Alzheimer’s disease (AD), identified as the predominant form of dementia in the elderly population, is characterized by the accumulation of extracellular amyloid-beta (Aβ) plaques and the formation of neurofibrillary tangles (NFTs), which result from the hyperphosphorylation of tau protein (Scheltens et al., 2021). According to the World Alzheimer Report 2023¹, the global prevalence of dementia is projected to increase from 55 million individuals in 2019 to 139 million by 2050, with AD comprising 60–80% of these cases. Notably, AD has become the seventh leading cause of mortality worldwide, highlighting its status as a major public health crisis, further intensified by aging populations and socioeconomic development².

Despite decades of research, therapeutic breakthroughs remain elusive. Current interventions are primarily categorized into two types: symptomatic therapies that temporarily relieve cognitive and behavioral symptoms, and disease-modifying therapies (DMTs), which aim to slow or halt the progression of the disease (Breijyeh and Karaman, 2020). Symptomatic agents, including acetylcholinesterase inhibitors (AChEIs) such as donepezil, rivastigmine, and galantamine, as well as the N-methyl-D-aspartate (NMDA) receptor antagonist memantine, offer limited symptomatic relief without addressing the underlying disease mechanisms. Novel DMTs, primarily focused on targeting Aβ pathology through various mechanisms, are generally considered more suitable for individuals with Mild Cognitive Impairment (MCI) or mild to moderate AD. For instance, aducanumab and lecanemab are monoclonal antibodies (MABs) binding to Aβ protofibrils and plaques, while donanemab selectively targets pyroglutamate-modified Aβ (Jucker and Walker, 2023). In contrast, GV-971 (sodium oligomannate) offers a novel multimodal approach by modulating gut-brain axis dysbiosis to reduce neuroinflammation and Aβ deposition. However, its comparative efficacy, safety, and clinical applicability compared to traditional therapies are still debated. DMTs have the potential to revolutionize clinical practice by impacting treatment strategies, diagnostic methods, and efficacy monitoring. The introduction of anti-amyloid MABs marks a transition toward precision medicine in AD, guided by genetic and biomarker insights (Liu et al., 2025).

Current evidence syntheses regarding AD treatments exhibit fragmentation. Traditional pairwise meta-analyses have typically compared individual agents within specific therapeutic classes. However, there is a paucity of studies that integrate both direct and indirect evidence across diverse interventions. For example, trials investigating DMTs frequently emphasize biomarker outcomes, whereas those evaluating symptomatic treatments often prioritize cognitive scales. This inconsistency poses challenges for cross-comparison, leaving clinicians uncertain about the optimal treatment hierarchies. Network meta-analysis (NMA), which concurrently synthesizes data from multiple interventions, is particularly well-suited to bridge this gap by providing a comprehensive ranking of therapies based on efficacy and safety endpoints. While DMTs such as aducanumab and lecanemab have shown efficacy in Aβ clearance during phase III trials, their cognitive benefits remain in limited and are associated with considerable risks, including amyloid-related imaging abnormalities (ARIA) (Hampel et al., 2023). Conversely, symptomatic treatments provide moderate yet consistent cognitive stabilization, albeit without altering the underlying disease progression. A comprehensive comparative analysis of these therapeutic strategies is currently absent. Safety concerns, such as ARIA associated with anti-amyloid MABs and gastrointestinal side effects linked to AChEIs, underscore the need for an integrated evaluation of their risk–benefit profiles. GV-971, which has demonstrated cognitive improvement in a phase III trial conducted in China (Xiao et al., 2021), though not yet subjected to global regulatory scrutiny, exerts multi-modal effects by modulating gut microbiota, reducing Aβ and tau pathology, and mitigating neuroinflammation, thereby reversing cognitive deficits (Wang et al., 2019). Therefore, this systematic review and NMA aims to assess the cognitive safety and efficacy of 10 distinct interventions: placebo, GV-971, aducanumab, lecanemab, donanemab, donepezil, rivastigmine (both patch and capsule forms), galantamine and memantine. The study intends to provide clinically actionable hierarchies to inform therapeutic decision-making. To our knowledge, this is the first NMA to compare both DMTs and symptomatic therapies with respect to cognitive, functional, and safety outcomes. This analysis is based on a systematic search of five databases and clinical trial registries, addressing critical questions such as whether DMTs offer significant advantages over traditional symptomatic treatments, and whether multimodal approaches, such as combining DMTs with AChEIs, can be justified based on current evidence.

