Effects of Fecal Microbiota Transplantation on Intestinal Microbial Characteristics and Clinical Phenotypes in Patients with Parkinson's Disease

Summary

This exploratory study of six patients with Parkinson’s disease who underwent fecal microbiota transplantation observed temporal changes in gut microbial diversity, taxonomic composition, and predicted functional profiles, alongside improvements in motor and non-motor symptoms. These findings provide preliminary evidence supporting further investigation of microbiota-based interventions.

Abstract

Alterations in the gut microbiota have been associated with Parkinson’s disease (PD), but longitudinal microbial changes after fecal microbiota transplantation (FMT) and their clinical associations remain poorly understood. This single-center retrospective observational study included 6 patients with PD, stratified into high- and low-severity subgroups based on disease duration (>6 years vs ≤6 years). Thirty-six fecal samples were collected before FMT and monthly for five months afterward. Microbial diversity, community structure, taxonomic composition, and predicted functional profiles were assessed using 16S ribosomal RNA gene sequencing. Analyses included alpha and beta diversity, taxonomic abundance, linear discriminant analysis effect size, Tax4Fun2-based functional prediction, and Spearman rank correlations between microbial features and clinical indicators. Descriptive analyses indicated differences in microbial richness, diversity, community structure, and predicted functions between severity subgroups and across post-FMT time points. At baseline, the low-severity subgroup had greater microbial richness and diversity than the high-severity subgroup, with relatively higher abundances of taxa including Bifidobacterium and Lactobacillus. One month after FMT, richness and diversity increased from baseline in the high-severity subgroup, accompanied by changes in taxonomic composition. Both subgroups showed time-associated variation in microbial diversity and predicted Kyoto Encyclopedia of Genes and Genomes pathway enrichment after FMT. Predicted functions included carbohydrate and amino acid metabolism, secondary metabolite biosynthesis, membrane transport, and signal transduction. Several operational taxonomic units correlated with indicators of motor impairment, constipation, sleep quality, functional status, and neuropsychiatric symptoms. FMT was therefore associated with longitudinal changes in gut microbial diversity, composition, and predicted functions, and specific microbial features were associated with motor and non-motor indicators. Given the small retrospective cohort, these findings are preliminary and warrant confirmation in larger controlled studies. Future studies should determine whether these microbial alterations are reproducible, persist beyond five months, reflect donor engraftment, and correspond to measurable clinical improvement after transplantation in PD.

Introduction

Parkinson’s disease (PD) is a common progressive neurodegenerative disorder characterized by bradykinesia, rigidity, resting tremor, and postural instability¹. While its pathogenesis remains incompletely elucidated, multiple factors, including neuroinflammation, environmental exposures, and genetic susceptibility, have been implicated in disease onset and progression². Epidemiological evidence indicates that PD is among the most common neurological disorders in older adults and is associated with a substantial burden of disability, with prevalence increasing with age³. Current treatments, particularly dopamine replacement therapy, can relieve motor symptoms to some extent, but their effects on disease progression and many non-motor symptoms remain limited, and treatment responses vary across individuals⁴. Therefore, further investigation of PD-related biological changes and potential therapeutic targets remains clinically important.

The gut microbial community plays an important role in maintaining intestinal barrier integrity, regulating immune responses, and supporting metabolic homeostasis⁵. Fecal microbiota transplantation (FMT) involves the transfer of fecal microbiota from a healthy donor to a recipient to restore gut microbial balance^6,7. Multiple animal studies have suggested that gut microbiota transplantation can influence neurobehavioral phenotypes^8,9. Given that microbiota from patients with depression or anxiety can induce corresponding behavioral changes in mice, microbiota from healthy donors might improve behavioral outcomes¹⁰. In addition, preliminary clinical observations have reported symptom improvement after FMT in some patients with PD and autism, suggesting that microbiota-based interventions may have potential relevance in neurological and neuropsychiatric disorders^11,12. Nevertheless, the mechanisms underlying these effects remain unclear, and clinical studies are required to establish the safety and efficacy of FMT in PD.

Although previous studies have suggested a relationship between the gut microbiota and PD, the longitudinal changes in gut microbial composition following FMT and their associations with clinical characteristics remain poorly understood. It remains unclear whether patients with different clinical severities show distinct microbial responses after FMT.

In this single-center retrospective observational study, we reviewed patients with PD who underwent FMT in routine clinical practice. Clinical scores before and after FMT were analyzed, and available fecal samples were further analyzed using 16S rRNA gene sequencing to explore gut microbial diversity, taxonomic composition, predicted functional profiles, and correlations with clinical indicators. This study aimed to characterize clinical outcomes and gut microbiota profiles after FMT in PD and to provide preliminary evidence for future prospective controlled studies.

Protocol

The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of The Second Affiliated Hospital of Nanjing Medical University (approval number: 2021-KY-100-01). Written informed consent was obtained from all participants.

This single-center retrospective observational study was conducted at the Department of Neurology, The Second Affiliated Hospital of Nanjing Medical University. From January 2022 to December 2023, a total of 32 patients with primary PD who met the Diagnostic Criteria for Parkinson’s Disease (2020 edition) were included. Ultimately, six patients who met strict selection criteria and completed all study procedures were included in subsequent analysis. Details of screening, exclusion, and follow-up are shown in Supplementary Figure 1.

Fecal sample collection and microbial community analysis

Fecal microbiota was obtained from healthy volunteers. The donor screening process was strictly conducted in accordance with the Chinese Expert Consensus on the Clinical Application of Fecal Microbiota Transplantation. Donor inclusion criteria: Age 18–40 years, with no history of chronic diseases; No history of gastrointestinal diseases, metabolic disorders, autoimmune diseases, or psychiatric disorders; No use of antibiotics, probiotics, immunosuppressants, or proton pump inhibitors in the past 3 months; No travel to epidemic areas, blood transfusions, or surgical history in the past 6 months. Blood and fecal screening to exclude infectious diseases and pathogens includes: Human immunodeficiency virus (HIV), Hepatitis B virus (HBV), Hepatitis C virus (HCV), Treponema pallidum, Clostridium difficile, Salmonella, Shigella, Campylobacter, pathogenic Escherichia coli, enteric viruses, and parasites. The volume of a single valid stool sample should be at least 50 g. Using an automated fecal microbiota purification system (see Table of Materials), 500 mL of sterile saline was added to every 100 g of fresh stool, three rounds of standardized centrifugation and washing (700 × g, 3 min), and immediately inoculated after preparing a suspension with sterile saline in a 1:2 volume ratio. Subjects received a single injection of 150 mL of suspension containing approximately 5 × 10^13 viable bacteria¹³.

