Effect of body shape index trajectories on kidney disease incidence and subsequent death in individuals with type 2 diabetes Authors: Chia-Ing Li, Chiu-Shong Liu, Chih-Hsueh Lin, Shing-Yu Yang, Cheng-Chieh Lin, Tsai-Chung Li Published: 2025 ABSTRACT AbstractBackgroundThis study aimed to identify three-year trajectories of body mass index (BMI), a body shape index (ABSI), waist circumference (WC), waist-to-hip ratio (WHR), waist-to-height ratio (WHtR), and body roundness index (BRI); to assess their association with incident chronic kidney disease (CKD) and mortality; and to investigate potential sex difference in these associations.MethodsThis retrospective longitudinal study included 8,445 patients with type 2 diabetes mellitus (T2DM) who were free of CKD at baseline, had ≥ 3 years of follow-up, and attended the Diabetes Care Management Program of a medical center in Taichung from 2001 to 2012 (with the follow-up period ending in 2021). Cluster analysis was performed to group three-year trajectory patterns, and Cox proportional hazard models were used to assess the associations.ResultsDuring a mean follow-up of 7.88 years, 2,901 (34.35%) participants developed CKD, and 856 (10.14%) died. An elevated risk of CKD incidence was associated with several trajectory clusters. Specifically, associations were found with: Cluster 1 (constant low-level) for ABSI, WC, WHtR, and BRI trajectories; Cluster 4 (high level with decreasing trend) for WC and BRI trajectories; and Cluster 5 (low level with increasing trend) for BMI, WC, WHtR, and BRI trajectories. Regarding mortality, an elevated risk was associated with: Cluster 1 (constant low-level) for WC and WHtR trajectories; Cluster 3 (constant high-level) for ABSI and BRI trajectories; and Cluster 5 (low level with increasing trend) for BMI trajectory.ConclusionLongitudinal trajectories of various body shape indices predict CKD incidence and mortality risk in T2DM.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13098-025-02048-1.Keywords: Type 2 diabetes, Body shape, Chronic kidney disease, Mortality Background Diabetes mellitus (DM) has become a global health issue [1]. Given its potential to lead to premature death and disability [2, 3], developing effective strategies to prevent and delay its progression is imperative. Renal impairment is one of the most serious complications of diabetes; notably, the prevalence of chronic kidney disease (CKD) among patients with type 2 diabetes (T2DM) is reported to be as high as 27% [4]. Furthermore, the global incidence of stage renal disease (ESRD) due to diabetes increased 22.1% in 2000 to 31.3% in 2015 [5]. Consequently, the early identification of individuals at high risk for CKD is crucial for timely intervention and improved outcomes. Obesity is an established risk factor for CKD, partly mediated by T2DM [6]. However, defining obesity solely by body mass index (BMI) presents limitations, primarily due to the inherent complexity of body composition and fat distribution. BMI fails to differentiate between fat mass and lean body mass, and it does not adequately capture central obesity, which is often considered to have a more profound impact on metabolic health [7]. Indeed, individuals with the same BMI can exhibit significant variations in body composition and, consequently, disease risk [8]. Reflecting this complexity, current research regarding the relationship between various obesity indicators and CKD risk in patients with T2DM yields inconsistent results. This discrepancy may arise from a narrow focus on only one or a few indicators and an underestimation of the dynamic nature of body composition. Existing literature suggests that longitudinal changes in body shape may serve as more effective predictors of CKD risk compared to single static measurements, exemplified by significant weight gain [9] or reductions in waist circumference (WC) [10]. While baseline measures, such as the fat mass index, have also demonstrated associations [11], these findings