Comparison of the Metabolomic Profiles of Irritable Bowel Syndrome Patients With Ulcerative Colitis Patients and Healthy Controls

New Insights Into Pathophysiology and Potential Biomarkers

Ammar Hassanzadeh Keshteli; Karen L. Madsen; Rupasri Mandal; Guy E. Boeckxstaens; Premysl Bercik; Giada De Palma; David E. Reed; David Wishart; Stephen Vanner; Levinus A. Dieleman

Disclosures

Aliment Pharmacol Ther. 2019;49(6):723-732. 

In This Article

Results

In the IBS cohort, most patients were either IBS-M (64.1%) or IBS-D (25.6%). The mean IBS-SSS was ~285 indicating that most patients were in the moderate to severe symptom range.[16] The majority of UC patients had a history of either left-sided colitis or pancolitis and their partial Mayo score was <3 (range: 0-2). Their median (interquartile range) values of FCP and CRP were 144.0 μg/g (77.0-343.4 μg/g) and 1.1 mg/L (0.6-3.1 mg/L) respectively. Comparisons of the demographic data (Table 1) show that the IBS patients were older and predominantly female relative to the UC patients and healthy controls. Ten (25.6%) IBS patients were reported to be on anti-depressant medications.

Metabolomic Profile of IBS vs UC Patients

Partial least squares discriminant analysis (PLS-DA) models showed that urinary metabolomic profiles differentiated IBS patients from patients with UC using metabolites identified by a combination of DI- LC- MS/MS and GC-MS assays (P < 0.001, Figure 1). The VIP score of fourteen metabolites were above 1.5 indicating their significant role in the discrimination between IBS and UC patients (Figure S1). The mean levels of these metabolites, which were either amino acids or organic acids are presented in Table 2. Based on logistic regression analysis, a model for the prediction of IBS vs UC patients was developed which included histidine, lactic acid, proline and Sumiki's acid. The associated ROC curve using these metabolites had an AUC of 0.99 (95% confidence interval [CI], 0.95-1.00) in the validation sets. In the validation set, the specificity and sensitivity of the model for prediction of IBS were 99.9% and 99.8% respectively.

Figure 1.

A, Principal component analysis plot and (B) Partial least squares discriminant analysis plot showing discrimination of patients with irritable bowel syndrome from ulcerative colitis patients in clinical remission using their urinary metabolomic profile assessed by direct infusion/liquid chromatography tandem mass spectrometry and gas-chromatography mass spectrometry assays (R 2 = 0.85, Q 2 = 0.76, P < 0.001)

Supplementary Figure 1.

Variable Importance in Projection (VIP) plot of major metabolites in urine responsible for the discrimination between irritable bowel syndrome (IBS) and ulcerative colitis (UC) patients.

Metabolomic Profile of IBS vs Healthy Controls

Irritable bowel syndrome patients had different urinary metabolic profiles compared with healthy controls using metabolites identified by DI- LC- MS/MS and GC-MS assays (P = 0.01, Figure 2). Metabolites that were the most useful in discriminating IBS patients from healthy controls are presented in Table 3 and Figure S2. Four metabolites including 2-methylsuccinic acid, PC ae C38:6, palmitic acid and PC aa C34:4 were used to generate a model to differentiate IBS patients from healthy controls using regression analysis. The associated ROC curve had an AUC of 0.89 (95% CI, 0.72-1.00) in the validation set. The specificity and sensitivity of the model to predict IBS were 57.1% and 100% in the validation set respectively.

Figure 2.

A, Principal component analysis plot and (B) Partial least squares discriminant analysis plot showing discrimination of patients with irritable bowel syndrome from healthy controls using their urinary metabolomic profile assessed by direct infusion/liquid chromatography tandem mass spectrometry and gas-chromatography mass spectrometry assays (R 2 = 0.79, Q 2 = 0.45, P = 0.008)

Supplementary Figure 2.

Variable Importance in Projection (VIP) plot of major metabolites in urine responsible for the discrimination between irritable bowel syndrome (IBS) patients and healthy controls (HC).

Metabolomic Profile of IBS-M vs IBS-D Patients

In this study, 35/39 (89.7%) IBS patients were classified as either IBS-M or IBS-D based on Rome III criteria and these patients were further analysed. They were well matched for age and gender (age = 49.9 ± 16.7 vs 49.1 ± 15.5 years, P = 0.84; females = 88.0% vs 80.0%, P = 0.61 respectively). As shown in Figure 3, urinary metabolomic profiles of IBS-M and IBS-D patients were not different from each other (P = 0.49). However, as shown in Table 4 patients with IBS-M had higher 3-methyladipic acid and lower suberic acid levels in comparison to IBS-D patients.

Figure 3.

A, Principal component analysis plot and (B) Partial least squares discriminant analysis plot showing no discrimination of patients with diarrhoea- predominant irritable bowel syndrome (IBS-D) from mixed irritable bowel syndrome (IBS-M) using their urinary metabolomic profiles (R 2 = 0.54, Q 2 = –0.6, P = 0.49)

Metabolomic Profiling and Severity of IBS

Total IBS-SSS score correlated significantly with a number of urinary metabolites including histamine, aspartic acid, methylmalonic acid, phosphatidylcholine diacyl C38:4 (PC aa C38:4), PC ae C36:2 (Table 5). These metabolites especially correlated with abdominal pain and/or abdominal distention domains of IBS-SSS.

Correlation of Urinary Metabolites With gut Microbes in UC and IBS Patients

Microbial composition of IBS and UC patients is presented in Figure S3. In UC patients, lysine levels were negatively correlated with abundance of Clostridiales (corrected r = −0.50) and proline level was correlated positively with abundance of Erysipelotrichales (corrected r = 0.39) (Figure S4). In IBS patients, oxoglutaric acid was correlated negatively with Bifidobacteriales (corrected r = −0.48) and palmitic acid was correlated positively with Bacteroidales (corrected r = 0.41) (Figure S5, Supplementary Table 1).

Supplementary Figure 3.

Microbial composition levels at phylum (A, B) and order levels (C, D) on stool samples in irritable bowel syndrome (A, C) and ulcerative colitis (B, D) patients.

Supplementary Figure 4.

Correlation between urinary metabolites and gut microbial composition in ulcerative colitis (UC) patients. Spearman's rank correlations were calculated between taxonomic data (order level) and urinary metabolites in UC patients that could discriminate them from irritable bowel syndrome patients. The correlations were filtered using |r| > 0.3 and subsequently correlation network was built. Blue lines represent negative correlations, and red lines represent positive correlations.

Supplementary Figure 5.

Correlation between urinary metabolites and gut microbial composition in irritable bowel syndrome patients (IBS). Spearman's rank correlations were calculated between taxonomic data (order level) and urinary metabolites in IBS patients that could discriminate them from ulcerative colitis patients or healthy volunteers, or metabolites responsible for the discrimination between diarrhea- predominant IBS and mixed IBS, or metabolites that were correlated with severity of IBS symptoms. The correlations were filtered using |r| > 0.3 and subsequently correlation network was built. Blue lines represent negative correlations, and red lines represent positive correlations.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....