Fecal bile acid levels of IBD patients in comparison with healthy controls

The study cohort consisted of 17 healthy controls and 62 patients with IBD (CD n = 38, UC n = 24). Patients with CD and patients with UC showed comparable levels of C-reactive protein (CRP) and fecal calprotectin (Table 1). Details of the study cohorts are summarized in Table 1.

Table 1 Characteristics of the study groups. There were no significant differences between these groups (Body mass index: BMI; Glomerular filtration rate: GFR)

The levels of the 18 bile acid species analyzed were similar in feces of females and males in the control cohort and the IBD patients. CA (r = 0.691, P = 0.013) and CDCA (r = 0.660, P = 0.023) positively correlated with age in the control group. Fecal bile acids did not correlate with the age of the patients (P > 0.05 for all).

In comparison between IBD patients and healthy controls, we identified lower levels of the glycine conjugated bile acid HDCA (GHDCA, P = 0.018) in stool of IBD patients and higher levels of CA (P = 0.047) (Fig. 1a and Table S1).

Bile acid levels showed high inter-individual variations (Fig. 1a), and therefore, % of single bile acid species relative to total bile acid levels were calculated. Here, %CA (P = 0.040) was higher and %LCA (P = 0.045) was lower in IBD. Accordingly, the proportion of primary bile acid levels relative to total bile acid concentrations was increased (P = 0.01) and that of the secondary bile acids relative to total bile acid concentrations of IBD patients was decreased (P = 0.01) in comparison to healthy controls (Fig. 1b, c).

Fig. 1

Comparison of fecal bile acid levels between patients with IBD and controls. a Concentrations of bile acid species in stool of healthy controls (left bars) and IBD patients (right bars). Data are shown in a logarithmic scale to improve the visualization of low abundant bile acid species; b Levels of primary bile acids relative to total bile acid concentrations of IBD patients and controls; c Levels of secondary bile acids relative to total bile acid concentrations of IBD patients and controls. * P < 0.05

Fecal bile acids of CD and UC patients

There are distinct differences between CD and UC patients such as the location and depth of inflammation [7, 34], and therefore, both disease entities were analyzed separately and compared to healthy controls. CD patients had lower fecal levels of GHDCA (P = 0.008) compared to controls. UC patients exhibited lower levels of DCA (P = 0.036) and LCA (P = 0.01) in stool compared to healthy controls, with total secondary bile acid levels also being reduced (P = 0.036) (Table S2 and Fig. 2b).

It should be noted that CD and UC patients were comparable in age, BMI, CRP, and fecal calprotectin levels (Table 1). In CD %GHDCA (P = 0.007) was lower in contrast to controls. In UC %TCA (P = 0.035) was higher and %DCA (P = 0.008) was reduced in comparison to the controls. Accordingly, %primary bile acids of UC patients was higher (P = 0.004) and %secondary bile acids (P = 0.004) was lower in contrast to controls.

Fecal bile acid levels varied between CD and UC patients; CD patients had higher fecal HDCA (P = 0.033) and secondary bile acids (P = 0.045) compared to UC patients (Fig. 2a, b and Table S2). In CD %GHDCA (P = 0.02) was higher and %GCDCA (P = 0.04) was lower in comparison to UC.

Most studies agree that fecal primary bile acids increase, and secondary bile acids decrease in IBD [9, 11], and thus the ratio of secondary to primary bile acids was calculated. This ratio was low in UC patients in contrast to controls (P = 0.004) but did not significantly differ between CD patients and healthy controls (P = 0.080) (Fig. 2c).

Fig. 2

Fecal bile acids of CD and UC patients. a Fecal bile acids in stool of CD patients (left bars) and UC patients (right bars). Data are shown in a logarithmic scale to improve the visualization of different levels; b Secondary bile acids in stool of controls, CD and UC patients; c Ratio of secondary to primary bile acids in stool of controls, CD and UC patients. Data are shown in a logarithmic scale to improve the visualization of different levels. * P < 0.05, ** P < 0.01

Correlations of fecal bile acid species with markers of inflammation

In the IBD cohort and the CD patients none of the bile acids correlated with serum CRP or fecal calprotectin (Table 2). In UC patients, DCA, TDCA, LCA, TLCA, UDCA and HDCA, and accordingly total levels of secondary bile acids, negatively correlated with CRP. GLCA, HDCA and secondary bile acids negatively correlated with fecal calprotectin in UC (Table 2).

