Recipients' native bacteria determine the outcome of FMT treatment in inflammatory bowel disease Authors: Di Zhao, Xinjun Wang, Ke Wang, Bo Yang, Huiyuan Zhu, Yue Xu, Chen Ye, Long Li, Xiaoqiong Lv, Shailan Zhou, Chunlian Ma, Xia Chen, Fang Yin, Yefei Zhu, Zhan Cao, Ning Li, Tao Zuo, Huanlong Qin, Qiyi Chen Published: 2025 ABSTRACT ABSTRACTBackgroundFecal microbiota transplantation (FMT) is a promising treatment for inflammatory bowel disease (IBD), achieving clinical response rate of ~50% for ulcerative colitis (UC), and Crohn's disease (CD). While prior research has emphasized donor selection and treatment protocols, the role of the patient's native intestinal microbiota in FMT outcomes remains underexplored.MethodsThis study analyzed a retrospective cohort of 96 IBD patients (45 CD, 51 UC) undergoing FMT, with 192 paired stool samples collected pre- and post-treatment, alongside 332 healthy donor samples from 18 donors. A prospective cohort of 45 IBD patients provided 45 baseline stool samples, and a validation cohort of 112 non-IBD patients contributed 224 paired samples. Retrospective cohort patients were monitored for 4 weeks to assess FMT responsiveness and 52 weeks for treatment effectiveness. Microbiome analysis identified enterotype-specific bacteria and native bacterial genera influence FMT outcomes. Random forest, permissivity, and mathematical models predicted treatment response, characterized microbiome remodeling, and defined microecological remission thresholds.ResultsThe FMT regimen was safe, with no serious adverse events reported. At week 4, the clinical response rates were 58.8% (26/45) for CD patients and 66.7% (34/51) for UC patients; by week 52, the remission rates were 82.4% (37/45) for CD patients and 84.4% (43/51) for UC patients. Microbiome analysis identified 54 bacterial genera linked to enterotype classification, 57 to UC response, and 93 to CD response. Notably, 38 high-frequency retentions of recipient native bacteria after FMT were predictive of FMT responsiveness. The permissivity model revealed a shift toward Bacteroidetes-dominated enterotypes in IBD patients post-FMT, which was validated in 112 non-IBD patients. The abundance ranges of recipients' native bacteria predictive of treatment responsewere determined by mathematical interpretation model.ConclusionThe patient's native microbiota significantly influences FMT efficacy in IBD, influencing microbiome remodeling and clinical outcomes, highlighting the importance of baseline microbial profiles in predicting FMT responsiveness and optimizing therapy.Keywords: Fecal microbiota transplantation, inflammatory bowel disease, native microbiota, clinical research Background Fecal microbiota transplantation (FMT) is increasingly recognized as a promising therapy for inducing remission in inflammatory bowel disease (IBD), particularly ulcerative colitis (UC) and Crohn's disease (CD). A meta-analysis of 13 randomized controlled trials (RCTs) involving 580 UC patients demonstrated that FMT significantly improves clinical remission rates compared to standard care (relative risk [RR] = 1.73, 95% confidence interval [CI] 1.41–2.12), with endoscopic remission also enhanced (RR = 1.74, 95% CI 1.24–2.44).1 In CD, a meta-analysis of 11 cohort studies and one RCT (228 patients) reported clinical remission in 57% of adults with active disease 2–4 weeks post-FMT (95% CI 49–64%, I² = 37%), alongside a significant reduction in Crohn's Disease Activity Index scores 4–8 weeks post-treatment (standardized mean difference = −0.66, 95% CI −1.12 to −0.20).2 However, achieving sustained endoscopic remission and ensuring long-term safety remain challenges.3-8 The efficacy of FMT hinges on the interplay between host and donor microbiota, which governs donor strain engraftment and therapeutic outcomes.9-12 Hosts with diverse microbiota or microbial profiles akin to the donor's exhibit better responses.13 Specific donor taxa, including Lachnospiraceae, Ruminococcus, Akkermansia muciniphila, and Bacteroides, are associated with positive outcomes, whereas high abundances of Fusobacterium, Sutterella, or Enterobacteriaceae in the host correlate with non-responsiveness.14 The host microbiota's resilience, nutrient competition, and adaptation to the mucosal redox environment further influence FMT success.9,15,16 Here, we analyzed microbiota dynamics in 96 IBD patients before and after FMT to elucidate their impact on treatment outcomes. Using a random forest model with native bacterial genera as predictors, we accurately forecasted responses in 45 prospectively enrolled IBD patients, which were validated over 14 weeks. By applying a “permissivity” model,17 we tracked microbiome trajectories driven by key gatekeeper microbes and assessed their role in microbiota plasticity and treatment response. These findings were corroborated in a cohort of 112 patients with gastrointestinal disorders, confirming the efficacy and safety of an optimized FMT protocol for complex IBD cases. Our results highlight the critical role of the pre-existing host microbiota in shaping FMT outcomes, offering insights for precision microbial therapies. Methods Study population Between January 2018 and January 2022, we enrolled 141 IBD patients who underwent FMT treatment at Shanghai Tenth People's Hospital. Of these, 96 patients completed FMT with follow-up assessments at 4 and 52 weeks. Additionally, 45 patients were prospectively recruited, completing FMT and a 14-week follow-up. Eighteen donors, each providing one fecal sample