Microbial and functional shifts between flare and remission in a single-center cohort of children with inflammatory bowel disease

Abstract

BACKGROUND

Gut microbial dysbiosis is central to the pathogenesis of inflammatory bowel disease (IBD). While gut microbiome differences between patients with and without IBD are well established, microbiome changes associated with disease activity and remission remain limited, particularly in paediatric populations.

AIM

To examine intra-individual taxonomic and functional gut microbiome changes during transition from active flare to remission under maintenance immunosuppression in a pilot single-center Singapore cohort of children with IBD.

METHODS

Paired stool samples and clinical data were collected from seven patients with paediatric IBD [5 Crohn’s disease (CD), 2 ulcerative colitis; ≤ 18 years] during active disease/flare (visit 1; Pediatric CD Activity Index/Pediatric Ulcerative Colitis Activity Index ≥ 10) and subsequent clinical remission (visit 2; Pediatric CD Activity Index/Pediatric Ulcerative Colitis Activity Index < 10). Samples underwent shotgun metagenomic sequencing for high-resolution taxonomic profiling and functional annotation of Kyoto Encyclopaedia of Genes and Genomes pathways.

RESULTS

Gut microbial diversity was reduced during flare compared to remission, with Actinobacteria abundance significantly higher in remission. Two distinct microbial clusters differentiated flare and remission states: The remission cluster was enriched with Bifidobacterium adolescentis, Bifidobacterium dentium, Lactobacillus gasseri, Faecalibacterium prausnitzii, while the flare state showed increased Klebsiella pneumoniae. Remission was further characterized by a downregulation of pathogenic microbes and an upregulation of beneficial microbes including a higher abundance of the butyrate producer Anaerostipes hadrus (P = 0.046). Microbial functional genes enriched in remission were predominantly associated with metabolic pathways including vitamin and cofactor biosynthesis, as well as carbohydrate, amino acid, and lipid metabolism.

CONCLUSION

The transition from flare to remission in Singaporean children with IBD is characterized by functional remodeling of the gut microbiome, which may contribute to recovery processes related to intestinal barrier integrity, cellular maintenance, and tissue repair. Targeted modulation of the gut microbiome may help sustain remission in paediatric IBD.

Key Words: Inflammatory bowel disease; Pediatrics; Metagenomic; Gut microbiome; Microbial shift; Functional shift

Core Tip: Using a longitudinal paired design, this paediatric inflammatory bowel disease metagenomic study shows that transition from active disease to remission is marked by a distinct gut microbiome with functional relevance. Remission was associated with enrichment of beneficial taxa (Anaerostipes hadrus, Bifidobacterium spp., Faecalibacterium prausnitzii, Lactobacillus gasseri) and metabolic pathways supporting short-chain fatty acid production, epithelial barrier repair, and immune modulation, whereas flare was dominated by Klebsiella pneumoniae. These findings highlight microbiome-derived functional pathways as promising targets for non-invasive biomarkers and microbiome-based therapeutic strategies to sustain remission in paediatric inflammatory bowel disease.

INTRODUCTION

Inflammatory bowel disease (IBD) is a chronic inflammatory disorder of the gastrointestinal tract, comprising two main phenotypes: Crohn’s disease (CD) and ulcerative colitis[1]. Of emerging importance is paediatric IBD (PIBD), which has seen a rapid rise in incidence across the Asia-Pacific region over the past decade[2]. This regional difference is clinically important, as Asian PIBD cohorts often exhibit distinct features, such as high rates of perianal disease[2] and higher proportions of very early-onset cases[3,4], which may be ascribed to geographic variations in host or environmental factors. The clinical spectrum is wide-ranging, encompassing a chronic relapsing-remitting pattern of gastrointestinal and extraintestinal manifestations, until definitive therapy is instituted[5]. While systemic immunomodulation remains the mainstay of management, microbiome modulation therapies have proven particularly efficacious in paediatric patients. For example, exclusive enteral nutrition in paediatric CD has demonstrated similar efficacy to corticosteroids for clinical remission but is superior in achieving mucosal healing[6]. A primary postulated mechanism for exclusive enteral nutrition is the reversal of gut microbial dysbiosis[6] and the restoration of an anti-inflammatory milieu[7]. Similarly, anti-inflammatory diets, such as the CD exclusion diet, have also shown therapeutic promise in maintaining clinical remission by modulating gut microbiota and removing pro-inflammatory dietary elements[8].

