Clinical, microbial, and experimental features associated with Grade C postoperative pancreatic fistula and pancreaticojejunostomy disruption

Introduction

Postoperative pancreatic fistula (POPF) remains one of the most challenging complications after pancreatoduodenectomy (PD), contributing substantially to postoperative morbidity, mortality, prolonged hospitalization, and healthcare costs (1). According to the 2016 International Study Group of Pancreatic Surgery (ISGPS) classification (2), biochemical leak represents pancreatic fluid leakage without clinical consequences, whereas Grade C POPF is characterized by severe complications including hemorrhage, organ failure, reoperation, and death. Although both entities arise from postoperative pancreatic leakage, their clinical outcomes differ dramatically. In contemporary series, biochemical leak is common, whereas Grade C POPF develops in only a small proportion of patients (3).

Several well-established factors, including soft pancreatic texture and a small pancreatic duct, increase the risk of POPF (4-6). However, these factors alone do not adequately explain why some patients progress from a relatively controlled pancreatic leak to catastrophic clinical deterioration. Clinically, severe POPF is frequently accompanied by pancreaticojejunostomy disruption, a structural failure of the pancreaticoenteric anastomosis. Previous studies describe salvage management of anastomotic dehiscence and postoperative CT evaluation of the pancreaticojejunostomy (7,8). Nevertheless, whether disruption merely reflects advanced disease or represents a critical event associated with progression from biochemical leak to Grade C POPF remains unclear (9).

Accumulating evidence suggests that infection plays an important role in clinically relevant POPF (10,11). Positive drain cultures and enteric microorganisms have been associated with increased rates of hemorrhage, sepsis, and mortality after PD (12,13). Nevertheless, the mechanisms linking microbial contamination to severe fistula progression remain incompletely defined. Loss of pancreaticojejunostomy integrity may permit direct communication between the intestinal lumen and the peritoneal cavity, resulting in enteric contamination, alterations of the local microbial environment, and amplification of inflammatory responses. However, the microbiological and immunological changes associated with this process have not been systematically investigated.

Macrophages are central regulators of innate immunity and tissue injury (14). In response to microbial stimulation, pro-inflammatory macrophages release cytokines such as interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-α), and interleukin-1 beta (IL-1β), which can perpetuate local inflammation and impair tissue repair (15). Excessive macrophage activation has been implicated in a variety of postoperative inflammatory complications, yet its role in severe POPF remains largely unknown. Furthermore, currently available pancreatic fistula (PF) models primarily reproduce pancreatic leakage and do not adequately mimic enteric contamination (16,17), a feature frequently observed in patients with pancreaticojejunostomy disruption.

Because biochemical leak and Grade C POPF both originate from postoperative pancreatic leakage but exhibit markedly different outcomes, characterizing their associated clinical and biological features may support earlier risk recognition. We hypothesized that Grade C POPF would be associated with loss of pancreaticojejunostomy integrity and corresponding enteric contamination, microbial differences, and inflammatory amplification. We therefore evaluated clinical factors associated with Grade C POPF, compared the peritoneal microbiota and inflammatory profiles of patients with biochemical leak and Grade C POPF, described pancreatic tissues obtained during completion pancreatectomy, and experimentally assessed pancreatic leakage with enteric contamination.

We integrated clinical, translational, and experimental evidence to characterize Grade C POPF. We present this article in accordance with the STROBE and ARRIVE reporting checklists (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0336/rc).

Methods

Study design and patient cohort

This study integrated clinical cohort analysis, microbiological and immunological characterization, human tissue evaluation, and experimental animal modeling. The overall design is shown in Figure 1. A prospectively maintained database of consecutive patients undergoing PD at West China Hospital of Sichuan University, from March 2022 to March 2024 was retrospectively analyzed. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The clinical cohort was approved by the Ethics Committee of West China Hospital of Sichuan University (Approval No. 2022 Ethical Review [804]), and translational biospecimen analyses were approved under Approval No. 2023 Ethical Review [2040]. Written informed consent was obtained from all patients. The study was registered in the Chinese Clinical Trial Registry (No. ChiCTR2200065046) during the enrollment period; registration occurred after accrual began but before accrual ended, without changes to the stated study objectives or analysis framework.

Figure 1 Study design and associative framework for Grade C POPF. (A) Participant flow, exclusions, and patient selection. (B) Clinical findings in the entire cohort. (C) Translational analyses comparing BL and Grade C POPF. (D) Experimental allocation and evaluation of pancreatic leakage with enteric contamination. (E) Proposed associative framework linking pancreaticojejunostomy disruption, enteric contamination, microbial differences, inflammation, and Grade C POPF. BL, biochemical leak; CI, confidence interval; IL-1α, interleukin-1 alpha; IL-1β, interleukin-1 beta; IL-6, interleukin-6; NOS, nitric oxide synthase; OR, odds ratio; PF, pancreatic fistula; POD, postoperative day; POPF, postoperative pancreatic fistula; TNF-α, tumor necrosis factor-alpha.

