Intranasal administration of stem cell derivatives for the treatment of AD animal models: a systematic review and meta-analysis

Abstract

Background

Alzheimer’s disease (AD) is a neurodegenerative disorder with increasing prevalence and limited efficacy of current therapies. Stem cell-derived therapies have attracted attention for their potential neurodegenerative and reparative effects. Intranasal administration provides a non-invasive route that bypasses the blood-brain barrier and delivers stem cell derivatives directly to the brain. Although studies have shown that the intranasal administration of stem cell derivatives alleviates symptoms in animal models of AD, a comprehensive meta-analysis evaluating their therapeutic efficacy is yet to be conducted. This study aims to evaluate the efficacy of intranasal stem cell derivatives therapies in AD animal models and provide a foundation for clinical translation.

Methods

We conducted a systematic literature search across four databases (Pubmed, Embase, Web of Science and Cochrane Library) using subject terms and complementary keywords. After applying inclusion and exclusion criteria, data were extracted using Origin 2024 software and analysed using Review Mange 5.4. The SYRCLE tool was applied to assess study quality and evaluate the potential risk of bias systematically.

Results

This meta-analysis of 14 studies investigated the efficacy of intranasal stem cell-derived therapies in animal models of Alzheimer’s disease (AD). Using predominantly mouse and rat models from diverse geographical locations and employing various stem cell types (bone marrow, umbilical cord blood, adipose tissue, hiPSCs), the analysis revealed a significant reduction in amyloid-beta (Aβ) deposition (SMD = -2.69, p < 0.0001). Subgroup analyses indicated that stem cell source, stem cell derivative types and Aβ detection method were not primary drivers of heterogeneity. Furthermore, treatment significantly reduced inflammatory markers IL-1β (SMD = -0.92, p = 0.008) and IBA-1 (SMD = -1.68, p = 0.006), suggesting an anti-inflammatory effect. Auxiliary outcomes CD68 and GFAP also exhibited decreased expression levels. Improved cognitive function was evident, as measured by increased target quadrant dwell time (MD = 10.17, p < 0.00001) and decreased escape latency (MD = -15.74, p = 0.003) in behavioral experiments, and enhanced recognition in the Novel Object Recognition Test (NORT) (SMD = 1.10, p = 0.006). Nissl staining demonstrated a significant reduction in neuronal cell death (SMD=-3.33,p < 0.00001),suggesting a role in neuronal repair.Together, these findings support the potential of intranasal stem cell-derived therapies to improve Alzheimer’s disease pathology, neuronal repair, and cognition in animal models.

Conclusion

Intranasal administration of stem cell derivatives has demonstrated efficacy in Alzheimer’s disease (AD) animal models, leading to reductions in amyloid-beta (Aβ) deposition, cognitive improvement, repair neurons and attenuation of inflammatory responses. However, limitations such as potential publication bias and heterogeneity among existing studies are noted. Due to insufficient data and study limitations, additional preclinical and clinical trials are required to confirm these results and explore the therapy’s long-term safety and efficacy.

Graphical abstract

Introduction

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder with a steadily increasing incidence, posing a significant challenge to global public health. The pathological hallmarks of AD include neuronal loss, synaptic dysfunction, the formation of amyloid-β (Aβ) plaques and neurofibrillary tangles, and the gradual deterioration of cognitive function [1–3]. The amyloid cascade hypothesis posits that Aβ deposition is a major driver of AD pathology, which has led to extensive research into targeting Aβ through pharmacological and cellular therapies [4, 5]. However, currently available treatments, including acetylcholinesterase inhibitors, N-methyl-D-aspartate receptor antagonists, and multi-targeted drugs (e.g. GV-971), can only alleviate symptoms, and no curative therapy has been established to date [6].

Stem cell derivatives, such as exosomes, extracellular vesicles, and soluble factors, have emerged as a promising therapeutic strategy for AD [7]. Exosomes maybe transpass the epithelial layer and migrate to lesion areas, thereby improving AD pathology and cognitive deficits [8, 9]. Exosomes are nanoscale vesicles (40–100 nm in diameter) released when intraluminal vesicles (ILVs) within multivesicular bodies (MVBs) fuse with the plasma membrane [10]. They contain proteins, nucleic acids, lipids, and other bioactive components, enabling intercellular communication. Extracellular vesicles consist of membrane-bound vesicles derived from various cellular sources, ranging from 40 to 1000 nm in diameter. This heterogeneous group includes exosomes, microvesicles, and apoptotic bodies [11, 12]. Soluble factors, such as growth factors and cytokines, are small proteins secreted by stem cells.Soluble factors exert their effects through paracrine and endocrine to modulate tissue repair and inflammatory responses [13] (Table 1). Four clinical trials have investigated the intranasal delivery of stem cell-derived exosomes for CNS disorders: NCT05152394, NCT06598202, NCT05158101, NCT04388982 (clinicaltrials.gov) (Table 2). Among these, NCT04388982 demonstrated the safety of intranasal administration of stem cell-derived exosomes in Alzheimer’s disease (AD) patients [14]. Due to their neuroregenerative and reparative properties, stem cell derivatives secreted by neural stem cells and mesenchymal stem cells have shown therapeutic potential in reducing Aβ deposition, modulating inflammatory responses, and promoting neuronal regeneration [15]. Additionally, transplanted stem cells enhance the therapeutic effects by promoting endogenous stem cells to replace damaged neurons, mitigating inflammatory responses, and protecting neuronal survival, all of which play a critical therapeutic role in Alzheimer’s disease [16, 17]. Furthermore, studies have demonstrated that stem cell-derived exosomes exhibit therapeutic effects comparable to their parental stem cells [18]. Therefore, the use of stem cell derivatives in AD treatment has contributed to a broader understanding of the disease’s multifactorial nature. This paradigm shift has driven the development of multi-targeted therapeutic strategies beyond conventional single-target approaches focused solely on Aβ, tau, or neuroinflammation [19, 20] (Table 3).