The complex pathogenesis of IBD is driven by an interplay of host genetic susceptibility, adverse environmental exposures, and gut microbial dysbiosis[9,10]. Patients with IBD exhibit reduced microbial richness and alpha-diversity[11,12], alongside altered microbial community composition (beta-diversity) compared to healthy controls[13]. Taxonomically, this dysbiosis is marked by a depletion of beneficial commensal bacteria - most notably within the phylum Firmicutes (e.g., Faecalibacterium prausnitzii, Roseburia and Ruminococcaceae)[14-16], which produce short-chain fatty acids (SCFAs) including butyrate, propionate and acetate. SCFAs fuel colonocytes, strengthen epithelial barrier integrity, and suppress inflammation by modulating immune responses. Their depletion impairs barrier function and promotes immune activation, thereby sustaining chronic intestinal inflammation[17]. Conversely, the IBD gut shows an expansion of potential pathobionts - commensal microbes with pathogenic potential - including members of the Proteobacterium phylum, such as Escherichia, Shigella, Fusobacterium, and Klebsiella[18]. These microbes drive differentiation of pro-inflammatory T-helper (Th) cell subsets, such as Th1 and Th17 cells[19], and secretion of pro-inflammatory cytokines, including tumor necrosis factor-alpha; interferon-gamma; interleukin (IL)-17; and IL-23[20]. Host genetic susceptibility, such as mutations in nucleotide-binding oligomerization domain 2, further exacerbates this process by impairing pathogen clearance (e.g., adherent-invasive Escherichia coli) and promoting dysregulated immune responses[22,23]. Environmental exposures - including antibiotic use and a high-fat, low-fiber diet - interact with the microbiome to worsen dysbiosis, reduce SCFA production, and favor pathobiont expansion, thereby contributing to the vicious cycle of barrier dysfunction, microbial imbalance, and chronic intestinal inflammation[24,25].

Active disease (flare) in IBD is defined by the presence of IBD-related symptoms and signs, accompanied by biochemical, radiological, and endoscopic indices of active inflammation. Therapeutic goals have now evolved beyond symptom resolution to aim for deeper levels of disease control, known as deep remission, which encompasses mucosal or endoscopic healing, transmural healing[26] and possibly ‘microbial healing’[27]. In a cross-sectional comparison of adults with IBD, those experiencing an active disease exhibited lower species diversity and richness, fewer butyrate-producing bacteria, and a higher abundance of potentially harmful bacteria compared with adults in remission[28]. In a paired fecal sample analysis from adults with IBD, 13 bacterial species showed differential abundance between active disease and remission states. For instance, CAG-269 sp001916005, Eubacterium F sp003491505, and Lachnospira sp000436475 were less abundant in remission, while Parabacteroides distasonis, Prevotellamassilia sp900540885, and CAG-495 sp001917125 increased. Together, these 13 species could distinguish between active disease and remission states of IBD[29]. Similarly, in a prospective cohort of children with IBD, gut microbial diversity was lower during active disease compared with remission, and microbial groups depleted during active disease tended to recover upon remission[30].