Table 2 Spearman correlation coefficients for the correlations of fecal bile acids with CRP and fecal calprotectin in IBD, CD and UC. * P < 0.05, ** P < 0.01, *** P < 0.001, not significant n.s

GLCA (P = 0.156), HDCA (P = 0.180), and total levels of secondary bile acids (P = 0.234) decreased with increasing fecal calprotectin in UC patients but these declines were not significant (Fig. 3a-c). However, the 8 UC patients with fecal calprotectin levels < 50 µg/ had higher fecal GLCA (P = 0.036), HDCA (P = 0.036) and a trend to increased levels of secondary bile acids (P = 0.066) compared to the 5 UC patients with calprotectin levels > 500 µg/g (Fig. 3a -c). In patients with CD, bile acid levels did not change with higher fecal calprotectin levels (data not shown).

Fig. 3

Relationship of fecal bile acids and fecal calprotectin. a HDCA levels in stool of UC patients with fecal calprotectin levels < 50 µg/g (8 patients), < 150 µg/g (8 patients), > 150 µg/g (2 patients) and > 500 µg/g (5 patients). Fecal calprotectin level of one patient was not documented; b GLCA in stool of UC patients categorized according to fecal calprotectin levels; c Secondary bile acids in stool of UC patients categorized according to fecal calprotectin levels; d %GCDCA in feces of UC patients in relation to fecal calprotectin levels. * P < 0.05

HDCA, GLCA and secondary bile acids were much higher in feces of the 3 CD patients compared to the 5 UC patients, all with calprotectin levels > 500 µg/g. These differences were, however, not significant probably because of low patient number (Fig. 4a - c).

Relative bile acid levels did not exhibit a correlation with CRP and fecal calprotectin in CD patients (Table 2). In UC, the relative proportions of GCA and GCDCA showed positive associations with CRP and fecal calprotectin (Table 2). Negative correlations of %DCA and CRP, and %HDCA and fecal calprotectin were observed. However, the percentage of GHDCA exhibited a positive association with fecal calprotectin (Table 2). The 8 UC patients with fecal calprotectin levels < 50 µg/ had lower %GCDCA (P = 0.020) than the 5 UC patients with calprotectin levels > 500 µg/g (Fig. 3d). Percent GCA (P = 0.077), %HDCA (P = 0.474), and %GHDCA (P = 0.076) did not significantly differ among UC patients with low and high calprotectin levels.

The 5 UC patients with calprotectin levels > 500 µg/g had higher %GCA and %GHDCA, and lower %GCDCA and %HDCA compared to the 3 CD patients with calprotectin levels > 500 µg/g. Here, the difference of %GCDCA between CD and UC patients was significant (Fig. 4d).

In the IBD cohort, %GCDCA positively correlated with fecal calprotectin (Table 2). Relative levels of primary bile acids positively, and that of secondary bile acids negatively correlated with fecal calprotectin (Table 2).

A receiver operating characteristic curve (ROC) was used to evaluate the predictability of GLCA, HDCA and secondary bile acids for discrimination of UC from CD patients in the whole cohort (Fig. 4e). The area under the ROC curve (AUROC) was 0.422 for GLCA, 0.296 for HDCA, and 0.292 for secondary bile acids to diagnose UC. The AUROC of %GCA was 0.661, for %GCDCA was 0.722, for %HDCA was 0.423 and for %GHDCA was 0.712. An AUROC of 0.7 to 0.8 is considered acceptable for the discrimination of two cohorts [35], and ROC curves for %GHDCA and %GCDCA are shown in Fig. 4e. The Youden index method was used to define the optimal cut-points for UC diagnosis, which were 42.8 × 10−6% for GHDCA (specificity: 79%, sensitivity: 67%) and 0.008% for GCDCA (specificity: 74%, sensitivity: 63%).