Eligible patients were aged ≥18 years and underwent elective PD with complete clinical, radiological, and pathological records. Of 642 patients assessed, 74 were excluded: 5 died by postoperative day (POD) 3, 21 lacked POD5–7 computed tomography (CT) imaging, 18 lacked an available peritoneal drainage sample, 12 underwent an intervention or had secondary intra-abdominal contamination before sample collection, and 18 had incomplete data. The remaining 568 patients were included and analyzed (Figure 1A).

POPF was defined and graded according to the 2016 ISGPS criteria (2). The primary clinical endpoint was Grade C POPF.

The 568 included patients comprised 198 without biochemical leak or Grade B/C POPF, 296 with biochemical leak, 54 with Grade B POPF, and 20 with Grade C POPF. The study included: (I) full-cohort analysis of factors associated with Grade C POPF; (II) microbiological and inflammatory comparison of 20 biochemical leak and 20 Grade C POPF patients; and (III) experimental evaluation of pancreatic leakage with enteric contamination in rats.

For translational analyses, 20 patients with Grade C POPF and 20 patients with biochemical leak were selected from the study cohort. Patients with biochemical leak were selected to achieve comparable baseline demographic, operative, and pathological characteristics.

Assessment of pancreaticojejunostomy disruption

Pancreaticojejunostomy disruption was assessed on contrast-enhanced CT performed routinely between POD5 and POD7. Disruption was defined as a gap >2 mm between the remnant pancreatic duct and jejunal mucosa along the duct-to-mucosa suture line (Figure 2) (8,18). Two abdominal radiologists with >8 years of pancreatic-imaging experience independently reviewed CT images while blinded to clinical outcomes and translational results. Before consensus, agreement was excellent [Cohen κ=0.86, 95% confidence interval (CI): 0.74–0.98]; disagreements were resolved by consensus.

Figure 2 Representative radiological findings of pancreaticojejunostomy disruption. (A) Intact pancreaticojejunostomy demonstrating continuity between the pancreatic duct remnant and jejunal mucosa. (B) Pancreaticojejunostomy disruption identified on contrast-enhanced computed tomography obtained between postoperative days 5 and 7, characterized by separation of the pancreatic duct remnant from the jejunal mucosa at the duct-to-mucosa anastomosis. Arrows indicate the pancreaticojejunostomy site.

Peritoneal fluid collection

Peritoneal fluid was collected from drainage tubes positioned adjacent to the pancreaticojejunostomy between POD5 and POD7. Samples were immediately transferred on ice and processed within 2 hours.

Following centrifugation at 3,000 rpm for 10 minutes at 4 ℃, supernatants and pellets were separated. Supernatants were aliquoted and stored at −80 ℃ for cytokine analyses. Pellets were snap-frozen in liquid nitrogen and stored at −80 ℃ for microbial DNA extraction and subsequent 16S ribosomal RNA (rRNA) sequencing.

16S rRNA sequencing and bioinformatic analyses

Microbial DNA was extracted from peritoneal fluid pellets, and the V3–V4 regions of bacterial 16S rRNA were sequenced on the Illumina MiSeq platform (OE Biotech, Shanghai, China). Reads were processed in QIIME2, amplicon sequence variants were generated with DADA2, and taxonomy was assigned using SILVA. Alpha diversity was assessed with Chao1 and Shannon indices; Bray-Curtis distances, principal coordinate analysis, and PERMANOVA were used for beta diversity. Differential-genus analyses were exploratory and were not adjusted for multiplicity. PICRUSt2 was used only to predict, not measure, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway representation. The available sequencing dataset did not contain dedicated extraction-blank or no-template-control libraries; consequently, contamination of this low-biomass material cannot be excluded and the microbiome findings were interpreted cautiously.

Multiplex cytokine profiling

Peritoneal cytokine concentrations were measured using a 48-plex bead-based Luminex assay (Bio-Plex Pro Human Cytokine Screening Panel, Bio-Rad Laboratories, Hercules, CA, USA) performed by Shanghai Uninovo Biotechnology Co., Ltd. (Shanghai, China). All samples were analyzed in duplicate. Cytokine concentrations were calculated using five-parameter logistic regression based on recombinant cytokine standard curves. Key cytokines analyzed in this study included interferon-gamma (IFN-γ), IL-6, interleukin-1 alpha (IL-1α), IL-1β, and TNF-α.

Human pancreatic tissue specimens

Pancreatic tissue specimens were obtained from two patients who underwent completion total pancreatectomy for uncontrolled Grade C POPF. In both patients, pancreaticojejunostomy disruption was confirmed by postoperative CT imaging and intraoperative findings at reoperation. Control pancreatic tissues were obtained from three patients who underwent pancreatic resection for lesions initially suspected to be pancreatic neuroendocrine tumors but ultimately diagnosed as intrapancreatic accessory spleens. Histologically normal pancreatic tissue located more than 2 cm from the lesion was used as control tissue.