Despite the well-established role of gut microbial dysbiosis in IBD pathogenesis, the gut microbial profile and functional potential underlying the active disease and remission states of IBD remains poorly understood. Most existing literature focus on adult or Western cohorts and often rely on cross-sectional 16S rRNA gene sequencing data to compare patients with and without IBD. Such approaches fail to capture the dynamic, intra-individual changes during the critical transition from active disease to clinical remission. Notably, there is a significant research gap in longitudinal, high-resolution functional metagenomic studies in PIBD, particularly among Asian populations. Our study addresses this gap by analyzing paired stool samples from a pilot cohort of Singaporean children with IBD collected during active disease and remission. This longitudinal, shotgun metagenomic approach enables characterization of both taxonomic shifts and changes in microbial functional potential, providing a novel foundation for identifying functional biomarkers and therapeutic targets to maintain remission.

MATERIALS AND METHODS

Study cohort

Children and adolescents aged ≤ 18 years of age with IBD were recruited from National University Hospital Singapore. The inclusion criteria were: (1) A paediatric gastroenterologist-verified diagnosis of IBD established via endoscopy, with phenotypic classification as per the Paris classification[31]; and (2) Availability of at least one stool sample collected during active disease/flare [visit 1 (V1)] and one during remission visit 2 (V2), as defined below. Written informed consent was obtained from all study participants. This study was approved by the National Healthcare Group Domain Specific Review Board (Approval No. 2017/00298).

Study design

This study employed a longitudinal design in which blood and stool samples were collected from the same paediatric participants with IBD at two sequential clinical time points. Active disease/flare (V1) was defined by a Pediatric CD Activity Index (PCDAI) or Pediatric Ulcerative Colitis Activity Index (PUCAI) score ≥ 10. Patients diagnosed with a flare at V1 were followed up according to their scheduled clinical appointments. If, at any follow-up, their PCDAI or PUCAI score was < 10, they were considered to be in clinical remission, and this time point was designated as V2. Disease activity status at each visit was determined by the treating clinician based on standardized clinical assessment. PCDAI or PUCAI scores were used to determine the active flare or remission state of patients with IBD, rather than stool calprotectin or plasma C-reactive protein (CRP) levels. PCDAI and PUCAI are composite clinical scores that combine symptoms, physical examination findings, growth, and laboratory data, whereas stool calprotectin and plasma CRP are biomarkers of inflammation that reflect mucosal healing rather than clinical remission. Clinical data at V1 and V2 were obtained from the patients’ hospital clinical records. To minimize confounding factors that could affect the gut microbiome, recent use of antibiotics, probiotics, or other medications, as well as recent infections, were recorded. Only patients who had not received antibiotics or probiotics and had no infections in the preceding month were recruited for the study.

Sample collection and handling

Blood and stool samples were collected as part of routine clinical care. Blood samples were obtained on the day of the clinical visit, concurrent with PCDAI or PUCAI scoring. Blood samples were analyzed for full blood count, erythrocyte sedimentation rate, and CRP, while stool samples were assessed for stool calprotectin. Stool samples were collected either at the hospital or at home using a provided stool collection kit (with no stabilizer) within one week of the clinical scoring. After stool collection, samples were immediately placed in the provided ice bag with two ice packs, and the courier was promptly notified for pick-up. Samples were transported to the laboratory within 2-4 hours for aliquoting and then stored at -80 °C until processing. Each aliquot was used only once for metagenomic analysis to minimize freeze-thaw cycles, which were limited to a single cycle. Prior to stool collection at V1 and V2, patients were instructed to maintain their usual diet, without any special restrictions or exclusion diets. No bowel preparation procedures were used for the stool collection.

DNA extraction and shotgun metagenomic sequencing

Total genomic DNA was isolated from stool samples using the Quick-DNA Fecal/Soil Microbe Midiprep Kit (Zymo Research, United States) following the manufacturer’s protocol. Briefly, 375 mg of thawed stool was combined with 6 mL of buffer in a bead tube and homogenized using a bead beater. After centrifugation at 4000 × g for 5 minutes, the bead column was removed, and 18 mL of lysis buffer was added to the supernatant. The resulting mixture was passed through a filter column, washed, and eluted with 100 μL of elution buffer by centrifugation at 10000 × g for 1 minutes. DNA yield and purity were evaluated using a Nanodrop spectrophotometer and Qubit fluorometer.