All recipients received FMT via colonoscopy, with the scope advanced to the cecum and ascending colon; the suspension was infused slowly and evenly through the biopsy channel to promote preferential colonization of the right colon. After infusion, patients remained in the right lateral decubitus position for at least 2 h. Preprocedurally, a low-residue diet was followed for 3 days, then a full liquid diet for 24 h before the procedure, with fasting from solids for 6 h and from liquids for 2 h prior to the infusion. On the evening before transplantation, a standard whole-bowel cleansing with polyethylene glycol electrolyte solution was performed until clear, stool-free effluent was obtained. Medication restrictions included withholding systemic antibiotics for at least 72 h before FMT and avoiding them for 1 week afterward. Proton pump inhibitors (PPIs) are discontinued 24–48 h before the procedure, and laxatives and probiotics are suspended on the day of transplantation.

During the procedure, heart rate, blood pressure, and oxygen saturation were continuously monitored, and any symptoms, such as abdominal distension, pain, or nausea, were observed; the infusion was immediately stopped if severe abdominal pain or hypotension occurred. For the first 7 days after transplantation, daily follow-up was conducted to record bowel movement frequency, stool consistency, and all adverse events, including abdominal pain, diarrhea, fever, and hematochezia. Long-term follow-up extended to 5 months, with scheduled stool sampling for gut microbiota analysis and clinical assessments using the Unified Parkinson’s Disease Rating Scale (UPDRS) and Wexner Constipation Score, alongside documentation of symptom changes and concomitant medication use, to comprehensively evaluate both efficacy and safety.

Fecal samples were collected at six time points: before FMT and at 1, 2, 3, 4, and 5 months after FMT. All participants received standardized instructions for stool sample collection. Approximately 10–20 g of mid-portion stool was collected using sterile forceps and placed into sterile collection containers. Samples were transported to the laboratory within 2–3 h after collection and aliquoted into sterile 1.5 mL microcentrifuge tubes (see Table of Materials) by trained researchers under aseptic conditions. All aliquots were immediately stored at −80 °C until further analysis.

After fecal sample collection, all specimens were transported on dry ice to a commercial sequencing provider (see Table of Materials) for 16S rRNA gene sequencing. Library preparation, sequencing, and initial quality control were performed according to the provider’s standard protocol. Genomic DNA was extracted from each fecal sample, and the V3–V4 hypervariable regions of the bacterial 16S rRNA gene were amplified by polymerase chain reaction (PCR) using primers 341F and 805R. Sequencing was performed on a benchtop short-read sequencing platform (see Table of Materials) using 2 × 300 bp paired-end chemistry to characterize the gut microbial profiles of patients before FMT and at different post-FMT follow-up time points. After sequencing, raw reads were processed to obtain high-quality sequences for downstream microbial analysis. Taxonomic classification was performed using a commercial 16S taxonomic assignment package and its associated microbial identification database (see Table of Materials). The resulting taxonomic profiles were used for subsequent analyses of microbial diversity, community composition, differential abundance, and predicted microbial function.

Clinical assessments and analytic labels

Patients with PD were stratified into two severity groups according to disease duration: a high-severity group (>6 years, designated “_H”) and a low-severity group (≤6 years, designated “_L”). Fecal samples collected before FMT were defined as baseline samples and labeled as FB_H or FB_L according to the corresponding severity group. Samples collected at 1, 2, 3, 4, and 5 months after FMT were labeled as FP1M_H, FP2M_H, FP3M_H, FP4M_H, and FP5M_H for the high-severity group, and FP1M_L, FP2M_L, FP3M_L, FP4M_L, and FP5M_L for the low-severity group. The assessed clinical parameters included the Wexner constipation score, Non-Motor Symptoms Scale (NMSS), Non-Motor Symptoms Questionnaire (NMSQ), Neuropsychiatric Inventory (NPI), Hoehn–Yahr stage, Activities of Daily Living scale (ADL), Pittsburgh Sleep Quality Index (PSQI), the original (UPDRS I–VI), and total UPDRS score. A combination of descriptive and inferential statistical methods was used to analyze the changes in clinical indicators of PD patients in the high-severity group (_H group) and the low-severity group (_L group) at six time points. To preliminarily explore group differences across all time points, the distributions of indicators between the _H and _L groups were compared at each time point using the Wilcoxon rank-sum test, and a more rigorous Linear Mixed Model (LMM) with patient-specific random intercepts was applied. Due to the limited sample size, the results of the mixed-effects model are provided only as exploratory references. The primary endpoint of this exploratory study was the change in gut microbial diversity and composition from baseline to post-fecal microbiota transplantation time points, with clinical changes evaluated as secondary endpoints.

OTU clustering, diversity analysis, and taxonomic abundance

Operational taxonomic units (OTUs) were used to summarize microbial community composition, and sequences with ≥97% similarity were clustered into the same OTU. Sequencing depth was evaluated using Shannon-based rarefaction curves, with curve plateaus indicating generally sufficient coverage for downstream diversity analyses.

Alpha diversity was assessed using the Chao1 and abundance-based coverage estimator (ACE) indices for microbial richness, and the Shannon and Simpson indices for overall diversity and evenness. Beta diversity was evaluated using weighted and unweighted UniFrac distances, followed by principal coordinate analysis (PCoA) to visualize differences in microbial community composition among groups. Between-group differences in alpha and beta diversity were assessed using non-parametric tests and Permutational Multivariate Analysis of Variance (PERMANOVA), respectively, as described in the statistical analysis section.

