Decreased Intestinal Microbiota Diversity is Associated With Increased Gastrointestinal Symptoms in Patients With Chronic Pancreatitis

Abstract

Objectives:

Chronic pancreatitis (CP) is characterized by abdominal pain, recurrent hospitalizations, frequent exposure to antibiotics, nutritional deficiencies, and chronic opioid use. Data describing the gut microbial community structure of patients with CP is limited. We aimed to compare gut microbiota of a group of patients with severe CP being considered for total pancreatectomy with islet autotransplantation (TPIAT) to those of healthy controls, and to associate these differences with severity of clinical symptoms.

Methods:

We collected stool from healthy donors (n = 14) and patients with CP (n = 20) undergoing workup for TPIAT, in addition to clinical metadata and a validated abdominal symptoms severity survey.

Results:

Patients with CP had significantly lower alpha diversity than healthy controls (P < 0.001). There was a significantly increased mean relative abundance of Faecalibacterium in healthy controls compared with CP (P = 0.02). Among participants with CP, those with lower alpha diversity reported worse functional abdominal symptoms (P = 0.006).

Discussion:

These findings indicate that changes in gut microbial community structure may contribute to gastrointestinal symptoms, and provide basis for future studies on whether enrichment of healthy commensal bacteria such as Faecalibacterium could provide clinically meaningful improvements in outcomes for CP patients undergoing TPIAT.

Introduction

Chronic pancreatitis (CP) and recurrent acute pancreatitis (RAP) are characterized by abdominal pain, frequent hospitalizations, nutritional deficiencies, and weight loss. When disease is severe and pain is refractory, a total pancreatectomy with islet autotransplantation (TPIAT) may be considered to reduce pain symptoms. Despite a large volume of emerging research on the association of intestinal microbiota with chronic diseases such as obesity, metabolic syndrome, and diabetes, little is known about the microbiome in patients with CP or after TPIAT. Patients with CP often have gastrointestinal symptoms, use opioids, are subject to progressive exocrine insufficiency, have dietary alterations, undergo repeated antibiotic exposure, and have small intestinal bacterial overgrowth,^1–4 all of which are potentially associated with intestinal dysbiosis.^3,5–9 Data describing intestinal (gut) microbial communities in individuals with chronic pancreatitis is limited, with only three studies in adults and one in children.^9–12 None of these prior studies have attempted to associate intestinal dysbiosis with clinical symptoms including gastrointestinal dysfunction, and only two of them have compared the gut microbial community structure of patients with CP to that of healthy controls.^9,11

There is evidence that bariatric surgery and other surgical procedures, such as pancreaticoduodenectomy, alter the intestinal microbiome.^13,14 Furthermore, changes in gut microbiota may contribute to certain metabolic and functional disorders of the gastrointestinal tract commonly observed after surgery and in non-surgical populations.^13,15,16 Changes in composition of gut microflora after TPIAT, however, have never been characterized.

We analyzed the stool microbiota from patients with CP referred for TPIAT. We compared the gut microbial community structure of CP patients with that of healthy controls, and within the CP group we associated intestinal dysbiosis with clinical characteristics of CP, including gastrointestinal symptomatology. We hypothesized that the intestinal microbiota of patients with CP would differ from those of normal controls, and that specific differences would be associated with symptom severity. Our first aim was therefore to compare gut microbial community structure of CP patients with that of healthy controls. Our secondary hypothesis was that the intestinal microbiome changes significantly after patients with CP undergo TPIAT. To this end, an exploratory aim of this study was to compare participants’ gut microbial structure from pre- to post-TPIAT. We believe that data generated from this study will provide an evidence base for developing interventions targeting gut microbiota for CP and TPIAT patients.

MATERIALS AND METHODS

Study Design

This was a prospective observational study of the intestinal microbiota in patients with chronic pancreatitis or recurrent acute pancreatitis being considered for and having undergone TPIAT. Data from healthy stool donors was used for comparison.