Table 1.

Definition of stem cell derivatives

Name	Definition	Solubility	Diameter	Source	Marker proteins	Compose	Function
Exosomes	Nanoscale vesicles released when intraluminal vesicles (ILVs) within multivesicular bodies (MVBs) fuse with the plasma membrane	Insoluble	40–100 nm	Endosomal System-Derived	TSG101,CD63	Nucleic Acids Lipids and Protein	Mediates intercellular material and information transfer, and exhibits multiple biological functions including anti-inflammatory, anti-apoptotic, and pro-neuroregenerative effects.
Extracellular vesicles	Membrane vesicles originating from various cellular sources	Insoluble	40–1000 nm	Endosomal System-Derived and Plasma Membrane-Derived	ARF6、CD63、SG10、ARF1	Nucleic Acids Lipids and Proteins	Mediates intercellular material exchange, activates intracellular signaling pathways, participates in cellular metabolism and signal transduction, and regulates gene expression and functionality in recipient cells.
Soluble factors	Free proteins and small molecules derived from stem cells through paracrine and endocrine mechanisms	Soluble	NO fixed diameter	N/A	N/A	The composition of growth factors, neurotrophic factors, growth factors and other factors	Modulates tissue repair and inflammatory responses

N/A: Not applicable

Table 3.

Characteristics of animal trials

References	Publication time	Country	Stem cell source	Cellular input method	Injection method	Dosage of stem cell derivatives	Total time of treatment	Interval time	Number of treatments	Types of animals	Animal year	Parameter
T. Harach [9]	2016	Switzerland	Bone marrow	Soluble Mesenchymal Stem cell factors	IN/IV	NA	3weeks	1day	21 times	APP/PS1 mouse	12 weeks	➀➂
Jiaxin Li [8]	2023	China	Bone marrow	MSC-EXO	IN	1ug/g	4weeks	3days	10 times	3×Tg-AD mouse	10 weeks	➀➁
Masoumeh Pourhadi [3]	2023	Iran	Umbilical cord	MSC-EXO	IN	1.4μg	2weeks	3days	5 times	Wistar rat	adult	➀➁➂
Allaura S. Cone [42]	2021	USA	Bone marrow	MSC-EV	IN	4 × 106	16weeks	4days	30 times	5XFAD mouse	6 weeks	➀➁➂
Hyunkyung Mo [32]	2023	South Korea	Umbilical cord	MSC-SE/CNSC-SE	IN	5μg/g	4weeks	7days	4 times	5XFAD mouse	12weeks	➀➁➂
Xinyi Ma [1]	2020	China	Adipose tissue	MSC-EVs	IN	1ug/g	2weeks	2days	7 times	APP/PS1 mouse	9months	➀➁➂
Seyed Amir Shaker [2]	2024	Iran	Adipose tissue	MSC-CM	IN	N/A	N/A	NA	N/A	Wistar rat	adult	➀➁➃
D. Yu. Zhdanova [43]	2020	Russia	Umbilical cord	MSC-EXO	IN	NA	3weeks	3days	6 times	NMRI mouse	6months	➁
Giulia Santamaria [44]	2020	Italy	Bone marrow	MSC-CS	IN/IV	1 × 106	4weeks	7days	4 times	APP/PS1 mouse	25months	➀➁➂
Morris Losurdo [46]	2020	Italy	Bone marrow	MSC-EV	IN	30ug	36 h	18 h	2 times	3xTgAD mouse	7months	➂
Mita Tsuneyuki [47]	2015	Japan	Bone marrow	MSC-CM	IN	0.15ug	4days	1days	8 times	ICR mouse	9 weeks	➁
Mehrnaz Izadpanah [37]	2020	Iran	Bone marrow	MSC-EV	IN	50ug	2weeks	3days	14 times	Wistar rat	adult	➀➁➂➃
Leelavathi.N.Madhu [45]	2024	USA	HiPSC	NSC-EV	IN	3 × 1010	8weeks	7days	8 times	5XFAD mouse	3 months	➂
Leelavathi.N.Madhu [48]	2024	USA	HiPSC	NSC-EV	IN	3 × 1010	2weeks	7days	2 times	5XFAD mouse	3 months	➀➂

N/A: Not reported, MSC-EXO: mesenchymal stem cell exosomes, MSC-CM mesenchymal stem cell culture medium, MSC-EV: extracellular vesicles of mesenchymal stem cells NSC-EVs: neural stem cell extracellular vesicles, MSC-CS: Mesenchymal Stem Cell-Conditioned Medium, MSC-SE/CNSC-SE: Mesenchymal stem cell/neural stem cell derivatives, hNSCs: human Neural Stem Cells, HiPSC: human Induced Pluripotent Stem Cells, IN: intranasal administration, IV: intravenous injection IH: intrahippocampal injection① behavioral test, Aβ deposition, inflammatory factor Neuron death count

Table 2.