Experimental animal model

Animal experiments were performed under a project license granted by the Institutional Animal Care and Use Committee of West China Hospital of Sichuan University (No. 202205003), in compliance with institutional guidelines for the care and use of animals. The experimental unit was one rat. Twenty-four male Sprague-Dawley rats (250–300 g) were allocated by a computer-generated random sequence to four groups (n=6/group): sham, PF, PF + intestinal fistula, and PF + intestinal fistula + antibiotics. Group size was based on prior comparable pancreatic-fistula models and feasibility; no formal a priori power calculation was performed. Surgeons were aware of allocation because the procedures differed, whereas histological, immunofluorescence, biochemical, and cytokine outcome assessors were blinded to group allocation. The protocol and analysis plan were prepared before the experiment but were not registered in a public animal-study registry.

Animals were fasted for 8 hours before surgery with free access to 5% glucose saline. General anesthesia was induced and maintained using inhaled isoflurane. Following a midline laparotomy, the stomach, spleen, and proximal small bowel were gently mobilized to expose the pancreatic body and tail. For the PF model, the pancreatic parenchyma was transected on the left side of the portal vein, disrupting the splenic branch of the pancreatic duct while preserving splenic vessels (Figure 3A) (16,19). No closure of the pancreatic stump was performed, allowing continuous pancreatic leakage into the abdominal cavity. To model enteric contamination associated with pancreatic leakage, an additional intestinal fistula was created in the PF + intestinal fistula group. A jejunal loop approximately 2 cm distal to the ligament of Treitz was mobilized and fixed adjacent to the pancreatic transection site. A small enterotomy (~3 mm) was created on the antimesenteric border, and a silicone catheter was inserted and secured to maintain a persistent intestinal fistula (Figure 3B). Animals in the PF + intestinal fistula + antibiotics group received a broad-spectrum antibiotic cocktail consisting of vancomycin (0.5 g/L), ampicillin (1 g/L), metronidazole (1 g/L), and neomycin sulfate (1 g/L) in drinking water for 7 days before surgery and continuously until sacrifice. This broad-spectrum antibiotic cocktail has been widely adopted as a standard approach for intestinal microbiota depletion and has been validated in multiple experimental models examining microbiota-mediated host immune responses (20,21). Antibiotic-containing water was freshly prepared and replaced daily. Following surgery, animals received subcutaneous fluid resuscitation (30 mL/kg) and were monitored daily. On POD3, animals were euthanized, and pancreatic tissues and peritoneal fluid were collected for further analyses. One animal in the PF + intestinal fistula group died on POD2 and was excluded; therefore, 24 animals were allocated and 23 were analyzed (sham n=6, PF n=6, PF + intestinal fistula n=5, PF + intestinal fistula + antibiotics n=6). All remaining animals reached the predefined POD3 endpoint.

Figure 3 Experimental modeling of pancreatic leakage with and without enteric contamination. (A) PF model established by pancreatic transection without closure of the pancreatic stump, resulting in continuous pancreatic leakage. The green arrow indicates the pancreatic transection site. (B) Pancreatic fistula plus intestinal fistula model (PF + intestinal fistula), designed to reproduce enteric contamination associated with pancreatic leakage. A jejunal fistula was created adjacent to the pancreatic transection site and maintained with a catheter. The green arrow indicates the pancreatic transection site, and the blue arrow indicates the intestinal fistula catheter. PF, pancreatic fistula.

Histological, immunohistochemical, and immunofluorescence analyses

Pancreatic tissues were fixed in 4% paraformaldehyde, paraffin embedded, and sectioned at 4 µm thickness. Histological injury was evaluated using hematoxylin and eosin (HE) staining.

Macrophage infiltration was assessed by CD68 immunohistochemistry, and CD68/iNOS immunofluorescence was performed using antibodies from Abcam. For the rat experiments, five randomly selected high-power fields per animal were quantified by assessors blinded to allocation. Human tissue fields were summarized descriptively at the patient level; because only two Grade C cases and three controls were available, no inferential test was performed for the human tissue comparison.

Enzyme-linked immunosorbent assay (ELISA) assays

Ascitic fluid samples collected on POD3 were centrifuged at 3,000 rpm for 15 minutes at 4 ℃. Supernatants were used for biochemical and cytokine analyses. Ascitic amylase concentrations were measured using an automated biochemical analyzer. Concentrations of IFN-γ, IL-6, TNF-α, and IL-10 were quantified using commercially available ELISA kits (ABclonal, Woburn, MA, USA) according to the manufacturer’s instructions.