biweekly, contributed a total of 332 samples. This clinical trial was registered as ChiCTR2000039915 and NCT05703477 and was approved by the Shanghai Tenth People's Hospital Ethical Review Board. A part of this clinical trial has been published previously.3,8 The Mayo score assesses UC activity (0–12; stool frequency, rectal bleeding, endoscopy, and physician's global assessment), with ≥2-point reduction indicating response and ≤2 with no subscore >1 indicating remission. The patients' inclusion criteria were: (1) patients diagnosed with IBD based on typical clinical, endoscopic, and histological criteria; (2) active UC patients with a Mayo score ≥3 (Mayo endoscopic score ≥2) and active CD patients with Crohn Disease Activity Index (CDAI) score >220, and all included patients had poor or partial response to 5-aminosalicylic acid, corticosteroids, immunomodulators, and/or biologics; (3) patients who provided written informed consent for FMT; and (4) patients with complete collection of feces samples and took colonoscopy at baseline and after 4 weeks. Participants were excluded if they accompanied (1) intestinal conditions that restrict enteral nutrition intake, e.g. fibrotic stenosis, hematochezia, or an uncontrolled fistula; (2) systemic diseases, e.g. malignant cancers, serious liver and kidney disease, cardiopulmonary failure, severely immune compromised and infections that require systemic antibiotics; (3) CD lesions that could not be identified by surveillance colonoscopy. The baseline characteristics of the eligible participants, including age, sex, diagnosis, disease duration, history of medication, circulatory inflammatory mediators, Mayo score/CDAI, and fecal calprotectin, were recorded. Laboratory examinations (routine blood examination, C-reactive protein, erythrocyte sedimentation rate, hepatorenal function, etc.) and small bowel MR enterography or small bowel CT enterography were used to assess the condition of the patients before FMT. All patients with concomitant biological therapy received a maintenance dose. Stool samples were collected before and 28 ± 3 d after the initiation of FMT therapy. Clinical response at week 4 was defined as follows: (1) For UC patients, a partial Mayo score of ≥2 points and a reduction in the rectal bleeding score of ≥1 point at the same interval; (2) for CD patients, a decrease in the CDAI of ≥100 points from baseline. Clinical remission at week 52 was defined as follows: (1) For UC patients, a total Mayo score of ≤2 and no subscore of >1; (2) For CD patients, CDAI < 150, and no signs of active inflammation on ileocolonoscopy. Failure to meet the above criteria or conversion of treatment (e.g. shift of biologics therapy or surgery) was considered nonresponse. Major adverse outcomes (MAOs) were defined as (1) disease relapse, (2) IBD-related unwanted hospitalization, (3) IBD-related surgery, and (4) the need for initiation or dose escalation of systemic steroids, immunosuppressants, small molecules, or biological therapy. Donor recruitment The standardized donor screening program is currently based on the Chinese Expert Consensus FMT Guideline,3,8,18 which recommends the evaluation of donor screening across the following six dimensions: physiology, psychology, personal history, stability, persistence, and tolerance to dietary restriction. The donors should be completely healthy, without any other illness, especially gastrointestinal disease or motility disorders, and had not been hospitalized for at least 3 months before FMT donation or not received antibiotics or proton pump inhibitors for at least 6 months before FMT donation. Preparation and procedure of FMT To prepare for FMT, 100 g of donor feces were placed into a sterile container and mixed with 300 mL of saline. The mixture was homogenized and strained through 2.0 mm and 0.5 mm mesh filters to remove solids. Then, sterile glycerol is added to achieve a 10% final concentration, and the mixture is frozen at −20 °C, where it can be stored for 1–8 weeks. Before undergoing FMT, patients undergo 3 d of antibiotic pretreatment (0.5% vancomycin twice per day), and the bowel cleanse using polyethylene glycol 12–24 h in advance. For the procedure, the fecal suspension is thawed at 37 °C and must be administered within 30 min to ensure viability. Fresh-frozen healthy microbes from donor stool are administered through a nasojejunal tube placed in the proximal jejunum, with a transplant volume of 100 mL each day for 6 consecutive days. Patients typically undergo at least two cycles of treatment spaced 21–28 d apart. During the treatment period, the use of antibiotics is generally discouraged to avoid disrupting the newly introduced microbiota. Microbiome sequencing For IBD patients, the first copy of the feces sample was collected after diagnosis, and the remaining copies were collected at 28 ± 3 d after the first FMT. Microbial DNA was extracted from a 200 mg fecal sample using the QIAamp PowerFecal Pro DNA Kit (QIAGEN), which contains a bead-beating step. Briefly, 200 mg of fecal sample and 800 mL of lysis buffer were added to a beads-containing tube and vortex at maximum speed for 10 min. A total of 350 mL of the supernatant was centrifuged for 1 min at 15000 × g and used in the subsequent steps according to the kit instructions. The DNA is finally eluted in a 100 mL of elution buffer for downstream applications. DNA was amplified using primers targeting the V4 region of the 16S rRNA gene (515F 50-GTGYCAGCMGCCGCGGTA-30, 806R 50-GGACTACNVGGGTWTCTAAT-30). PCR was run in a VeritiTM 