Representative clinical trials of intranasal therapy using stem Cell-Derived products for CNS diseases

NCT number	Condition	Phase	Stem cell source	Stem cell derivative type	Status	Interventional Model	Outcome Measure	Time Frame
NCT05152394	Parkinson’s Disease	Phase I	Human umbilical cord	MSCs-Exos	Recruiting	Single Group Assignment	Safety (adverse events)	Four year follow-up
NCT06598202	Amyotrophic Lateral Sclerosis	Phase I/II	Human umbilical cord	MSCs-Exos	Recruiting	Sequential Assignment	Number of participants who experienced dose-limiting Toxicities (DLTs)	24 h, 4±1 Weeks
NCT05158101	Stroke	Phase I	Human umbilical cord	MSCs-Exos	Recruiting	Single Group Assignment	Safety (adverse events)	Four year follow-up
NCT04388982	Alzheimer Disease	Phase I/II	Allogenic Adipose	MSCs-Exos	Unknown	Sequential Assignment	Number of participants with treatment-related abnormal laboratory values of Liver or kidney function	12 weeks

Resource of the clinical trial is from ClinicalTrials.gov

MSCs-Exos, mesenchymal stem cell derived exosomes

Table 3 Characteristics of animal trials

Compared with conventional drug delivery methods such as intravenous and brain-targeted injections, intranasal administration offers a unique pathway for drug transport across the blood-brain barrier (BBB) [2, 21]. Clinical applications of intranasal delivery include midazolam administration for epilepsy management and budesonide treatment for COVID-19 [22, 23]. Additionally, clinical trials have investigated the potential of intranasal insulin delivery for the treatment of Alzheimer’s disease (AD)(NCT00581867). Furthermore, intranasal administration of stem cells has been explored in clinical experiments for conditions such as cerebral palsy, and preclinical studies have suggested its therapeutic potential for Alzheimer’s disease (AD) [24–26]. Traditional stem cell delivery routes encompass intravenous, subcutaneous, and intraperitoneal injections [27, 28]. In cardiovascular medicine, intra-cardiac and local intra-arterial injections with cell carriers are employed [29, 30]. However, unlike these methods, intranasal administration enables stem cells to traverse the BBB and accumulate in the brain. This route is also effective for delivering stem cell-derived products to the brain [25]. Evidence indicates that stem cell-derived exosomes may cross the BBB via the olfactory and trigeminal nerve pathways to reach the hippocampus or sites of brain injury [3]. Specifically, mesenchymal stem cell exosomes delivered intranasally have been observed to migrate to the brain as early as 1 h post-administration [31, 32]. Accumulation in the hippocampal CA3 region has been reported within 6 h [33], and localization at damaged sites within 24 h [34]. Intranasal administration may be a therapeutic prospect for stem cell derivatives targeting neurodegenerative diseases of the central nervous system [32]. Thus, intranasal administration presents significant advantages, including rapid and efficient targeting, making it highly promising for clinical translation.

Currently, common methods for delivering stem cell derivatives, such as intracerebroventricular and intravenous injections, facilitate direct transport to the brain. However, their invasiveness and the risk of serious complications pose significant challenges to clinical translation [35–37]. In contrast, intranasal administration provides a less invasive and potentially safer alternative, reducing the risk of complications such as bleeding and intracranial tumor formation [38, 39].

Studies have demonstrated that intranasal administration of stem cell derivatives can reduce Aβ deposition, improve the pathological environment of the AD brain, and significantly alleviate cognitive deficits in animal models. Although animal models have been widely used to evaluate the safety and efficacy of intranasal stem cell derivatives therapy, there is a notable lack of meta-analyses synthesizing evidence on its effectiveness in AD models. Therefore, this systematic review and meta-analysis aim to assess the efficacy of intranasal stem cell derivatives therapy in experimental AD models and provide guidance for its clinical translation and further development.

Methods

Search strategy

We systematically searched PubMed, EMBASE, the Cochrane Library, and Web of Science to identify studies on intranasal stem cell derivative therapy for Alzheimer’s disease (AD). Our comprehensive strategy combined Medical Subject Headings (MeSH) terms, including “mice,” “stem cells,” “Alzheimer’s disease,” and “intranasal drug delivery,” along with relevant free-text terms such as “mother cell,” “progenitor cells,” “colony-forming unit,” “stem cell derivatives,” “intranasal administration,” “administration, intranasal drug,” “administration, nasal,” “intranasal transplantation,” “intranasal porting,” “intranasal transplant,” “Mouse,” “Mice, Laboratory,” “Mouse, Laboratory,” “Mouse, Swiss,” “Mice, Swiss,” “Mouse, House,” “Mice, House,” “Rat,” “Rat, Laboratory,”and “Rat, Norway.”

Additionally, we incorporated AD-related terms, including “Alzheimer-type dementia (ATD),” “Alzheimer syndrome,” “Alzheimer dementia,” “dementia, senile,” “Alzheimer-type senile dementia,” “primary senile degenerative dementia,” “Alzheimer sclerosis,” “acute confusional senile dementia,” “early-onset Alzheimer disease,” “presenile Alzheimer dementia,” “late-onset Alzheimer disease,” “focal-onset Alzheimer’s disease,” and “familial Alzheimer disease (FAD).”

We also screened reference lists of relevant articles and ClinicalTrials.gov for additional studies to ensure a comprehensive review.

Inclusion and exclusion criteria

We followed the Population, Intervention, Comparison, Outcome, and Study Design (PICOS) model to define our eligibility criteria:

1. Population: Studies involving Alzheimer’s disease (AD) animal models, including transgenic AD models (e.g., APP/PS1, 5XFAD), toxin-induced models (e.g., streptozotocin), or Aβ injection-induced models;

2. Intervention: Intranasal administration of stem cell derivatives;

3. Comparison: Saline or phosphate-buffered saline (PBS) injections or AD animal models;

4. Outcomes: The primary outcome was β-amyloid (Aβ) deposition.Or secondary outcomes included time to target quadrant, time to escape, IL-1β levels, IBA expression, and cell death counts. CD68, GFAP expression, and the novel object recognition outcome were also considered auxiliary outcomes;

5. Study Design: Parallel controlled trials;

6. Language: Published in English;

Exclusion criteria

Studies were excluded if they:

1. used non-pathological animal models;

2. were non-experimental studies (e.g., meta-analyses, reviews, commentaries, case reports, or case series).