Statistical analysis

Continuous variables are presented as mean ± standard deviation or median (interquartile range), as appropriate; categorical variables are frequencies and percentages. Student t, Mann-Whitney U, χ², or Fisher exact tests were used as appropriate. Crude odds ratios (ORs) were audited against the 2×2 counts in Table 1; Table 2 reports their univariable-logistic Wald CIs and P values. Variables with P<0.10 in univariable analysis were entered into Firth penalized multivariable logistic regression to reduce sparse-data and separation bias; ORs and 95% CIs are reported. Cytokine comparisons used Welch two-sample t-tests with Benjamini-Hochberg false-discovery-rate adjustment. Because several cytokine distributions were skewed, the five prespecified cytokines were also assessed with Mann-Whitney U tests as a sensitivity analysis. Differential-genus comparisons were exploratory without multiplicity adjustment. No inferential testing was performed for the human tissue series. Animal groups were compared by one-way analysis of variance (ANOVA) with Tukey post hoc tests. Analyses used SPSS 27.0 and R 4.3.0; two-sided P<0.05 indicated statistical significance.

Results

Clinical characteristics and risk factors for Grade C POPF

Of 642 consecutive patients assessed, 74 were excluded: 5 died by POD3, 21 lacked POD5–7 CT, 18 lacked an available drainage sample, 12 underwent an intervention or had secondary contamination before sampling, and 18 had incomplete data. Thus, 568 patients were included. Among them, 296 (52.1%) developed biochemical leak, 54 (9.5%) Grade B POPF, and 20 (3.5%) Grade C POPF. The Grade C group (n=20) was compared with the heterogeneous non-Grade C group (n=548).

Pancreaticojejunostomy disruption was identified in 13/20 Grade C patients and 17/548 non-Grade C patients (65.0% vs. 3.1%, P<0.001). Corrected crude ORs were 6.75 (95% CI: 1.96–23.31) for soft texture, 4.98 (95% CI: 1.78–13.91) for duct <3 mm, 4.00 (95% CI: 1.51–10.56) for non-PDAC/CP pathology, 2.74 (95% CI: 1.12–6.74) for blood loss >500 mL, and 58.01 (95% CI: 20.54–163.80) for disruption. In the Firth multivariable model, disruption remained strongly associated with Grade C POPF (OR 21.84, 95% CI: 7.12–72.63, P<0.001); soft texture (OR 2.91, P=0.07), duct <3 mm (OR 1.82, P=0.32), non-PDAC/CP pathology (OR 2.47, P=0.11), and blood loss >500 mL (OR 2.08, P=0.17) were not statistically significant (Table 2).

Because pancreaticojejunostomy disruption was strongly associated with Grade C POPF, we next sought to characterize biological differences between biochemical leak and Grade C POPF to explore mechanisms underlying severe fistula progression.

Grade C POPF is associated with peritoneal microbiota dysbiosis

To investigate biological alterations associated with severe fistula progression, peritoneal fluid samples from 20 patients with biochemical leak and 20 patients with Grade C POPF were subjected to 16S rRNA sequencing. Baseline characteristics are presented in Table S1. Alpha-diversity analyses demonstrated lower microbial richness and diversity in Grade C POPF (Figure 4A,4B). At the genus level, Klebsiella and Fusobacterium were enriched, whereas Blautia, Lactobacillus, and Lachnospiraceae_NK4A136_group were depleted compared with biochemical leak (Figure 4C-4G). Beta-diversity analysis showed a modest between-group difference with overlapping distributions on principal coordinate analysis based on Bray-Curtis distances (PERMANOVA P=0.005; Figure 4H). PICRUSt2-based functional prediction suggested reduced representation of several microbial metabolic pathways (Figure 4I). These exploratory findings indicate peritoneal microbial differences between Grade C POPF and biochemical leak.

Figure 4 Peritoneal microbiota alterations associated with progression from BL to Grade C POPF. (A,B) Alpha-diversity analyses showing significantly reduced microbial richness (Chao1 index) and diversity (Shannon index) in Grade C POPF compared with BL. (C-G) Differentially abundant genera between groups. Grade C POPF demonstrated enrichment of Klebsiella and Fusobacterium, whereas Lactobacillus, Blautia, and Lachnospiraceae_NK4A136_group were significantly depleted. (H) Principal coordinate analysis based on Bray-Curtis distances showing a statistically detectable but visually overlapping between-group difference (PERMANOVA P=0.005). (I) Heatmap of KEGG pathway representation predicted by PICRUSt2. Grade C POPF exhibited reduced abundance of multiple microbial metabolic pathways, including phosphotransferase system, galactose metabolism, amino acid biosynthesis, lysine biosynthesis, alanine, aspartate and glutamate metabolism, methane metabolism, porphyrin metabolism, glyoxylate and dicarboxylate metabolism, quorum sensing, and 2-oxocarboxylic acid metabolism. *, P<0.05; ***, P<0.001. BL, biochemical leak; KEGG, Kyoto Encyclopedia of Genes and Genomes; PC, principal component; POPF, postoperative pancreatic fistula.