Recruitment

Adult patients with chronic pancreatitis being considered for TPIAT were approached at the time of initial consultation at our institution for participation in the study. Our institution’s definitions of chronic and recurrent acute pancreatitis, along with factors that determine candidacy for TPIAT, have been published previously.¹⁷ Baseline characteristics including demographics, medical history, pancreatitis history, opioid use, and antibiotic use within the three months before stool sample collection were recorded. For an exploratory aim of comparing CP to post-TPIAT, five patients who ultimately underwent TPIAT provided stool samples between 6–8 weeks postoperatively. All patients provided informed consent to participate and the study was approved by the University of Minnesota Institutional Review Board.

Healthy controls for comparison were recruited from the University of Minnesota stool donor program.¹⁸ These donors are screened for a variety of viral, bacterial, and parasitic enteric pathogens and meet rigorous inclusion and exclusion criteria previously described and in accordance with the Investigational New Drug Application 15071 sponsored by the University of Minnesota Microbiota Therapeutics Program. Healthy donors in this program have no history of chronic medical conditions, are not taking medications, have never undergone abdominal surgery, and have not taken antibiotics for at least three years before donation.

Fecal Sample Collection and Processing

All patients referred to the University of Minnesota under consideration for TPIAT are asked to provide a stool sample for fecal elastase quantification, to assess pancreatic exocrine function. Participants were asked to collect stool for study purposes at the same time as their clinical lab sample. Patients who provided stool samples after TPIAT did so at a routine post-operative clinic appointment. Stool samples for the study were kept at −20°C until study staff were able to transfer them to −80°C for storage, where they stayed until DNA extraction was performed.

We extracted DNA from 250 mg of thawed stool using the DNeasy PowerSoil DNA isolation kit (QIAGEN, Hilden, Germany) in accordance with the manufacturer’s instructions using the QIAcube inhibitor removal technology (IRT) protocol. Concentration of DNA was quantified using a Qubit 4 Fluorometer (Thermo-Fisher Scientific, Inc., Waltham, Mass). Extracted DNA was stored at −20°C until sequencing. Methods for extracting DNA were the same for healthy donor samples.

Amplicon Sequencing

The V4 hypervariable region of the 16S rRNA gene was amplified using methods previously described by our group.¹⁹ Pair-ended sequencing was done at a read length of 300 nucleotides on the Illumina MiSeq platform (Illumina, San Diego, Calif) by the University of Minnesota Genomics Center (Minneapolis, Minn). Healthy donors’ stool DNA was sequenced using the same methods, with each sample run in triplicate. Negative (sterile water) controls were included on each sequencing run.

Bioinformatics

Sequence data were processed and analyzed using mothur software, version 1.41.1 (Creative Commons, Mountain View, Calif.).²⁰ Operational taxonomic units were defined at 99% sequence similarity, and taxonomic classifications were made against the Ribosomal Database Project version 14 release (Creative Commons, Mountain View, Calif).²¹ Compositional data were non-normalized. For statistical comparisons, samples were rarefied to 5500 sequence reads in all samples by random subsample.²² Sequence data were deposited in the NCBI Sequence Read Archive under BioProject accession number SRP250716.

Clinical Characteristics

We collected clinical data thought to affect the microbiome, including recent opioid use (within three months), recent antibiotic use (within three months), diet (regular or tube feeds), smoking status (never smoker or current/prior smoker, including the use of cannabis for medical or recreational purposes), alcohol use, and presence of diabetes.

Association of Microbiome Characteristics With Symptoms of Gut Dysfunction

To associate characteristics of the microbiome with clinical symptoms of gut dysfunction, the Irritable Bowel Syndrome Symptom Severity Score (IBS-SSS) survey was administered to all participants at the time of stool sample collection ±3 days. The IBS-SSS is a clinically validated questionnaire for assessing symptoms of gut dysfunction, including abdominal pain and bloating, constipation/diarrhea, bowel movement frequency, overall patient satisfaction with their bowel habits, and symptom interference with daily life.^23,24 The survey is scored from 0–500 points, with 0 representing no symptoms and 500 representing symptoms that are severe, disruptive, and constant.

