Serum lipidome points to a blood test for bile acid diarrhoea
Bile acid diarrhoea is treatable, but diagnosis has long depended on tests that are difficult to access and slow to deploy in routine care. In a Gut study, investigators used serum metabolomics and machine learning to build a blood test-based diagnostic score designed to distinguish patients with bile acid diarrhoea from controls and from patients with non-alcoholic fatty liver disease.
The clinical problem is familiar. Patients with chronic diarrhoea are often evaluated repeatedly before bile acid diarrhoea is considered, and the disorder can be missed because the established diagnostic pathway is cumbersome. A single-blood-sample strategy has obvious appeal if it can separate true disease from other common gastrointestinal and metabolic conditions without sacrificing accuracy.
Metabolomic separation from controls and liver disease
The study included serum from 50 treatment-naive patients with bile acid diarrhoea diagnosed by the gold standard 75selenium homotaurocholic acid test, 56 feature-matched controls, and 37 patients with non-alcoholic fatty liver disease. Investigators profiled the samples with mass spectrometry spanning 1295 metabolites, then used machine learning to derive a BAD Diagnostic Score.
The metabolomic signal was not subtle. Serum profiles from patients with bile acid diarrhoea differed significantly from those of controls and from those of the non-alcoholic fatty liver disease cohort. In the discovery set, 70 metabolites showed discriminatory performance with an area under the receiver-operating curve above 0.80, a threshold that suggests strong separation between groups rather than random clustering around a common metabolic background.
One sentence captures the practical relevance: the disease left a measurable fingerprint in circulating lipids and related metabolites.
A compact five-metabolite model carried most of the signal
From the broader metabolomic field, logistic regression modeling selected five metabolites: decanoylcarnitine, cholesterol ester (22:5), eicosatrienoic acid, L-alpha-lysophosphatidylinositol (18:0), and phosphatidylethanolamine (O-16:0/18:1). Together, these compounds distinguished bile acid diarrhoea from controls with a sensitivity of 0.78 and a specificity of 0.93.
Those operating characteristics matter because bile acid diarrhoea has been underrecognized partly because clinicians lack an easy, reliable blood-based rule-in test. High specificity is especially useful in this setting: a diagnostic tool that falsely labels too many patients as affected would quickly lose credibility in busy outpatient practice, where chronic diarrhoea has many competing explanations.
The model remained independent of age, sex, and body mass index, which strengthens the argument that the signal reflects disease biology rather than a demographic or anthropometric pattern. It also distinguished bile acid diarrhoea from non-alcoholic fatty liver disease regardless of fibrosis stage, an important safeguard because liver disease can reshape metabolism and complicate interpretation of lipid-based assays.
Why bile acid biology may leave a lipid signature
The metabolic profile is biologically plausible. Bile acid diarrhoea arises when bile acids reach the colon in excess, driving secretion and motility changes that produce watery stools. Disordered bile acid handling also intersects with hepatic lipid metabolism, enterohepatic signaling, and mitochondrial fuel pathways. A serum lipidome that tracks this disturbance may therefore capture downstream effects of altered bile acid synthesis, transport, and feedback regulation.
Several of the selected molecules sit squarely in pathways that could reflect altered energy handling and membrane lipid turnover. Carnitine-related intermediates, cholesterol esters, and phospholipid species often shift when hepatic and intestinal lipid trafficking is perturbed. In that sense, the diagnostic score does not merely classify symptoms; it appears to detect a metabolic imprint of the underlying disorder.
Performance relative to existing blood-based tests
The new score outperformed blood test-based assays currently under development, including 7-alpha-hydroxy-4-cholesten-3-one and fibroblast growth factor 19. Those comparators are important because they reflect the current direction of travel in noninvasive bile acid testing, yet the new model separated cases more effectively in this dataset.
Specificity was particularly strong, and that could influence how clinicians think about triage. A patient with chronic diarrhea, normal or nonspecific routine labs, and an elevated score might move more quickly toward treatment or confirmatory testing, rather than cycling through repeated noncommittal evaluations. In a condition that is both debilitating and treatable, reducing diagnostic delay is not a cosmetic gain; it changes when patients finally reach therapy.
What the cohort does and does not prove
The study was carefully structured, but it still represents a development and discrimination analysis rather than a definitive practice-changing trial. The model was built in a relatively modest cohort, and the strongest numbers come from the discovery framework used to identify the metabolite combination. Wider validation in external populations will determine whether the same score keeps its accuracy across different laboratories, symptom profiles, and real-world referral patterns.
Even so, the central message is hard to miss. A single serum sample produced a metabolite-based diagnostic score that identified bile acid diarrhoea with strong specificity and useful sensitivity, and it did so better than the blood tests already under development.
For clinicians confronting chronic unexplained diarrhoea, that opens a credible path toward a more accessible diagnostic workflow. The remaining challenge is to show that the score performs consistently outside the study setting and can be integrated into routine gastroenterology practice without losing the clarity it demonstrated here.