Characteristics of patients
Forty-seven patients were enrolled in our study, and their bile fluids were collected and analyzed. Of these, 19 underwent ERCP for recurrent CBD stones (the SR group), and the other 28 patients (the NSR) underwent ERCP for the management of benign cancer (n=13), biliary stricture (n=12), and CBD stones for the first time (n=3). Patient baseline characteristics were similar in the two groups. Patient baseline characteristics are summarized in Table 1 (Table 1).
Table 1
Patient Baseline Characteristics of SR and NSR groups
Variables
|
Total (n=47)
|
SR (n=19)
|
NSR (n=28)
|
p*
|
Age (years) §
|
75 (42-99)
|
78 (42-99)
|
75 (42-88)
|
0.44
|
Sex, male, n (%)
|
25 (53.2%)
|
7 (36.8%)
|
18 (64.3%)
|
0.07
|
HTN, presence, n (%)
|
21 (44.7%)
|
8 (42.1%)
|
13 (61.9%)
|
0.78
|
DM, presence, n (%)
|
12 (25.5%)
|
6 (31.6%)
|
6 (21.4%)
|
0.44
|
Dyslipidemia, presence, n (%)
|
19 (40.4%)
|
8 (42.1%)
|
11 (39.3%)
|
0.85
|
WBC (/ul)
|
8,779.4
(3,050-29,130)
|
8,152.6
(3,710-18,480)
|
9,204.6
(3,050-29,130)
|
0.51
|
CRP (mg/dL)
|
5.1
(0.1-28.7)
|
7.0
(0.1-24.0)
|
3.7
(0.1-28.7)
|
0.13
|
Total bilirubin (mg/dL)
|
4.4
(0.2-27.9)
|
2.5
(0.3-10.4)
|
5.6
(0.2-27.9)
|
0.09
|
AST (IU/L)
|
154.9
(12.0-663.0)
|
151.2
(12.0-535.0)
|
157.4
(19.0-663.0)
|
0.91
|
ALT (IU/L)
|
136.5
(8.0-688.0)
|
139.4
(8.0-510.0)
|
134.5
(10.0-688.0)
|
0.92
|
ALP (IU/L)
|
368.6
(41.0-3,435.0)
|
135.7
(41.0-267.0)
|
526.7
(45.0-3,435.0)
|
0.05
|
Abbreviation: HTN, hypertension; DM, diabetes mellitus; WBC, white blood cell count; CRP, c-reactive protein; AST, alanine aspartatetransferase; ALT, alanine aminontransferase; ALP, alkaline phosphatase
§, median (range)
*, P values were calculated using the t-test or the Chi-square test
Bile microbiome compositions in SR and NSR groups
Figure 1 shows microbiome compositions at the major phylum and genus levels in the SR and NSR groups. At the phylum level (Fig. 1A), five groups of microbiomes (proteobacteria, firmicutes, bacteroidetes, fusobacteria, actinobacteria) were detected in the bile fluid samples of the 47 patients at an average percentage composition of ≥ 1% of total microbiomes. In both groups, proteobacteria and firmicutes accounted for most of the microbiomes. In the SR group, fusobacteria, bacteroidetes, and actinobacteria were detected in decreasing order of abundance, while in the NSR group, bacteroidetes, fusobacteria, and actinobacteria were detected in of abundance. Phyla that accounted for < 1% of the total microbiome were designated as ‘others’. At the genus level of the proteobacteria phylum, enterobacteriaceae_unclassified was the most common in both groups. At the genus level, firmicutes phylum, enterococcus, and enterococcaceae_unclassified were most abundant in the SR group, whereas in the NSR group, streptococcus and enterococcus were most abundant (all in decreasing order). (Fig. 1B) As was performed at the phylum level, genera occupying < 1% were collected and expressed as ‘others’.