Taxonomic composition was summarized at the phylum, class, order, family, and genus levels based on 16S rRNA gene sequencing data. Using a community-ecology analysis package for R (see Table of Materials), identify and visualize the top 10 most abundant taxa at each taxonomic level. Taxonomic profiles were displayed using bar plots, and selected taxa were further compared among groups using box plots.

Differential abundance analysis of microbial taxa

Differentially abundant taxa among predefined groups were identified using linear discriminant analysis effect size (LEfSe). LEfSe integrates the Kruskal–Wallis test, Wilcoxon test, and linear discriminant analysis (LDA) to identify taxa with both statistical significance and discriminatory value. Taxa with P < 0.05 and an LDA score (log₁₀) > 3 were considered differentially abundant. The relative abundances of discriminative taxa were visualized using taxonomic bar plots.

Functional prediction and clinical correlation analysis

Predicted functional profiles of the gut microbiota were inferred from 16S rRNA gene sequencing data using Tax4Fun2 and annotated according to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Differentially enriched predicted pathways were identified using LEfSe, with P < 0.05 and an LDA score (log₁₀) > 3 as thresholds. Because these functional profiles were inferred rather than directly measured by metagenomic or metabolomic approaches, they were interpreted as predicted functional differences. Spearman’s rank correlation analysis was performed to explore associations between microbial features and clinical indicators.

Statistical analysis

Statistical analyses were performed using R software (see Table of Materials). Clinical variables were summarized as medians and ranges (minimum–maximum). Alpha diversity indices were compared between groups using the Wilcoxon rank-sum test, while beta diversity was assessed via PERMANOVA with 999 permutations. Differentially abundant taxa and predicted functional pathways were identified using LEfSe, with P < 0.05 and LDA score (log₁₀) > 3 as the significance thresholds. For longitudinal differential-abundance analysis, a mixed-effects model with patient-specific random intercepts was applied, followed by post-hoc pairwise comparisons with Benjamini-Hochberg false discovery rate correction. Spearman's rank correlation analysis was used to evaluate associations between gut microbial features and clinical indicators, with Benjamini–Hochberg false discovery rate correction applied to correlation P values. A two-sided P < 0.05 was considered statistically significant. Multiple-comparison correction was applied to correlation and taxonomic abundance analyses. Given the small sample size and repeated longitudinal sampling design, the statistical analyses were considered exploratory, and the results were interpreted cautiously.

Results

Patient selection and baseline characteristics

Six patients with PD were included in the final analysis, including three males and three females. The median age was 69.0 years (range, 61–75), and the median body mass index was 22.6 kg/m². The cohort included two rigidity-dominant, two tremor-dominant, and two mixed clinical phenotypes. Baseline clinical characteristics are shown in Table 1.

Clinical characteristics differed between the high- and low-severity PD subgroups (Table 1). In this exploratory cohort, the total UPDRS score was higher in the FB_H group than in the FB_L group, consistent with the predefined severity classification, indicating more severe overall motor impairment. Differences were also observed in several motor and non-motor clinical parameters, including UPDRS subscales, PSQI, Wexner constipation score, ADL scale, and NMSS. These findings suggest that the predefined severity subgroups differed not only in motor disease stage but also in several non-motor and functional characteristics.

Individual patient trajectories were visualized using line charts for key clinical indicators (Figure 1). As illustrated, a consistent downward trend in NMSQ, NMSS, NPI, PSQI, UPDRS Total, and Wexner scores was observed in all six patients following FMT, regardless of initial disease severity. Notably, patients in the _H group exhibited steeper clinical improvement than those in the _L group, despite higher baseline scores. Clinical comparisons showed consistent differences between the high- and low-severity groups. Most time-point-specific comparisons were significant, all pooled clinical indicators differed between groups, and mixed-effects models confirmed lower symptom burden and better daily function in the low-severity group.

Adverse events in PD patients after FMT

Adverse events recorded after FMT were summarized in Table 2. The most frequently reported events were flatulence, diarrhea, aggravated constipation, and nausea/vomiting, each occurring in two patients. Abdominal pain was reported in one patient. No fever was recorded, and no patient discontinued FMT because of adverse events. These findings suggest that gastrointestinal symptoms were the most commonly documented adverse events in this small cohort; however, the safety results should be interpreted with caution given the limited sample size.

Sufficient diversity coverage and identification of group-specific OTUs

Rarefaction curve analysis showed that the Shannon index gradually approached a plateau as sequencing depth increased, suggesting that sequencing depth was generally sufficient for assessing microbial diversity across samples (Figure 2A). A total of 80 OTUs were identified across all samples. This relatively low number of OTUs is consistent with the sequencing depth and the limited sample size in this exploratory study. Venn diagram analysis identified both shared and unique OTUs among the 12 groups (Figure 2B). A total of 80 OTUs were detected, including 53 OTUs shared by all groups. The number of unique OTUs in each group was as follows: 2 in FB_H, 12 in FB_L, 4 in FP1M_H, 0 in FP1M_L, 1 in FP2M_H, 0 in FP2M_L, 4 in FP3M_H, 0 in FP3M_L, 0 in FP4M_H, 1 in FP4M_L, 1 in FP5M_H, and 2 in FP5M_L. Among these groups, FB_L had the highest number of unique OTUs, whereas FP1M_L, FP2M_L, FP3M_L, and FP4M_H showed no unique OTUs in this analysis. These findings suggest that the PD subgroups shared a core set of OTUs, while some groups exhibited limited group-specific OTU patterns.

Microbial diversity and predicted functional profiles differ between PD severity subgroups

Alpha diversity was higher in the FB_L group than in the FB_H group, as reflected by increased Chao1, ACE, Shannon, and Simpson indices (Figure 3A). UniFrac-based PCoA showed apparent separation between the two severity subgroups, suggesting differences in gut microbial community structure (Figure 3B). At the phylum level, FB_L showed relatively higher abundances of Firmicutes and Actinobacteria, whereas FB_H was characterized by higher abundances of Bacteroidetes and Proteobacteria (Figure 3C). At the genus level, Bacteroides, Megasphaera, and Escherichia were more abundant in FB_H, while Bifidobacterium and Lactobacillus were more abundant in FB_L (Figure 3D).