This analysis integrates both indirect and direct evidence from 23 randomized controlled trials (RCTs), aiming to inform clinical practice and outline optimal strategies for the subsequent management of AD.

The NMA map is shown in Figure 3, and detailed NMA results are shown in Supplementary Figure 1.

Over the past three decades, the FDA has authorized the use of several pharmacological treatments for AD, including tacrine, donepezil, rivastigmine, galantamine, and memantine (Ogos et al., 2024). These pharmacological agents primarily target symptomatic relief; however, they exhibit limitations as they predominantly mitigate symptoms without addressing the underlying etiological factors, such as facilitating neuronal regeneration or the clearance of Aβ plaques and hyperphosphorylated tau (Reddi Sree et al., 2025). Professor Jeffery Cummings (Cummings and Fox, 2017) first defined DMTs in 2017 as an emerging therapeutic strategy that targets the core pathophysiological mechanisms of AD to alter its disease trajectory, alleviate symptoms, and delay progression. These strategies encompass the reduction of Aβ deposition, the regulation of tau hyperphosphorylation, the inhibition of neuroinflammation, and the preservation of neuronal survival and synaptic function. As of 2024, the AD drug development pipeline comprises 164 clinical trials evaluating 127 distinct agents. Notably, 75% of these trials are concentrated on DMTs, predominantly involving biologics and small molecules, which target fundamental pathologies such as Aβ plaque formation, tau protein hyperphosphorylation, neuroinflammation, synaptic plasticity dysfunction, and neurotransmitter receptor modulation (Cummings et al., 2024).

Historically, meta analyses on AD drug efficacy and safety centered on symptomatic treatments (Dou et al., 2018; Guo et al., 2020), omitting increasingly prominent DMTs and limiting comprehensive evaluation of treatment strategies. Previous NMA, such as Qiao et al. (2024), focused on anti-amyloid MABs while excluding symptomatic therapies, precluding a holistic assessment. Moreover, several drugs (e.g., bapineuzumab, solanezumab, gantenerumab, crenezumab) in their analysis later failed in clinical trials, thereby calling into question the robustness of the findings. To address these, this NMA comprehensively compares 10 interventions (DMTs and symptomatic treatments) across cognitive, functional, and safety outcomes. DMTs include three anti-amyloid MABs (aducanumab, lecanemab, donanemab) and the gut microbiota-targeting agent GV-971; symptomatic treatments are common clinical agents (e.g., AChEIs, NMDA receptor antagonists).