96-Well Thermal Cycler PCR system (Thermo Fisher Scientific) using the following program: 95 °C for 3 min, followed by 21 cycles of 95 °C for 30 s, 56 °C for 30 s, and 72 °C for 30 s, with a final extension at 72 °C for 5 min. The mixed amplicons were pooled, and sequencing was conducted at Shanghai Biotecan Pharmaceuticals Co., Ltd. (Shanghai, China) using an Illumina NovaSeq 6000 Sequencing system (Illumina, USA) according to the manufacturer's instructions. The sequences were assigned with 97% similarity (Greengenes database: http://greengenes.lbl.gov) into mother (v.1.39.5), and taxonomy was assigned to the Greengenes database for the comparisons. For alpha diversity, a “summary. single​​​” script was used to calculate the Shannon index via the Mothur software package. The co-occurrence network was established based on the MB distance matrix using the R package SPIEC-EASI. To identify taxa with differing relative abundances between the two groups, linear discriminant analysis (LDA) effect size (LEfSe) analysis was performed on a website (http://huttenhower.sph.harvard.edu/galaxy). The cutoff value was the absolute LDA score (log 10) >2.0 with a p < 0.05. Based on the prognostic-associated genera obtained by LEfSe analysis, more than 60% of the prognostic-associated genera colonization events (including consistent presence or absence before and after treatment) were linked to the recipient's native bacteria, calculated as the ratio of genera with cohort consistency >50% to the total number of prognostic-associated genera (Supplementary Figures 2 and 4). Enterotype identification was performed on the website (http://enterotypes.org/). The 16S rRNA sequencing data of all participants in the study have been uploaded to the National Center for Biotechnology Information platform, with the accession number PRJNA1269753 (https://dataview.ncbi.nlm.nih.gov/object/PRJNA1269753?reviewer=dbcqo7sa8aeocrcsmna1lmudfo). Calculating the gut microbiome index We used the gut microbiome index (GMI)19 to elucidate the alterations in intestinal microbiota homeostasis following FMT intervention. The collective abundance of remission-related microbes (ψMR,i) is defined as where RMR,i is the richness of the remission-related microbes (MR) in sample i, |MR| is the set size of these microbes, IMR is the index set, and nj,i is the relative abundance of bacterial j in sample i. For the unremission-related microbes (MU) in the same sample i, the collective abundance ψMU,i can be calculated in the same way. Next, the GMI was defined as follows: Identification of IBD high-frequency colonization native bacteria Microbiome abundance data were normalized to 1, and bacterial genera with a relative abundance >0.00001 were included in the analysis for native bacteria identification. The bacterial genera with relative abundance >0.00001 before and after FMT treatment were first identified and selected. Subsequently, the native bacteria were separately identified in 51 UC patients and 45 CD patients, with a coincidence rate of 50%. The intersection of the two sets was considered the IBD native bacteria. Finally, we specifically focused on high-frequency colonization of native bacteria, defined as those detected in at least 50% of IBD recipients before FMT and still present in at least 50% of patients after treatment. Spearman correlation analysis confirmed the association between 38 high-frequency colonization native bacteria and patients' clinical indicators. LEfSe analysis identified genera linked to FMT prognosis in four subgroups: UC_ETB, UC_ETF, CD_ETB, and CD_ETF. The impact of native bacteria on FMT prognosis was systematically evaluated using scoring. The enrichment of native bacteria was scored as follows: Remission_baseline (1 point), Remission_FMT (2 points), Donor (3 points), Unremission_baseline (−1 point), and Unremission_FMT (−2 points). A total score >3 was categorized as “Great”, and <−3 as “Bad”. Random forest machine learning classification model construction The random forest machine learning model was built by the EnrichVisBox platform (www.omicsolution.com/wkomics/passwd/randomforest/). For the construction of the prognostic evaluation model of the prospective cohort, the abundance data of 38 IBD high-frequency colonization native bacteria were used as variables. The “permissivity” model Enterotype analysis was conducted on 332 donor samples to identify characteristic bacteria of ET_B and ET_F using LEfSe analysis. Based on this, the microbiome data of IBD patients were categorized into three groups: ET_B, ET_F, and others. Then, bacterial genera unrelated to FMT prognosis in the enterotype groups were excluded and assigned to the “other” group. A binary coordinate system was defined, and the abscissa represents the relative abundance difference between ET_F and ET_B classes (σ = ∑ET_F − ∑ET_B), and the ordinate represents the relative abundance of other genera (Δ = 1 − ∑ET_F − ∑ET_B). Concerning the concept of “permissivity”,17 the microbiome plasticity (i.e. the degree of difficulty of the microbiota to reproduce the previous FMT events) under different proportions of native bacteria can be calculated. Its mathematical model is as follows: where vt⃑ represents the vector corresponding to each patient's microbiome trajectory during FMT treatment, that is, (σt+1−σt, Δt+1−Δt); v0⃑ represents the vector from the baseline of each IBD cases to the point of an arbitrary enterotype bacteria proportion, that is, (σ−σt, Δ−Δt). It should be emphasized that this original “permissivity” model