LEfSe analysis identified severity-associated taxa that were differentially abundant. Bacteroidetes/Bacteroides-related taxa and Proteobacteria-related taxa were enriched in FB_H, whereas Bifidobacterium-, Actinobacteria-, Firmicutes-, Clostridia-, and Parabacteroides-related taxa were enriched in FB_L (Figure 3E). Predicted KEGG pathway analysis also showed distinct functional profiles between groups. FB_H was mainly enriched in predicted pathways related to global metabolism, carbohydrate and energy metabolism, signal transduction, two-component systems, and lipopolysaccharide biosynthesis. In contrast, FB_L was enriched in predicted pathways related to secondary metabolite biosynthesis, amino acid metabolism, nucleotide metabolism, replication and repair, quorum sensing, membrane transport, and ABC transporters (Figure 3F). These findings indicate that PD severity subgroups differed in microbial diversity, taxonomic composition, and predicted functional potential.

One month after FMT, the FP1M_H subgroup showed higher microbial diversity and distinct predicted functional profiles

Compared with FB_H, FP1M_H showed higher alpha diversity, as indicated by increased Chao1, ACE, Shannon, and Simpson indices (Figure 4A). UniFrac-based PCoA showed apparent separation between FB_H and FP1M_H, suggesting a shift in gut microbial community structure one month after FMT (Figure 4B). At the phylum level, FP1M_H had relatively higher abundances of Firmicutes, Actinobacteria, and Verrucomicrobia, whereas FB_H showed higher abundances of Bacteroidetes and Proteobacteria (Figure 4C). At the genus level, Bifidobacterium, Akkermansia, Enterococcus, and Lactobacillus were more abundant in FP1M_H, while Bacteroides, Megasphaera, and Escherichia were more abundant in FB_H (Figure 4D).

LEfSe analysis identified group-associated differentially abundant taxa. Bacteroidetes/Bacteroides-related taxa were enriched in FB_H, whereas Firmicutes-, Actinobacteria-, Clostridia-, Bifidobacterium-, and Acidaminococcus-related taxa were enriched in FP1M_H (Figure 4E). Predicted KEGG pathway analysis showed that FB_H was enriched in pathways related to signal transduction, two-component systems, carbohydrate metabolism, lipid metabolism, energy metabolism, and cofactor/vitamin metabolism. In contrast, FP1M_H was enriched in predicted pathways related to amino acid metabolism, nucleotide metabolism, starch and sucrose metabolism, replication and repair, translation, secondary metabolite biosynthesis, quorum sensing, ABC transporters, and membrane transport (Figure 4F). These findings suggest that, in the high-severity subgroup, FMT was associated with increased microbial diversity and shifts in taxonomic composition and predicted functional potential.

Microbial richness, taxonomic composition, and predicted functional profiles differed between the FB_L and FP1M_L groups

Compared with FB_L, FP1M_L showed higher microbial richness, as indicated by increased Chao1 and ACE indices, whereas Shannon and Simpson indices did not differ significantly between groups (Figure 5A). UniFrac-based PCoA showed apparent separation between FB_L and FP1M_L, suggesting differences in microbial community composition after FMT (Figure 5B). At the phylum level, FP1M_L showed relatively higher abundances of Actinobacteria, Verrucomicrobia, Synergistetes, and Euryarchaeota, whereas Firmicutes and Bacteroidetes were more abundant in FB_L (Figure 5C). At the genus level, Bifidobacterium, Akkermansia, and Parabacteroides were more abundant in FP1M_L, while Bacteroides and Lactobacillus were more abundant in FB_L (Figure 5D).

LEfSe analysis showed enrichment of Firmicutes-, Clostridia-, and Ruminococcaceae-related taxa in FB_L, whereas Bacteroidetes-, Bacteroidales-, Parabacteroides-, Porphyromonadaceae-, Synergistetes-, Alistipes-, and Rikenellaceae-related taxa were enriched in FP1M_L (Figure 5E). Predicted KEGG pathway analysis indicated that FB_L was enriched in pathways related to signal transduction, two-component systems, carbohydrate metabolism, lipid metabolism, amino sugar and nucleotide sugar metabolism, starch and sucrose metabolism, and the phosphotransferase system. FP1M_L was enriched in predicted pathways related to global metabolic pathways, amino acid metabolism, translation, biosynthesis of amino acids, secondary metabolite biosynthesis, and genetic information processing (Figure 5F). These results indicate that, in the low-severity subgroup, FMT was associated mainly with increased microbial richness and changes in taxonomic composition and predicted functional potential.

Gut microbial diversity, taxonomic composition, and predicted functional profiles varied among high-severity PD subgroups at different post-FMT time points

Alpha diversity analysis among the five high-severity PD subgroups (FP1M_H to FP5M_H) showed variation in microbial richness and diversity across time points (Figure 6A). Among these subgroups, FP2M_H showed the lowest Chao1, ACE, Shannon, and Simpson indices, whereas FP1M_H, FP3M_H, FP4M_H, and FP5M_H showed relatively higher values. Beta diversity analysis using weighted and unweighted UniFrac distances revealed separation among the five subgroups in the PCoA plots, indicating differences in gut microbial community composition across post-FMT time points (Figure 6B). At the phylum level, the top 10 most abundant taxa are shown in Figure 6C. Bacteroidetes appeared more abundant in FP3M_H, FP4M_H, and FP5M_H, whereas Firmicutes and Actinobacteria were more abundant in FP1M_H. Verrucomicrobia appeared relatively enriched in FP2M_H. At the genus level, Bacteroides was relatively more abundant in FP3M_H, FP4M_H, and FP5M_H, whereas Akkermansia and Escherichia were more abundant in FP2M_H. Bifidobacterium was relatively enriched in FP1M_H (Figure 6D). LEfSe analysis identified distinct differentially abundant taxa across several high-severity subgroups (Figure 6E), including Firmicutes-, Actinobacteria-, Bifidobacterium-, and Clostridia-related taxa in FP1M_H; Proteobacteria- and Enterobacteriaceae-related taxa in FP2M_H; Bacteroidetes- and Bacteroides-related taxa in FP3M_H; and Synergistetes-related taxa in FP4M_H. Predicted functional profiling based on KEGG pathway annotation showed subgroup-specific enrichment patterns (Figure 6F). FP1M_H showed predicted enrichment in starch and sucrose metabolism and biosynthesis of other secondary metabolites, FP2M_H showed predicted enrichment in metabolism of cofactors and vitamins and energy metabolism, and FP5M_H showed predicted enrichment in signal transduction and two-component systems. These findings suggest that the high-severity PD subgroups showed temporal variation in microbial diversity, taxonomic composition, and predicted functional profiles after FMT.