Technique of Total Pancreatectomy With Islet Auto-transplantation

Surgical technique of TPIAT has been described previously by our group.^17,25 For the current study, TPIAT was performed by a single surgeon (SC) in patients under 18 years of age, and by three primary surgeons (V.A.K., G.J.B., T.L.P.) in patients 18 years of age and over. Briefly, pancreatectomy is performed with pylorus preservation along with 1–2 cm of the duodenum (D1). The pancreas is dissected with attention to maintaining its blood supply until the time of removal, after which it is immediately placed in cold preservation solution and transported to the islet isolation lab. While the pancreas is processed, the gastrointestinal tract is reconstructed, most often with a Roux-en-Y duodenojejunostomy and choledochojejunostomy. Some of our patients are referred after undergoing a Whipple or other pancreas drainage procedure, and in these cases, reconstruction is tailored to the patient’s pre-existing anatomy.

Post-operative Management

For the participants who provided post-TPIAT stool samples, these were collected between 4–8 weeks after surgery, in the same fashion as pre-TPIAT samples. The IBS-SSS survey was administered again within three days of post-TPIAT sample collection.

All patients undergoing TPIAT receive broad-spectrum antibiotics within 30 minutes before incision, and are re-dosed as appropriate depending on the length of time in the operating room. The pancreas preservation solution and the final islet product are routinely sent for Gram stain and culture, and return positive for about 60% of patients based on previously published data.²⁶ For 48 hours post-operatively, while cultures are pending, patients receive prophylactic broad-spectrum antibiotics. If cultures return positive the antibiotic regimen is tailored to sensitivity results. Antibiotics are occasionally administered for other reasons as clinically appropriate, such as for surgical site or urinary tract infections. All perioperative antibiotic use was recorded for study participants.

All patients undergoing TPIAT are given an enteral feeding tube at the time of surgery. A majority will continue on at least some amount of tube feedings for 4–6 weeks postoperatively, with or without supplemental oral intake. For the purpose of the study, tube feeding regimens were noted at the time of post-operative stool sample collection.

A majority of patients referred for TPIAT use opioid pain medications chronically. They are slowly weaned off opioids as clinically appropriate following surgery. All opioid use at the time of post-operative stool sample collection was recorded for study participants.

Total pancreatectomy results in type 3c diabetes mellitus, and islet auto-transplantation is undertaken to lessen the effects of post-surgical diabetes without the need for immunosuppression. All TPIAT patients at our institution are managed with insulin during the initial islet engraftment phase (0–12 weeks following surgery) to maintain tight glucose control between 80 and 125–140 mg/dL. Therefore, all participants in the study were on insulin (without additional oral hypoglycemic medications) at the time of post-operative stool sample collection.

Statistical Analysis

Alpha and beta diversity were calculated using mothur software.²⁰ Alpha diversity was estimated with Shannon’s diversity index. Beta diversity was evaluated with Bray-Curtis dissimilarity matrices, and ordination of Bray-Curtis distances was done using principal coordinate analysis (PCoA). Spearman’s rank correlation was used to identify genera related to ordination position, and differences in composition among groups were evaluated with non-parametric analysis of similarity (ANOSIM).²⁷ Linear discriminant analysis (LDA) effect size (LEfSe)²⁸ was used to characterize differences in operational taxonomic units between groups. A P value of 0.05 was considered statistically significant, and Bonferroni’s correction for multiple comparisons was used when appropriate. Quantitative data is expressed as mean (standard deviation), or mean ± standard error (SE) and analyzed using the t-test, or one-way analysis of variance for multiple groups. Associations between clinical metadata from participants with chronic pancreatitis and Shannon’s index were determined by both simple and multiple linear regression. Regression models, along with descriptive statistics, were analyzed using SAS software (Version 3.8, ©2018 SAS Institute Inc., Cary, N.C.).