For shotgun metagenomic library preparation, genomic DNA was randomly fragmented and subjected to end repair, A-tailing, and ligation with Illumina sequencing adapters. Adapter-ligated fragments were polymerase chain reaction -amplified, size-selected, and purified. Libraries were quantified using Qubit and real-time polymerase chain reaction, and fragment size distribution was assessed by Bioanalyzer. Pooled libraries were sequenced on the Illumina NovaSeq 6000 platform. All library preparation, sequencing, and data processing were performed by NovogeneAIT Genomics (Singapore).

Data processing

Raw sequencing reads generated from the Illumina NovaSeq 6000 platform were pre-processed using Readfq to remove low-quality sequences and adapter contaminants. Host-derived DNA was eliminated by aligning the cleaned reads to the human reference genome (GRCh38/hg38) using Bowtie2 with the parameters “end-to-end”, “sensitive”, “-I 200”, and “-X 400”. The remaining high-quality reads were assembled into contigs with MEGAHIT, and gene prediction was performed using MetaGeneMark. Predicted open reading frames exceeding 100 bp were clustered with cluster database at high identity with tolerance to generate a non-redundant gene catalogue, which was used as the reference for downstream quantification. Clean reads from each sample were then mapped back to this catalogue using Bowtie2, and gene abundance was calculated based on read counts. Genes with fewer than two mapped reads were excluded from subsequent analyses. Read counts were normalized, and sequencing depth was adjusted to match the sample with the lowest read number to enable cross-sample comparison.

Taxonomic and functional profiling were conducted using the normalized data. Shannon diversity indices and principal coordinate analysis plots were calculated in R (version 4.3.2) with the package’s vegan, ggplot2, and ggrepel. Visualizations, including diversity indices, phylum-level bar plots, species-level dot plots, and fold-change heatmaps, were generated using GraphPad Prism (version 10.0.2). Bacterial taxa that exhibited ≥ 2-fold changes in relative abundance when comparing V2 (remission) with V1 (active disease) were selected for functional annotation. Their associated enzyme commission numbers were mapped to biological pathways using the Kyoto Encyclopaedia of Genes and Genomes database. The relative abundance of microbial functional pathways was calculated as the proportion of gene counts assigned to each pathway relative to the total gene count across all pathways.

Statistical analysis

All statistical analyses were performed in IBM SPSS Statistics (version 29.0). Paired comparisons between V1 and V2 within the same participants were conducted using the Wilcoxon signed-rank test. Given the exploratory nature of this pilot study and the small sample size, P-values were initially calculated without adjustment. Correction for multiple comparisons using the false discovery rate was subsequently applied where appropriate; however, no comparisons remained statistically significant after adjustment. Therefore, results are presented as exploratory findings. A two-sided P < 0.05 was considered statistically significant. The statistical methods of this study were reviewed by Ooi DSQ and Tay CJX from National University of Singapore.

RESULTS

Clinical characteristics of patients with IBD

Seven paediatric patients fulfilled the inclusion criteria, of whom 5/7 had CD, 4/7 were male, 6/7 were of Indian ethnicity. At V1 (flare), 2/7 were receiving immunosuppressive medication while all 7 of them had received immunosuppressive treatment by V2 (remission). The median age (interquartile range) at enrollment was 13.0 (10.4-15.0) years. V2 occurred at a median (interquartile range) time of 14.3 (12.9-22.2) months after V1 (Table 1).

Table 1 Clinical characteristics of pediatric inflammatory bowel disease patients during flare and remission.