Gut microbial diversity, taxonomic composition, and predicted functional profiles varied among low-severity PD subgroups at different post-FMT time points

Alpha diversity analysis among the five low-severity PD subgroups (FP1M_L to FP5M_L) showed variation in microbial richness and diversity across time points (Figure 7A). FP2M_L showed relatively lower Chao1 and ACE indices, whereas FP4M_L and FP5M_L showed relatively higher Shannon and Simpson indices, indicating variation in richness and evenness among the low-severity subgroups. Beta diversity analysis using weighted and unweighted UniFrac distances revealed separation among the five subgroups in the PCoA plots, indicating differences in gut microbial community composition across post-FMT time points (Figure 7B). At the phylum level, the top 10 most abundant taxa are shown in Figure 7C. Bacteroidetes and Firmicutes were the dominant phyla across groups, while Actinobacteria and Verrucomicrobia appeared relatively enriched in FP1M_L, and Firmicutes appeared relatively more abundant in FP3M_L. At the genus level, Bifidobacterium and Akkermansia were relatively more abundant in FP1M_L, whereas Bacteroides was relatively more abundant in FP2M_L (Figure 7D). LEfSe analysis identified subgroup-associated differentially abundant taxa across several low-severity subgroups (Figure 7E), including Ruminococcaceae-related taxa in FP1M_L, Firmicutes- and Clostridiales-related taxa in FP2M_L, Synergistetes-related taxa in FP3M_L, and Bacteroidetes-, Parabacteroides-, Actinobacteria-, and Bifidobacterium-related taxa in FP4M_L. Predicted functional profiling based on KEGG pathway annotation showed subgroup-specific enrichment patterns (Figure 7F). FP1M_L showed predicted enrichment in biosynthesis of secondary metabolites, FP2M_L showed predicted enrichment in pathways related to signal transduction, two-component systems, environmental information processing, cellular community-prokaryotes, and carbohydrate metabolism, FP3M_L showed predicted enrichment in membrane transport, quorum sensing, replication and repair, and the phosphotransferase system, FP4M_L showed predicted enrichment in genetic information processing, and FP5M_L showed predicted enrichment in metabolism and metabolic pathways. These findings suggest that the low-severity PD subgroups showed temporal variation in microbial diversity, taxonomic composition, and predicted functional profiles after FMT.

Associations between specific OTUs and clinical indicators in PD

To explore the associations between gut microbial alterations and clinical features in PD, the top 50 OTU-clinical parameter correlations were visualized after applying multiple-comparison correction, corresponding to 23 unique OTUs (Figure 8). Several OTUs showed similar correlation patterns across multiple clinical indicators. For example, OTU165, OTU67, OTU230, OTU154, OTU72, and several other OTUs showed positive correlations with Wexner score, Hoehn–Yahr stage, PSQI, and multiple UPDRS-related measures, while showing negative correlations with ADL scale and NPI. In contrast, OTU10 (Parabacteroides) and OTU45 (Fusobacterium) showed an opposite correlation pattern for several clinical variables. OTU165, which was annotated at the phylum level as Firmicutes, showed positive correlations with constipation severity, disease stage, sleep disturbance, and several UPDRS components, while showing negative correlations with ADL scale and NPI. Similarly, OTU67 (Porphyromonadaceae) was also correlated with several motor and non-motor clinical indicators. Overall, these findings suggest that specific OTUs were associated with multiple clinical measures in PD, although these correlations should be interpreted with caution and regarded as hypothesis-generating rather than as evidence of causality.

DATA AVAILABILITY:

Supplementary materials for this study are available at DOI: https://doi.org/10.5281/zenodo.21620775.

Figure 1: Longitudinal changes in clinical scores after fecal microbiota transplantation in patients with PD. Line plots show changes in NMSQ, NMSS, NPI, PSQI, UPDRS total, and Wexner constipation scores from baseline before FMT (FB) to 1, 2, 3, 4, and 5 months after FMT (FP1M–FP5M) in the high-severity group (H) and low-severity group (L). Data points represent group-level summary values, and error bars indicate the corresponding variability/range. Please click here to view a larger version of this figure.

Figure 2: Rarefaction curves and OTU comparison among groups. (A) Shannon rarefaction curves showing the relationship between sequencing depth and microbial diversity. The x-axis represents sequencing depth, and the y-axis represents the Shannon diversity index. The gradual plateau of the curves suggests that the sequencing depth was generally sufficient for diversity assessment. (B) Venn diagram showing shared and unique OTUs among the 12 groups. The number in the center represents OTUs shared by all groups, while the numbers in the outer regions represent OTUs unique to each group. Please click here to view a larger version of this figure.

Figure 3: Microbial diversity, taxonomic composition, and predicted functional differences between the FB_H and FB_L groups. (A) Alpha diversity analysis based on Chao1, ACE, Shannon, and Simpson indices. (B) Principal coordinate analysis (PCoA) based on weighted and unweighted UniFrac distances. (C,D) Relative abundance of the top 10 gut microbial taxa at the phylum (C) and genus (D) levels, shown by stacked bar plots and box plots. (E) LEfSe analysis showing differentially abundant taxa between the FB_H and FB_L groups. (F) LEfSe analysis of predicted KEGG functional pathways. Positive and negative LDA scores indicate pathways enriched in different groups. *P < 0.05, **P < 0.01, ***P < 0.001. Please click here to view a larger version of this figure.