Grade C POPF exhibits inflammatory activation and accumulation of CD68+ iNOS+ macrophages (Figure 5)

Figure 5 Inflammatory activation and accumulation of CD68⁺ iNOS⁺ macrophages in Grade C POPF. (A) Representative gross specimens, HE staining, CD68 immunohistochemistry, and CD68/iNOS IF staining of control pancreatic tissues and pancreatic tissues obtained from patients undergoing completion total pancreatectomy for uncontrolled Grade C POPF. Both Grade C POPF patients had confirmed pancreaticojejunostomy disruption. Control pancreatic tissues were obtained from three patients with intrapancreatic accessory spleens and histologically normal pancreas (magnification ×200). (B) Heatmap showing differential expression of cytokines measured in peritoneal fluid using a 48-plex Luminex assay. (C) Descriptive quantification of CD68⁺ iNOS⁺ macrophages in pancreatic tissues; no inferential test was performed for the two cases and three controls. (D-H) Concentrations of IFN-γ, IL-6, IL-1α, IL-1β, and TNF-α in peritoneal fluid. Grade C POPF demonstrated significantly elevated levels of multiple pro-inflammatory cytokines compared with biochemical leak. ****, P<0.0001. HE, hematoxylin and eosin; IF, immunofluorescence; IFN-γ, interferon-gamma; IL-1α, interleukin-1 alpha; IL-1β, interleukin-1 beta; IL-6, interleukin-6; POPF, postoperative pancreatic fistula; TNF-α, tumor necrosis factor-alpha.

To evaluate local inflammatory responses associated with severe fistula progression, cytokine concentrations in peritoneal fluid were quantified using multiplex Luminex assays. Compared with biochemical leak, Grade C POPF demonstrated higher levels of IFN-γ, IL-6, IL-1α, IL-1β, and TNF-α (all false-discovery-rate-adjusted P<0.001; Figure 5D-5H). These five findings remained significant in the Mann-Whitney U sensitivity analysis (all P<0.001). Heatmap analysis of the entire exploratory cytokine panel showed a broadly pro-inflammatory profile in Grade C POPF (Figure 5B), with full results reported in Table S2.

To illustrate the tissue phenotype, pancreatic specimens from two patients who underwent completion total pancreatectomy for uncontrolled Grade C POPF were examined descriptively. Both patients had pancreaticojejunostomy disruption confirmed by postoperative imaging and intraoperative findings. Control pancreatic tissues were obtained from three patients with intrapancreatic accessory spleens. Histological examination illustrated extensive tissue destruction and inflammatory infiltration in the Grade C specimens compared with histologically normal pancreatic tissue (Figure 5A). Immunohistochemistry and immunofluorescence descriptively showed greater CD68-positive infiltration and CD68⁺ iNOS⁺ macrophage accumulation (Figure 5C). No patient-level hypothesis testing was performed because the tissue sample comprised only two cases and three controls; these findings are illustrative and cannot be attributed specifically to disruption.

Experimental enteric contamination recapitulates inflammatory features of Grade C POPF

To investigate whether enteric contamination contributes to the inflammatory phenotype observed in Grade C POPF, a rat model combining PF and intestinal fistula was established (Figure 6). Compared with the PF group, animals with combined PF and intestinal fistula exhibited more severe pancreatic injury characterized by extensive acinar destruction, inflammatory cell infiltration, and increased macrophage accumulation on histological examination (Figure 6A-6C). Immunofluorescence staining further demonstrated significantly increased accumulation of CD68⁺ iNOS⁺ macrophages in the PF + intestinal fistula group (Figure 6D). Consistent with these histological findings, ascitic fluid concentrations of IFN-γ, IL-6, and TNF-α were significantly elevated in animals with combined pancreatic and intestinal fistulas compared with those with PF alone (Figure 6E-6G). Ascitic amylase levels were also markedly increased, indicating aggravated pancreatic leakage and tissue injury (Figure 6H). Importantly, antibiotic treatment substantially attenuated pancreatic injury, reduced inflammatory cytokine production, decreased accumulation of CD68⁺ iNOS⁺ macrophages, and lowered ascitic amylase levels. These findings are consistent with an inflammatory contribution of enteric contamination following pancreatic leakage. Because antibiotics can exert direct antibacterial and anti-inflammatory effects, attenuation cannot be attributed solely to microbiota depletion.