Parameters	P_003	P_010	P_015	P_019	P_020	P_022	P_023
Gender	Male	Male	Female	Female	Female	Male	Male
Age (years)	16.3	3.2	11.8	9	14.2	18.5	12.8
Ethnicity	Chinese	Indian	Indian	Indian	Indian	Indian	Malay
IBD phenotype	UC	UC	CD	CD	CD	CD	CD
Disease location (Paris classification)	E1	E4	L4b	L3	L3	L3	L3
CD phenotype			B1	B1	B1	B1	B1
V1 (active disease/flare)
Disease activity score (PUCAI/PCDAI)	60	80	35	52.5	55	32.5	32.5
Weight (kg)	58.5	15	35.6	27	35.6	41	30.9
Height (cm)	174	99	150.7	135	150	167.1	145.2
Hemoglobin (g/dL)	12.9	5.5	11.4	8.6	12.2	11.3	10.2
White blood cell (109/L)	4.5	12.36	7.87	9.28	15.31	8.14	8.3
Platelets (109/L)	184	62	547	508	437	408	424
ESR (mm/hour)	13	10	19	89	68	23	80
CRP (mg/L)		38	165		162
Stool calprotectin (μg/g)	32	> 1000	> 1000		729	> 1000	> 1000
Existing immunosuppression	None	Steroid	None	None	Steroid; Aza	None	None
V2 (remission)
Time from V1 (months)	4	11.8	73	14.3	13.9	14.6	29.8
Disease activity score	0	0	0	7.5	0	5	0
Weight (kg)	59.35	18.2	53	34	46.7	47	44.1
Height (cm)	174	101	153.5	153.5	156.5	169	150.5
Hemoglobin (g/dL)	14.8	10.1	11.3	10.7	12.6	12.5	11.6
White blood cell (109/L)	12.32	4.54	3.88	7.6	5.83	7.46	9.5
Platelets (109/L)	184	453	351	271	268	291	282
ESR (mm/hour)	9	10	14	14	7		15
CRP (mg/L)		< 5			< 5	25
Stool calprotectin (μg/g)	32	176	< 30	264	838		250
Current immunosuppression	5ASA	Infliximab	Aza	MTX	Infliximab	Aza	Aza

PUCAI: Paediatric Ulcerative Colitis Activity Index; PCDAI: Paediatric Crohn’s Disease Activity Index; CD: Crohn’s disease; V1: Visit 1; ESR: Erythrocyte sedimentation rate; CRP: C-reactive protein; V2: Visit 2; P: Patient; UC: Ulcerative colitis; E1: Proctitis; E4: Pancolitis; L4b: Isolated small bowel disease; L3: Ileocolonic disease; B1: Inflammatory only; Aza: Azathioprine; 5ASA: 5-aminosalicylate; MTX: Methotrexate.

At V2, the disease activity score was significantly reduced compared to V1 [0.00 (0.00-2.50) vs 52.5 (33.8-57.5), P = 0.018]. Although reductions in CRP levels [5.00 (5.00-15.0) mg/L vs 162 (100-164) mg/L] and stool calprotectin concentrations [213 (32.0-264) vs μg/g vs 1000 (729-1000) μg/g] were also observed at V2 relative to V1, these changes did not reach statistical significance (Table 2).

Table 2 Paired comparison of clinical parameters between flare and remission, median (interquartile range).

Parameters	V1	V2	P value
Disease activity score	52.5 (33.8-57.5)	0.00 (0.00-2.50)	0.018
Hemoglobin (g/dL)	11.3 (9.40-11.8)	11.6 (11.0-12.5)	0.028
White blood cell (109/L)	8.30 (8.00-10.8)	7.46 (5.19-8.55)	0.271
Platelets (109/L)	424 (296-473)	282 (269-321)	0.345
ESR (mm/hour)	23.0 (16.0-74.0)	12.0 (9.00-14.0)	0.043
CRP (mg/L)	162 (100-164)	5.00 (5.00-15.0)	0.180
Stool calprotectin (μg/g)	1000 (729-1000)	213 (32.0-264)	0.144

Paired comparisons between visit 1 and visit 2 were performed using the Wilcoxon signed-rank test. V1: Visit 1; V2: Visit 2.