Figure 4: Microbial diversity, taxonomic composition, and predicted functional differences between the FP1M_H and FB_H groups. (A) Alpha diversity analysis based on Chao1, ACE, Shannon, and Simpson indices. (B) Principal coordinate analysis (PCoA) based on weighted and unweighted UniFrac distances. (C,D) Relative abundance of the top 10 gut microbial taxa at the phylum (C) and genus (D) levels, shown by stacked bar plots and box plots. (E) LEfSe analysis showing differentially abundant taxa between the FP1M_H and FB_H groups. (F) LEfSe analysis of predicted KEGG functional pathways. Positive and negative LDA scores indicate pathways enriched in different groups. *P < 0.05, **P < 0.01, ***P < 0.001. Please click here to view a larger version of this figure.

Figure 5: Microbial and functional differences between the FB_L and FP1M_L groups. (A) Alpha diversity analysis of the FB_L and FP1M_L groups. (B) Beta diversity analysis of the FB_L and FP1M_L groups. (C,D) Relative abundance of gut microbial taxa in the FB_L and FP1M_L groups at the phylum (C) and genus (D) levels. (E) Comparative analysis of species differences between the FB_L and FP1M_L groups. (F) Predicted microbial functions in the FB_L and FP1M_L groups. Please click here to view a larger version of this figure.

Figure 6: Microbial diversity, taxonomic composition, and predicted functional differences among the FP1M_H, FP2M_H, FP3M_H, FP4M_H, and FP5M_H groups. (A) Alpha diversity analysis based on Chao1, ACE, Shannon, and Simpson indices. (B) Principal coordinate analysis (PCoA) based on weighted and unweighted UniFrac distances. (C,D) Relative abundance of the top 10 gut microbial taxa at the phylum (C) and genus (D) levels, shown by stacked bar plots and box plots. (E) LEfSe analysis showing differentially abundant taxa among the five high-severity post-FMT subgroups. (F) LEfSe analysis of predicted KEGG functional pathways. Positive LDA scores indicate pathways enriched in the corresponding groups. Please click here to view a larger version of this figure.

Figure 7: Microbial diversity, taxonomic composition, and predicted functional differences among the FP1M_L, FP2M_L, FP3M_L, FP4M_L, and FP5M_L groups. (A) Alpha diversity analysis based on Chao1, ACE, Shannon, and Simpson indices. (B) Principal coordinate analysis (PCoA) based on weighted and unweighted UniFrac distances. (C,D) Relative abundance of the top 10 gut microbial taxa at the phylum (C) and genus (D) levels, shown by stacked bar plots and box plots. (E) LEfSe analysis showing differentially abundant taxa among the five low-severity post-FMT subgroups. (F) LEfSe analysis of predicted KEGG functional pathways. Positive LDA scores indicate pathways enriched in the corresponding groups. Please click here to view a larger version of this figure.

Figure 8: Spearman correlation analysis between OTUs and clinical indicators. Rows represent OTUs and columns represent clinical features. Each cell shows the Spearman correlation coefficient between a specific OTU and a clinical indicator. Red indicates a positive correlation, and blue indicates a negative correlation, with color intensity reflecting the strength of the association. Numeric values in the cells represent correlation coefficients. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. Please click here to view a larger version of this figure.

Patient	1	2	3	4	5	6
Gender (F/M)	M	F	M	F	M	F
Age (years)	65	72	68	75	61	70
BMI	23.5	21.8	24.2	20.5	22.1	23
PD subtype	Rigidity	Tremor	Hybrid	Rigidity	Tremor	Hybrid
Course of disease	5	8	6	10	3	7
Wexner score	12	15	10	18	8	13
UPDRS total	45	61	49	72	36	55
NMSS total	45	62	50	70	38	55
NMSQ	10	14	11	16	8	12
PSQI	8	11	9	13	6	10
ADL	90	75	85	70	95	65
NPI	5	9	6	12	3	7
Constipation duration	4	7	5	9	2	6
LEDD (mg/d)	600	800	700	900	400	750
Laxative use	No	Intermittent use	No	Long-term use	No	Intermittent use
Probiotic use	No	Yes	No	Yes	No	No
Antibiotic use	No	No	Yes	No	No	No
Hoehn–Yahr stage	2	3	2.5	3.5	1.5	3
Severity group	low	high	low	high	low	high

Table 1: Baseline demographic and clinical characteristics of patients with Parkinson’s disease, stratified by disease duration. Patients were classified into the high-duration group (>6 years; FB_H) or low-duration group (≤6 years; FB_L). Data should be presented as median (range) for continuous variables and number (%) for categorical variables, where applicable. Abbreviations: ADL, Activities of Daily Living; BMI, body mass index; NMSS, Non-Motor Symptoms Scale; NMSQ, Non-Motor Symptoms Questionnaire; NPI, Neuropsychiatric Inventory; PD, Parkinson’s disease; PSQI, Pittsburgh Sleep Quality Index; UPDRS, Unified Parkinson’s Disease Rating Scale.

Adverse events	Patients (n, %)
Flatulence	2 (33.3%)
Abdominal pain	1 (16.7%)
Diarrhea	2 (33.3%)
Aggravated constipation	2 (33.3%)
Fever	0
Nausea/vomiting	2 (33.3%)
Termination of FMT due to adverse events	0

Table 2: Adverse events following fecal microbiota transplantation in patients with Parkinson’s disease. The table presents the number and percentage of patients experiencing each adverse event during follow-up after fecal microbiota transplantation. No fever or treatment discontinuation due to adverse events was reported. Abbreviations: FMT, fecal microbiota transplantation; PD, Parkinson’s disease.

Supplementary Figure 1: Participant selection and patient flow. Flowchart of the sample screening process for Parkinson’s disease patients undergoing FMT and participating in 16S rRNA sequencing. n represents the number of participants.Please click here to download this file.