Figure 6 Experimental enteric contamination recapitulates inflammatory features observed in Grade C POPF. (A) Representative operative photographs, HE staining, and CD68 immunohistochemical staining of pancreatic tissues from sham, PF, PF + intestinal fistula, and PF + intestinal fistula + antibiotics groups. (B) Representative immunofluorescence staining showing CD68 (green), iNOS (red), and DAPI (blue) in pancreatic tissues. (C) Quantification of CD68-positive macrophages. (D) Quantification of CD68⁺ iNOS⁺ macrophages. (E-G) Concentrations of IFN-γ, IL-6, and TNF-α in ascitic fluid measured by ELISA. (H) Ascitic amylase levels in each experimental group. Combined pancreatic fistula and intestinal fistula resulted in increased pancreatic injury, inflammatory cytokine production, ascitic amylase levels, and accumulation of CD68⁺iNOS⁺ macrophages, whereas antibiotic treatment attenuated these changes. *, P<0.05; **, P<0.01; ***, P<0.001. ELISA, enzyme-linked immunosorbent assay; HE, hematoxylin and eosin; IF, immunofluorescence; IFN-γ, interferon-gamma; IL-6, interleukin-6; PF, pancreatic fistula; POPF, postoperative pancreatic fistula; TNF-α, tumor necrosis factor-alpha.

Discussion

The present study addressed why most postoperative pancreatic leaks remain clinically inconsequential whereas a minority progress to Grade C POPF. Integrating clinical observations, microbiome analyses, human tissue evaluation, and experimental modeling, we identified pancreaticojejunostomy disruption as the factor most strongly associated with Grade C POPF. Grade C POPF was accompanied by microbial and inflammatory differences compared with biochemical leak. Experimental introduction of enteric contamination into a pancreatic-fistula model reproduced several inflammatory features, but the antibiotic findings did not isolate a microbiota-specific effect.

Although pancreatic leakage after PD is common, severe fistula-related complications remain relatively uncommon (22). In our cohort, Grade C POPF occurred in only 20/568 patients. In Firth penalized analysis, pancreaticojejunostomy disruption showed the strongest association with Grade C POPF (OR 21.84, 95% CI: 7.12–72.63). The other candidate predictors did not reach statistical significance. Soft pancreatic texture is a recognized risk factor in previous studies (4), but it did not reach statistical significance in our sparse-event Firth model. Because CT was obtained during postoperative evolution, this association does not establish that disruption preceded or caused Grade C POPF.

The temporal relationship between disruption and Grade C POPF cannot be established from POD5–7 CT. Pancreatic leakage, local inflammation, tissue ischemia, and anastomotic failure may interact bidirectionally. Loss of pancreaticojejunostomy integrity could permit enteric contamination (23,24), but disruption could also reflect advanced local inflammation. Our findings therefore support an association and a biologically plausible framework rather than a causal sequence.

For the translational comparison, biochemical leak was selected because it represents pancreatic leakage without severe clinical consequences, whereas Grade C POPF represents the severe endpoint. Grade B POPF is clinically relevant but generally does not meet the organ-failure, reoperation, or death criteria that define Grade C. It was not included in the focused translational comparison, which was designed to contrast biochemical leak with the severe Grade C endpoint; consequently, the full intermediate continuum was not assessed. Previous studies have documented positive drain cultures and enteric microorganisms in clinically relevant POPF (10,25,26). Grade C POPF samples showed lower diversity and differential representation of several genera compared with biochemical leak. These findings are exploratory, particularly because dedicated negative-control libraries were unavailable and genus-level multiplicity was not adjusted, and they do not establish that microbial differences caused clinical progression.

The microbiological changes were accompanied by profound inflammatory activation. Grade C POPF was associated with significantly elevated concentrations of IFN-γ, IL-6, IL-1α, IL-1β, and TNF-α, indicating activation of innate immune pathways. Importantly, these findings were mirrored in human pancreatic tissues. Pancreatic specimens obtained from two patients who underwent completion total pancreatectomy for uncontrolled Grade C POPF, both with confirmed pancreaticojejunostomy disruption, demonstrated extensive inflammatory destruction and marked accumulation of CD68⁺ iNOS⁺ macrophages. Given the central role of macrophages in coordinating cytokine production and tissue injury, these observations suggest that macrophage-driven inflammatory amplification may contribute to persistent pancreatic damage and failure of local healing (14,27). Because our analyses were based on CD68 and iNOS staining, the data are most appropriately interpreted as evidence of M1-like macrophage accumulation rather than definitive macrophage polarization.

A strength of this study is the experimental evaluation of enteric contamination in the setting of pancreatic leakage. Most available pancreatic-fistula models reproduce leakage alone and therefore do not capture enteric contamination (17,19,28). By combining pancreatic and intestinal fistulas, the rat model created simultaneous peritoneal exposure to pancreatic secretions and intestinal contents, thereby simulating a key biological feature of pancreaticojejunostomy leakage. Although it did not recreate the complete human anastomotic environment, the model provides a clinically relevant experimental approximation. This produced greater pancreatic injury, cytokine concentrations, ascitic amylase levels, and CD68⁺ iNOS⁺ macrophage accumulation than PF alone. Broad-spectrum antibiotic treatment attenuated these findings, but its direct antibacterial and anti-inflammatory actions preclude attributing the effect specifically to microbiota depletion. Confirmatory germ-free or microbiota-transfer studies are needed.