With the successive approvals of anti-amyloid MABs, DMTs for AD are increasingly prominent, yet their clinical application relies on accurate AD diagnosis. The National Institute on Aging and Alzheimer’s Association (NIA-AA) ATN framework characterizes AD pathophysiology by positive brain Aβ deposition (Jack et al., 2018). Aβ positron emission tomography (PET) imaging, via specific tracers, enables noninvasive localization and quantification of cerebral Aβ plaques, crucial for early diagnosis and therapeutic guidance (Rabinovici et al., 2023). The Centiloid (CL) scale standardizes Aβ PET quantification, ensuring comparability across multicenter studies, tracers, and pipelines, and is widely used in global trials (Klunk et al., 2015). In the CLARITY AD trial, lecanemab demonstrated a reduction in Aβ load by 59.12 CL units, with a mean clearance falling below the 22.99 CL positivity threshold, indicating a potential delay in disease progression of approximately 2 to 3 years. Data from the 2024 Alzheimer’s Association International Conference (AAIC) 3-year follow-up of the CLARITY AD trial revealed sustained improvements in Aβ-PET imaging and the plasma Aβ_42/40 ratio compared to placebo, underscoring the cumulative benefits of continued lecanemab administration. In the TRAILBLAZER-ALZ 2 trial, the donanemab cohort exhibited an 87.0 CL reduction in brain Aβ levels at week 76, accompanied by reductions of 39.3 and 21.3% in plasma phosphorylated tau217 (p-tau217) and glial fibrillary acidic protein (GFAP), respectively. Regarding aducanumab, Aβ PET imaging indicated 48% amyloid negativity in the EMERGE trial and 31% in the ENGAGE trial; in EMERGE, cerebrospinal fluid (CSF) Aβ₄₂ levels increased in a dose-dependent manner, with concomitant decreases in total tau, phosphorylated tau181 (p-tau181), and plasma p-tau181. Similarly, in the ENGAGE trial, CSF Aβ₄₂ levels rose dose-dependently with reductions in plasma p-tau181. Nonetheless, the approval of anti-amyloid MABs based solely on the reduction of Aβ plaques is scientifically unjustified, as changes in biomarkers do not consistently correlate with clinical benefits.

Due to the rising attention on early disease stages, DMTs that focus on Aβ are becoming increasingly popular. In the early stages of AD, the disruption of continuous Aβ production and its efficient clearance results in the pathological aggregation of Aβ into misfolded structures. Mabs targeting Aβ are designed to identify and eliminate these harmful aggregates to delay or prevent their formation. Approved anti-amyloid MABs vary in their targets, focusing on different structural regions and aggregation states of Aβ peptides (Kim et al., 2025). Aducanumab, a humanized IgG1 MAB, targets the N-terminus of Aβ with a preference for fibrillary aggregates over monomers, aiding in Aβ clearance via microglial phagocytosis (Dhillon, 2021). Approved by the FDA in 2021 as the first Aβ-targeting therapy, its use is debated. In phase III trials, EMERGE showed a 22% cognitive decline reduction with high doses, while ENGAGE failed its primary endpoint. Both trials noted decreased amyloid plaques and plasma p-tau181, but high doses led to more AEs like ARIA-E, headaches, and brain microhemorrhages. Despite the results of EMERGE, further validation of the efficacy and safety of aducanumab is needed (Budd Haeberlein et al., 2022). Lecanemab is a humanized antibody that crosses the blood–brain barrier to target and bind with Aβ protofibrils, oligomers, and fibrils. Its Fab segment forms complexes with Aβ, while the Fc segment engages microglial receptors to promote phagocytosis, clearing toxic Aβ and reducing neuronal damage. Lecanemab also inhibits Aβ aggregation and slows tau pathology progression (Shi et al., 2022). Approved by the FDA on July 6, 2023, lecanemab showed superior efficacy in slowing cognitive decline compared to placebo in an 18-month phase III trial (van Dyck et al., 2023), with significant changes in CDR-SB scores (p < 0.001). It also reduced brain amyloid burden and improved cognitive and functional outcomes. Safety concerns included infusion-related reactions in 26.4% of recipients and ARIA-E in 12.6%, mostly mild to moderate and resolving within four months. Researchers have highlighted potential risks for women and ApoE4 carriers, underscoring the need for careful approval and monitoring (Kurkinen, 2023). Donanemab is a humanized IgG1 MAB that targets N-terminal pyroglutamate-modified Aβ proteins in amyloid plaques, facilitating their removal via microglial phagocytosis. It reduces brain amyloid burden by targeting existing plaques but does not prevent new plaque formation or the growth of existing ones (Buccellato et al., 2023). Approved by the FDA on July 2, 2024, donanemab significantly slowed cognitive and functional decline in Aβ-positive early AD in the phase III TRAILBLAZER-ALZ 2 trial (Dyer, 2024). Nearly half of the early-stage patients exhibited no clinical progression after 1 year, with subgroup analysis revealing a 60% slower decline compared to placebo (Sims et al., 2023).

