Assessing whether rectal swabs reflect appendiceal microbiota profiles in acute appendicitis: a 16S rRNA-based comparative study

Abstract

Background

Acute appendicitis is associated with characteristic changes in the intestinal microbiota, but direct sampling of appendiceal contents is invasive and cannot be performed in healthy controls. We therefore evaluated whether rectal swabs could partially capture appendiceal microbiome signatures in adults with acute appendicitis.

Methods

In a prospective cross-sectional study, we enrolled adults with acute appendicitis and healthy volunteers between October 2023 and December 2024. Four types of samples were collected: feces from healthy controls (HC), appendiceal luminal contents from patients with acute appendicitis (AC), intraoperative rectal swabs from patients with acute appendicitis (RS), and initial postoperative feces from patients with acute appendicitis (IF; first stool within 24 h after surgery). 16 S rRNA gene (V3–V4) sequencing was performed, and reads were processed with QIIME2. Alpha and beta diversity, differential taxonomic composition, and PICRUSt2-based functional predictions were compared across matrices. Genus-level and functional concordance between paired AC–RS samples was assessed.

Results

After quality control, 64 AC, 34 RS, 24 IF, and 29 HC samples were included. Phylogenetic diversity (PD whole-tree) was higher in AC and RS than HC, with AC also higher than RS; IF showed lower PD than AC. Bray–Curtis principal coordinate analysis showed AC forming a distinct cluster separated from HC and RS along PC1, whereas IF overlapped with HC and RS. AC, RS, and IF were enriched for Escherichia/Shigella and Fusobacterium and depleted in butyrate-producing genera such as Faecalibacterium compared with HC. In the 21 paired AC–RS cases, genus-level relative abundances and several predicted functional pathways showed concordance, indicating that RS captured many but not all appendiceal dysbiosis features.

Conclusions

Our findings suggest that intraoperative rectal swabs may partially reflect appendiceal microbiome alterations at the genus and pathway levels and may serve as a minimally invasive adjunct for microbiome profiling in acute appendicitis. However, these associations are inferred from 16 S amplicon data in a modestly sized, antibiotic-exposed cohort and should be validated using shotgun metagenomics in larger, clinically stratified populations.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13099-025-00794-1.

Introduction

Acute appendicitis is a common surgical emergency typically managed by appendectomy and empiric antibiotics [1]. However, conventional cultures of appendiceal or peritoneal samples often fail to identify a clear pathogen, particularly in uncomplicated cases, limiting guidance for targeted therapy [2]. Recent evidence suggests that appendicitis reflects a community-level dysbiosis of the gut microbiota rather than infection by a single overt pathogen [3–5]. For example, anaerobic oral organisms such as Fusobacterium and Parvimonas are frequently enriched in inflamed appendices, correlating with disease severity, while beneficial gut commensals (e.g. Faecalibacterium, Blautia, Lachnospiraceae) are depleted [3]. These findings imply that shifts in the appendiceal microbiome contribute to inflammation and suggest new opportunities for understanding appendicitis pathogenesis.

Profiling the appendiceal microbiota is challenging. Healthy appendices cannot be sampled ethically, and appendiceal lumen conditions may differ from those of stool. In practice, many studies use feces from healthy volunteers as a baseline proxy for gut microbiota, but this has limitations because the appendix niche is unique. Indeed, culture-independent analyses have demonstrated that the appendix harbors a distinctive community – for example, certain taxa are consistently more abundant in healthy appendices than in the rectum [6]. This makes it difficult to establish the “normal” appendix microbiome for comparison. At the same time, obtaining patient samples in a timely way is problematic: perioperative antibiotics and the usual delay in intraoperative bowel movements mean that stool may not be available or may be altered by treatment.

Rectal swabs have emerged as a practical alternative for gut microbiome sampling. They are minimally invasive, can be collected at or near the time of surgery, and have been shown to capture the major features of the fecal microbiota [7]. Studies in other settings report that swab-derived communities closely mirror stool. For instance, in intensive care patients no significant differences in microbial diversity or resistance gene profiles were observed between rectal swabs and paired fecal samples [8]. Similarly, Budding et al. [9] found that rectal swabs yield reproducible gut microbiome profiles and are “ideally suited for clinical diagnostics and large-scale studies”. These results suggest that standard perioperative rectal swabs might provide a rapid window into the gut microbiome without waiting for stool. In the context of appendicitis, such swabs could potentially reveal the disease-associated microbial shifts in real time.

In this prospective cross-sectional 16 S rRNA gene sequencing study, our primary objective was to assess whether intraoperative rectal swab microbiota reflect appendiceal luminal microbiota and could be used as a practical, minimally invasive proxy for appendiceal microbiome profiling in adults with acute appendicitis. Secondary objectives were: (1) to describe appendicitis-associated alterations in gut microbial diversity, taxonomic composition, and predicted function compared with healthy feces; (2) to explore early postoperative changes in fecal microbiota using the first stool passed within 24 h after appendectomy; and (3) to relate postoperative patterns to both appendiceal and rectal swab profiles. We hypothesized that the dysbiotic signature of appendicitis observed in appendiceal contents would also be detectable in rectal swabs, whereas early postoperative stool would show only partial recovery toward a healthy-like community.