3. were in vitro studies;

4. had missing outcome data that could not be calculated, extrapolated, or obtained;

Study selection

Search results from four databases were imported into NoteExpress for duplicate removal. Titles and abstracts were screened to exclude irrelevant studies. Two researchers independently select the studies, and a third researcher was consulted to resolve any disagreements.

Data extraction

We extracted key information from the included studies, including the first author, country, publication time, types of animals, year of animals, stem cell source, dosage of stem cell derivatives, injection method, total time of treatment, number of treatments, interval time and outcomes. Data were collected for primary, secondary, and auxiliary outcomes. The primary outcome was β-amyloid (Aβ) deposition. Secondary outcomes included escape latency, target quadrant residence time, IL-1β levels, IBA-1 expression, and cell death count. Auxiliary outcomes consisted of GFAP expression, CD68 levels, and the novel object recognition outcome.

For studies presenting data in graphical form, we used OriginPro 2024 for data extraction. Two researchers independently select the studies, and a third researcher was consulted to resolve any disagreements. Standard deviations (SD) were calculated from standard errors (SE) using the formula √N × SE when not directly reported. If the study did not specify the sex of the animals, data from both male and female animals were included by default. Studies lacking essential information were excluded.

Quality assessment

Two researchers independently assessed the quality of the included studies by reviewing abstracts and full-text extracts. Two researchers independently select the studies, and a third researcher was consulted to resolve any disagreements.The quality of the animal studies was evaluated using the SYRCLE risk of bias tool [40]. The entire process, including the meta-analysis, was conducted in accordance with PRISMA guidelines for systematic reviews.

Data analysis

The primary outcome for data analysis was β-amyloid (Aβ) plaque accumulation. Secondary outcomes included escape latency, time spent in the target quadrant, IL-1β levels, and IBA-1 expression. Standardized mean differences (SMD) or mean differences (MD), along with 95% confidence intervals (CI), were calculated for each outcome to assess the effect of stem cell treatment. Inter-study heterogeneity was evaluated using the I² statistic. If I² ≤ 50%,indicated homogeneity and fixed-effect models were employed, or random-effect models were used instead. For studies with I² >50%, indicating high heterogeneity, a random-effects model was used [41]. To ensure the accuracy and reliability of our findings, we conducted a sensitivity analysis.The individual studies were deleted in turn according to the order of publication, so as to analyze the main and secondary research results. RevMan software was employed to generate forest plots and risk-of-bias plots for the meta-analysis.

During subgroup analysis, study heterogeneity was assessed using the I² statistic. If I² >50% within a subgroup, a random-effects model was applied. If high overall heterogeneity persisted after subgroup analysis, meta-regression analysis was performed to explore potential sources of variation, considering factors such as animal age and cell dosage.

For result presentation, Review Manager 5.4 software was used to generate forest plots and funnel plots. Effect sizes (SMD/MD) were adjusted for small-sample bias using Hedges’ g for standardized mean differences, while mean differences for continuous variables were reported in their original units. Funnel plots were utilized to assess publication bias. All statistical tests were two-tailed, with P < 0.05 considered statistically significant. If data were derived from a single-gender cohort, sensitivity analysis was conducted to evaluate potential biases. Additionally, if included studies reported data in inconsistent units, Review Manager 5.4 software was used to standardize and combine data using SMD for subsequent analysis and comparison.

We conducted a subgroup analysis of the main outcomes, and the subgroup factors include stem cell source, Aβ detection method and stem cell derivative type. For the multiple comparison, the Bonferroni correction method was used to adjust the significance level to the overall alpha/the number of comparisons (i.e. 0.05/3, α = 0.016667). In addition, when analyzing the effects of inflammation, since multiple independent outcomes are involved (such as IL-1β, IBA-1, etc.), Bonferroni correction is also used, setting the adjusted tothe overall alpha/the number of comparisons (i.e. 0.05/4, α = 0.0125).

Risk of bias assessment

Two researchers independently select the studies, and a third researcher was consulted to resolve any disagreements. The SYRCLE tool was used to evaluate the risk of bias in preclinical animal studies, focusing on selection bias (sequence generation, baseline characteristics, allocation concealment), implementation bias (randomization of animal placement, investigator blinding), measurement bias (randomized outcome assessment, assessor blinding), attrition bias (incomplete outcome data), reporting bias (selective outcome reporting), and other potential biases.

Allocation sequence adequacy was assessed by determining whether the allocation sequences for the control and experimental groups were sufficient. Studies with adequate sequences were classified as low risk, while those with insufficient sequences were categorized as high risk. Baseline characteristic consistency was evaluated to ensure that animals in both groups were in similar conditions before treatment; inconsistencies led to a high-risk classification. Allocation concealment was determined by whether the intervention assignment was concealed from relevant personnel. If allocation concealment was maintained, the study was rated as low risk; otherwise, it was considered high risk.

Randomization of animal placement was verified to ensure unbiased distribution. Studies with proper randomization were categorized as low risk, whereas non-random placement resulted in a high-risk classification. Random assessment of outcomes was checked to confirm whether randomization was applied to data collection; if not, the study was marked as high risk. Blinding of outcome evaluators was assessed, with blinded studies considered low risk and non-blinded studies classified as high risk.

Data completeness was evaluated, with incomplete datasets classified as high risk. Selective reporting bias was assessed by determining whether the reported outcomes aligned with the study objectives. Studies with consistent reporting were considered low risk, whereas inconsistencies led to a high-risk classification. Other biases, such as those arising from animal mortality or additional treatments, were also considered.