RESULTS

Baseline Demographics

Participants with CP (n = 20) and healthy stool donors (n = 14) were similar in terms of age, sex, and body mass index. Participants with CP had a variety of pancreatitis etiologies consistent with our usual TPIAT population, with hereditary CP as the majority. Three participants with CP (15%) had baseline pancreatic exocrine insufficiency (evidenced by stool fecal elastase <200 μg/g). Six (30%) had a history of tobacco use. One (5%) had diabetes at the time of stool sample collection. Sixteen (80%) had used opioids within three months of stool sample collection, and 7 (35%) had used antibiotics within three months of sample collection. Baseline mean relative abundance of genera in stool samples from chronic pancreatitis patients is displayed in Figure 1A.

FIGURE 1.

Relative abundances of gut microbial genera differ between chronic pancreatitis patients and healthy controls (HC). A, Mean relative abundances of most abundant genera for each individual patient with chronic pancreatitis. B, Cumulative mean relative abundance of genera for healthy controls and chronic pancreatitis patients.

Patients With Chronic Pancreatitis Have a Distinct Gut Microbial Composition Compared With Healthy Controls

Relative abundances of gut microbial genera differ between chronic pancreatitis patients and healthy controls (Fig. 1B). We performed LEfSe analysis to determine the features most likely to explain differences between the two groups of patients. Relative abundance of the genera Bacteroides and Escherichia/Shigella were significantly greater in patients with CP (LEfSe LDA >3.5, P = 0.03 and 0.02, respectively), whereas abundance of the genera Faecalibacterium and Coprococcus, and Clostridium XVIII were greater in healthy controls (HC) (LEfSe LDA >3.5, P = 0.001–0.02). Analysis of similarity demonstrated that beta diversity differed significantly between CP and HC (R² = 0.36, P = 0.002, Fig. 2). Principal coordinate analysis (PCoA) for comparison between the CP and HC microbial community structure confirmed the differences driven by Bacteroides and Escherichia/Shigella in CP (P(partial correlation [corr.]) < 0.001 and 0.01) and Faecalibacterium, Coprococcus, and Clostridium XVIII in HC (P(corr.) < 0.001–0.004).

FIGURE 2.

Beta diversity of gut microflora differs between patients with chronic pancreatitis and healthy controls. Principal coordinate analysis of chronic pancreatitis patients (blue) and healthy controls (orange) demonstrates significant separation of the two groups. R² = 0.36, ANOSIM P = 0.002. Bacteroides, Escherichia/Shigella, Faecalibacterium, Coprococcus, and Clostridium XVIII (P(corr.) ≤ 0.01) significant by Spearman’s rank correlation.

Alpha diversity was significantly lower in patients with CP (mean Shannon index, 3.57) than in HC (mean Shannon index, 3.96) (mean difference ± SE, −0.38 ± 0.09, P < 0.001) Among patients with CP, alpha diversity of those who had used antibiotics within three months of stool sample collection (mean Shannon index, 3.51) did not differ significantly from those who had not (mean Shannon index, 3.77) (mean difference ± SE, 0.26 ± 0.16, P = 0.1). Those who had used opioids within three months of stool sample collection (mean Shannon index, 3.65) also did not differ significantly from those who had not (mean Shannon index, S3.76) (mean difference ± SE, 0.12 ± 0.20, P = 0.6).

Association Between Clinical Symptoms of Gut Dysfunction and Alpha Diversity Among Patients With Chronic Pancreatitis

The IBS-SSS was significantly associated with alpha diversity (as measured by Shannon index) in participants with CP, so that participants who reported worse functional gastrointestinal symptoms (higher IBS-SSS) had, on average, lower Shannon index (less richness and evenness of the gut microbial community) (Pearson’s correlation r = −0.62, P = 0.006, Fig. 3). However, among candidate predictors of IBS-SSS (Table 1), recent use of opioids, patient sex, and the 4 categories combining opioid use and sex were also associated with a significantly higher IBS-SSS; after adjusting for these characteristics, the association between Shannon index and IBS-SSS was no longer statistically significant (P = 0.15). The effects of opioid use differed significantly between men and women, so that women with recent opioid use had the highest survey scores on average, while men without recent opioid use had the lowest survey scores (Kruskal-Wallis test P = 0.007).