Methods

Study design and sample collection

This was a prospective cross-sectional study conducted at the Department of General Surgery, the Seventh Affiliated Hospital of Xinjiang Medical University (Urumqi, China). Between October 2023 and December 2024, we prospectively recruited adults with clinically and radiologically diagnosed acute appendicitis undergoing surgical appendectomy, together with a group of healthy individuals without appendicitis who provided fecal samples as a reference cohort. A total of 122 participants were included, comprising 93 patients with acute appendicitis and 29 healthy individuals. Appendicitis was confirmed clinically and by imaging, and all patients underwent surgical appendectomy. Cases with periappendiceal abscess or perforation were not included in this study, consistent with routine clinical practice at our center during the study period. Presence of a fecalith was extracted from the operative and/or pathology reports. In three cases, fecaliths were described as small/non-obstructing, and no luminal obstruction was documented. The study protocol was approved by the institutional ethics committee, and written informed consent was obtained from each subject (or guardian, if applicable) prior to sample collection.

In patients with acute appendicitis, intraoperative specimen collection followed a standardized protocol. From each appendicitis patient, an intraoperative sample of appendiceal luminal content was aseptically collected immediately after the appendix was removed (before any bowel resection or opening of the specimen for pathology). In addition, two further sample types were obtained. Rectal swab specimens were collected intraoperatively—after anesthesia induction and administration of prophylactic antibiotics (first dose 30 min before incision), but before bowel closure. Early postoperative fecal samples were defined as the first stool passed within 24 h after surgery and were collected into sterile containers as soon as they became available. Because of the effects of anesthesia, analgesics, and antibiotics, not all patients produced a usable stool within the hospitalization period; rectal swabs were therefore an important proxy sample for many patients.

Healthy fecal samples were obtained from adult volunteers with no history of appendicitis and no acute illness. These individuals were recruited to broadly resemble the appendicitis cohort in age and sex distribution and had not used systemic antibiotics in the preceding 3 months. All stool samples (from both healthy volunteers and patients) were collected into sterile containers, and rectal swabs were placed into sterile tubes, immediately placed on ice after collection, and transported to the laboratory within 2 h. Upon arrival, samples were aliquoted (for stool) or the swab tip was cut into a sterile microcentrifuge tube, and all specimens were stored at − 80 °C until DNA extraction. Clinical metadata (such as patient age, sex, and antibiotic use/duration) were recorded for all participants.

For analysis, microbiota specimens were categorized into four sample-type groups: feces from healthy controls (HC), appendiceal luminal contents from patients with acute appendicitis (AC), intraoperative rectal swabs from patients with acute appendicitis (RS), and initial postoperative feces from patients with acute appendicitis (IF). The final sample sizes for each group were as follows: HC (n = 29), AC (n = 64), RS (n = 34), and IF (n = 24). Only 21 patients contributed both appendiceal content and rectal swab samples (AC–RS); this was the only paired subset used for within-patient analyses. All remaining samples were unpaired and were analyzed in cross-sectional comparisons. Each participant was therefore assigned a unique study identifier, and all samples from that individual were labeled with the same identifier plus a suffix indicating the sample type. Laboratory processing and 16 S rRNA gene sequencing were performed at the sample level. In cross-sectional analyses, each sample was treated as an observation, whereas within-patient comparisons (e.g., AC vs. RS pairs) were restricted to patients with paired samples and analyzed using paired or within-individual distance measures as appropriate. A detailed flow chart describing patient screening, sample collection, exclusion of low-quality sequencing data, and final numbers in each group is provided in Fig. 1.

Fig. 1.

Flow diagram of participant enrollment, sample collection, quality control, and inclusion across matrices (HC, AC, RS, IF)

Per institutional practice during the study period, appendicitis patients typically received a cephamycin (most commonly cefmetazole; occasionally cefoxitin) or a third-generation cephalosporin (e.g., ceftazidime), with β-lactam/β-lactamase inhibitor regimens (cefoperazone–sulbactam or piperacillin–tazobactam) used at the surgeon’s discretion. Selected cases additionally received nitroimidazoles (metronidazole/ornidazole), fluoroquinolones (levofloxacin/ciprofloxacin), fosfomycin, aztreonam, or carbapenems (meropenem). The first dose was administered approximately 30 min before incision. Agent-level dosing was not captured in a structured form; therefore, no stratified analysis by antibiotic class was performed in the primary microbiome analyses.

Sample size considerations

A priori power analysis was not performed because this was an exploratory microbiome study and reliable effect-size estimates for the target comparisons were not available. Post hoc, the minimum detectable between-group effect sizes (Cohen’s d; two-sided α = 0.05) were approximately 0.63 (AC vs. HC), 0.71 (RS vs. HC), 0.77 (IF vs. HC), 0.59 (AC vs. RS), 0.67 (AC vs. IF), and 0.75 (RS vs. IF), indicating sensitivity to medium-to-large effects.