collectively underscore the critical importance of examining temporal changes in body shape to comprehensively understand kidney health trajectories in patients with T2DM and to investigate the effect of sex differences on these dynamic associations. While studies have reported various body shape indices in general population (e.g., WC, waist-to-hip ratio (WHR), hip-to-height ratio (WHtR), a body shape index (ABSI) and body roundness index (BRI) [12–17], a comprehensive assessment of these specific indices within the context of T2DM patients remains relatively limited. Notably, indices such as the ABSI and BRI have demonstrated superior accuracy in assessing mortality risk compared with traditional anthropometric measurements, such as WC and BMI [12–17]. To address this research gap, the present study aims to analyze the three-year trajectories of multiple body shape indices (including BMI, ABSI, WC, WHR, WHtR, and BRI) in patients with T2DM. Furthermore, this study investigates the impact of sex differences on the associations between these body shape indices and kidney health and mortality, thereby providing critical insights into how longitudinal changes in body composition influence CKD progression and subsequent mortality in patients with T2DM. Methods Study population and data collection This retrospective cohort study was drawn from the diabetic care management program (DCMP) at a single tertiary medical center in Taichung, Taiwan. The DCMP, initiated by the Taiwan Bureau of National Health Insurance in 2002, is a case management program for diabetic patients, involving care from physicians, nurses, and nutritionists. A total of 26,413 patients were enrolled in the DCMP at this center between 2002 and 2020. Patient recruitment was systematically conducted by physicians in the endocrinology and metabolism as well as family medicine outpatient clinics, inviting all eligible patients diagnosed with T2DM according to the criteria established by the American Diabetes Association (International Classification Disease, Ninth Revision, Clinical Modification, abbreviated as ICD-9 CM Code of 250). From this initial DCMP cohort, we included subjects with normal renal function at baseline who had at least one body shape measurement per year for three consecutive years between 2002 and 2020. This criterion allowed for the calculation of a three-year body shape trajectory. Subjects were excluded if they had a follow-up duration of less than one year after the three-year trajectory measurement period or had missing data regarding BMI, baseline characteristics, diabetes-related variables, comorbidities, or laboratory blood test results. This selection process resulted in a final study sample of 8,445 individuals. The flowchart for recruitment procedures of study subjects in this study is presented in Supplemental Fig. 1. Ethics approval was obtained from the Ethical Review Board of China Medical University Hospital (CMUH113-REC1-198). The index date was defined as the date of the first recorded BMI measurement in the database for each participant. Baseline characteristics were collected from the integrated care database within one year before the index date. These characteristics included sociodemographic factors, duration of T2DM, age of onset, diabetes-related factor and biomarker, comorbidity, types of anti-diabetes medications, hypertension medications, cardiovascular medications, and hyperlipidemia medications. The follow-up period for assessing outcomes began immediately after the completion of each participant’s initial three-year body shape trajectory measurement period and extended until December 31, 2021. The mean follow-up time for CKD incidence and death among the 8,445 subjects was 4.47 and 7.88 years, respectively. The development of CKD was assessed using information on renal function biomarkers (including albuminuria and estimated glomerular filtration rate [eGFR]), which were assessed at least annually in accordance with diabetes case management