Gut microbial diversity and taxonomic composition

There was a non-significant trend toward higher gut microbial diversity during V2 compared with V1, as measured by the Shannon index [2.80 (2.67-2.92) vs 2.56 (0.66-2.74); P = 0.165; Figure 1]. At the phylum level, the abundance of Pseudomonadota was significantly decreased during remission [7591 (1797-20788) vs 58878 (23733-149567), P < 0.05). In contrast, no statistically significant differences were observed between V1 and V2 for other major phyla, including Actinomycetota, Bacillota, Bacteroidota, Fusobacteriota, Pseudomonadota and Verrucomicrobiota. A proportion of reads (approximately 10%) could not be classified at the genus level and were grouped as “unclassified” bacteria, likely reflecting limitations of current reference databases and the presence of novel taxa.

Figure 1 Diversity and taxonomic composition of gut microbiota in inflammatory bowel disease during flare (visit 1) and remission (visit 2). A: Shannon diversity index; B: Phylum-level diversity; C: Phylum-level abundance. V1: Visit 1; V2: Visit 2. ^aP < 0.05.

Gut microbiota composition across disease activity states

Principal coordinates analysis of gut microbiota composition revealed two distinct clusters corresponding to disease states of flare (V1) and remission (V2; Figure 2). Microbiota from participants in remission clustered separately from those in active disease. During active disease, microbes such as Klebsiella pneumoniae were found in the cluster, whereas remission was associated with taxa including Bifidobacterium adolescentis, Bifidobacterium dentium, Lactobacillus gasseri, and Faecalibacterium prausnitzii.

Figure 2 Multidimensional scaling plots of gut microbiota composition during flare (visit 1) and remission (visit 2) states of inflammatory bowel disease. A: Overall gut microbiota composition across both flare and remission states; B: Gut microbiota composition among participants during the flare state (visit 1); C: Gut microbiota composition among participants during the remission state (visit 2). MDS: Multidimensional scaling; V1: Visit 1; V2: Visit 2.

Differential gut microbial abundance in remission relative to active disease

Analysis of the gut microbiome in remission relative to active disease states of IBD revealed significant shifts in microbial composition (Figure 3). Pathogenic microbes associated with IBD were downregulated in remission, whereas beneficial microbes were upregulated. Notably, Anaerostipes hadrus was significantly enriched in remission relative to active disease (P = 0.046).

Figure 3 Differentially abundant gut microbes in remission (visit 2) compared with flare (visit 1) states of inflammatory bowel disease. A: Heatmap showing gut microbes exhibiting significant differential abundance (2-fold change) in remission (visit 2) compared with flare (visit 1) for each individual participant; B: Log₂ fold change of gut microbes that are significantly upregulated or downregulated (2-fold change) in remission (visit 2) compared with flare (visit 1). Red box indicates pathogenic microbes associated with inflammatory bowel disease and green box indicates beneficial microbes associated with inflammatory bowel disease; C: Individual gut microbe showing statistically significant differential abundance (2-fold change) in remission (visit 2) compared with flare (visit 1). V1: Visit 1; V2: Visit 2.

Microbial functional gene profiles in remission relative to active disease

Majority of the identified functional genes were associated with microbes upregulated in remission. Functional annotation revealed that 7.09% of genes were involved in vitamin and cofactor biosynthesis, 8.79% in carbohydrate metabolism including SCFA production, 11.92% in amino acid metabolism including branched-chain amino acid pathways, and 8.64% in lipid and fatty acid metabolism (Figure 4). Collectively, these four pathways comprised 36.44% of the genes encoded by microbes that were differentially abundant in remission relative to active disease.