DNA extraction and 16 S rRNA sequencing

Microbial DNA was extracted from all samples using standardized protocols suitable for complex biological specimens. For stool samples, approximately 200 mg of feces was used, while entire swab heads were processed for rectal swabs. We employed a combination of mechanical disruption (bead-beating) and chemical lysis to ensure effective recovery of bacterial DNA, including from Gram-positive organisms. The extracts were purified using a commercial DNA isolation kit optimized for stool (QIAamp Fast DNA Stool Mini Kit; cat. no. 51604), following the manufacturer’s instructions. DNA concentration and purity were assessed by spectrophotometry, and integrity was verified on a 1% agarose gel. Extracted DNA samples were stored at − 20 °C until amplification.

The V3–V4 hypervariable region of the 16 S rRNA gene was amplified for each DNA sample. PCR was performed using universal primers (forward primer 341 F: 5′-CCTAYGGGRBGCASCAG-3′, reverse primer 806R: 5′-GGACTACNNGGGTATCTAAT-3′ or analogous primers targeting the same region) with Illumina adapter overhangs. Each 25 µL PCR reaction contained ~ 50 ng of template DNA, 0.2 µM of each primer, and a high-fidelity DNA polymerase with appropriate buffer and dNTPs. Thermal cycling consisted of an initial denaturation at 95 °C for 3 min; 30 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 45 s; and a final extension at 72 °C for 5 min. Amplification success was confirmed by agarose gel electrophoresis, yielding a product of ~ 460 bp. Negative PCR controls (no template) were included to check for contamination. Triplicate PCR reactions per sample were performed and pooled to mitigate amplification bias.

Amplicon libraries were prepared using a two-step indexing protocol. In the first PCR (described above), locus-specific primers amplified the 16 S V3–V4 region. In a second PCR, dual indices and Illumina sequencing adapters were attached using a limited number of cycles. Indexed libraries were purified and quantified (using a Qubit fluorometer). Equimolar amounts of each library were then pooled. The final library pool was denatured and diluted as per Illumina’s guidelines and sequenced on an Illumina MiSeq platform (RRID: SCR_016379) using a 2 × 300 bp paired-end kit (MiSeq Reagent Kit v3; RRID: SCR_016379). This configuration yields overlapping paired reads that cover the entire V3–V4 amplicon. A 10% PhiX spike-in was added to the sequencing run to increase base diversity.

Sequence data processing and taxonomic analysis

Raw reads were quality filtered using Trimmomatic (v0.33; RRID: SCR_011848) and primer sequences were trimmed with Cutadapt (v1.9.1; RRID: SCR_011841). Paired-end reads were merged with USEARCH (v10; RRID: SCR_027438) and chimeras were removed using UCHIME (v8.1; RRID: SCR_008057). Amplicon sequence variants (ASVs) were inferred using DADA2 (RRID: SCR_023519) within QIIME2 (v2020.6; RRID: SCR_021258), while OTUs were clustered at 97% identity via UPARSE for comparative purposes. Taxonomy was assigned with a Naïve Bayes classifier trained on the SILVA 138 database (RRID: SCR_006423) using QIIME2’s classify-sklearn plugin, with a confidence threshold of 0.7. Given the limited species-level resolution of V3–V4 16 S amplicons, we primarily report taxa at the genus level, and any species labels are putative; otherwise features were collapsed to the genus level.

Diversity and statistical analyses

Alpha diversity was evaluated using Chao1, Shannon, and Simpson indices, based on rarefied ASV tables. Group differences were tested using Kruskal–Wallis (or ANOVA when assumptions were met) with Benjamini–Hochberg FDR correction across taxa; significance was defined as q < 0.05. Beta diversity was computed using Bray–Curtis dissimilarities and visualized by principal coordinates analysis (PCoA). Statistical significance was tested using PERMANOVA (999 permutations). Differentially abundant taxa were identified using Metastats (RRID: SCR_014610; pairwise; q-values reported) and LEfSe (RRID: SCR_014609; Kruskal–Wallis α = 0.05; Wilcoxon α = 0.05; LDA ≥ 4.0) across groups. All analyses were performed using QIIME2, R (vegan package; RRID: SCR_011950), and Python (RRID: SCR_008394).

Functional prediction

Functional profiles were inferred using PICRUSt2 (RRID: SCR_022647) based on ASV tables, with pathway annotation referencing KEGG Level 2 categories. Group comparisons (e.g., RS vs. HC) were conducted using Welch’s t-test or Mann–Whitney U-test, with false discovery rate correction. Pathways with adjusted p < 0.05 were considered significantly different. Functional shifts were interpreted in relation to microbial metabolism, stress adaptation, and antimicrobial resistance.