guidelines. All-cause mortality data were obtained from the Center for Application of Health and Welfare Statistics of the Ministry of Health and Welfare, and participant survival status was determined by linkage with the National Death Registry dataset. Measurements Body indices All anthropometric measurements were conducted according to the standardized clinical care guidelines of the medical center’s diabetic care program to ensure data consistency and quality. Body height and weight were measured using calibrated, automated digital stadiometers and scales located in the outpatient clinic area. Patients were measured without shoes and in light clothing. WC and hip circumference were measured by trained nursing staff. Following protocol, WC was measured at the midpoint between the lower margin of the last palpable rib and the top of the iliac crest, and hip circumference was measured at the widest portion of the buttocks. All anthropometric data were recorded by the nurses and entered into the electronic diabetes care information system during the patient’s visit. This study used six body indices: BMI, representing general adiposity; ABSI, WC, WHR, and WHtR, all representing abdominal adiposity; and BRI indicating visceral adiposity. Body weight, height, and WC measurements were retrieved from the DCMP database for each patient. The formulas for calculating these indices are as follows: BMI = Weight (kg)/Height (m)^2 BMI = Weight (kg)/Height (m)^2 ABSI = WC (cm)/[BMI^(2/3) x Height (m)^(1/2)] ABSI = WC (cm)/[BMI^(2/3) x Height (m)^(1/2)] WHR = WC (cm)/Hip circumference (cm) WHR = WC (cm)/Hip circumference (cm) WHtR = WC (cm)/Height (cm) WHtR = WC (cm)/Height (cm) BRI = 364.2–365.5 × √[1- (WC (cm)/2π)^2/(0.5 × Height (cm))^2] BRI = 364.2–365.5 × √[1- (WC (cm)/2π)^2/(0.5 × Height (cm))^2] Sociodemographic factors and biomarkers These variables included age, sex, diabetes duration, smoking habits (yes/no), and alcohol consumption (yes/no). Diabetes duration was calculated by dividing the number of days between enrollment in the DCMP and the date of diabetes onset by 365. The following values, measured at the index date, were also retrieved: glycated hemoglobin (HbA1c), systolic blood pressure (SBP), diastolic blood pressure (DBP), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), fasting plasma glucose (FPG), and triglycerides (TG). Kidney function Urine albumin-to-creatinine ratio (uACR) was used as a marker of the albumin excretion rate. The eGFR was estimated based on serum creatinine levels according to the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [18], where eGFR (mL/min/1.73 m2) = 141 × min(Scr/κ,1)α × max(Scr/κ,1)−1.209 × 0.993Age × 1.018 (if female) × 1.159 (if black), where Scr is serum creatinine, κ is 0.7 for females and 0.9 for males, α is −0.329 for females and − 0.411 for males, min indicates the minimum of Scr/κ or 1, and max indicates the maximum of Scr/κ or 1. Comorbidities and diabetic complications Data on the following comorbidities, ascertained from the DCMP, were included in the analysis: hypertension and hyperlipidemia. Diabetic acute complications considered were diabetic ketoacidosis, hyperglycemic hyperosmolar state, and severe hypoglycemia. Chronic diabetic complications assessed included stroke, peripheral neuropathy, nephropathy, diabetic retinopathy, peripheral vascular disease, amputation, and diabetic foot. These comorbidities and complications were ascertained from records covering the one-year period prior to the index date. Each comorbidity and complication were categorized as present (yes) or absent (no). Outcome ascertainment The primary outcomes were CKD incidence and all-cause mortality, each ascertained independently. CKD incidence was defined at least two confirmed measurements of eGFR and albuminuria. CKD onset was determined according to one of the following three criteria, consistent with the Kidney Disease: Improving Global Outcomes 2012 clinical practice guidelines [19]: (1) uACR ≥ 