If bias-related information was not explicitly reported, the study’s risk status was classified as unclear. The overall risk rating was considered low if all categories were rated as low risk, unclear if at least four categories were rated as unclear, and high if at least four categories were rated as high risk. For studies identified as high risk, sensitivity analysis was performed by excluding these studies and recalculating the combined effect size to assess the robustness of the findings.

Result

Search results

Our systematic literature search identified 124 records: 29 from PubMed, 18 from Embase, 77from Web of Science, and 0 from the Cochrane Library (Fig. 1). After screening titles and abstracts, 36 full-text articles met the inclusion and exclusion criteria. Of these, 14 studies investigating intranasal stem cell derivatives administration in Alzheimer’s disease (AD) animal models were included in the quantitative analysis. These 14 studies were selected because 10 reported data on primary outcomes, while 4 provided information secondary outcomes [1–3, 8, 9, 32, 37, 42–48].

Fig. 1.

PRIMSA flow diagram of the search strategy for this systematic review and meta analysis

Ten studies reported data on the primary outcome, β-amyloid (Aβ) accumulation. Among the five secondary outcomes, seven studies assessed time spent in the target quadrant, five reported escape latency, six measured IL-1β levels, seven examined IBA-1 expression, and two analyzed cell death counts. Additionally, three auxiliary outcomes were included: two studies assessed the novel object recognition test (NORT) outcome, three measured CD68 levels, and two evaluated GFAP expression.

Study characteristics and quality

The 14 included studies consisted of 3 rat studies and 11 mouse studies. Among the animal models, 3 studies used Wistar rats, 3 used APP/PS1 mice, 4 used 5XFAD mice, 2 used 3xTgAD mice, and 1 study each utilized NMRI mice and ICR mice. These studies were conducted in diverse locations: 4 in Iran, 2 in China, 3 in the USA, and 1 each in Italy, Japan, Russia, Switzerland, and South Korea Table 3.

Ten studies employed different methods to detect Aβ deposition: 4 used Western blotting, 1 used Congo red staining, 2 used immunohistochemistry with the anti-Aβ antibody 6E10, 1 used Thioflavin-S staining, and 2 used immunofluorescent staining. The sources of stem cell derivatives varied, with 7 studies using bone marrow-derived stem cells, 3 using human umbilical cord blood-derived stem cells, 2 using adipose tissue-derived stem cells, and 2 using human-induced pluripotent stem cells (hiPSCs) Table 3.

Risk of bias

Figure 2 summarizes the risk of bias assessment across all included studies. Two domains—selection bias and reporting bias—were categorized as unclear risk, while measurement bias was rated as high risk in 4 studies, primarily due to insufficient blinding of researchers and outcome assessors. The remaining domains, including implementation bias and other potential biases, generally showed a low risk.

Fig. 2.

The risk of bias assessment for the study was conducted using the SYRCLE tool, categorizing each item in the reviewed literature. The figure illustrates the risk classification for each category: low risk of bias (green), high risk of bias (red), and unclear risk (yellow)

Figure 3 provides a more detailed breakdown of these assessments. Implementation bias was relatively evenly distributed between low and unclear risk categories. Selection bias exhibited a higher proportion of unclear risk, with fewer studies classified as high risk. In contrast, measurement bias showed a substantial proportion of low-risk ratings, with fewer studies categorized as unclear. All other bias domains were assessed as low risk.

Fig. 3.

Risk of bias assessment of studies using the SYRCLE tool based on the listed categories. The figure displays the percentage of studies classified as having a low risk of bias (green), an unclear risk (yellow), and a high risk of bias (red)

Aβ deposition

A meta-analysis of 10 studies [1–3, 8, 9, 32, 37, 42, 44, 45] (Fig. 4) was conducted to evaluate the effect of intranasal stem cell derivative therapy on Aβ deposition in AD model animals. Due to high heterogeneity (I² = 81%), a random-effects model was employed. The results indicated a statistically significant reduction in Aβ deposition in the treatment group compared to controls (SMD = -2.69, 95% CI [-3.94, -1.43], Z = 4.19, P < 0.0001). The combined sample included 70 animals in the stem cell derivative group and 67 in the control group, totalling 137 animals. The overall effect, visualized in the forest plot (Fig. 4), is represented by a diamond positioned to the left of the null line, with no overlap, strongly supporting a significant reduction in Aβ deposition. This finding aligns with previous studies that reported reduced Aβ plaque accumulation following stem cell derivative treatment, including a substantial reduction in plaques compared to untreated controls [32]. To assess the stability of Aβ deposition outcomes for the primary outcome, we further conducted sensitivity analyses by systematically excluding each study in sequence. Regarding the pooled standardized mean difference (SMD), Aβ deposition was not significantly affected by any individual study.

Fig. 4.

Forest plot showing mean effect sizes and 95% confidence intervals (CI) for the comparison of study Aβ deposition accumulation between stem cell derivative treatment and control groups

Given the high heterogeneity observed in the overall analysis, subgroup analyses were conducted to explore potential sources of variation. First, we examined the impact of the stem cell source (Fig. 5). Subgroup analysis of bone marrow-derived stem cells (I² = 60%) and other stem cell sources (I² = 89%) both revealed substantial heterogeneity, which persisted when combined (Fig. 5A). Further stratification into bone marrow (I² = 60%), umbilical cord (I² = 96%), and adipose tissue/other (I² = 81%) subgroups also demonstrated persistent heterogeneity (Fig. 5B). These findings suggest that the stem cell source is unlikely to be a major contributor to heterogeneity.

Fig. 5.