assessed the similarity between two vectors, vt⃑, and v0⃑, using cosine similarity, which measures the angle between them. This approach captures the directional alignment of the vectors but overlooks their modulus, which reflects the extent of changes in a patient's microbiota trajectory. To address this limitation, we modified the “permissivity” formula to incorporate both the similarity and the modulus of the vectors. The revised “permissivity” is defined as the Euclidean norm of the vector difference vt⃑− v0⃑, representing the distance from the starting point of v0⃑ to the endpoint of vt⃑. This new measure accounts for both the directional relationship and the magnitude of the microbiota's trajectory shift. The mathematical expression is given below: Therefore, the larger the p-value, the easier it is to repeat the FMT repair event that occurred in the past, which means the higher the plasticity of the patient's microbiome. The explanatory model of FMT response rules We propose three hypotheses: (i) the direction of microbiota changes contributes to the FMT response; (ii) the closer the proportion of the native bacteria is to the previous cases, the higher the probability of the same event occurring; (iii) due to the presence of native bacteria, the greater the range of changes in the microbiota, the more difficult it is. Therefore, the eventuality of an FMT response case can be quantified by a mathematical model as follows: where vt+⃑(i) represents the vector corresponding to each remission patient's microbiome trajectory; vt−⃑(i) represents the nonremission patient's microbiome trajectory; and v0+⃑(i) and v0−⃑(j) represent the vectors from the baseline of each remission cases or nonremission case, respectively, to the point of arbitrary native bacteria proportion. Since remission cases and nonremission cases are independent events that do not affect each other, α and β can be approximately regarded as equal to 1 and −1, respectively. The E-value < 0 indicates the response to FMT treatment, and the smaller the value, the higher the eventuality. Statistical analysis SPSS 19.0 software was used for statistical analysis. ANOVA was employed to assess statistical differences in α diversity and the gut microbiome index across groups, PERMANOVA was used to evaluate statistical differences in β diversity, and the Wilcoxon rank-sum test was applied to compare statistical differences in the gut microbiome index of patients before and after treatment. Only p-value < 0.05 was considered statistically significant. The data were plotted using the online tool Chiplot (chiplot.online). The schematic diagrams were drawn using the online tools MedPeer (image.medpeer.cn) and Biorender (www.biorender.com). Results Treatment outcomes in IBD patients following FMT We followed 96 IBD patients (51 with UC, 45 with CD) for 52 weeks to assess FMT outcomes (Figure 1a,b; Supplementary Table 1). At week 4, the clinical response rates were 58.8% (26/45) for CD patients and 66.7% (34/51) for UC patients. By week 52, remission rates reached 82.4% (37/45) for CD patients and 84.4% (43/51) for UC patients (Figure 1c,d). Among 81 patients with malnutrition and confirmed nutritional risk, 61.7% (50/81) achieved a clinical response at week 4, with 66.7% (54/81) in remission by week 52. Adverse events occurred in 21.0% (17/81) of these patients, primarily nasopharyngeal discomfort post-tube placement (12.3%, 10/81) and abdominal distension (9.9%, 8/81). In 17 patients with incomplete intestinal obstruction, the response and remission rates were both 70.6% (12/17) at weeks 4 and 52, respectively, with adverse events in 17.6% (3/17), mainly increased diarrhea frequency (11.8%, 2/17) and nasopharyngeal discomfort (11.8%, 2/17). Among the five patients with intra-abdominal infection and intestinal fistula, 80% (4/5) achieved response and remission at weeks 4 and 52, with no adverse events. In 19 patients receiving concurrent biologic therapy, the response and remission rates were 57.9% (11/19) at week 4 and 63.2% (12/19) at week 52, with adverse events in 21.1% (4/19), primarily abdominal distension (50%, 2/4) and increased diarrhea (50%, 2/4). No serious adverse events (SAEs), such as IBD exacerbation, unplanned hospitalizations, or surgery, occurred within the treatment subgroups (Supplementary Table 1) (Supplementary Table 1). Therapeutic regimens, cohort composition, and follow-up results for FMT treating IBD. (a) FMT treatment scheme for IBD patients. FMT, fecal microbiota transplantation. (b) Cohort composition: Ninety-six IBD patients were included retrospectively, and 45 IBD patients were included prospectively. The follow-up time was 4 and 52 weeks. (c) Follow-up results of a retrospective cohort of 96 patients. The treatment response results are presented separately for patients with different subtypes, including UC and CD, with or without C. difficile, the Bacteroides enterotype, and the Firmicutes enterotype. (d) Colonoscopy results before and after treatment in symptom-remitted UC patients, symptom-unremitted UC patients, symptom-remitted CD patients, and symptom-unremitted CD patients. Each group shows the representative colonoscopy results of three patients, and the results before and after treatment corresponded up and down. Microbiota enterotype classification and native bacteria's role in FMT outcomes Using between-class analysis, we classified IBD patients and healthy donors into two enterotypes: Bacteroides-dominated (ET_B) and Firmicutes-dominated (ET_F), based on the HMP and