Results

Patient and sample characteristics

A total of 122 participants were enrolled in this study, including 93 patients diagnosed with acute appendicitis and 29 healthy individuals. Among the 93 patients with acute appendicitis, intraoperative appendiceal content was successfully collected from 84 (90.3%); in the remaining 9 cases, research sampling of luminal contents was not performed because of operating-theatre workflow limitations (for example, the appendix was sent directly for routine pathology before research sampling could be undertaken). Of the 84 AC specimens, 64 (76.2%; 64/93, 68.8% of all appendicitis cases) passed amplification and sequencing quality control and were included in the final analysis. Intraoperative rectal swabs were obtained from 35 of the 84 patients (41.7%), with 34 of 35 (97.1%) yielding valid sequencing data. Fecal samples were available from 24 of 93 (25.8%) appendicitis patients, all of which were successfully sequenced. Fecal samples from all 29 healthy individuals (29/29, 100%) were successfully sequenced and served as the healthy comparison group (Table 1). Among patients contributing AC samples, pathology subtyping identified 26 simple and 38 suppurative appendicitis. A fecalith was reported in 3 cases and was described as small/non-obstructing. Given the small sample sizes within these subgroups—and the limited paired subset (AC–RS, n = 21)—we did not perform stratified analyses by pathology subtype or fecalith status.

Table 1.

Cohort clinical characteristics by group

Sample group	Sample size (n)	Age range (≤ 30), n (%)	Age range (30–50), n (%)	Age range (≥ 50), n (%)	Mean Age ± SD (years)	Male, n (%)	Female, n (%)	Treatment duration (days) mean ± SD
Healthy Feces	29	8 (27.6%)	13 (44.8%)	8 (27.6%)	39.7 ± 12.8	14 (48.3%)	15 (51.7%)	NA
Appendiceal Content	64	28 (43.8%)	22 (34.4%)	14 (21.9%)	37.4 ± 14.7	28 (43.8%)	36 (56.3%)	6.0 ± 2.8
Intraoperative rectal swab	34	11 (32.4%)	12 (35.3%)	11 (32.4%)	40.7 ± 16.5	14 (41.2%)	20 (58.8%)	5.4 ± 3.4
First postoperative feces	24	12 (50.0%)	8 (33.3%)	4 (16.7%)	34.0 ± 13.7	4 (16.7%)	20 (83.3%)	6.7 ± 2.9

Because stool was difficult to obtain in acutely ill patients (constipation and discharge before defecation were common), systematic three-way pairing of appendiceal content, rectal swab, and stool was not feasible. However, 21 patients contributed both appendiceal content and rectal swab samples, forming the paired subset (n = 21) used for within-patient comparisons.

Key demographic and clinical characteristics for participants contributing each sample type are summarized in Table 1. All appendicitis patients received perioperative antibiotics with treatment duration of approximately 5–7 days.

Microbial taxonomy and composition across sample types

We profiled microbial taxonomic composition across the four matrices—healthy feces (HC), appendiceal content (AC), intraoperative rectal swabs (RS), and first postoperative feces (IF)—at multiple ranks. Figure 2 summarizes the mean relative abundances (with variability across samples) at the phylum, order, genus, and putative species levels to facilitate direct visual comparison across matrices.

Fig. 2.

Taxonomic composition of gut microbiota across sample groups. Relative abundances of taxa are shown for HC (healthy controls), AC (appendiceal content), RS (intraoperative rectal swabs), and IF (first postoperative feces). (A1) Major phyla. (A2) Low-abundance phyla displayed on an expanded y-axis to improve visualization. (B1) Major orders. (B2) Low-abundance orders displayed on an expanded y-axis to improve visualization (including Campylobacterales). (C) Genus-level relative abundance. (D) Species-level assignments (putative, based on 16S rRNA amplicon data). Bars represent mean relative abundance, with error bars indicating variability across samples. Taxa shown in the low-abundance panels (A2 and B2) are not duplicated in the corresponding major-taxa panels (A1 and B1)

At the phylum level (Fig. 2A1–A2), both AC and RS were dominated by Bacteroidota and Firmicutes, with visibly higher contributions of Bacteroidota, and Proteobacteria and lower contributions of Actinobacteriota compared with HC. In contrast, IF showed a more distinct postoperative pattern, with a conspicuously higher proportion of Actinobacteriota and minimal Fusobacteriota relative to the intraoperative matrices. Low-abundance phyla that are difficult to discern on the main y-axis are displayed separately on an expanded scale (Fig. 2A2), including Campylobacterota and several other minor lineages.

At the order level (Fig. 2B1–B2), RS broadly recapitulated the AC profile, supporting RS as a practical proxy for capturing appendicitis-associated microbial shifts. In both AC and RS, the overall composition was dominated by Bacteroidales and several inflammation-associated orders, with a relative reduction of common fecal commensal-associated orders (e.g., Lachnospirales/Oscillospirales) compared with HC. In contrast, first postoperative feces (IF) showed a more distinct postoperative pattern, with relatively higher contributions from Lactobacillales and Bifidobacteriales and lower contributions from Bacteroidales than in AC and RS, consistent with an antibiotic-exposed, early postoperative gut environment. To improve readability, lower-abundance orders are displayed separately on an expanded y-axis (Fig. 2B2), including Fusobacteriales, Pseudomonadales, Campylobacterales and other minor orders that are difficult to discern in the main plot and are detected predominantly in the appendicitis-related matrices.