Plot of forest subgroup analysis of the effect of different sources of stem cell derivatives on Aβ accumulation. (A) The primary source analyzed was bone marrow, and comparisons were made with other sources (B) Comparative analyses were performed for bone marrow, umbilical cord, and other sources

Next, we performed subgroup analyses based on the method used to detect Aβ deposition (Fig. 6). Studies using Western blotting (I² = 74%) and other staining methods (I² = 83%) exhibited considerable heterogeneity, which remained high when combined (Fig. 6A). Further subdivision into Western blotting, anti-Aβ 6E10 immunohistochemistry, Congo red staining, and other methods revealed a similar pattern, suggesting that detection method is also unlikely to be a significant source of heterogeneity (Fig. 6B).

Fig. 6.

Forest plot of the subgroup analysis of Aβ accumulation studies, stratified by Aβ detection method. (A) Comparison between Western blotting and other staining methods. (B) Comparison among four methods: Western blotting, Antibody 6E10 immunohistochemistry, Congo red staining, and other staining techniques

In our initial subgroup analyses examining stem cell sources and Aβ detection methods, the results suggested these factors were unlikely to be major contributors to between-study heterogeneity. To further investigate potential sources of heterogeneity, we conducted additional subgroup analyses by stem cell derivative types across the seven included studies: four extracellular vesicle studies (I²=57%), two exosome studies (I²=93%), and one soluble factor study. These findings suggest that stem cell derivative types are unlikely to be the major contributor to heterogeneity (Fig. 7).

Fig. 7.

Forest plot of subgroup analyses for Aβ accumulation studies, stratified by stem cell derivative types. Comparisons were made across three distinct categories: extracellular vesicles, exosomes, and soluble factors

Cognitive function and Spatial memory ability

To evaluate the effects of intranasal administration of stem cell derivatives on cognitive and memory functions in AD model animals, we primarily analyzed escape latency and target quadrant dwell time in the behavioral experiments, with the New Object Recognition Test (NORT) outcome serving as an auxiliary outcome.

A total of seven studies were included in the meta-analysis of target quadrant dwell time [2, 3, 8, 32, 37, 42, 43].The heterogeneity test revealed a high degree of variability among studies (I² = 83%), necessitating the use of a random-effects model to combine effect sizes. The analysis demonstrated a significant increase in target quadrant dwell time in animals treated with stem cell derivatives compared to controls (MD = 10.17, 95% CI [6.46, 13.89], P < 0.00001) (Fig. 8). This finding suggests that intranasal stem cell derivative therapy effectively enhances spatial learning in AD model animals.

Fig. 8.

Forest plot showing mean effect sizes and 95% confidence intervals (CI) for the effect of the study on spatial memory capacity between the stem cell derivative treatment and control groups

For escape latency, five studies were included in the meta-analysis [2, 3, 8, 32, 37]. The heterogeneity test indicated substantial variability between studies (I² = 76%), leading to the use of a random-effects model. The results showed a significantly shorter escape latency in animals treated with stem cell derivatives compared to controls (MD = -15.74, 95% CI [-26.07, -5.41], P = 0.003) (Fig. 9). These findings suggest that intranasal administration of stem cell derivatives can effectively improve cognitive function in AD model animals.

Fig. 9.

Forest plot showing mean effect sizes and 95% confidence intervals (CI) for the effect of the study on the level of cognitive functioning in the stem cell derivative treated and control groups

To strengthen the methodological robustness of our secondary outcome assessments (target quadrant dwelling time and escape latency), we employed an iterative leave-one-out approach for sensitivity evaluation. The pooled MD estimates demonstrated no substantive alterations across exclusion iterations, with neither cognitive function outcome nor memory performance indices exhibiting significant fluctuations attributable to any singular dataset., confirming the statistical robustness of these cognitive parameters against individual study influences.

To further confirm the cognitive benefits of intranasal stem cell derivative therapy, we analyzed the NORT outcome as an auxiliary outcome. Two studies [42, 44] were included in the analysis. The results revealed a significant increase in the new object recognition outcome in the treatment group compared to controls (SMD = 1.10, 95% CI [0.31, 1.89], P = 0.006) (Fig. 10). This further supports the conclusion that intranasal administration of stem cell derivatives significantly enhances cognitive function and spatial memory in AD model animals.

Fig. 10.

Forest plot showing the mean effect sizes and 95% confidence intervals (CI) for the study’s new target discrimination outcome, comparing stem cell derivative-treated and control groups

Inflammatory effects

To evaluate the impact of intranasal administration of stem cell derivatives on inflammation, we assessed interleukin-1β (IL-1β) and IBA-1 expression levels as primary outcomes, while CD68 and GFAP expression levels served as secondary outcomes.

A total of six studies were included in the meta-analysis of IL-1β expression levels [32, 37, 44–46, 48]. The heterogeneity test indicated moderate variability among studies (I² = 50%), prompting the use of a random-effects model for effect size combination. The analysis revealed a significant reduction in IL-1β expression levels in animals treated with intranasal stem cell derivatives compared to controls (SMD = -0.92, 95% CI [-1.60, -0.24], P = 0.008) (Fig. 11).

Fig. 11.

Forest plot showing mean effect sizes and 95% confidence intervals (CI) for anti-inflammatory effects of IL-1 β expression levels between stem cell derivative treated and control groups

For IBA-1 expression levels, seven studies were incorporated into the meta-analysis [1, 9, 32, 44–46, 48]. The heterogeneity test showed substantial between-study variability (I² = 84%), necessitating the use of a random-effects model. The results demonstrated a significant decrease in IBA-1 expression levels in the treatment group compared to the control group (SMD = -1.68, 95% CI [-2.90, -0.47], P = 0.006) (Fig. 12).

Fig. 12.

