Abstract
BACKGROUND: Diabetic gastrointestinal autonomic neuropathy (DGAN) is a common yet underrecognized manifestation of diabetic neuropathy. However, the clinical correlations and associated alterations of serum bile acid (BA) metabolism with DGAN remain unclear. AIM: To identify potential metabolite biomarkers capable of identifying DGAN among patients with type 2 diabetes mellitus (T2DM). METHODS: This cross-sectional study included 26 patients with clinically defined DGAN and 69 patients with uncomplicated T2DM. Fifteen individual BAs were quantified in fasting serum using liquid chromatography-tandem mass spectrometry. Gastrointestinal symptoms were scored using the Gastrointestinal Symptom Rating Scale. Spearman's correlation, multivariable logistic regression, receiver operating characteristic, and decision curve analyses were used to explore the associations and build a predictive nomogram. RESULTS: Orthogonal partial least squares discriminant analysis of serum BAs revealed clear separation between the T2DM and DGAN groups, identifying taurolithocholic acid (TLCA) as the key discriminator (variable importance in projection > 1, fold change < 0.5). Univariate logistic regression identified age, body mass index, hemoglobin, fasting/2-h C-peptide, albumin, and TLCA levels as protective factors and the urinary albumin-to-creatinine ratio as a risk factor. After multivariate adjustment age, fasting C-peptide levels, and TLCA levels remained independently associated with DGAN. Receiver operating characteristic analysis yielded areas under the curve of 0.651, 0.760, and 0.678 for age, fasting C-peptide, and TLCA, respectively, and the three variables collectively had an area under the curve of 0.970 (95% confidence interval: 0.937-1.000). Decision curve analysis confirmed the clinical net benefit across threshold probabilities of 9%-68%. The derived nomogram displayed excellent calibration and net clinical benefits for the model. CONCLUSION: These findings highlighted the association between altered BA profiles and DGAN in patients with T2DM. Combining BA profiling with conventional clinical data could facilitate the early identification of DGAN and offer new insights into early screening and BA-targeted interventions. While these findings offer valuable insights, they should nevertheless be viewed as hypothesis-generating and require further validation in larger, multicenter cohorts.