MetaHIT datasets,20,21 (Supplementary Figure 1a; Supplementary Table 2). ET_B was enriched with 26 genera, including Bacteroidetes, designated ET_B microbes, whereas ET_F exhibited elevated levels of 28 genera, including Prevotella_9, designated as ET_F microbes (Supplementary Figure 1b–d). The α diversity of ET_F donors was significantly higher than that of ET_B donors, and the 4-week response rate of ET_F patients was higher than that of ET_B patients, highlighting enterotype-specific microbiome differences critical for the FMT response (Figure 1c and Supplementary Figure 1). Linear discriminant analysis effect size (LEfSe) identified 57 and 93 genera associated with FMT response in UC and CD patients, respectively (Supplementary Figure 2). The gut microbiome index (GMI)19 revealed no significant changes in microbiota homeostasis post-FMT (Supplementary Figure 3; Equations 1 and 2). During FMT-driven microbiota remodeling, 65% of the recipient microbiota maintained a consistent presence or absence of pre- and post-FMT (Supplementary Figure 4). Bacteria present before FMT persisted post-treatment, while those absent pre-FMT did not emerge post-FMT, which is consistent with prior studies.12,22 We defined these consistently present microbes as recipient native bacteria, which accounted for ~60% of FMT response-related genera. These native bacteria comprised 63% of the microbiota in ET_B UC, 59% in ET_F UC, 65% in ET_B CD, and 62% in ET_F CD subgroups, highlighting their dominance across enterotypes. Recipient native bacteria with >50% prevalence in IBD patients both pre- and post-FMT were defined as high-frequency native bacteria and were selected for further analysis. We identified 38 high-frequency recipient native bacteria in 96 IBD patients and assessed their associations with FMT outcomes (Figure 2a). Among these genera, Faecalibacterium, Roseburia, and Butyricoccus exhibited the strongest response to FMT treatment, whereas Alistipes demonstrated the most significant inhibitory effect. Using these bacteria's abundance data, a random forest model trained on the 96-patient cohort predicted outcomes in a validation cohort of 45 IBD patients with comparable baseline characteristics (Supplementary Tables 3,4, Supplementary Figure 5). After 14 weeks, the predicted remission group (n = 29) achieved a 79.3% remission rate (23/29), which was significantly higher than the 56.3% (9/16) in the non-remission group (Figure 2b). Confirming that recipient native bacteria composition predicts FMT prognosis. Gut native bacteria determine patients' FMT treatment response. (a) Thirty-eight native bacteria had a prevalence greater than 50% in the UC or CD cohort. The 14 genera marked with light green backgrounds were generally enriched in the donor or IBD remission groups. The four bacterial genera marked with a light red background were enriched only in the IBD non-remission groups. The genera marked with a light blue background cannot determine their contribution to FMT. (b) Forty-five IBD patients were prospectively recruited for FMT treatment, and their prognosis was predicted by a random forest model before treatment. After 14 weeks, we followed up on the patients' actual prognosis and presented the concordance rate of the predicted results. (c–j) Ternary diagrams showing microbiota trajectories in UC and CD patients by ET_B/ET_F enterotypes and FMT response. Vectors indicate pre- to post-FMT changes; the heatmap reflects permissivity. (k–n) Microbiota trajectory before and after treatment in 112 non-IBD patients treated with FMT. (o) Schematic showing a patient's baseline microbiota (dot: ET_B, ET_F, other recipient native bacteria proportions). The dashed lines indicate hypothetical shifts aligning with prior FMT cases. Red/black lines represent actual FMT response/nonresponse trajectories, with vector angles reflecting response probability. (p–q) Predicted FMT treatment response in IBD and non-IBD cohorts. The heatmap intensity values indicate the likelihood of FMT response based on the baseline microbiome composition. The values <0 predict a treatment response, while values >0 predict a nonresponse (r). Binary image shows the presence or absence of Actinomyces and Lactobacillales_unclassified in all IBD patients. (s) The scatter diagram presents the relative abundance of Escherichia–Shigella and Faecalibacterium in all IBD patients. Enterotype-specific native bacteria shape FMT outcomes The recipient microbiota was classified into ET_B (26 genera, including Bacteroides), ET_F (28 genera, including Prevotella_9), and non-enterotype genera via LEfSe analysis. High-frequency recipient native bacteria accounted for 87%–92% (IQR) of ET_B and 98–100% (IQR) of ET_F microbe relative abundance, underscoring their dominance in enterotype-specific microbiota (Supplementary Figure 2). Microbiota trajectories pre- and post-FMT were mapped using ternary diagrams, starting at 100% native bacteria. The recipient microbiota was classified into ET_B (Bacteroides-dominated), ET_F (Firmicutes-dominated), and other genera (unrelated to enterotype or FMT response) via LEfSe analysis (Supplementary Figure 1). Ternary diagrams plotted normalized abundances, with edges representing ∑ET_B, ∑ET_F, and other genera. For “permissivity” calculations, ternary coordinates were converted to binary coordinates (σ = ∑ET_F − ∑ET_B, Δ = 1 − ∑ET_F − ∑ET_B), with microbiota plasticity visualized as heatmap intensity for 96 IBD patients (Figure 2c–j). Each arrow vector is used to illustrate the shift from the