FIGURE 3.

Alpha diversity, measured by the Shannon index, is associated with IBS-SSS survey results in patients with chronic pancreatitis. A higher score on the survey indicates worse, more frequent symptoms reported by the patient. For each 1.0 increase in the Shannon index, the survey score drops 308 points on average (standard error ± 98.2). Pearson’s correlation r = −0.62; P = 0.006.

TABLE 1.

Baseline Clinical Characteristics Influence Reported IBS-Type Symptoms Among Patients With Chronic Pancreatitis

	Mean Survey Score ± Standard Error	P
Shannon index	−308.2* ± 98.3	0.006
Years of pancreatitis pain	−2.0* ± 6.8	0.8
Opioid use within 3 mo of sample collection		0.02
No (n = 4)	160.0 ± 62.5
Yes (n =14)	313.6 ± 27.7
Etiology of pancreatitis		0.4
Alcohol (n = 1)	280.0
Hereditary (n = 7)	255.0 ± 48.1
Idiopathic (n = 7)	252.9 ± 47.5
Obstructive (n = 3)	398.3 ± 52.0
Age, y	1.4* ± 2.3	0.6
Sex		0.03
Female (n = 8)	348.8 ± 39.2
Male (n = 10)	224.0 ± 34.1
Any smoking history		0.4
No (n = 12)	262.9 ± 41.1
Yes (n = 6)	312.5 ± 30.2
Prior pancreas surgery		0.8
No (n = 14)	275.7 ± 34.7
Yes (n = 4)	292.5 ± 57.5
Fecal elastase, μg/g		0.7
<200 (n = 2)	247.5 ± 32.5
>200 (n = 14)	288.2 ± 36.7
Antibiotic use within 3 months of sample collection		0.4
No (n = 11)	258.2 ± 40.8
Yes (n = 7)	312.9 ± 39.0
Sex-by-opioid use categories		0.007
Female, recent opioids (n = 6)	391.7 ± 29.3
Male, recent opioids (n = 8)	255.0 ± 29.9
Female, no recent opioids (n = 2)	220.0 ± 90.0
Male, no recent opioids (n = 2)	100.0 ± 90.0

Univariate analysis of candidate predictors of symptoms of gastrointestinal dysfunction and their associations with the IBS-symptom severity score (IBS-SSS) in patients with chronic pancreatitis, tested with ANOVA for categorical predictors and simple linear regression for continuous variables. Bold font indicates statistically significant P value.

^*Mean change per unit

Association Between Alpha Diversity and Baseline Characteristics Among Patients With Chronic Pancreatitis

There were no statistically significant predictors of the Shannon index among selected baseline characteristics of CP patients, including antibiotic use within three months of sample collection, opioid use within three months, or exocrine pancreatic insufficiency, defined as fecal elastase <200 μg/g (P ≥ 0.05 for all simple associations, Supplemental Table 1).

Total Pancreatectomy With Islet Autotransplantation Leads to Significant Shifts in the Gut Microbial Composition of Patients With Chronic Pancreatitis

Relative abundances of gut microbiota differed between chronic pancreatitis patients and TPIAT recipients (Fig. 4). Stool samples from TPIAT recipients (n = 5) had significantly greater mean relative abundances of the genus Bacteroides and Clostridium XIVa clade compared with all chronic pancreatitis patient samples (LEfSe LDA >2.2, P < 0.001 and 0.04, respectively). In comparison, chronic pancreatitis samples had significantly greater mean relative abundances of the genera Roseburia and Fusicatenibacter compared with post-TPIAT samples (LEfSe LDA >3.5, P = 0.02 and 0.03, respectively). Beta diversity was significantly different between CP and post-TPIAT patients (R² = 0.41, P = 0.002). Roseburia and Fusicatenibacter in CP (Spearman’s P(corr.) < 0.001 and 0.001) contributed to these differences (Fig. 5). The increase in relative abundance of Bacteroides and Clostridium XIVa also appeared to drive the shift in gut microbial composition from pre- to post-TPIAT on PCoA (P(corr.) = 0.002 and 0.05, respectively).