At the genus level (Fig. 2C), HC displayed a commensal-enriched profile dominated by Bacteroides, with relatively higher Faecalibacterium and Blautia compared with the other groups; Bifidobacterium was also present at a moderate level. In AC, the overall profile shifted away from these commensal genera, with lower Faecalibacterium and Blautia and a more prominent contribution from the Escherichia/Shigella group and Fusobacterium (with Pseudomonas also noticeable). RS showed a slightly distinct pattern characterized by higher Prevotella and several anaerobic genera (e.g., Anaerococcus, Peptoniphilus and Finegoldia), while Enterococcus remained low in RS (as well as in AC). IF exhibited increased Bifidobacterium together with higher Parabacteroides and Collinsella, and Enterococcus was relatively more abundant in IF than in AC/RS (while still being present in HC). Streptococcus was detected across all groups with domination in IF. Overall, substantial inter-individual variability was observed across genera, as reflected by the error bars. At the putative species level (Fig. 2D), HC was characterized by higher Faecalibacterium prausnitzii and other commensal-associated taxa, whereas AC showed higher signals for disease-associated species including Escherichia coli and Fusobacterium nucleatum. RS overlapped with AC for several disease-linked features but also exhibited matrix-specific enrichments, particularly anaerobic taxa such as Finegoldia magna and Prevotella copri, that were low or absent in HC. Importantly, IF did not revert to a healthy-like profile; instead, it showed persistently reduced F. prausnitzii and relatively higher contributions from taxa such as Bifidobacterium pseudocatenulatum and oral-associated species (e.g., Peptostreptococcus stomatis and Streptococcus mitis), supporting a mixed early postoperative state rather than recovery toward HC.

To minimize inter-individual variation, we repeated the taxonomic comparisons in the 21 patients who provided paired AC–RS samples. Across class, family, and genus levels (Supplementary Fig. S5A–C), the paired analysis reproduced the main trends: AC and RS shared the same dominant lineages and showed concordant shifts versus HC, including higher Bacteroidia (with increased Prevotellaceae) and lower Actinobacteriota (notably Bifidobacteriaceae). Within this paired subset, AC and RS remained more similar to each other than to HC for the principal disease-associated lineages, consistent with the patterns seen in the full cohort (Supplementary Fig. S5).

Bacterial richness (alpha diversity) and community structure (beta diversity) across groups

Alpha diversity measures revealed subtle but important differences across the four groups. Richness estimators (Chao1 and ACE) were comparable among HC, AC, RS, and IF, with no statistically significant pairwise differences (all p > 0.05) (Supplement Fig. 1). In contrast, diversity indices that account for evenness showed selective reductions in the appendicitis samples. Shannon diversity in HC was comparable to RS and higher than AC, with IF significantly decreased (Fig. 3A). Simpson’s index showed little difference between HC and the other groups, but RS differed significantly from both AC and IF (p ≈ 0.04–0.042) (Fig. 3B). Phylogenetic diversity (PD whole tree) captured larger disparities. Both AC and RS exhibited significantly higher PD than HC (p < 0.0001), and IF was lower than AC (p = 0.0013), similar to HC (Fig. 3C). Taken together, richness and evenness were largely preserved in appendicitis in AC and RS (but not in IF). In contrast, PD significantly increased in AC and RS versus HC. Along with these results, in the paired AC–RS subset (Figure S4A–C), alpha-diversity differences were not significant among all the three groups.

Fig. 3.

Alpha and beta diversity of gut microbiota across groups. (A) Shannon diversity index; (B) Simpson index of dominance (higher values indicate lower diversity); (C) Faith’s phylogenetic diversity (PD, whole-tree). (D) Boxplot of Bray–Curtis dissimilarities comparing between-group versus within-group sample pairs. (E) Principal Coordinates Analysis (PCoA) of Bray–Curtis distances (PC1 vs PC2)

Beta diversity analyses demonstrated clear community compositional shifts associated with appendicitis. Boxplots of Bray–Curtis and Jaccard distances showed greater between-group dissimilarities than within-group, and PERMANOVA confirmed that overall community composition differed significantly among the four groups (adonis p < 0.01 for all distance metrics) (Supplement Fig. 2). The average Bray–Curtis distance between any appendicitis sample (AC, RS, or IF) and an HC sample was markedly higher than distances among HC samples (Fig. 3D). Unweighted UniFrac and Jaccard metrics (sensitive to presence/absence of lineages) showed especially strong separation between HC and the appendicitis-related groups (Supplement Fig. 2). In the Bray–Curtis PCoA (Fig. 3E), AC formed a distinct cluster separated from HC and RS along PC1, whereas IF overlapped with HC and RS (and HC and RS also showed partial overlap). Consistently, in the paired AC–RS subset (n = 21), Bray–Curtis analyses and PCoA also showed significant separation between the two matrices (PERMANOVA R² = 0.145, p = 0.001; Supplementary Fig. S4D–E). Across ordination plots, AC remained the most clearly separated group, whereas IF occupied an intermediate region overlapping with HC and RS.