Fig. 4

Bile acid levels of patients with fecal calprotectin levels > 500 µg/g and receiver operating characteristic curve. (a) HDCA; (b) GLCA and (c) Secondary bile acids in feces of CD and UC patients with fecal calprotectin levels > 500 µg/g; (d) %GCA, %GCDCA, %HDCA and %GHDCA of patients with CD (left bars) and UC (right bars), all with fecal calprotectin levels > 500 µg/g; (e) Receiver operating characteristic curve for diagnosis of UC including all IBD patients

Relation of fecal bile acids with stool consistency

Excess bile acids in the colon can cause diarrhoea but the association of fecal bile acids and stool consistency in IBD is unclear [13, 21, 36]. The type of the stool was documented by the patients using the Bristol stool chart, where type 1 and 2 are constipation, type 3 and 4 normal stool, type 5 and 6 diarrhoea and type 7 watery stool. There was a negative correlation of fecal DCA (r = -0.588, P = 0.015), GDCA (r = -0.682, P = 0.001) and LCA (r = -0.637, P = 0.005) with the Bristol stool scale in UC. Accordingly, there was a negative correlation of total secondary bile acids with the Bristol stool scale (r = -0.643, P = 0.004). Such correlations were not existent in CD patients (DCA: r = -0.035, P = 0.833, GDCA: r = 0.119, P = 0.280 and LCA: r = -0.180, P = 0.280). DCA, GDCA and LCA declined in stool of UC patients with higher water content but did not change in feces of CD patients (Fig. 5a - h). In UC, GDCA levels of patients with normal and watery stool differed significantly (Fig. 5a). There was also a trend for lower levels of secondary bile acids in UC but not CD patients with diarrhoea (Fig. 5g, h).

CRP and fecal calprotectin levels did not correlate with the Bristol stool chart in UC (r = 0.435, P = 0.259 and r = 0.404, P = 0.336, respectively).

Fig. 5

Relationship of fecal bile acids and stool consistency. a GDCA levels in stool of UC patients with constipation (1 patient), normal stool (6 patients), diarrhoea (12 patients) and watery stool (5 patients); b GDCA levels in stool of CD patients with constipation (4 patients), normal stool (9 patients), diarrhoea (21 patients) and watery stool (4 patients); (c) DCA in stool of UC patients described in a; d DCA in stool of CD patients described in b; e LCA in stool of UC patients described in a; f LCA in stool of CD patients described in b; g Secondary bile acids in stool of UC patients described in a; h Secondary bile acids in stool of CD patients described in b. * P < 0.05

Association of fecal bile acids and medications

In IBD therapy, anti-TNF antibodies are commonly utilized [5]. In our cohort, 13 CD and 9 UC patients received this treatment. Total and relative levels of bile acids were similar in anti-TNF antibody treated and non-treated UC and CD patients (P > 0.05 for all).

Corticosteroids were administered to 9 CD and 8 UC patients, with those treated exhibiting similar levels of bile acids as the non-treated patients (P > 0.05 for all). The bile acid profile was comparable between these two groups. Mesalazine (8 CD and 13 UC) was associated with higher GDCA (P = 0.046) and with lower %GCA (P = 0.032) in UC (Fig. 6a, b). UC patients treated and not-treated with this drug had similar levels of fecal calprotectin and CRP (P = 0.111 and P = 0.981, respectively). Anti-interleukin 12/23 antibody therapy (12 CD and 6 UC), and azathioprine (4 CD and 2 UC) treatments were not associated with changes of fecal bile acid levels (data not shown).

Fig. 6

Associations of fecal bile acids with mesalazine therapy of UC patients. a GDCA levels of UC patients treated or not treated with mesalazine; b GCA levels relative to total bile acid concentration of UC patients treated or not treated with mesalazine. * P < 0.05