300 mg/g; (2) eGFR < 60 mL/min/1.73 m² and uACR ≥ 30 mg/g; or (3) eGFR < 45 mL/min/1.73 m². CKD incidence was followed from the index date to the date of a confirmed CKD event for CKD-incident subjects, and from the index date to the end of 2021 or death (whichever occurred first) for non-CKD-incident subjects. All-cause mortality was defined as death from any cause, with the follow-up period starting from the index date until the end of the study period (2021) or death (whichever occurred first). Statistical analysis Body indices measurements, collected at baseline and at three-year intervals from patients with T2DM who had at least two measurements, were used to calculate body indices trajectories. First, multilevel models were used to estimate individual growth curves of body indices over the three-year period. These models included an intercept, linear time, and quadratic time terms as predictors. Second, cluster analysis using Ward’s method and the cubic clustering criterion was conducted to identify distinct patterns in the three-year trajectories of body indices. This analysis classified patients into distinct cluster groups. Third, the body indices trajectories for each cluster group were graphically depicted, and the groups were assigned descriptive names. Finally, the incidence of CKD and all-cause mortality were compared across the identified trajectory groups. The association of membership in each body index trajectory group with CKD incidence and all-cause mortality was assessed using Cox proportional hazards models. Separate models were constructed for each body index trajectory. Unadjusted and adjusted Cox proportional hazards models were used to evaluate the association of cluster group membership (defined by body index trajectory) with the incidence of CKD and all-cause mortality. Covariates were selected for inclusion in the multivariate models using a significance level of 0.25 for entry and 0.05 for retention [20, 21]. Kaplan–Meier plots were generated to illustrate the cumulative incidence of CKD and all-cause mortality. Furthermore, the interaction of sex and diabetes duration with the relationship between body shape index trajectory and each outcome (CKD incidence and all-cause mortality) was examined. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) with a significance level of 0.05 (two sided). Results A total of 8,445 patients with T2DM (mean age 57.88 ± 12.00 years old, 53.56% male) were enrolled in this study. During mean follow-up periods of 4.47 years for CKD and 7.88 years for all-cause mortality, the corresponding incidence rates were 76.77 and 12.87 per 1,000 person-years, respectively. Table 1 presents baseline sociodemographic and clinical factors in subjects grouped according to CKD incidence and death status. Compared with patients who did not develop CKD, those who did were older, had a longer duration of diabetes, were more likely to be female, and exhibited a higher prevalence of alcohol consumption, regular exercise, concurrent use of oral hypoglycemic agents and insulin injections, diabetic complications (including amputation), and the use of hypertension and cardiovascular medications. Conversely, they showed a lower prevalence of hyperlipidemia medication use. Furthermore, individuals with CKD had higher mean levels of HbA1c, FPG, TG, and LDL-C and a lower mean level of HDL-C. A similar pattern was observed for mortality. Table 1Baseline characteristics according to chronic kidney disease (CKD) incidence and mortality in patients with type 2 diabetes (n = 8,445)VariablesCKD incidencep valueMortalityp valueNo(n = 5,544)Yes(n = 2,901)No(n = 7,589)Yes(n = 856) Sociodemographic factors Sex0.020.06Men3020 (54.47)1503 (51.81)4038 (53.21)485 (56.66)Women2524 (45.53)1398 (48.19)3551 (46.79)371 (43.34)Age (years) †56.08 ± 11.8360.4 ± 11.83< 0.00156.65 ± 11.7965.68 ± 10.83< 0.001 Lifestyle behaviors Smoking863 (15.57)461 (15.89)0.721160 (15.29)164 (19.16)0.004Alcohol drinking321 (5.79)235 (8.10)< 0.001483 (6.36)73 (8.53)0.02Exercising2701 (48.72)1499 (51.67)0.013764 (49.6)436 (50.93)0.48 Diabetes-related variables Duration of diabetes (years)†3.97 ± 5.615.86 ± 6.83< 0.0014.29 ± 5.787.48 ± 8.06< 0.001Types of diabetes treatment< 0.001< 0.001Diet or exercise363 (6.55)143 (4.93)476 (6.27)30 (3.50)Oral hypoglycemic drug4719 (85.12)2432 (83.83)6434 (84.78)717 (83.76)Insulin injection126 (2.27)46 (1.59)154 (2.03)18 (2.10)Both336 (6.06)280 (9.65)525 (6.92)91 (10.63) Diabetic complications Hypertension1133 (20.44)989 (34.09)< 0.0011793 (23.63)329 (38.43)< 0.001Hyperlipidemia875 (15.78)582 (20.06)< 0.0011273 (16.77)184 (21.5)< 0.001Stroke83 (1.50)86 (2.96)< 0.001128 (1.69)41 (4.79)< 0.001Coronary artery disease148 (2.67)135 (4.65)< 0.001228 (3.00)55 (6.43)< 0.001Angina52 (0.94)100 (3.45)< 0.001118 (1.55)34 (3.97)< 0.001Severe hypoglycemia4 (0.07)24 (0.83)< 0.00122 (0.29)6 (0.70)0.06Peripheral neuropathy112 (2.02)267 (9.20)< 0.001265 (3.49)114 (13.32)< 0.001Nephropathy35 (0.63)127 (4.38)< 0.001103 (1.36)59 (6.89)< 0.001Diabetes retinopathy1233 (22.24)1008 (34.75)< 0.0011850 (24.38)391 (45.68)< 0.001Amputation6 (0.11)11 (0.38)0.0210 (0.13)7 (0.82)< 0.001DKA17 (0.31)14 (0.48)0.2828 (0.37)3 (0.35)1.00HHNK10 (0.18)30 (1.03)< 0.00125 (0.33)15 (1.75)< 0.001 Medication use Hypertension medications1695 (30.57)1223 (42.16)< 0.0012517 (33.17)401 (46.85)< 0.001Hyperlipidemia medications1208 (21.79)430 (14.82)< 0.0011532 (20.19)106 (12.38)< 0.001Cardiovascular medications939 (16.94)583 (20.10)< 0.0011315 (17.33)207 (24.18)< 0.001Biomarker†HbA1c (%)7.65 ± 1.777.98 ± 1.89< 0.0017.73 ± 1.88.04 ± 1.96< 0.001FPG (mg/dL)143 ± 47.62158.19 ± 61.36< 0.001146.37 ± 49.75164.59 ± 75.63< 0.001TC (mg/dL)179.15 ± 44.08185.86 ± 42.28< 0.001181.25 ± 43.51183.32 ± 44.260.19TG (mg/dL)158.84 ± 215.77160.85 ± 224.190.69159.75 ± 212.52157.58 ± 267.350.82HDL-C (mg/dL)45.05 ± 11.8342.72 ± 11.92< 0.00144.39 ± 11.8342.97 ± 12.580.003LDL-C (mg/dL)104.4 ± 34.24111.72 ± 35.28< 0.001106.62 ± 34.4109.56 ± 37.810.03T test was used for continuous variables to calculate p-valuesChi-square test or Fisher’s Exact test was used for categorical variables to calculate p-values†: mean ± SD; DKA: diabetic ketoacidosis; HHNK: hyperglycemic hyperosmolar nonketotic coma; FPG: fasting plasma glucose; TC: total cholesterol; TG: triglyceride; HDL-C: high density lipoprotein-cholesterol; LDL-C: low density lipoprotein-cholesterol Baseline characteristics according to chronic kidney disease (CKD) incidence and mortality in patients with type 2 diabetes (n = 8,445) T test was used for continuous variables to calculate p-values Chi-square test or Fisher’s Exact test was used for categorical variables to calculate p-values †: mean ± SD; DKA: diabetic ketoacidosis; HHNK: hyperglycemic hyperosmolar nonketotic coma; FPG: fasting plasma glucose; TC: total cholesterol; TG: triglyceride; HDL-C: high density lipoprotein-cholesterol; LDL-C: low density lipoprotein-cholesterol As illustrated in Supplemental Fig. 2, five distinct trajectory patterns were identified for BMI, ABSI, WC, WHR, WHtR, and BRI. Trajectory models, derived using latent class growth mixture modeling, revealed five patterns: cluster 1, “constant low-level”; cluster 2, “middle level with increasing trend” or “constant middle-level”; cluster 3, “constant high-level”; cluster 4, “high level with decreasing trend”; and cluster 5, “low level with increasing trend”. Table 2 summarizes the association of these body shape trajectory patterns with CKD incidence, with the group exhibiting the lowest adjusted risk serving as the reference. Compared with individuals in the BMI cluster 3 (constant high-level), elevated hazard ratios (HRs) for CKD were observed in cluster 2 (middle level with increasing trend; HR: 1.21; 95% confidence interval [CI], 1.04–1.40) and cluster 5 (low level with increasing trend; 1.47, 1.24–1.74). Relative to individuals in cluster 2 (middle level with increasing trend or constant middle–level), increased hazards of CKD incidence were observed in