Disturbances of the Perioperative Microbiome Across Multiple Body Sites in Patients Undergoing Pancreaticoduodenectomy

Abstract

Objective: The goals of this study were to characterize bacterial communities within fecal samples, pancreatic fluid, bile, and jejunal contents from patients undergoing pancreaticoduodenectomy (PD) and to identify associations between microbiome profiles and clinical variables.

Methods: Fluid was collected from the pancreas, common bile duct, and proximal jejunum from 50 PD patients. Postoperative fecal samples were also collected. The microbial burden within samples was quantified with droplet digital polymerase chain reaction. Bacterial 16S ribosomal RNA gene sequences were amplified, sequenced, and analyzed. Data from fecal samples were compared with publicly available data obtained from volunteers.

Results: Droplet digital polymerase chain reaction confirmed the presence of bacteria in all sample types, including pancreatic fluid. Relative to samples from the American Gut Project, fecal samples from PD patients were enriched with Klebsiella and Bacteroides and were depleted of anaerobic taxa (eg, Roseburia and Faecalibacterium). Similar patterns were observed within PD pancreas, bile, and jejunal samples. Postoperative fecal samples from patients with a pancreatic fistula contained increased abundance of Klebsiella and decreased abundance of commensal anaerobes, for example, Ruminococcus.

Conclusions: This study confirms the presence of altered bacterial populations within samples from PD patients. Future research must validate these findings and may evaluate targeted microbiome modifications to improve outcomes in PD patients.

Figure 1. Bacterial load per sample assessed via ddPCR counts

A highly conserved region of the bacterial 16S rRNA gene was targeted for biomass quantification via droplet digital PCR. Dots represent counts from technical replicates, which are plotted behind a standard boxplot. Samples were analyzed from six patients with complete sample sets, i.e. patients for whom quality DNA sequencing reads were obtained from all body sites (fecal, pancreas, bile, and jejunal samples). Different colors represent samples from different patients.

Figure 2. Alpha diversity comparisons of microbial communities within samples from pancreaticoduodenectomy patients and adult citizen scientists participating in the American Gut Project

Shown are the Chao1 indicators for each sample group. Diversity is highest in fecal samples from both PD patients and AG participants. Lower diversity is observed in the pancreas, bile, and jejunum samples.

Figure 3. Beta diversity comparisons of microbial communities within samples from pancreaticoduodenectomy patients and AG particpants

Displayed are principal component analyses of weighted UniFrac distances between (A) all samples analyzed, including fecal samples from AG participants and (B) intraoperative samples from the pancreas, bile, and jejunum of PD patients. Axis labels indicate the proportion of variance explained by each principal coordinate axis. In panel (A), fecal samples from AG participants and fecal samples from PD patients cluster separately within PCoA space (PERMANOVA, p<<0.001, R = 0.02). As shown in panel (B), samples from the pancreas, bile, and jejunum overlap significantly within PCoA space rather than clustering according to body site.

Figure 4. Mean relative abundances of bacterial taxa in samples from pancreaticoduodenectomy patients and AG participants

Taxonomic assignments are shown at the level of (A) phylum and (B) genus. Note the expansion of Klebsiella (phylum Proteobacteria) in all PD samples relative to fecal samples from AG citizen scientists. A complete list of taxa found to be enriched or depleted within PD samples is provided in Supplemental Table 4.

Figure 5. Abundance of bacterial taxa in fecal samples from PDA patients with and without postoperative pancreatic fistula

Shown is the abundance of (A) Klebsiella and (B) Parabacteroides in fecal samples from 227 healthy adult volunteers, 42 PDA patients without fistula, and 8 PDA patients with fistula. For both genera, the mean relative abundance among patients with and without fistula was significantly different (p < 0.05).