Predicted functional profiles of the microbiota

Predictive metagenomic profiling (KEGG pathway analysis) revealed pronounced functional shifts in the microbiome during appendicitis (Fig. 4A). Comparisons between appendiceal content and rectal swabs from appendicitis patients (AC vs. RS) showed that the bacterial communities within the inflamed appendix were functionally enriched in pathways related to motility, environmental sensing, and resistance. Specifically, AC had significantly higher abundances of genes for bacterial cell motility (e.g. flagellar assembly and chemotaxis pathways) and signal transduction than the corresponding fecal samples (RS). Pathways associated with antimicrobial resistance were also elevated in AC, including genes for multidrug efflux pumps and resistance to anti-microbial and anti-neoplastic compounds. On the other hand, core cellular functions were comparatively suppressed in AC. The protein biosynthesis machinery (indicated by the KEGG Translation category) and related processes like protein folding were significantly under-represented in AC relative to RS. Likewise, AC showed lower relative abundances of pathways related to nucleotide metabolism and DNA replication/repair than RS.

Fig. 4.

Predicted functional (KEGG pathway) shifts associated with appendicitis. Comparisons of selected KEGG Level 2 pathway abundances: (A) appendiceal content (AC) versus rectal swabs (RS) and (B) RS versus healthy feces (HC). Bars show mean pathway proportions, and points with horizontal lines indicate mean differences with 95% confidence intervals. Group abbreviations are as in Figure 1. Pathways within each panel are ordered by increasing adjusted p-value (from top to bottom), rather than alphabetically; therefore, the vertical order of pathways differs between panels A and B

When RS from appendicitis patients were compared with HC stool, clear differences in KEGG Level 2 pathway representation were observed (Fig. 4B). For example, genes involved in nucleotide and amino acid metabolism were significantly overrepresented in RS (difference + 0.58% for nucleotide pathways, p≪0.001), alongside increases in protein synthesis functions such as ribosomal assembly and chaperones (Translation + 0.64%, p < 0.001). In contrast, pathways related to biosynthesis and diverse metabolism were diminished in RS compared to HC. Key catabolic functions like carbohydrate and lipid metabolism were significantly reduced in the appendicitis stool (e.g. carbohydrate metabolism − 0.69%, p < 0.001; lipid metabolism − 0.10%, p < 0.001). Similarly, routes for producing secondary metabolites and vitamins were downshifted (e.g. secondary metabolite biosynthesis − 0.18%, p < 0.00001). The RS community also had lower representation of motility and community interactive processes (e.g. biofilm formation genes, flagellar components), as reflected in a ~ 35% drop in the Cell motility category versus healthy controls (p < 0.005).

To incorporate the early postoperative stool samples into the functional analysis, we additionally compared KEGG Level 2 pathway profiles between AC and HC, HC and IF, and RS and IF (Supplementary Figure S6A–C). The AC vs. HC contrast largely recapitulated the appendix-related functional shifts seen in the main analysis, with AC enriched for pathways related to bacterial infectious diseases, xenobiotic biodegradation and antimicrobial resistance, and relatively depleted in broad metabolic and biosynthetic categories such as carbohydrate metabolism and the biosynthesis of secondary metabolites. In contrast, only a limited set of pathways differed significantly between HC and IF, mainly involving energy metabolism, metabolism of other amino acids, immune system functions and a parasitic infectious-disease module, consistent with an early postoperative community that still carries an inflammatory and treatment-related functional imprint. Comparisons between RS and IF showed significant differences in energy, carbohydrate and lipid metabolism, immune system–related pathways, environmental adaptation and nucleotide metabolism, indicating that intraoperative rectal swabs and early postoperative stool reflect related but non-identical functional states of the gut microbiota after appendicitis and surgery. Because the IF group was relatively small (n = 24), these IF-based pathway comparisons should be interpreted as exploratory.

Rectal swabs as a proxy for appendiceal microbiota

To determine whether rectal swabs reflect the appendiceal microbiota during acute appendicitis, we compared genus-level profiles of appendiceal content (AC) and rectal swab (RS) samples. An OTU-level Venn diagram (Fig. 5A) showed substantial but incomplete overlap between AC and RS, with a shared core alongside many matrix-specific OTUs (Fig. 5A). Consistently, genus-level differential abundance analyses showed that key appendicitis-associated genera (e.g., Prevotella, Blautia) were identified as significantly altered in both AC and RS relative to healthy feces (HC) in same directions (Supplement File 2). In line with these results, LEfSe biomarker analysis revealed overlapping taxa distinguishing appendicitis from health in both AC and RS (Fig. 5B). Many genera identified as biomarkers in AC (vs. HC) also emerged as biomarkers in RS (vs. HC), underscoring shared microbial signatures between the two sample types during appendicitis.

