Total bile acid as an independent indicator for early-stage diabetic kidney disease: a cross-sectional study

The dynamic changes of total bile acid (TBA) metabolism during the progression of diabetic kidney disease (DKD) remain unclear. We aimed to investigate the non-linear relationship between serum TBA levels and different clinical stages of DKD. The research team performed a retrospective study. A total of 485 subjects were enrolled in this cross-sectional study. The diagnostic criteria for diabetes are based on the 1999 WHO standards. Stratification based on the urinary albumin-to-creatinine ratio (UACR) and estimated glomerular filtration rate (eGFR) classified the type 2 diabetes mellitus(T2DM) patients into patients with uncomplicated T2DM and T2DM with DKD. The DKD group was further divided into 3 groups based on UACR and eGFR. The study population includes normal glycemia (NC, n = 87), uncomplicated type 2 diabetes mellitus (T2DM, n = 126), and DKD stages 1 to 3 ( n = 118, 80, and 74, respectively). Differences in TBA levels were compared using non-parametric tests. Multivariate logistic regression and receiver operating characteristic (ROC) curves were utilized to evaluate the association of TBA with early-stage DKD. Serum TBA levels exhibited a significant non-linear, inverted U-shaped pattern across the groups, statistically confirmed by a significant quadratic term in polynomial regression ( P for non-linearity = 0.013). TBA levels were slightly elevated in uncomplicated T2DM (median 2.25 µmol/L), peaked abruptly at DKD stage 1 (median 3.65 µmol/L, P < 0.001 vs. T2DM), and gradually declined by DKD stage 3 (median 1.95 µmol/L). Multivariate logistic regression identified TBA as a factor independently associated with the presence of early-stage DKD (fully adjusted Odds Ratio = 1.394, 95% CI: 1.193–1.629, P < 0.001), after adjustment for age, diabetes duration, blood pressure, HbA1c, BMI, lipid profiles, eGFR, fatty liver and antidiabetic/statin medications. Bootstrap internal validation (1000 iterations) yielded a mean apparent AUC of 0.818 (95% CI: 0.761–0.868); after correction for optimism, the bias-corrected AUC was 0.731, indicating good internal reproducibility that nonetheless requires confirmation in an external cohort. ROC analysis revealed an area under the curve (AUC) of 0.699, with an exploratory Youden-derived threshold of 2.80 µmol/L for distinguishing DKD stage 1 from uncomplicated T2DM; given the modest AUC and wide confidence intervals, this threshold requires validation in an independent external cohort before any clinical use. Serum TBA does not change linearly with renal function decline but is markedly elevated specifically during the early clinical stage of DKD. Serum TBA is independently associated with early-stage DKD and may represent a potential complementary biomarker. Given the cross-sectional design, these findings are hypothesis-generating, and TBA should be regarded as a candidate marker requiring validation in prospective, externally validated cohorts before any clinical application.