These findings have several potential clinical implications. First, postoperative CT evaluation should not be viewed solely as a tool for detecting fluid collections but also as a means of identifying structural failure of the pancreaticojejunostomy. Second, the presence of pancreaticojejunostomy disruption may identify a subgroup of patients at particularly high risk for progression to Grade C POPF and may justify intensified surveillance and more aggressive source control strategies. Finally, our results suggest that future therapeutic approaches should focus not only on drainage and hemorrhage management but also on limiting enteric contamination and modulating excessive inflammatory responses triggered by microbiota-immune interactions.

Several limitations should be acknowledged. This was a single-center observational study with only 20 Grade C events; conventional logistic-regression estimates are therefore vulnerable to sparse-data bias and separation. CT assessment on POD5–7 was partly concurrent with clinical evolution and cannot establish the temporal or causal direction between disruption and Grade C POPF. Requiring POD5–7 CT and excluding patients who underwent an intervention before sample collection may have selected against fulminant or early-intervened Grade C cases. The non-Grade C comparator was heterogeneous, and the extreme-group translational comparison did not represent the full clinical continuum. Drain fluid is a low-biomass specimen susceptible to skin-, drain-, and reagent-derived contamination. The 16S rRNA analyses provide limited functional resolution, and PICRUSt2 pathways are predictions rather than measurements. The human tissue series was small, used clinically different controls, and was descriptive; it cannot isolate effects attributable specifically to disruption. The rat model did not recreate the complete human pancreaticojejunostomy environment; however, its combined pancreatic and intestinal fistulas simulated concurrent pancreatic and enteric leakage relevant to anastomotic fistula. Finally, broad-spectrum antibiotics can exert direct antibacterial, systemic, and anti-inflammatory effects; without germ-free experiments or microbiota transplantation, their effects cannot be attributed solely to microbiota depletion.

Conclusions

In conclusion, pancreaticojejunostomy disruption was strongly associated with Grade C POPF. Grade C POPF was accompanied by peritoneal microbial and inflammatory differences compared with biochemical leak. The animal model simulated key features of pancreaticojejunostomy leakage and supports the possibility that enteric contamination contributes to inflammatory aggravation.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE and ARRIVE reporting checklists. Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0336/rc

Data Sharing Statement: Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0336/dss