Fig. 5.

Shared and distinguishing taxa among HC, AC, RS, and IF (Venn overlap and LEfSe cladogram). (A) OTU-level Venn diagram summarizing the overlap of taxa among HC, AC, RS, and IF. (B) Cladogram from LEfSe analysis (LDA > 2.0, p < 0.05) highlighting taxa enriched in each group; outer nodes are taxa, colored by the group in which they are over-represented (blue = AC, green = HC, orange = IF, red = RS)

Genus-level composition visualizations further support the relationship between rectal swabs and appendiceal content. The heatmap (Figure S3A) and stacked bar plots (Figure S3B) show a shared set of dominant genera in AC and RS—e.g., Escherichia/Shigella and Prevotella (with Anaerococcus relatively prominent in RS)—together with reduced commensals such as Faecalibacterium and Blautia compared with HC. However, samples did not collapse into a single mixed cluster, and matrix-specific differences remained.

Discussion

In this prospective cross-sectional microbiome study of adults with acute appendicitis and healthy individuals, we observed clear matrix- and timepoint-specific alterations in gut bacterial communities rather than a uniform “appendicitis signature”. Appendiceal content and rectal swabs shared similar appendicitis-associated taxonomic shifts, characterized by higher Bacteroidota and Proteobacteria and lower Actinobacteriota and butyrate-producing commensals (e.g., Faecalibacterium and Blautia) compared with healthy feces. In contrast, early postoperative feces showed a mixed, antibiotic-perturbed configuration (e.g., higher Actinobacteriota/Bifidobacteriales and Streptococcus and minimal Fusobacteriota) rather than a return toward the healthy profile. Predicted functional pathways also differed between matrices, with appendiceal content more enriched in motility, chemotaxis and antimicrobial-resistance functions and rectal swabs showing a narrower metabolic repertoire than healthy feces. Although confidence intervals around some diversity and differential-abundance estimates were wide, the direction and relative magnitude of these differences were consistent across analyses. Taken together, our findings indicate that rectal swabs provide a minimally invasive window into appendix-related dysbiosis in acute appendicitis, capturing key genus-level disease signatures, but they do not fully recapitulate the global community structure of the appendiceal microbiota.

At the compositional level, our findings are broadly consistent with and extend previous work on the appendiceal microbiome in acute appendicitis. Recent lumen-based and multi-site studies have repeatedly reported overrepresentation of Bacteroides, Escherichia/Shigella, Fusobacterium, Prevotella and Campylobacter in inflamed appendices, often accompanied by depletion of Bifidobacterium and other Actinobacteriota and butyrate-producing Lachnospiraceae, compared with non-inflamed appendices or healthy controls [4]. Our observation that AC and RS share enrichment of Bacteroidetes and Fusobacteriales and loss of Actinobacteriota and Faecalibacterium therefore fits well with this emerging picture of a polymicrobial, dysbiotic state involving both gut-associated taxa and oral pathobionts [3]. In terms of diversity, several studies based on appendiceal tissue or rectal swabs have found reduced alpha diversity and loss of health-associated commensals in appendicitis compared with controls, whereas others have reported only modest or no differences depending on the metric used [10]. Our data, showing similar richness but lower evenness and altered phylogenetic diversity in appendicitis-related matrices, suggest that disease may be characterized less by simple loss of taxa and more by an imbalance in dominance structure, with expansion of a subset of opportunistic lineages against a background of residual commensal diversity.

The predicted functional shifts observed in our PICRUSt2 analyses complement these taxonomic patterns. PICRUSt2 and related pipelines infer metagenomic potential from 16 S data and have been widely used to reveal enrichment of motility, chemotaxis, environmental sensing and antimicrobial-resistance functions in dysbiotic, inflammation-associated gut communities [11]. In line with this broader literature, AC in our cohort showed higher inferred capacity for bacterial motility, chemotaxis, two-component signalling and resistance functions than RS, whereas RS compared with HC was enriched in nucleotide and amino-acid metabolism, protein synthesis and membrane transport but depleted for carbohydrate and lipid metabolism, secondary metabolite biosynthesis and cell motility. These profiles are compatible with a community under combined inflammatory and antibiotic selection pressure: motile, stress-tolerant taxa occupy the inflamed appendix niche, while the rectal lumen microbiota displays a narrower, more competition-oriented metabolic repertoire and reduced cooperative ecosystem functions. Importantly, however, functional predictions from amplicon data remain probabilistic and should be interpreted cautiously, particularly in a setting where perioperative antibiotics and surgery may both reshape community structure [12]. Our appendix–rectum comparisons also add nuance to the growing literature on rectal swabs as a sampling tool, as studies in healthy volunteers and various patient populations have reported that rectal swabs recover alpha- and beta-diversity and broad phylum-level composition similar to stool, supporting their use as a practical alternative when faecal sampling is challenging [7]. By directly comparing AC, RS and stool across the cohort, and by performing within-patient analyses in the 21 paired AC–RS cases, we show that RS can indeed capture key genus-level signatures of appendix dysbiosis, consistent with earlier rectal-swab-based appendicitis cohorts [10], but that matrix-specific differences in global community structure and predicted function remain and should be considered when rectal swabs are used as proxys for the appendiceal lumen.