FIGURE 4.

Relative abundance of gut microflora at the genus level differs between chronic pancreatitis patients and TPIAT recipients. A, Cumulative mean relative abundance of genera for patients with chronic pancreatitis (n = 20) and patients after TPIAT (n = 5). B, Mean relative abundances of abundant genera for 5 patients with chronic pancreatitis before (“pre”) and 6 weeks after TPIAT (“post”).

FIGURE 5.

Beta diversity of gut microbiota differs between patients with chronic pancreatitis and TPIAT recipients. Principal coordinate analysis of chronic pancreatitis (blue/triangle) and post-TPIAT (yellow/circle). R² = 0.41, ANOSIM P = 0.002. Bacteroides, Clostridium XIVa, Roseburia and Fusicatenibacter (P(corr.) ≤ 0.05) significant by Spearman’s rank correlation.

Despite these changes, the alpha diversity of gut microbes (Shannon index) did not change significantly from pre- to post-TPIAT (mean change ± SE, −0.36 ± 0.17, P = 0.1). Also, post-TPIAT Shannon index was not significantly associated with post-TPIAT IBS-SSS survey results (mean change in survey score ± SE, +230 points for each 1 point increase in Shannon index ± 364 points; P = 0.6), however post-TPIAT survey results were only available for 4 patients.

DISCUSSION

Changes in gut microbial community structure have been proposed to contribute to chronic disease in humans, but to date there are limited data on its characteristics in patients with chronic pancreatitis or the association of intestinal dysbiosis with clinical features of chronic pancreatitis. Patients with chronic pancreatitis have recurrent opioid and antibiotic exposure, exocrine insufficiency, and chronic inflammation, all of which could contribute to microbiome changes. The present study identifies features that distinguish the gut microbial community of patients with chronic pancreatitis from that of healthy controls, and suggests that these differences may contribute to the incidence and severity of abdominal symptoms.

Gut microbial communities of patients with chronic pancreatitis differ significantly from those of healthy controls. Consistent with prior studies, we observed significantly lower alpha diversity in patients with CP compared with healthy controls. In their study of 30 patients with CP (16 without diabetes and 14 with diabetes) and 10 HC, Jandhyala et al¹¹ described decreased alpha diversity among the participants with CP, after specifically excluding patients with recent antibiotic use. Among these CP patients, those with diabetes had a lower alpha diversity than those without, but the authors did not include an analysis of other clinical characteristics associated with alpha diversity. Wang et al⁹ also found decreased alpha diversity in their group of 30 children with chronic pancreatitis compared with healthy controls after excluding patients with recent antibiotic or probiotic exposure. However, neither of these prior studies considered other CP characteristics including opioid use or exocrine pancreatic insufficiency. Our study did not find a significant difference in alpha diversity among patients with CP based on clinical factors such as recent antibiotic use, opioid use, or exocrine pancreatic insufficiency. The potential mechanisms underlying reduced alpha diversity remain unclear. The medical complexity of patients with CP make it difficult to quantify potential mechanisms from a small sample size such as ours.

In our study, a second difference between the gut microbiota of patients with CP and those of HC was the relative abundance of genera Bacteroides and Faecalibacterium. Bacteroides was significantly increased in patients with CP, while Faecalibacterium was more abundant in HC. Jandhyala et al found a decrease in Faecalibacterium in patients with CP compared to their HC, though it did not reach statistical significance, and Wang et al reported a significantly decreased abundance of Faecalibacterium in CP relative to HC. Previous evidence indicates that Faecalibacterium is a beneficial commensal organism that produces butyrate (a source of nutrition for colonocytes) and other short-chain fatty acids via fermentation.¹⁴ Members of Faecalibacterium also have anti-inflammatory effects in the gastrointestinal tract.^29,30 The consistent findings of lower relative abundance of this genus across studies in CP patients may provide some basis for the hypothesis that modulating the gut microenvironment to promote enrichment of Faecalibacterium could be beneficial in CP.