Figure 4 Microbial functional pathways associated with significantly altered microbial genes in remission (visit 2) compared with flare (visit 1) states of inflammatory bowel disease. A: Top 30 microbial functional pathways associated with microbial genes exhibiting significant differential abundance (≥ 2-fold change) in remission (visit 2) compared with flare (visit 1); B: Major metabolic pathway categories, including vitamin and cofactor biosynthesis, carbohydrate metabolism, amino acid metabolism, and lipid and fatty acid metabolism, associated with microbial genes significantly altered in remission (visit 2) compared with flare (visit 1). SCFA: Short-chain fatty acid; BCAA: Branched-chain amino acids.

DISCUSSION

Our findings indicate a trend toward higher gut microbiota diversity in children with IBD during remission, and the overall microbial profile effectively distinguishes between active disease and remission states. The active disease state was characterized by an increased abundance of pathogenic bacteria such as Klebsiella pneumoniae, whereas the remission state was enriched with beneficial microbes, including Bifidobacterium adolescentis, Bifidobacterium dentium, Lactobacillus gasseri, and Faecalibacterium prausnitzii. Furthermore, pathogenic taxa were differentially downregulated, while beneficial microbes such as Anaerostipes hadrus were upregulated during remission compared to active disease. Functional analysis of microbial genes identified pathways related to vitamin and cofactor biosynthesis, carbohydrate metabolism, amino acid metabolism, and lipid and fatty acid metabolism.

Patients with IBD have been shown to exhibit reduced gut microbial diversity compared to individuals without IBD[12,32,33]. Similarly, in Korean children with IBD, the Chao 1 Diversity Index was lower during active disease (flare) than during remission[30]. These findings align with our results, indicating that the reduction in microbial diversity may contribute to disease activity of IBD. Among the different phyla, we observed a significant decrease in Pseudomonadota during remission in our PIBD cohort. Aldrian et al[34] performed a meta-analysis of PIBD microbiome studies and found that the abundance of Pseudomonadota increased with disease severity, supporting its association with intestinal inflammation. In contrast, He et al[35] reported that the phylum Proteobacteria (now classified as Pseudomonadota) was reduced in patients with IBD in remission compared with those with active disease.

Two distinct microbial clusters were identified, distinguishing the flare and remission states in our PIBD cohort. The remission cluster was characterized by an increased abundance of putatively beneficial microbes that promote gut barrier integrity and are associated with favorable health outcomes - Bifidobacterium adolescentis, Bifidobacterium dentium, Lactobacillus gasseri, and Faecalibacterium prausnitzii[36,37]. The reduction of Faecalibacterium prausnitzii is a well-established marker of IBD-associated dysbiosis[16,38,39], and Zhao et al[40] confirmed its negative association with disease activity in a meta-analysis of 1669 individuals with IBD. Similarly, Diederen et al[41] reported significantly higher abundances of both Faecalibacterium prausnitzii and Bifidobacteria adolescentis following diet reintroduction among paediatric patients with CD who responded to exclusive enteral nutrition, compared with non-responders. These findings suggest that these microbes may support clinical remission, and a more favorable response to microbial-modulating interventions such as exclusive enteral nutrition.

Conversely, Klebsiella pneumoniae emerged as the dominant pathobiont in the flare cluster, consistent with evidence supporting its role in IBD. A multi-cohort study by Federici et al[42] identified a specific Klebsiella pneumoniae clade strongly associated with IBD exacerbation and severity across four countries (France, Israel, United States, and Germany), characterized by unique antibiotic resistance and mobilome features. Mechanistically, Zhang et al[43] demonstrated in a mouse model that Klebsiella pneumoniae induces intestinal inflammation via activation of the caspase-11 pathway in gut epithelial cells, leading to IL-18 release. Furthermore, we observed that microbes downregulated during remission relative to flare were predominantly pathogenic species linked to adverse outcomes[44], whereas those upregulated were beneficial commensals associated with intestinal health[45]. Notably, the butyrate-producing Anaerostipes hadrus was significantly more abundant in remission, consistent with previous reports showing its reduced abundance in active IBD across both paediatric and adult cohorts[30,46].