recipient's pre-FMT microbiota composition to the post-FMT composition. Using the “permissivity” model,17 we quantified microbiota plasticity to assess recipient native bacteria's impact on FMT microbiota remodeling (Equations (3) and (4)). The ternary heatmap, generated by the “permissivity” model, displays color intensity representing recipient microbiota plasticity under varying proportions of ET_B, ET_F, and other native bacteria, with higher values indicating a greater likelihood of FMT-driven remodeling. Using the “permissivity” model, we found that the recipient microbiota exhibited highest plasticity when lacking ET_B or ET_F native bacteria, enabling maximal FMT-driven remodeling. High plasticity occurred with high ET_B and low ET_F proportions, followed by balanced ET_B/ET_F proportions (50:50), which showed lowest plasticity (Figure 2c–j). In ET_B UC patients, the pre- and post-FMT microbiota of remission cases occupied high-plasticity “hot spot” regions, while non-remission cases were in low-plasticity “freezing point” regions (Figure 2c,d, Supplementary Figure S6a,b). ET_F UC patients generally showed lower plasticity, but remission cases shifted to ET_B-like high-plasticity regions post-FMT (Figure 2e,f). Linking recipient native bacteria plasticity to FMT outcomes. Similar patterns were observed in CD patients with ET_B (Figure 2g,h, Supplementary Figure S6c, d) but not in ET_F CD patients, likely due to the limited sample size (Figure 2i,j). Similar to the UC findings, a validation cohort of 112 non-IBD gastrointestinal disease patients with ET_B-dominated recipient native bacteria showed high plasticity and a 78% FMT response rate, with remission cases shifting post-FMT to ET_B-like high-plasticity regions (Figure 2k–n, Supplementary Figure S6e,f). These findings highlight that high-frequency recipient native bacteria and their plasticity critically shape FMT efficacy across IBD and non-IBD cohorts. Modeling native bacteria response to FMT We developed a model to predict FMT remission probability using recipient enterotype native bacteria proportions (ET_B, ET_F, non-enterotype genera) from 96 IBD patient trajectories (Figure 2o and Equations (5)–(7)). Assumption 1 links directional shifts (e.g., to ET_B-like high-plasticity regions) to enhanced remission. Assumption 2 ties similarity to remission-associated compositions (e.g., high ET_B) to consistent outcomes. Assumption 3 posits that greater baseline deviations from known remission events reduce remission likelihood. The E-value (Equations (5)–(7)) quantifies deviations from remission/non-remission profiles, with lower values (<0) indicating a higher remission probability. Ternary diagrams revealed overlapping remission probability ranges for IBD and non-IBD patients, with high ET_B and low-to-absent ET_F native bacteria being critical for FMT response in both IBD (Figure 2p) and non-IBD (Figure 2q) patients. Our model validated the recipient native bacteria‒plasticity relationship, predicting FMT remission in 96 IBD patients (Figure 2o–q). To explore its ecological basis, we analyzed microbial interactions. Trends of Lactobacillales_unclassified and Faecalibacterium, both closely associated with IBD remission,23,24 In UC and CD remission cases, Actinomyces boosted Lactobacillales_unclassified (lactate-producing) prevalence to 91.2% and 93.3%, versus 36.4% and 56.2% without Actinomyces (Figure 2r). Conversely, Escherichia–Shigella and Faecalibacterium (SCFA-producing) abundances shifted oppositely (Figure 2s). These findings suggest that symbiotic or competitive interactions among native bacteria were formed, influencing recipient microbiota remodeling. Building on prior FMT experience, we aim to advance future FMT clinical practice toward greater precision. To this end, we investigated the abundance ranges of native bacteria associated with high expectations for FMT treatment response. We designated ∑ET_B and ∑ET_F as the X- and Y-axes, respectively, with the E-value as the Z-axis, and computed the spatial distribution of E-values for all ∑ET_B and ∑ET_F compositions using Equation (7). The E-value distributions in IBD and non-IBD cohorts showed consistent pattern, with low values (<0) at 80% ∑ET_B and 40% ∑ET_F, indicating higher FMT remission probability (Figure 3a,b). Compared with non-IBD patients, the lowest E-value in the IBD patient group was lower than that in the non-IBD group, suggesting that prognostic factors in IBD patients are more intricate. Additionally, the absence of a distinct 40% ∑ET_F trough indicated challenges in ET_F-driven remission. Using an 80% minimum E-value threshold, high FMT responsiveness in IBD patients occurred at 65.5–82.5% ∑ET_B and 0–3.9% ∑ET_F, with a minimal E-value of −1.35. In non-IBD patients, low ET_B/ET_F compositions (0–11.7% ∑ET_B, 26.1–45.3% ∑ET_F) predicted the highest remission probability, with a minimal E-value of −2.11. (a) Predicted remission probability for the IBD cohort. (b) Predicted remission probability for the non-IBD cohort. The heatmaps display E-values (Equation (7)) for recipient native bacteria compositions, with lower values (<0, darker) indicating higher FMT remission probability and higher values (>0, lighter) indicating a lower probability. Discussion Although traditional FMT has made significant strides in treating IBD, the approximately 50% remission rate underscores the need for more refined strategies to enhance efficacy. Our prior studies, involving screening of 8483 donors and 5-y follow-up of 8547 donors, explored