Participants in our study who reported worse functional gastrointestinal symptoms on the IBS-SSS had significantly lower alpha diversity of their gut microbial communities. These findings are complicated, however, by further analyses suggesting that sex differences and opioid use also impact symptom scores. Women who used opioids within three months of stool sample collection reported the worst symptoms, and when taking these factors into account, the association of symptom scores with Shannon index was no longer statistically significant. These findings indicate that it may be difficult to separate sex differences and the side effects of opioids from other modifiable risk factors (such as decreased alpha diversity of the microbiome) as major causes of functional abdominal symptoms in patients with chronic pancreatitis. Further study in a larger group of participants to allow for sub-analysis by sex, or additional measures of abdominal pain and dysmotility, may lead to a better understanding of the relative contributions of the gut microbial community structure to gastrointestinal symptoms in patients with CP.

Although other investigators have previously reported changes in the gut microbiota that take place after gastrointestinal surgical procedures, the effects of TPIAT on gut microbial community structure have never been described. We were able to obtain post-TPIAT stool samples from five patients for comparison with their pre-operative samples to generate preliminary data for use in developing future observational and interventional studies of post-TPIAT intestinal dysbiosis. As expected, TPIAT resulted in a significant change in the composition of gut microflora, compared with both CP and HC samples. During and after TPIAT, patients undergo anatomic and functional changes to their gastrointestinal tract, in addition to broad-spectrum antibiotic exposure, several weeks on tube feedings, and fluctuations in use of opioid and other pain medications. Given these circumstances, it is not surprising to see a shift in their gut microbes. We observed greater relative abundances of Clostridium XIVa and Bacteroides in our post-TPIAT patients, compared with pre-TPIAT. Interestingly, this is similar to some, but not all, findings from other investigators who have published data on post-surgical gut microbial changes. Rogers et al described microbiome changes in 50 patients after pancreaticoduodenectomy (PD; Whipple’s procedure), and compared them to healthy controls.¹⁴ They found, similar to our results, an increase in the relative abundance of Bacteroides at the genus level after PD. However, they also reported an increased abundance of the genus Parabacteroides after PD, which is similar to our findings in the CP group prior to TPIAT.

Although our post-TPIAT patients’ gut microbial community structure remained statistically different from that of healthy controls, the increase in abundance of Bacteroides is encouraging. Kelly et al³¹ found that increased relative abundance of Bacteroides was associated with both healthy stool donors and those receiving fecal microbiota transplant from healthy donors for successful treatment of recurrent C. difficile infections. Bacteroides is generally considered a beneficial commensal organism that reduces inflammatory peptide levels in the gastrointestinal tract and synthesizes essential nutrients such as vitamin K.^31–34 To establish and validate the clinical relevance of our findings in this small group of patients, other studies of the post-TPIAT gut microbiota are needed, preferably including robust clinical metadata and associations with objective measurements of clinically relevant symptoms, as discussed above. For example, due to our small sample size, it is still unclear whether alpha diversity increases or decreases after TPIAT, and whether changes in relative abundance of certain genera drive meaningful improvements in clinical symptomatology.

Additional limitations of this study include the single-center design and the relatively small number of participants. We also did not have the resources to collect stool samples at multiple time points prior to TPIAT, and doing this in a future study might provide a more accurate overall description of gut microbial community structure in patients with CP by accounting for day-to-day and week-to-week variations. Future directions include analysis of the gut microbial community structure following TPIAT.

In conclusion, the present study suggests an altered gut microbial environment in patients with chronic pancreatitis, with implications for clinical symptoms of gastrointestinal dysfunction associated with CP. Similar to prior studies, our patients with CP had decreased amounts of beneficial commensal organisms such as Faecalibacterium, and lower alpha diversity compared with healthy controls. We report the novel finding that lower alpha diversity was associated with worse gastrointestinal functional symptoms, although these symptoms differed depending on sex and opioid use. Our findings provide a basis for future trials in patients with chronic pancreatitis to determine whether enrichment of Faecalibacterium or other beneficial commensal organisms results in clinically meaningful improvement in outcomes, and merit further research.