All patients received perioperative broad-spectrum antibiotics (with the first prophylactic dose administered 30 min before incision), which represents an important ecological perturbation of the gut microbiota and a potential confounder when interpreting disease-associated changes. AC and RS were collected intraoperatively shortly after this first dose, so antibiotic effects on these matrices are likely to reflect early, short-term shifts, whereas IF (first stool within 24 h) integrates both ongoing postoperative dosing and the immediate recovery period and therefore carries a stronger antibiotic imprint. Experimental and clinical studies have shown that even brief courses of antibiotics can rapidly reduce alpha diversity, deplete short-chain fatty acid–producing commensals such as Bifidobacterium and Faecalibacterium, and promote expansion of stress-tolerant Proteobacteria together with an increased burden of antimicrobial-resistant genes, with some alterations persisting for weeks or months [13–19]. In our cohort, IF exhibited a mixed early postoperative configuration with increased Actinobacteriota (notably Bifidobacteriales/Bifidobacterium) and Streptococcus and minimal Fusobacteriota, while phylogenetic diversity in IF was lower than AC and broadly similar to HC; together with the enrichment of predicted antimicrobial-resistance and stress-response functions in AC and RS, these findings likely represent a composite of appendicitis-related inflammation and antibiotic-induced dysbiosis rather than pure disease effects alone.

This study has limitations. First, the 16 S rRNA V3–V4 amplicon provides limited species-level resolution; accordingly, taxa are interpreted primarily at the genus level and any species labels are treated as putative rather than definitive. Second, functional shifts were inferred using PICRUSt2 and should be considered hypothesis-generating; confirmation with shotgun metagenomics and/or metabolomics is warranted. Third, all appendicitis samples were obtained after antibiotic initiation (first dose administered 30 min before skin incision): rectal swabs were collected intraoperatively, whereas IF represented the first postoperative stool within 24 h. Immediate antibiotic exposure and differences in specimen matrices (rectal swab vs. appendiceal luminal contents) may confound between-group differences; we mitigated this by emphasizing comparisons within shared windows (e.g., AC vs. RS intraoperatively, IF ≤ 24 h), but residual confounding cannot be excluded. Fourth, not all patients contributed every specimen type, and only 64 of 93 appendicitis cases (68.8%) yielded analyzable appendiceal content after sequencing quality control, resulting in uneven group sizes, partial pairing, and a risk of selection bias. Fifth, this was a single-center cohort without external validation, and host covariates (e.g., inflammatory biomarkers, histology) were not jointly modeled with microbiome features. Despite unified wet-lab and bioinformatic pipelines, unmeasured technical variation or residual batch effects remain possible. Also, numbers within individual pathology subtypes and fecalith strata were very small; although clear graphical display can aid interpretation, it cannot compensate for the resulting lack of statistical power, so we refrained from formal stratified comparisons in this dataset and instead report overall appendicitis patterns only.

Conclusion

In summary, our findings suggest that rectal swab samples may serve as a minimally invasive adjunct for appendiceal microbiome profiling rather than a full proxy. These findings indicate that rectal swabs capture appendix-related genus-level signatures but do not reproduce the appendix’s global community structure. As such, RS is a non-invasive adjunct for detecting disease-relevant microbial signals when direct appendiceal sampling or timely stool collection is not feasible, rather than a full substitute for appendix profiling. Early postoperative stool reflects partial microbiome rebound, underscoring the need to consider timing when interpreting recovery. Future multicenter studies with longitudinal follow-up and integrative ‘omics will be valuable to validate these signatures, refine swab-based diagnostic adjuncts, and map the trajectory of microbiome restoration after appendectomy.

Author contributions

Yuan-yuan Liu: guarantor of integrity of the entire study, study concepts, study design, manuscript preparation, and manuscript editing.Fei Xia: literature research, clinical studies, data acquisition.Yimuran Reyimu: literature research, clinical studies, data acquisition.Nuermamaiti Amidula: clinical studies.Yang Yang: data analysis, statistical analysis.Jing-tao Zhou: definition of intellectual content, manuscript review.

Funding

This work was supported by the Natural Science Foundation of Xinjiang Uygur Autonomous Region (Grant No. 2023D01A93).

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of The Seventh Affiliated Hospital of Xinjiang Medical University (Approval No. 20230505-04, dated May 5, 2023).

Competing interests

The authors declare no competing interests.