factors influencing FMT success, including donor microbiota health, recipient pretransplantation preparation, formulation of the transplanted microbiota, content of live microbes, and FMT treatment duration.3,8,11,12,22,25 Through a decade of investigation, we improved the efficacy of FMT for gastrointestinal disorders, including but not limited to IBD, from 42.8% to 68.7%, while reducing the adverse event rate from 30.7% to 20.1%.3 Building on this, we introduced enteral nutrition access as an inclusion criterion for FMT, restoring intestinal function through integrated therapies to create a nutrient-rich gut environment conducive to colonization by healthy donor microbiota, further enhancing FMT response in IBD patients. This criterion has been incorporated into the Expert consensus on clinical application management of fecal microbiota transplantation (2022 edition)26 and the Consensus of Chinese experts on gut microbiota and fecal microbiota transplantation in inflammatory bowel disease (2025 edition).27 Based on our prior work, we implemented a comprehensive FMT protocol in 141 IBD patients, with 96 completing a 52-week follow-up and fecal microbiome analysis. This protocol achieved a clinical efficacy rate of 62.5% at week 4, with a remission rate of 83.3% by week 52. In this study, we pioneered the application of FMT to patients with complex complications, including intestinal strictures, partial intestinal obstruction, fistulas, and intra-abdominal infections, demonstrating safety and efficacy over 52 weeks. Optimizing enteral nutrition (achieving ≥50% of target intake) before FMT through surgical or pharmacological interventions improved patient tolerance and reduced adverse reactions. Notably, the enteral nutrition access strategy is critical, as it not only fosters a nutrient-rich gut environment conducive to donor strain colonization but also normalizes baseline intestinal function across patients with varying clinical characteristics. This enables the study of FMT response based on native gut microbiota composition, which is independent of the influence of intestinal function. To optimize FMT efficacy, we employed a random forest model to identify recipient-specific native microbial signatures predictive of treatment success, stratifying 29 of 45 patients into a high-efficacy subgroup with a 79.3% remission rate at week 14. This approach highlights the value of patient selection strategies based on inherent microbiota profiles. The influence of the native microbiota on FMT response can be attributed to the inherent stability of gut enterotypes. Due to the combined effects of diet, genetics, and environmental factors, the gut microbiome in Han Chinese populations is predominantly characterized by two microbial enterotypes: ET_B, which is dominated by Bacteroides, and ET_F, which is dominated by Prevotella_9 and characterized by a diverse range of Firmicutes, including Ruminococcus, Clostridiales, and Blautia.20,21 Bacteroides dominance is associated with diets rich in protein and animal fats, whereas Prevotella dominance is correlated with carbohydrate-rich diets. Critically, Bacteroides and Prevotella core taxa exhibit mutually exclusive relationship in the gut, making direct conversion between these enterotypes challenging. Our prior studies found that, owing to colonization resistance from IBD native microbiota, the healthy donor microbiota typically achieves less than 30% colonization efficiency in recipients,22,25 providing both theoretical and practical evidence for predicting FMT response based on the native microbiota. The ecological barrier to FMT success is shaped by both host-intrinsic and microbial factors, with nutrient competition and niche occupancy being central.12 Prior to FMT, the recipient's dominant enterotype microbiota occupies optimal ecological niches and efficiently utilizes specific nutrients, such as mucin or dietary fiber. During FMT, the introduced healthy microbiota must compete with the native microbiota for growth resources to exert therapeutic effects. Research from Zhao Fangqing's group confirmed that the presence of an enterotype barrier is a key cause of FMT failure, a pattern consistent with both IBD and recurrent Clostridioides difficile infection (rCDI).28 They observed that the ratio of engrafted to resident bacteria was significantly lower in FMT failure groups, with minimal structural changes in the microbiota pre- and post-treatment, indicating that healthy donor bacteria failed to breach the recipient's enterotype barrier. Our prior studies, which analyzed 2,142 samples from 31 FMT cohorts, revealed that overcoming the gut ecological transfer barrier requires donor–recipient microbiota complementarity: implanted bacteria must maintain sufficient phylogenetic distance from native bacteria to avoid ecological competition for identical nutrients and niches.22 Building on this, our study further defined the enterotype microbial range for FMT response in IBD patients, specifically, 65.5–82.5% ∑ET_B and 0–3.9% ∑ET_F. Our data revealed a significant shift from ET_F to ET_B in patients with Prevotella-dominated microbiomes post-FMT, which correlated reproducibly with the clinical response. This shift suggests that FMT, combined with interventions such as antibiotics or nutritional support, can destabilize Prevotella-dominated states, facilitating transitions toward Bacteroides-dominated states associated with reduced inflammation. Notably, Prevotella-dominated ET_F microbiomes were associated with elevated tumor necrosis