximately 1.2 points in CDR-SB despite treatment (32). A similar pattern of worsening was observed in the TRAILBLAZER-ALZ 1 trial of donanemab (33). In our study, however, no mean change in CDR-SB was observed and slight improvement in MMSE was noted over a comparable period. Moreover, their high costs, intravenous administration requirements, and limited availability hinder widespread adoption, particularly in countries without full insurance coverage. In contrast, Ginkgo biloba is accessible, affordable, and orally administered, making it a pragmatic option for early intervention. Furthermore, the ethical and logistical challenges of conducting large, randomized trials in biomarker-confirmed MCI populations make real-world data increasingly valuable. Our study, based on registry-derived information and supported by biomarker assays, strengthens the internal validity of our findings.

Despite these promising results, several limitations must be acknowledged. First, the retrospective and non-randomized nature of this study inherently limits causal inference. Factors such as health behaviors, comorbidities, caregiver support, and other unmeasured variables might have influenced outcomes. Although we attempted to minimize these effects by comparing clinically similar groups, we did not apply formal statistical adjustments for baseline characteristics, and thus the possibility of residual confounding should be considered when interpreting between-group differences. Second, our sample size, while adequate for exploratory analysis, may not provide sufficient power to detect subtler subgroup effects or rare adverse events. Third, adherence to the assigned treatments was not objectively measured, although follow-up visits confirmed continued use in most cases. Fourth, while MDS-Oaβ is a promising biomarker, it should be acknowledged that MDS-Oaβ is not yet a globally standardized biomarker. To date, numerous peer-reviewed international studies have been published using MDS-Oaβ, demonstrating consistent analytic validity and substantial clinical utility in various contexts. Although its analytical performance can be sensitive to preanalytical factors—such as freeze–thaw cycles, sample handling, and storage duration—continuous efforts have been made to optimize and standardize these procedures. While there are currently no globally accepted regulatory standards or universal clinical thresholds for MDS-Oaβ, the method has already been approved by regulatory authorities in some countries (e.g., the Ministry of Food and Drug Safety, MFDS in Korea) and is actively utilized in clinical research. Thus, given its growing evidence base, MDS-Oaβ is emerging as a promising biomarker for Alzheimer’s disease, warranting further prospective validation studies. Lastly, we did not assess other potential biomarkers such as plasma p-tau or neurofilament light chain (NfL), which could provide additional insights into disease progression. Nonetheless, the findings of this study carry practical implications. In healthcare systems where early AD treatments are not accessible or reimbursed, Ginkgo biloba may represent a viable first-line option for biomarker-confirmed MCI. Its tolerability profile and multi-mechanistic action make it especially suitable for early intervention strategies. For clinicians, the availability of real-world biomarker data may also encourage more personalized and proactive management of MCI. Future research should aim to validate these findings through prospective randomized controlled trials, ideally incorporating multimodal biomarkers (e.g., tau PET, digital cognitive tests, and CSF markers). Additionally, studies comparing Ginkgo to emerging AD treatments in head-to-head designs could clarify its position in the evolving therapeutic landscape.

This study provides preliminary but compelling evidence that Ginkgo biloba monotherapy may offer clinical and biological benefits in amyloid PET–positive MCI. By stabilizing cognitive performance and reducing amyloid oligomerization in plasma, Ginkgo may represent a cost-effective and accessible option for early-stage intervention in the Alzheimer’s continuum.