The association between the extracellular water-to-total body water ratio and albuminuria in Chinese type 2 diabetes mellitus patients

Subjects

Between August 2022 and May 2024, patients with T2DM admitted to our institution were continuously screened. The inclusion criteria were as follows: (1) diagnosis of T2DM; (2) age ≥ 18 years; (3) underwent BIA for body fluid composition analysis; and (4) tested UACR. The exclusion criteria were as follows: (1) incomplete data, (2) presence of other kidney diseases, and (3) severe liver disease, (4) muscle disorders, (5) acute infection, (6) decompensated heart failure. This study was approved by the Institutional Review Board of the Medical Ethics Committee of Tangdu Hospital, Fourth Military Medical University, China (No. K202207-05), and complied with the principles of the Declaration of Helsinki. All participants provided written informed consent prior to participation in the study.

Data collection

Demographic data, including the duration of T2DM, smoking history (defined as smoking at least one cigarette per day for at least six months), and alcohol consumption history (defined as consumption for at least one year), were collected. The medication history of the patients was recorded, including the use of antidiabetic drugs and antihypertensive drugs, such as insulin, metformin hydrochloride, sodium-glucose cotransporter 2 inhibitors (SGLT-2i), glucagon-like peptide 1 receptor agonists (GLP-1RA), renin–angiotensin–aldosterone system (RAAS) inhibitors, calcium channel blockers (CCB), and diuretics. Height and weight were measured by a trained nurse in the morning while the patients were fasting. Blood pressure was measured following a 10-minute rest period in a seated position. To ensure accuracy, blood pressure was measured at least twice for each participant, with at least one-minute interval between measurements. The average of these two readings was then calculated and recorded as the participant’s blood pressure for the study.

Laboratory indicators, including glycated hemoglobin (HbA1c), creatinine, and lipid profile including total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), fasting plasma glucose (FBG), potassium, sodium, hematocrit were collected, after an overnight fast. eGFR was calculated using the CKD-EPI formula (Levey et al., 2009b), eGFR = a × (Scr/b) c × (0.993)^age. a: female = 144, male = 141; b: female = 0.7; male = 0.9; c: female, Scr ≤ 0.7 mg/dl = −0.329, Scr > 0.7 mg/dl = −1.209; male, Scr ≤ 0.9 mg/dl = −0.411; Scr > 0.9 mg/dl = −1.209. TyG Index = ln (fasting triglycerides (mg/dL) × FBG (mg/dL)/2). Urine albumin and creatinine concentrations were measured from midstream urine after waking up, and the UACR was calculated. The participants were devided into three groups: normo-albuaminuria (UACR < 30 mg/g), microalbuaminuria (30–300 mg/g) an0d macroalbuaminuria (UACR > 300 mg/g).

Body composition parameters, including extracellular water (ECW), intracellular water (ICW), total body water (TBW), ECW/TBW ratio, body fat percentage, waist-to-hip ratio, and skeletal muscle mass, were measured using a bioelectrical impedance analyzer (BIA; Model: Biospace, Seoul, South Korea; Serial No.: 720). Prior to measurement, participants were required to fast for ≥8 h and abstain from water intake for ≥4 h. They were further instructed to avoid strenuous exercise, sauna use, and excessive fluid consumption within 24 h before testing. Standardized measurement protocols were followed: participants wore lightweight clothing, and women of reproductive age were advised to schedule assessments outside their menstrual periods.

Statistical analysis

All statistical analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Continuous variables were expressed as means ± standard deviations, and categorical variables were expressed as proportions. One-way analysis of variance (ANOVA) was used for normally distributed continuous variables, while nonparametric tests were used for variables that were not normally distributed. Chi-square tests were used for categorical variables. Multivariate logistic regression analysis was employed to examine factors affecting UACR in T2DM patients. Finally, two predictive models for macroalbuminuria were constructed, nomograms were depicted and used to evaluate the prediction accuracy through ROC curve analysis, and were evaluated by the concordance index (C-index) and the calibration curve. P-value < 0.05 was considered statistically significant.

Baseline characteristics

A total of 1,034 patients with T2DM were enrolled in this study, the baseline characteristics were shown in Table 1. Among the participants, 640 (58.4%) had normo-albuminuria, 244 (23.6%) had microalbuminuria, and 150 (14.5%) had macroalbuminuria. Patients with microalbuminuria and macroalbuminuria had a longer duration of diabetes, were more often male, had higher blood pressure, more extracellular fluid, higher creatinine levels, lower eGFR, higher lipid levels, and a higher proportion use insulin, diuretics and antihypertensive medications (P < 0.05) (Table 1).

Comparative analysis of ECW/TBW across various anatomical regions among three UACR groups

To further investigate whether the variations in ECW/TBW among different albuminuria groups were associated with specific anatomical regions, patients were categorized into two groups based on the median ECW/TBW for distinct body segments, including the left upper limb, right upper limb, trunk, and left lower limb, and the differences in ECW/TBW across distinct UACR groups were expressed as percentages. The results indicated that as UACR increased, the ECW/TBW across all anatomical regions exhibited a gradual increase (Fig. 1) (Table S1), suggesting that the relationship between ECW/TBW and UACR has no anatomical regional specificity.