Discussion

FMT is a microbiota-based intervention that transfers fecal microbiota from a healthy donor to the recipient’s gut to restore intestinal microbial balance¹⁴. Growing evidence suggests that the gut microbiota plays an important role in maintaining nervous system homeostasis¹⁵. Evidence from the bidirectional gut-brain axis indicates that the gut microbiota may influence central nervous system inflammation, neurotransmitter metabolism, and behavioral responses in PD^16,17. Microbiota transplantation has therefore attracted attention as a potential intervention, although its clinical application in neuropsychiatric and neurodegenerative diseases remains exploratory¹⁸. This single-center retrospective observational study evaluated clinical outcomes, safety, and exploratory profiles of the gut microbiota in patients with PD who underwent fecal microbiota transplantation (FMT) in routine clinical practice. We observed differences in gut microbial richness, diversity, community structure, and predicted functional profiles across PD severity subgroups and post-FMT time points. These findings suggest that gut microbial profiles may vary with disease severity and may change dynamically after FMT intervention. Therefore, these results should be viewed solely as hypothesis-generating and require validation in larger, adequately powered studies.

Constipation is one of the most common and burdensome non-motor symptoms in PD¹⁹. Constipation may reduce quality of life, increase the need for laxatives, and interfere with the absorption of dopaminergic medications. In the present cohort, Wexner constipation scores decreased after FMT in both analytical subgroups, although the magnitude and persistence of improvement differed across follow-up time points. This pattern suggests a temporal association between FMT and improvement in bowel symptoms.

We also observed changes in several motor and non-motor clinical measures, including NMSS, PSQI, UPDRS III, and total UPDRS scores. These findings are noteworthy because PD is a multisystem disorder in which gastrointestinal dysfunction, sleep disturbance, autonomic symptoms, neuropsychiatric symptoms, and motor impairment frequently coexist. A potential explanation is that modulation of the gut microbiota may influence gastrointestinal motility, systemic inflammation, gut barrier integrity, microbial metabolites, and signaling along the microbiota–gut–brain axis. Nevertheless, motor and non-motor scores in PD are affected by multiple factors, including disease stage, dopaminergic therapy, sleep quality, mood, autonomic dysfunction, rehabilitation status, and daily activity. Therefore, the changes in UPDRS, NMSS, PSQI, and ADL-related measures should be interpreted as exploratory clinical observations.

Our findings are broadly consistent with previous studies linking gut microbiota to PD-related clinical heterogeneity. A recent study reported that fecal microbiota transplantation from healthy donors could improve motor and non-motor symptoms in patients with Parkinson’s disease, accompanied by changes in gut microbiota composition and functional pathways^13,20. Although impulse control disorders did not significantly affect α and β diversity in PD patients in one previous study, specific microbial taxa were enriched in patients with impulse control disorders and showed potential functional differences in pathways such as xenobiotic degradation and niacin metabolism²¹. Another study reported that PD patients with anxiety exhibited altered gut microbiota composition and β-diversity, further supporting the possible association between intestinal flora and PD-related non-motor symptoms²². In this study, microbial diversity and composition differed between high- and low-severity subgroups, and several post-FMT subgroups showed distinct microbial patterns. These results support the possibility that gut microbiota may be involved in the multidimensional clinical manifestations of PD, although causal relationships remain to be confirmed.

Several taxa identified in this study may be biologically relevant to PD-related changes in the gut microbiome. Certain bacteria may influence PD pathophysiology through immune and metabolic pathways²³. Firmicutes is one of the major Gram-positive bacterial phyla in the human gut and includes members involved in short-chain fatty acid production and intestinal barrier maintenance²⁴. In this study, OTU165, annotated at the phylum level as Firmicutes, was one of the OTUs most frequently associated with clinical variables. However, broad taxonomic groups such as Firmicutes contain functionally diverse members, and their biological interpretation should be approached with caution. Fecal metagenomic analysis has shown that Lactobacillus and its associated microbial epitopes from the Firmicutes phylum are enriched in PD patients and are associated with inflammatory markers and metabolic pathways such as propionate fermentation²⁵. Bifidobacterium is a bacterial genus with potential anti-inflammatory properties and may contribute to gut microbial regulation, whereas Bacteroides is an important commensal genus whose dysbiosis has been associated with impaired intestinal barrier function and increased inflammatory responses^26,27. In the present study, Bacteroides was relatively more abundant in FP2M_L, whereas Bifidobacterium was relatively more abundant in FP1M_L. In a rat model of PD, supplementation with Bifidobacterium breve Bif11 improved motor and cognitive impairment, reduced inflammatory factors and oxidative stress in the midbrain, and restored short-chain fatty acid content and intestinal barrier function²⁸. Previous studies have also reported altered Bacteroides abundance in PD patients, accompanied by changes in short-chain fatty acid precursor metabolism and associations with disease severity²⁹. These findings indicate that the taxa observed in this study may have potential biological significance, but higher-resolution taxonomic analysis and functional validation are needed.

Functional prediction analysis suggested that different PD subgroups showed distinct predicted KEGG pathway enrichment patterns. The FB_H group showed predicted enrichment in pathways such as carbohydrate metabolism, oxidative phosphorylation, microbial metabolism in diverse environments, and lipopolysaccharide-related processes, which may reflect altered microbial metabolic potential in the high-severity group. Oxidative phosphorylation is a major process by which mitochondria generate energy and maintain normal neuronal function³⁰. Disturbance of oxidative phosphorylation may lead to energy deficiency, oxidative stress, and neuroinflammation, which have been implicated in depression and PD³¹. Studies have suggested that the gut microbiome may influence mitochondrial dysfunction by regulating neurotransmitter synthesis, mitophagy, and oxidative stress homeostasis³². PD-associated genes may also impair mitochondrial complex I function, reduce ATP production, and exacerbate disease progression by disrupting the Akt pathway³³. In addition, lipopolysaccharide (LPS), an important component of Gram-negative bacteria, may trigger neuroinflammatory responses, and animal studies have shown that LPS exposure can activate microglia, promote the release of inflammatory factors, damage neural structures, and induce Parkinsonian-like behavioral changes³⁴. In this study, some low-severity and post-FMT subgroups also showed predicted enrichment in pathways related to amino acid metabolism, secondary metabolite biosynthesis, membrane transport, replication and repair, and short-chain fatty acid-related metabolism. However, because these pathway results were inferred from 16S rRNA sequencing rather than directly measured by metagenomic or metabolomic approaches, they should be regarded as exploratory.