Microbial community heterogeneities in the SR and control groups
The significances of taxonomic compositional differences were determined using the ANOVA range test and Turkey’s HSD test. At the phylum level, no significant difference was observed between the compositions of microorganisms in the SR and NSR groups (Fig. 2A). However, enterococcus and enterococcaceae_unclassified were significantly more abundant at the genus level in SR group. (Fig. 2B)
Linear discriminant analysis effect sizes (LEfSe) were determined to identify major differences between the biliary microbiomes of the SR and NSR groups. Based on linear discriminant analysis (LDA) selection, enterococcus (LDA score 5.2) and enterococcaceae_unclassified (LDA score 4.7) proportions were significantly greater in the SR group at the genus level. (Fig. 3) Alpha diversity metrics were assessed to quantify heterogeneities of microbial communities in samples. A comparison of alpha diversity scores of the SR and NSR groups using the ANOVA range test and Turkey’s HSD test failed to reveal any significant difference (Fig. 4). We attributed this null result to compromised patient health due to the presence of pancreatic or biliary diseases. AMOVA testing and Non-metric Multidimensional Scaling (NMDS) ordination of beta-diversity analyses at the ASV (Amplicon Sequence Variant) level showed no significant differences between the SR and NSR groups (p=0.1) (Fig. 5)
Relationships between the Bile and Duodenal Microbiome and CBD stone recurrence
Several studies have reported that recurrent CBD stone is related to duodenal biliary reflux. To analyze the microbiomes of duodenal tissues and elucidate the origin of bile enterococcus in the SR group, we analyzed duodenal mucosa biopsy and bile fluid samples obtained from 5 patients in the SR group.
A comparison of the microbiome compositions of duodenum and bile fluid of these five patients showed that for each patient, proteobacteria compositions were almost identical in bile fluid and duodenal mucosa from the genus to phylum levels. (Fig. 6) Furthermore, proteobacteria and firmicutes largely accounted for compositions at the phylum level in bile fluid and duodenal mucosa. (Fig. 6A) At the genus level analysis, the most common genera in bile fluid and duodenum tissues were enterobacteriaceae_unclassified in the proteobacteria phylum and enterococcus in the firmicutes phylum. (Fig. 6B)
Taxonomic compositions were analyzed using the ANOVA range test and Turkey’s HSD test at the genus level. No significant (p=0.66, Welch’s t-test) enterococcus compositional difference was observed between bile fluid and duodenal mucosa at the genus level. (Fig 7A)
Additionally, alpha-diversities of bile fluid and duodenal mucosa were determined using Chao1, Shannon and Simpson diversity indices. (Fig 7B) The microbiome compositions of the bile fluid and duodenal mucosal samples collected from the five SR patients were analyzed, and Chao1 and Shannon indices were significantly different, but Simpson indices were not. (Chao1: p = 0.047; Shannon: p = 0.034; Simpson: p = 0.241) These results indicated that microbiome compositions and environments are similar in the bile duct and duodenum in recurrent CBD stone patients. Subsequently, we compared enterococcus compositions in bile fluid and duodenum, which differed in the SR group, in order to investigate the possibility that enterococcus in bile fluid originated from duodenum. We conducted NMDS ordination based on Bray–Curtis’s dissimilarities in genetic diversity in a microbial community to cluster 16S rRNA sequences into OTUs based on sequence similarities. (Fig 8) NMDS analysis showed no significant difference between the microbial environments of bile fluid and duodenal biopsy samples in the 5 SR patients. (p=0.46). These results supported the notion that enterococcus in bile fluid originated from duodenum.
Microbial community heterogeneity between the bile and duodenal microbiomes of recurrent CBD stone patients and the duodenal microbiome of Controls
Finally, to compare the microbiome compositions of duodenal mucosae in the SR group with controls (gastric ulcer patients, n=12) without any kind of biliary disease, duodenal mucosa biopsy samples were collected and analyzed.
NMDS ordination of beta-diversity analyses at the ASV level demonstrated a significant difference between the duodenal compositions of the SR group and controls, indicating the environments represented differed. The beta diversity analysis of microbiome diversity based on weighted Braycurtis-distance also showed the duodenal microbial environments in SR patients and controls significantly different. (Fig. 9)