Analysis of the relationship between ECW/TBW and UACR groups

Multifactorial logistic regression models examined the association between ECW/TBW and UACR. In comparison to the normo-albuminuria group, ECW/TBW demonstrated a significant correlation with macroalbuminuria (P < 0.001), while no significant association was observed with microalbuminuria. Similarly, in comparison to microalbuminuria group, ECW/TBW remained significantly associated with macroalbuminuria (P < 0.001) (Table 2), indicating a robust association between ECW/TBW and macroalbuminuria. Given that no significant difference in ECW/TBW was observed between the normo-albuminuria and microalbuminuria groups, further analysis was conducted to elucidate the relationship between ECW/TBW and macroalbuminuria by dividing patients into two groups: one comprising those with normo-albuminuria plus microalbuminuria and another consisting of those with macroalbuminuria. Binary logistic regression models were employed to assess differences in ECW/TBW between these two cohorts; the results consistently revealed a significant association between ECW/TBW and macroalbuminuria (P < 0.001) (Table 3).

Furthermore, to investigate whether the association between ECW/TBW and macroalbuminuria differs among patients with varying grades of renal function, participants were categorized into two groups based on eGFR: one group with eGFR < 60 mL/min/1.73 m² and another with eGFR ≥ 60 mL/min/1.73 m². The relationship between ECW/TBW and macroalbuminuria was assessed in these two groups using binary logistic regression models. It was found that in patients with eGFR ≥ 60 mL/min/1.73 m², ECW/TBW demonstrated a significant association with macroalbuminuria (P < 0.001). Conversely, in patients with eGFR < 60 mL/min/1.73 m², no significant association between ECW/TBW and macroalbuminuria was identified; this indicates that in T2DM patients with preserved renal function (eGFR grades 1–2), a significant association exists between ECW/TBW and macroalbuminuria (Table 3).

Development and validation of an individualized prediction model

Based on multivariate logistic regression models, two predictive models were developed to estimate the risk of macroalbuminuria, targeting individuals with normoalbuminuria and microalbuminuria, respectively. Model A predicts the progression from normoalbuminuria to macroalbuminuria (Fig. 2), while Model B predicts the progression from microalbuminuria to macroalbuminuria (Fig. 3). Both models incorporated variables including age, sex, duration of type 2 diabetes, use of SGLT2 inhibitors, systolic blood pressure, and the ECW/TBW ratio. Corresponding nomograms were constructed for each model (Figs. 2A and 3A), where each variable’s contribution to macroalbuminuria risk is represented by a specific point value. The total risk score was calculated by summing the points of all variables, providing an individualized estimate of macroalbuminuria risk in patients with type 2 diabetes.

Model performance was evaluated using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA). ROC analysis demonstrated that Model A achieved an AUC of 0.795 (95% CI [0.752–0.838]; Fig. 2B), while Model B achieved an AUC of 0.761 (95% CI [0.711–0.812]; Fig. 3B), indicating robust predictive accuracy. Calibration curves showed excellent agreement between predicted probabilities and observed outcomes (Figs. 2C and 3C). Hosmer-Lemeshow tests confirmed good model calibration (Model A: P = 0.469; Model B: P = 0.197). DCA further evaluated the clinical utility of the nomograms, demonstrating high net clinical benefit for predicting macroalbuminuria across a threshold probability range of 0% to 80% (Figs. 2D and 3D). These findings underscore the strong predictive performance, clinical applicability, and decision-support value of the two nomogram models.

Innovation

This study is innovative as it is the first to demonstrate the predictive ability of the ECW/TBW for albuminuria in T2DM patients across different CKD stages, particularly its ability to independently predict macroalbuminuria in the early stages of CKD. By integrating clinical indicators such as age, diabetes duration, and blood pressure, the prediction accuracy was significantly enhanced. Furthermore, this study validated the utility of the extracellular water ratio, determined by BIA, as a convenient and low-cost predictive tool for albuminuria, making it especially suitable for resource-limited settings. These findings provide new insights into the early screening and management of renal injury in T2DM patients.

Limitations

Although this study provides a novel approach for the early screening of DKD, there are several limitations. Firstly, as a cross-sectional study, it does not establish a causal relationship. Secondly, while BIA offers the advantages of simplicity and high repeatability, its accuracy can be affected by the patient’s condition (e.g., oedema). Future studies should validate the causal relationship between the extracellular water ratio and albuminuria through prospective cohort studies and assess the utility of this indicator in different T2DM subgroups. Additionally, although this study confirmed the predictive value of the extracellular water ratio in T2DM patients with better renal function, its role in advanced CKD patients remains unclear. Future research should explore how to integrate noninvasive tools like BIA with traditional indicators (e.g., UACR and eGFR) to develop a multidimensional early screening model for T2DM patients with advanced renal impairment, optimizing the diagnosis and treatment of DKD. We have also taken an interest in the correlation between phase angle and chronic kidney disease. Regrettably, our InBody device does not measure this parameter. We anticipate that our forthcoming research will delve into the relationship between phase angle and diabetic nephropathy.