This study also evaluated associations between gut microbial features and clinical indicators reflecting motor and non-motor symptoms in PD. The Wexner score reflects defecation function; the Hoehn–Yahr stage and UPDRS assess motor impairment severity; PSQI and ADL measure sleep quality and daily functional ability, respectively; and NPI assesses neuropsychiatric symptoms such as anxiety and depression^35,36,37. Several OTUs were associated with clinical scales related to constipation, sleep disturbance, motor impairment, functional status, and neuropsychiatric symptoms. Previous randomized controlled studies have reported that FMT may improve autonomic function and gastrointestinal symptoms in PD patients and increase gut microbiome complexity, suggesting that FMT may be a feasible adjunctive intervention³⁸. Another study also found that gut microbial diversity was altered in PD patients, with Firmicutes and Actinobacteria positively correlated with UPDRS III, NMSS, Wexner, and 39-item Parkinson’s Disease Questionnaire PDQ-39 scores, whereas Bacteroidetes showed negative correlations with these indicators³⁹. Nevertheless, the correlations observed in the present study do not establish causality, and the directionality of these associations should be interpreted cautiously.

Safety is an essential consideration in evaluating FMT for neurodegenerative diseases. In this cohort, reported adverse events were mainly gastrointestinal, including flatulence, diarrhea, aggravated constipation, nausea/vomiting, and abdominal pain. No fever was recorded, and no patient discontinued FMT due to adverse events. These findings suggest that FMT was generally tolerated in this small cohort. However, the safety assessment was based on retrospective clinical records and a limited number of patients. Rare but clinically important complications, including infection, aspiration, donor-derived pathogen transmission, metabolic complications, and delayed adverse events, cannot be excluded. Future prospective studies should include standardized definitions of adverse events, active safety surveillance, donor-screening documentation, and long-term follow-up.

Several limitations should be acknowledged, including the small sample size (n = 6), lack of a control group, repeated measures design with potential pseudoreplication and regression to the mean, and incomplete documentation of potential confounders such as diet, constipation treatments, medication adjustments, and antibiotic or probiotic use, although participants were advised to avoid the latter, undocumented exposures cannot be ruled out and may have influenced the gut microbiota. In addition, family history of Parkinson’s disease or other neurological disorders was not systematically recorded, precluding assessment of genetic or familial contributions. Functional predictions were based on 16S rRNA data rather than direct metagenomic or metabolomic measurements, and several OTUs were annotated only at broad taxonomic levels, limiting biological interpretability. These limitations warrant cautious interpretation of these findings.

Nonetheless, this exploratory study suggests that FMT was associated with changes in constipation-related symptoms, selected motor and non-motor clinical scores, gastrointestinal adverse events, and exploratory gut microbiota profiles in patients with PD. The findings provide preliminary real-world evidence and support further investigation of microbiota-directed interventions in PD with constipation. However, given the small sample size, retrospective design, repeated-measures structure, and the absence of a control group, the results should be interpreted with caution. Prospective controlled studies with standardized FMT protocols, predefined clinical endpoints, rigorous safety monitoring, and integrated multi-omics analyses are needed to clarify the clinical value, safety, and biological mechanisms of FMT in PD.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
MiSeq System	Illumina, Inc.	SY-410-1003	Benchtop short-read sequencing platform used for 16S rRNA V3–V4 amplicon sequencing with 2 × 300 bp paired-end chemistry.
Metagenome@KIN	World Fusion Co., Ltd.	v2.2.1; https://www.w-fusion.com/metagenome	Commercial software used for 16S rRNA taxonomic assignment and microbial-community analysis.
Microbial Identification Database (bacterial)	TechnoSuruga Laboratory Co., Ltd.	DB-BA, database v10.0; https://www.tecsrg.co.jp/services/enki/	Bacterial 16S rDNA reference database used for microbial identification.
341F primer	BGI Genomics Co., Ltd.	Custom oligonucleotide; no catalogue number	Forward primer sequence: 5′-CCTACGGGNGGCWGCAG-3′.
805R primer	BGI Genomics Co., Ltd.	Custom oligonucleotide; no catalogue number	Reverse primer sequence: 5′-GACTACHVGGGTATCTAATCC-3′.
16S rRNA gene sequencing service	BGI Genomics Co., Ltd. (Wuhan, China)	https://www.bgi.com/us/sequencing-services/rna-sequencing-solutions/16s18s-its-sequencing/	Commercial sequencing provider; performed library preparation, sequencing and initial quality control.
Eppendorf Safe-Lock microcentrifuge tubes, Biopur, 1.5 mL	Eppendorf SE	'0030121589	Sterile polypropylene tubes used to aliquot fecal samples for storage at −80 °C.
LEfSe	Segata Lab/bioBakery	v1.1.2; https://github.com/SegataLab/lefse	Linear discriminant analysis effect size software used to identify differentially abundant taxa and predicted pathways.
Tax4Fun2	Tax4Fun2 development team	v1.1.5; https://github.com/fjossandon/Tax4Fun2	R package used to infer Kyoto Encyclopedia of Genes and Genomes functional profiles from 16S rRNA data.
microeco R package	Chi Liu and contributors	v1.15.0; https://cran.r-project.org/package=microeco	Community-ecology analysis and visualization package for microbial abundance and diversity analyses.
R software	R Foundation for Statistical Computing	Version not reported; https://www.r-project.org/	Statistical computing environment used for clinical and microbiome analyses; the exact version should be confirmed by the authors.
GenFMTer automated fecal microbiota purification system	Nanjing FMT Medical Co., Ltd.	Model: GenFMTer; https://en.fmtmed.com/	Automated system used for standardized centrifugation, washing and preparation of the fecal microbiota suspension.