In this NMA, aducanumab demonstrated superior efficacy in improving cognitive and daily living abilities, aligning with its potent Aβ plaque clearance mechanism. It is worth noting that the initial statements discussed the approvals of aducanumab, lecanemab and donanemab which were based on their ability to reduce Aβ levels, superior to their impact on cognitive function. While the drugs may perform well in Aβ levels, their improvement in patients’ cognitive function in real-world applications may not be as expected (Imbimbo et al., 2023; Richard et al., 2021). Symptomatic agents showed moderate, consistent benefits across cognitive scales, underscoring their role in stabilizing cholinergic and glutamatergic neurotransmission. In terms of mental symptoms, memantine emerged as the top intervention for neuropsychiatric symptoms, likely due to its NMDA receptor modulation mitigating agitation and aggression. Rivastigmine capsule outperformed other AChEI in global clinician assessments, possibly due to dual acetylcholinesterase/butyrylcholinesterase inhibition (Inglis, 2002).

Researchers have found that dysbiosis of the gut microbiota can lead to the accumulation of amino acids such as phenylalanine and isoleucine in the bloodstream. These metabolic products promote the differentiation and proliferation of peripheral immune cells, activating M1-type microglia, which exacerbates neuroinflammation and subsequently accelerates the pathological progression of AD (Wang et al., 2019). GV-971 is an orally given blend of acidic linear oligosaccharides sourced from marine brown algae, created by Shanghai Green Valley Pharmaceuticals, and received first approval in China in November 2019 for use in the management of mild to moderate AD. GV-971 has demonstrated the ability to modify gut microbiota, potentially alleviating the impact of compromised peripheral immunity on AD pathogenesis (Xiao et al., 2021). In a randomized, double-blind, placebo-controlled, multicenter phase III clinical trial conducted in China, GV-971 improved the ADAS-cog scores in patients over a 36-week treatment period, but no significant differences were detected among the drug and placebo groups regarding secondary endpoints, including ADCS-ADL and NPI (Syed, 2020). Several factors may contribute to these results: Firstly, GV-971 early clinical trial data in China have been published, but data from the international multicenter phase III trial are not yet available. The Chinese studies, with small sample sizes and high population heterogeneity, restrict the relevance of the results. Secondly, the ADCS-ADL, an informant-based inventory designed to evaluate everyday activities in AD (Galasko et al., 1997), may experience information bias and measurement bias due to cultural differences. Additionally, the baseline NPI scores were very low (with a mean of 3), limiting the dynamic range necessary to detect measurable improvements. Finally, the limited timeframe of the experiment could have restricted the observation of changes from baseline.

In safety assessments, symptomatic treatments have been used clinically for many years with a relatively low incidence of adverse effects, which are primarily concentrated in the gastrointestinal systems. The side effects of DMTs are primarily a series of discomforts caused by ARIA. The safety profile of aducanumab within the EMERGE and ENGAGE investigations was consistent and in accordance with prior research findings. The most frequently observed AE was ARIA-E, detected via brain MRI. In cases of ARIA-E, severe symptoms have indeed occurred, including intense seizures that required hospitalization. The analysis of the comprehensive safety dataset indicates that the predominant AE observed in the 10 mg/kg aducanumab cohort was ARIA-E, with an incidence rate of 35.2% (362 cases), of which 26.0% (96 cases) presented with related symptoms such as headache (Salloway et al., 2022). As a critical AE associated with anti-amyloid MABs, ARIA-E requires rigorous monitoring and clinical management throughout the treatment course (Budd Haeberlein et al., 2022). In patients receiving donanemab treatment, ARIA-E was observed, with the majority of cases being predominantly asymptomatic. While donanemab demonstrated superiority over placebo in the composite endpoint integrating measures of cognition and activities of daily living, secondary outcomes did not reach statistical significance. Lecanemab is also associated with similar AEs. Additionally, a pharmacovigilance study based on the FDA Adverse Event Reporting System database identified new and unexpected lecanemab-related AEs that were not previously reported in regulatory trials, such as tremors, migraines, pancreatic cancer, et al. Certain patient subgroups, particularly those receiving polypharmacy for AD, as well as those taking aspirin, proton pump inhibitors, statins, antidepressants, or benzodiazepines, may be at a higher risk of experiencing severe AEs (Xing et al., 2025). In light of this, clinicians must place significant emphasis on the aforementioned AEs by closely monitoring the vital signs of participants and systematically assessing potential risks.