The microbial functional genes associated with microbes more abundant in remission were predominantly involved in metabolic pathways, with vitamin and cofactor biosynthesis, carbohydrate metabolism, amino acid metabolism, and lipid/fatty acid metabolism accounting for over one-third of the differentially abundant genes. Carbohydrate metabolism, particularly SCFA production (butyrate, acetate, propionate), highlights a core function of the remission microbiome: Maintaining intestinal homeostasis. SCFAs such as butyrate suppress pro-inflammatory pathways (e.g., nuclear factor kappa B activation)[47] and promote differentiation of anti-inflammatory regulatory T-cells[48]. Clinical data from a longitudinal adult IBD cohort also showed that butyrate and its synthesis substrates in the stool metabolome were significantly associated with remission following anti-tumor necrosis factor biologic therapy[49]. Arginine biosynthesis supports intestinal homeostasis by producing ornithine, which enhances epithelial barrier function, and contributes to the host nitric oxide pool, modulating myeloid and lymphoid cell activation to reduce inflammation[50]. Glycerophospholipid metabolism indicates microbial growth and proliferation, as these lipids are essential for membrane synthesis, and microbial lipid products (e.g., sphingolipids) can modulate host immunity and reinforce barrier integrity[51,52]. Additionally, pathways in the folate-mediated one-carbon pool (DNA synthesis)[53] and nicotinate metabolism (intestinal epithelial barrier and mitochondrial function)[54,55] provide essential micronutrients and cofactors for both commensal microbes and host intestinal repair. Collectively, these functional shifts demonstrate that the remission microbiome is metabolically active in supporting intestinal barrier integrity, cellular maintenance, and tissue repair, thereby promoting eubiosis and sustaining disease remission.

The strengths of this study include being the first PIBD cohort study in Southeast Asia to employ a longitudinal paired design across two defined clinical states, enabling the assessment of intra-individual variation. Furthermore, integration of taxonomic and functional metagenomic data provided mechanistic insights into the remission-associated microbiome. However, several limitations should be acknowledged. First, the sample size was small (n = 7), which limited statistical power and may explain why observed differences did not reach statistical significance. The small cohort also precluded correlation analyses between microbial diversity or abundance with clinical parameters such as stool calprotectin and CRP[56]. Nonetheless, even with this small cohort, we observed distinct gut microbial clusters that differentiated active disease and remission states, providing preliminary evidence for disease-associated microbial shifts. Each participant was sampled at two sequential time points, and the use of shotgun metagenomic sequencing allowed for a detailed, high-resolution analysis of microbial composition and functional potential despite the limited sample size. Future studies with larger cohorts are needed to validate these findings, confirm statistical significance, and facilitate subgroup analyses based on race, sex, disease phenotype, drug exposure and other co-variates. Second, the study did not include a healthy control group, which limits the ability to determine whether the gut microbiota profile of children with IBD in remission resembles that of healthy children. However, the primary aim of this study was to perform a paired comparison of microbial profiles within the same participants between active disease and remission, with the goal of identifying microbial taxa and functional pathways that may serve as potential therapeutic targets to attain remission in PIBD.

CONCLUSION

In conclusion, our study shows that the transition from active IBD to remission in a pilot single-center cohort of Singaporean children is accompanied by a functionally significant shift in the gut microbiome. Microbial functional genes associated with remission highlight key metabolic pathways supporting epithelial barrier repair, cellular energy, and immune modulation. These findings identify specific microbial taxa and pathways as tangible targets for therapy. Future work should leverage this functional data to develop non-invasive diagnostic biomarkers and design next-generation live biotherapeutics or postbiotic formulations, advancing precision medicine for sustained PIBD remission.

ACKNOWLEDGEMENTS

We thank Dr Dimple Rajgor for her assistance in formatting and submitting the manuscript for publication.