Forest plot showing the mean effect size and 95% confidence intervals (CI) of the anti-inflammatory effects of IBA-1 expression levels between the stem cell derivative treated and control groups

To enhance the methodological rigor of our investigation into the stability of secondary outcomes (IL-1β and IBA-1 expression levels), we implemented a sequential exclusion approach for sensitivity analysis, systematically omitting each included study in successive iterations to assess potential influences on overall result consistency. In the pooled standardized mean difference (SMD) analysis, inflammatory response profiles were not significantly impacted by any single study.

To further validate the anti-inflammatory effects of intranasal stem cell derivative therapy, we analyzed CD68 and GFAP expression levels as secondary outcomes. Data from two studies [42, 44] were available for GFAP, while three studies [44–46]provided data for CD68. The results showed a significant reduction in GFAP expression levels (MD = -0.52, 95% CI [-0.86, -0.18], P = 0.003) and CD68 expression levels (SMD = -1.14, 95% CI [-1.88, -0.40], P = 0.003) in animals treated with stem cell derivatives (Figs. 13 and 14).

Fig. 13.

Forest plot showing mean effect sizes and 95% confidence intervals (CI) for GFAP levels studied between the stem cell derivative treated and control groups

Fig. 14.

Forest plot showing the mean effect sizes and 95% confidence intervals (CI) for CD68 levels, comparing stem cell derivative-treated and control groups

Nissl staining

Nissl staining was used to assess the extent of cell death. Two studies [2, 37] reported quantitative data on cell death, which were analysed to determine the effects of stem cell derivative treatment. The results revealed a significantly lower number of dead cells in the stem cell derivative treatment group compared to the control group (SMD = -3.33, 95% CI [-4.74, -1.92], P < 0.00001) (Fig. 15).

Fig. 15.

Forest plot showing mean effect sizes and 95% confidence intervals (CIs) for cell death counts in stem cell derivative treated and control groups

Sensitivity analysis

To ensure the accuracy and reliability of our findings, we conducted a sensitivity analysis. Do this for the primary and secondary research results by sequentially removing individual studies based on their publication order. In the combined standardized mean difference (SMD) analysis, we found that Aβ deposition, andinflammation levels were not significantly influenced by between-study differences. Similarly, in the combined mean difference (MD) analysis, cognitive function data and spatial memory capacity were unaffected by between-study variations.

Discussion

This study conducted a systematic review and meta-analysis integrating evidence from 14 preclinical investigations on intranasal administration of stem cell-derived agents in Alzheimer’s disease (AD) animal models, providing a theoretical foundation for developing multitarget therapeutic strategies via the nasal route. Current AD interventions predominantly focus on single-target pharmacological approaches, which demonstrate limited clinical efficacy, thereby underscoring the scientific imperative to explore synergistic multitarget modalities [49–51]. The results revealed that intranasal delivery of stem cell derivatives significantly reduced β-amyloid (Aβ) plaque deposition, suppressed neuroinflammatory responses, and markedly improved cognitive function and spatial learning performance in AD models. These multidimensional therapeutic effects suggest that stem cell-derived agents may bypass the blood-brain barrier through specific delivery mechanisms, concurrently targeting multiple pathological pathways including Aβ clearance, neuroimmune modulation, and inflammatory cytokine regulation.

Meta-analysis of Aβ deposition demonstrates that intranasal stem cell-derived therapy can significantly reduce Aβ deposition. The mechanism of Aβ reduction may involve stem cell exosomes downregulating presenilin-1 (PS1) expression, inhibiting β-site amyloid precursor protein cleaving enzyme 1 (BACE1) activity, and activating Aβ-degrading enzymes (neprilysin, neprilysin, and insulin-degrading enzyme), thereby decreasing intracerebral accumulation of Aβ plaques [52–54].Meta-analysis of inflammatory effects indicates intranasal stem cell-derived therapy significantly reduces IL-1β and IBA-1 expression, confirming its anti-inflammatory effects. The potential anti-inflammatory mechanisms involve mesenchymal stem cell exosomes (MSC-exos) exerting multi-target synergistic effects through their protein and miRNA components: regulating immune cell phenotype transformation (e.g., promoting macrophage/microglia polarization from pro-inflammatory M1 to anti-inflammatory M2 phenotype and enhancing Treg cell activity), blocking key inflammatory signaling pathways (NF-κB/NLRP3), and remodeling cytokine microenvironment, downregulating TNF-α/IL-6 while upregulating IL-10/TGF-β; simultaneously delivering regulatory miRNAs like miR-124-3p and miR-21 to suppress inflammation-related gene expression [55–58].Meta-analysis of neuronal effects shows intranasal stem cell-derived therapy reduces neuronal cell death and contributes to neuronal repair. The neurorestorative mechanism may involve stem cell exosomes carrying growth differentiation factor 15 (GDF-15) to activate the AKT/GSK-3β/β-catenin signaling pathway in neurons, upregulate neprilysin (NEP) and insulin-degrading enzyme (IDE) expression, promote Aβ degradation while inhibiting apoptosis and inflammatory responses [59–61].Ultimately alleviating Aβ-induced neuronal damage and providing a potential novel therapeutic strategy for neurodegenerative diseases like Alzheimer’s disease.