Our analysis found that new experimental drugs often show better therapeutic results than established ones, possibly due to optimistic perceptions or biases like selective outcome reporting in early trials. This “newness advantage” suggests inherent biases in DMT trials, needing further validation through IPD-based NMA. Chin J Intern Med recently published two consensus documents that systematically outline DMT strategies for AD, integrating the latest clinical evidence and expert insights. The consensus underscores that simultaneously targeting multiple pathological mechanisms, including Aβ, Tau, and neuroinflammation, may enhance therapeutic efficacy. It also advocates for personalized treatment approaches, such as tailoring interventions based on tau pathology burden and ApoE genotype (Neurologist Branch, Chinese Medical Doctor Association; The Expert Group for Expert Consensus on Disease-Modifying Therapy for Alzheimer’s Disease, 2025; Expert Consensus Review Committee on Disease-Modifying Treatments for Early Alzheimer’s Disease, 2025). However, as DMTs transition into clinical practice, several challenges emerge. For instance, community physicians and internists face difficulties in adopting these novel therapies, including limited awareness and experience in managing their side effects. Additionally, there is currently no standardized guidance on whether patients can switch between different anti-amyloid MABs, highlighting the need for further exploration. Consequently, clinicians and patients should critically evaluate the perceived superiority of newly introduced medications by integrating evidence hierarchies and mitigating cognitive biases in therapeutic decision-making.

This study represents the most recent and largest evidence-based NMA to date, which for the first time evaluates the efficacy of FDA-approved and internationally recognized therapeutic agents for AD over multiple years, encompassing both traditional symptomatic treatments and DMTs. A total of 23 RCTs were included in this analysis, all of which were deemed to be of high quality. Furthermore, the findings of this study are characterized by their authenticity and comprehensiveness. Consequently, this NMA provides comprehensive and rigorous evidence-based recommendations for the treatment and management of patients with AD.

However, there were some limitations to this NMA. Firstly, due to the limited number of head-to-head studies, this NMA primarily relied on indirect estimates, which may affect the accuracy of the results. Secondly, due to different evaluation tools, some scales have poor sensitivity and specificity, which may lead to bias. Thirdly, the inability to obtain sufficient IPD in the RCTs necessitated analysis at a general level, thereby leaving the potential for confounding factors unaddressed. Additionally, the included RCTs vary in participant characteristics, sample size, intervention targets, frequency, and time, which may lead to heterogeneity in results and thus reduce the strength of clinical evidence. Finally, the meta-analysis data were derived solely from publicly available scientific literature, and the publication bias regarding negative results and non-statistical data should be considered, prompting readers to interpret these findings with caution.

In summary, NMA suggests that aducanumab holds the greatest potential for cognitive and clinical improvements in patients with MCI and early AD, as evidenced by assessments including the MMSE, ADAS-cog, and ADCS-ADL. In contrast, lecanemab provides moderate benefits, while donanemab proves less effective. However, memantine, the traditional symptomatic treatment, remains the preferred option for alleviating neuropsychiatric symptoms in AD patients. The safety profile of DMTs requires further clinical validation. Clinicians must carefully consider biomarker status, disease stage, and safety profiles to optimize personalized treatment strategies for AD. Additionally, owing to the restricted quantity of investigations incorporated in certain interventions, the findings necessitate careful interpretation. Subsequent inquiries ought to prioritize the execution of high-caliber, extensive, and protracted RCTs to substantiate the validity of these findings.