factor alpha (TNFα), adiponectin, and high-density lipoprotein cholesterol (HDL-C), alongside reduced saturated fat and C-reactive protein (CRP) levels, which is consistent with distinct inflammatory profiles that may influence IBD severity and treatment outcomes.17 This shift may reflect a rebalancing of the microbial community, potentially aligning with a core microbiome signature characterized by stable interactions between two competing bacterial guilds: a beneficial guild enriched in dietary fiber digestion and short-chain fatty acid production and a detrimental guild linked to proinflammatory traits. In IBD, the Prevotella-dominated ET_F might indicate a dysbiotic state favoring the detrimental guild, while the post-FMT shift to ET_B could suggest a restoration of the beneficial guild, contributing to reduced inflammation and improved clinical outcomes. A limitation of this study is the absence of a placebo control group, which limits the ability to directly compare remission rates with vancomycin monotherapy or no intervention. This design choice was driven by ethical and practical considerations, as antibiotic monotherapy without microbiota restoration may exacerbate gut dysbiosis or contribute to antimicrobial resistance, posing potential risks to patients. To address this limitation and provide higher-level evidence, we initiated a multicenter, open-label randomized controlled trial (ClinicalTrials.gov, NCT07149441) in early 2025 to rigorously evaluate the efficacy and safety of this treatment strategy. Future studies with controlled designs will further elucidate the specific contributions of FMT to clinical outcomes in IBD patients. In conclusion, our findings suggest that FMT is an effective strategy for managing complex IBD. The intestinal microbiota's health determines FMT prognosis in IBD patients. A shift to a Bacteroides-dominated ET_B microbiome enriched in fiber digestion and short-chain fatty acid production improves outcomes and reduces inflammation. Conversely, the Prevotella-dominated ET_F reflects dysbiosis, hindering success. ET_B represents a healthier microbiota in IBD patients, and promoting its establishment through FMT and pre-treatment interventions is key to optimizing efficacy. Supplementary Material Supplementary material Figure S1. Overview of the microbiome and enterotype results of the cohort. Funding Statement This study was supported by National Natural Science Foundation of China (No: 82300753, No: 81972221, No: 81730102, and No: 82330093); the National Key R&D Program (No: 2024YFA1307100/2024YFA1307101); Gut Microbiota Research Project of Gusu College, Nanjing Medical University (No: GSKY20250807); China postdoctoral science foundation (No: 2022M722412); Top priority project of Shanghai (No: 076478684Q/2023-00154); Major project of Zhangjiang (No:ZJ2022-ZD-005); Clinical Research Plan of SHDC (No: SHDC2020CR4026, No: SHDC2020CR1030B, and No: SHDC2022CRS041); Project of Shanghai Science and Technology Commission (No: YDZX20213100003690); Project of Shanghai Municipal Health Commission (No: 19YF1431000 and No: ZHYY-ZXYJHZX-202105); and National Special Fund for the Construction of Key Clinical Specialties. Supplemental material Supplemental data for this article can be accessed at https://doi.org/10.1080/19490976.2025.2600055. Disclosure of potential conflicts of interest No potential conflicts of interest were disclosed. Acknowledgments We especially wish to thank Biotecan Medical Diagnostics Co., Ltd. for the assistance in 16S rRNA gene sequencing experiments. We gratefully acknowledge the Metagenome database for providing computational resources and technical support. Author contributions DZ, XW, QX, QC, and HQ conceived the ideas and experimental design. DZ, BY, HZ, CY, LL, FY, QY, QC, ZC and XL provided the clinical samples. SZ, CM, XC, YZ, and MW collected the baseline information of our cohort. XW analyzed the data. KW was responsible for the construction of the mathematical models. DZ and XW wrote the manuscript. XW, NL, TZ, and HQ jointly supervised this project. All the authors read and approved the final manuscript. Data availability statement The detailed baseline information of our cohort is available in the Supplementary information. The raw sequencing data are available at the National Center for Biotechnology Information server under study accession numbers PRJNA1269753. Any additional information required to reanalyze the data reported in this work is available from the lead contact upon request. Ethics approval and consent to participate The study was approved by the Ethics Committee of the Shanghai Tenth People's Hospital. All patients provided written informed consent for this study. All methods in this study were carried out in accordance with the Declaration of Helsinki. Permission to use the patient's samples was obtained from the Ethics Committee of the Shanghai Tenth People's Hospital. The ClinicalTrials ID is NCT05703477. Consent for publication Not applicable. References Associated Data Supplementary Materials Supplementary material Figure S1. Overview of the microbiome and enterotype results of the cohort. Data Availability Statement The detailed baseline information of our cohort is available in the Supplementary information. The raw sequencing data are available at the National Center for Biotechnology Information server under study accession numbers PRJNA1269753. Any additional information required to reanalyze the data reported in this work is available from the lead contact upon request.