The systematic review and meta-analysis revealed high heterogeneity, with several contributing factors.First, significant diversity in stem cell types across included studies, including umbilical cord mesenchymal stem cells (UC-MSCs) and adipose-derived mesenchymal stem cells (ASCs). Research evidence shows inherent differences in proliferation kinetics and secretome profiles among stem cells from different sources [62, 63].UC-MSCs exhibit superior proliferative capacity compared to ASCs, with secretion profiles characterized by high expression of macrophage inflammatory protein-2 (MIP-2), interleukin-6 (IL-6) and growth-regulated oncogene (GRO), suggesting future studies should conduct standardized analyses on specific stem cell subtypes to clarify their clinical translation potential [64]. Second, stem cell derivatives constitute complex bioactive systems containing extracellular vesicles (EVs), exosomes and soluble factors. Experimental studies demonstrate distinct molecular compositions and mechanisms among these components: exosomes primarily mediate anti-inflammatory, anti-apoptotic and neuroregenerative effects; extracellular vesicles participate in metabolic reprogramming by modulating signaling pathways; soluble factors play central roles in tissue repair and immune microenvironment regulation (as shown in Table 1). It is necessary to establish a standardized classification system based on component functions and systematically evaluate the efficacy and targets of each subclass in large-scale preclinical studies.

Finally, variability in animal models may be a major contributor to inter-study heterogeneity. The included Alzheimer’s disease (AD) animal models encompass different genotypes such as 5XFAD and 3×Tg-AD, which exhibit distinct pathological progression: 5XFAD mice show detectable Aβ plaque deposition at 2 months and synaptic dysfunction at 4 months, whereas 3×Tg-AD mice develop Aβ deposition at 6 months and progress to tau pathology stage at 12 months [65]. These differences in pathological onset windows may systematically bias treatment efficacy evaluation by affecting the biological effects of intervention timing. To enhance the translatability of preclinical research data, subsequent experiments should employ standardized AD animal models with well-defined pathological phenotypes (e.g. uniformly using 5XFAD or APP/PS1 models) and implement interventions during critical pathological windows (e.g. initial stage of Aβ deposition), thereby improving cross-study comparability and providing more consistent evidence for clinical translation research.

There are several potential limitations to consider in this review and meta-analysis.First of all, this meta-analysis is limited to mouse and rat models and may not fully replicate the pathological and physiological conditions of human AD patients.Second, most of the studies included in this analysis involved male rodents, for which data on female studies were inadequately provided.There are articles reporting that there are gender differences in the onset of AD affected by the differential expression of sex hormones and sex chromosome genes. Estrogen has a protective effect on pathological changes in AD.Estrogen promotes the normal transport of amyloid precursor protein (APP) and reduces Aβ deposition; Reduces tau hyperphosphorylation.These mechanisms work together to provide protection against AD [66].

To reduce research heterogeneity and enhance clinical translatability, future analyses should be conducted under the control of a single variable.Firstly, for mesenchymal stem cells derived from specific tissues (such as umbilical cord derived UC-MSCs), systematically compare the therapeutic efficacy differences of different derivative subtypes (exosomes, soluble factors, extracellular vesicles).Secondly, under the premise of fixed derivatives (e.g., exosomes), the similarities and differences in the efficacy of different tissue-derived stem cells (UC-MSCs vs. ASCs vs. bone marrow MSCs) were compared horizontally.

Conclusion

Through a systematic review and meta-analysis, our findings indicate that the intranasal administration of stem cell derivatives effectively reduces Aβ amyloid plaque accumulation, prolongs the time spent in the target quadrant, shortens escape latency, and significantly enhances cognitive function and spatial memory in animal models. Additionally, this therapy demonstrates anti-inflammatory effects by reducing IL-1β and IBA-1 expression levels. Furthermore, it promotes neuronal survival by nourishing neural tissue and decreasing neuronal death. While these findings highlight the potential of intranasal stem cell derivative therapy for Alzheimer’s disease, further validation through additional clinical trials and animal studies is essential to confirm its efficacy and therapeutic potential.

Abbreviations

AD

Alzheimer’s disease

SMD

Standardized mean difference

MD

Mean difference

Aβ

Amyloid-β

MSC-EXO

Mesenchymal stem cell exosomes

MSC-CM

Mesenchymal stem cell culture medium

MSC-EV

Mesenchymal stem cell extracellular vesicles

NSC-EVs

Neural stem cell extracellular vesicles

MSC-CS

Mesenchymal stem cell-conditioned medium

MSC-SE

Mesenchymal stem cell derivatives

NSC-SE

Neural stem cell derivatives

hNSCs

Human neural stem cells

HiPSC

Human induced pluripotent stem cells

IN

Intranasal administration

IV

Intravenous injection

IH

Intrahippocampal injection

BBB

Blood-brain barrier

MWM

Morris water maze

IL-1β

Interleukin-1β

Author contributions

Jun Ma and Huixian Cui conceived and designed the study. Zilin Hua, Zijing Zhou, Zewei Fu selected the articles and extracted and cross-checked the data. Haixia Chang, Nan Zhou Herman Yao Akogo, Jiayin Yang, Meixuan Yu, Yujie Jiang, Siyi Lan contributed to the statistical analysis. Zilin Hua wrote the first draft of the manuscript. Jun Ma and Huixian Cui revised and discussed the final edition. The authors read and approved the final manuscript.

Funding

This study was funded by the Natural Science Foundation of China (81801278), the China Scholarship Council (201608130015), the Natural Science Foundation of Hebei Province (H2019206637, H2015206409, H2023206266), the Key Natural Science Foundation of Hebei Province (H2020206557), the Overseas Researcher Program of the Hebei Provincial Department of Human Resources and Social Security (C20190509), the Hebei Province High-Level Talent Funding Project (B2022003030), the Hebei Medical University Chunyu Project (CYQD2023005), and the Industry-University-Research Project of Universities Stationed in Hebei in collaboration with Shijiazhuang (241200073 A).

Data availability

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval

This study did not require ethics approval.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Artificial intelligence

The authors declare that they have not use AI-generated work in this manuscript.

Competing interests

The authors declare no competing interests.