Peer Review File: Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0336/prf

Funding: This work was supported by Sichuan Province Science and Technology Planning Project (No. 2023YFS0316).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0336/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The clinical cohort was approved by the Ethics Committee of West China Hospital of Sichuan University (Approval No. 2022 Ethical Review [804]), and translational biospecimen analyses were approved under Approval No. 2023 Ethical Review [2040]. Written informed consent was obtained from all patients. Animal experiments were performed under a project license (No. 202205003) granted by the Institutional Animal Care and Use Committee of West China Hospital of Sichuan University, in compliance with institutional guidelines for the care and use of laboratory animals.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Smits FJ, van Santvoort HC, Besselink MG, et al. Management of Severe Pancreatic Fistula After Pancreatoduodenectomy. JAMA Surg 2017;152:540-8. [Crossref] [PubMed]
2. Bassi C, Marchegiani G, Dervenis C, et al. The 2016 update of the International Study Group (ISGPS) definition and grading of postoperative pancreatic fistula: 11 Years After. Surgery 2017;161:584-91. [Crossref] [PubMed]
3. Luu AM, Krasemann L, Fahlbusch T, et al. Facing the surgeon's nightmare: Incidence and management of postoperative pancreatic fistulas grade C after pancreaticoduodenectomy based on the updated definition of the International Study Group of Pancreatic Surgery (ISGPS). J Hepatobiliary Pancreat Sci 2020;27:171-81. [Crossref] [PubMed]
4. Kang JS, Park T, Han Y, et al. Clinical validation of scoring systems of postoperative pancreatic fistula after pancreatoduodenectomy: applicability to Eastern cohorts? Hepatobiliary Surg Nutr 2019;8:211-8. [Crossref] [PubMed]
5. Teng Y, Wang Y, Yang S, et al. Surgery-related factors for pancreatic fistula after pancreatectomy: an umbrella review. Hepatobiliary Surg Nutr 2025;14:442-59. [Crossref] [PubMed]
6. Schuh F, Mihaljevic AL, Probst P, et al. A Simple Classification of Pancreatic Duct Size and Texture Predicts Postoperative Pancreatic Fistula: A classification of the International Study Group of Pancreatic Surgery. Ann Surg 2023;277:e597-608. [Crossref] [PubMed]
7. Kent TS, Callery MP, Vollmer CM Jr. The bridge stent technique for salvage of pancreaticojejunal anastomotic dehiscence. HPB (Oxford) 2010;12:577-82. [Crossref] [PubMed]
8. Seo JW, Hwang HK, Lee M, et al. Normal Postoperative Computed Tomography Findings after a Variety of Pancreatic Surgeries. Korean J Radiol 2017;18:299-308. [Crossref] [PubMed]
9. Shrikhande SV, Sivasanker M, Vollmer CM, et al. Pancreatic anastomosis after pancreatoduodenectomy: A position statement by the International Study Group of Pancreatic Surgery (ISGPS). Surgery 2017;161:1221-34. [Crossref] [PubMed]
10. Yamashita K, Kato D, Sasaki T, et al. Contaminated drainage fluid and pancreatic fistula after pancreatoduodenectomy: A retrospective study. Int J Surg 2018;52:314-9. [Crossref] [PubMed]
11. Xiang C, Chen Y, Liu X, et al. Prevention and Treatment of Grade C Postoperative Pancreatic Fistula. J Clin Med 2022;11:7516. [Crossref] [PubMed]
12. Yang F, Jin C, Li J, et al. Clinical significance of drain fluid culture after pancreaticoduodenectomy. J Hepatobiliary Pancreat Sci 2018;25:508-17. [Crossref] [PubMed]
13. Ciprani D, Capretti G, Sorrentino M, et al. Clinical usefulness of routinely performed drain fluid cultures after pancreaticoduodenectomy. A new ally to predict patients' outcomes? Pancreatology 2024;24:178-83.
14. Liu YC, Zou XB, Chai YF, et al. Macrophage polarization in inflammatory diseases. Int J Biol Sci 2014;10:520-9. [Crossref] [PubMed]
15. Sica A, Mantovani A. Macrophage plasticity and polarization: in vivo veritas. J Clin Invest 2012;122:787-95. [Crossref] [PubMed]
16. Tanaka T, Kuroki T, Adachi T, et al. Development of a novel rat model with pancreatic fistula and the prevention of this complication using tissue-engineered myoblast sheets. J Gastroenterol 2013;48:1081-9. [Crossref] [PubMed]
17. Kemmochi A, Tamura T, Shimizu Y, et al. A novel hydrogel sheet prevents postoperative pancreatic fistula in a rat model. J Hepatobiliary Pancreat Sci 2021;28:192-201. [Crossref] [PubMed]
18. Lee HJ, Kim JW, Hur YH, et al. Multidetector CT findings differ between surgical grades of pancreatic fistula after pancreaticoduodenectomy. Eur Radiol 2019;29:2399-407. [Crossref] [PubMed]
19. He W, Wang Y, Li X, et al. Sealing the Pandora's vase of pancreatic fistula through entrapping the digestive enzymes within a dextrorotary (D)-peptide hydrogel. Nat Commun 2024;15:7235. [Crossref] [PubMed]
20. Routy B, Le Chatelier E, Derosa L, et al. Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors. Science 2018;359:91-7. [Crossref] [PubMed]
21. Rakoff-Nahoum S, Paglino J, Eslami-Varzaneh F, et al. Recognition of commensal microflora by toll-like receptors is required for intestinal homeostasis. Cell 2004;118:229-41. [Crossref] [PubMed]
22. Marchegiani G, Bassi C. Prevention, prediction, and mitigation of postoperative pancreatic fistula. Br J Surg 2021;108:602-4. [Crossref] [PubMed]
23. Li GQ, Zhang T, Yang WG, et al. Gut microbiota patterns associated with somatostatin in patients undergoing pancreaticoduodenectomy: a prospective study. Cell Death Discov 2020;6:94. [Crossref] [PubMed]
24. Loos M, Strobel O, Legominski M, et al. Postoperative pancreatic fistula: Microbial growth determines outcome. Surgery 2018;164:1185-90. [Crossref] [PubMed]
25. Kawai M, Okada KI, Miyazawa M, et al. Positive drain fluid culture on postoperative day 1 predicts clinically relevant pancreatic fistula in early drain removal with higher drain fluid amylase after pancreaticoduodenectomy. Surgery 2023;173:511-20. [Crossref] [PubMed]
26. Schmitt FCF, Brenner T, Uhle F, et al. Gut microbiome patterns correlate with higher postoperative complication rates after pancreatic surgery. BMC Microbiol 2019;19:42. [Crossref] [PubMed]
27. Murray PJ, Allen JE, Biswas SK, et al. Macrophage activation and polarization: nomenclature and experimental guidelines. Immunity 2014;41:14-20. [Crossref] [PubMed]
28. Mikamori M, Gotoh K, Kobayashi S, et al. Self-assembling peptide hydrogel SPG-178 as a pancreatic fistula-preventing agent. Langenbecks Arch Surg 2022;407:189-96. [Crossref] [PubMed]