cluster 1 (constant low–level) for ABSI (1.14, 1.03–1.26), WC (1.20, 1.10–1.31), WHtR (1.20, 1.09–1.33), and BRI (1.16, 1.05–1.28); in cluster 4 (high level with decreasing trend) for WC (1.31, 1.04–1.65) and BRI (1.22, 1.04–1.44); and cluster 5 (low level with increasing trend) for WC (1.23, 1.06–1.42), WHtR (1.17, 1.04–1.31), and BRI (1.19, 1.06–1.33). No significant association was observed between WHR trajectory patterns and CKD incidence. Table 2Hazard ratios of chronic kidney disease (CKD) incidence for body index trajectories in patients with type 2 diabetes (n = 8,445)Variables n casePerson-yearsIncidence rate †Multivariate model 1Multivariate model 2Multivariate model 3HR (95% CI)p valueHR (95% CI)p valueHR (95% CI)p valueBMI TrajectoriesCluster 1: Constant low-level4906170422597.0875.410.97 (0.88, 1.08)0.600.97 (0.87, 1.07)0.541.02 (0.92, 1.13)0.75Cluster 2: Middle level with increasing trend9083103731.7283.071.26 (1.09, 1.45) 0.002 1.21 (1.05, 1.4) 0.009 1.21 (1.04, 1.40) 0.01 Cluster 3: Constant high-level15464706931.2867.811.001.001.00Cluster 4: High level with decreasing trend5642102515.2183.491.10 (0.93, 1.29)0.271.11 (0.95, 1.31)0.201.14 (0.97, 1.35)0.11Cluster 5: Low level with increasing trend5212072011.19102.921.51 (1.28, 1.78) < 0.001 1.34 (1.14, 1.59) < 0.001 1.47 (1.24, 1.74) < 0.001 ABSI TrajectoriesCluster 1: Constant low-level277391712603.4172.761.15 (1.04, 1.27) 0.005 1.13 (1.03, 1.25) 0.01 1.14 (1.03, 1.26) 0.01 Cluster 2: Middle level with increasing trend224772410168.8571.201.001.001.00Cluster 3: Constant high-level246686210411.0982.801.06 (0.96, 1.17)0.261.03 (0.93, 1.14)0.571.00 (0.90, 1.10)0.98Cluster 4: High level with decreasing trend4091771824.8297.001.26 (1.06, 1.48) 0.007 1.16 (0.99, 1.37)0.081.16 (0.98, 1.36)0.09Cluster 5: Low level with increasing trend5502212778.3279.541.18 (1.02, 1.38) 0.03 1.16 (1.00, 1.35)0.061.13 (0.97, 1.32)0.11WC TrajectoriesCluster 1: Constant low-level277697112219.5879.461.13 (1.04, 1.23) 0.005 1.12 (1.03, 1.21) 0.01 1.20 (1.10, 1.31) < 0.001 Cluster 2: Constant middle-level3887129417807.572.671.001.001.00Cluster 3: Constant high-level11923484817.1172.241.12 (0.99, 1.26)0.061.13 (1.00, 1.27)0.051.11 (0.98, 1.25)0.10Cluster 4: High level with decreasing trend14179750.16564105.311.35 (1.08, 1.70) 0.01 1.30 (1.03, 1.63) 0.03 1.31 (1.04, 1.65) 0.02 Cluster 5: Low level with increasing trend4492092192.1495.341.27 (1.10, 1.48) 0.001 1.23 (1.06, 1.42) 0.007 1.28 (1.10, 1.48) 0.001 WHR TrajectoriesCluster 1: Constant low-level16425287367.3271.671.00 (0.90, 1.11)0.991.02 (0.92, 1.13)0.741.07 (0.96, 1.19)0.21Cluster 2: Constant middle-level3862129317496.9173.901.001.001.00Cluster 3: Constant high-level23158019772.3681.971.08 (0.99, 1.19)0.081.07 (0.98, 1.17)0.161.02 (0.93, 1.12)0.62Cluster 4: High level with decreasing trend182921008.4891.231.13 (0.92, 1.40)0.251.12 (0.90, 1.38)0.311.08 (0.87, 1.33)0.50Cluster 5: Low level with increasing trend4441872141.487.331.12 (0.96, 1.31)0.161.12 (0.96, 1.30)0.161.14 (0.97, 1.33)0.11WHtR TrajectoriesCluster 1: Constant low-level20937179483.6175.601.13 (1.02, 1.25) 0.02 1.13 (1.02, 1.25) 0.02 1.20 (1.09, 1.33) 0.004 Cluster 2: Constant middle-level247778411105.7870.591.001.001.00Cluster 3: Constant high-level22597199375.2776.691.09 (0.98, 1.20)0.111.09 (0.99, 1.21)0.091.08 (0.98, 1.20)0.14Cluster 4: High level with decreasing trend4622102425.5686.581.11 (0.95, 1.30)0.181.11 (0.96, 1.30)0.171.09 (0.93, 1.27)0.30Cluster 5: Low level with increasing trend11544715396.2687.281.19 (1.06, 1.34) 0.003 1.17 (1.04, 1.31) 0.008 1.16 (1.03, 1.30) 0.01 BRI TrajectoriesCluster 1: Constant low-level20376849329.1673.321.08 (0.98, 1.19)0.111.07 (0.97, 1.18)0.181.16 (1.05, 1.28) 0.003 Cluster 2: Constant middle-level3163104114295.472.821.001.001.00Cluster 3: Constant high-level18995867783.2475.291.05 (0.95, 1.17)0.311.06 (0.95, 1.17)0.301.05 (0.95, 1.16)0.38Cluster 4: High level with decreasing trend3421711684.7101.501.26 (1.07, 1.48) 0.006 1.23 (1.05, 1.45) 0.01 1.22 (1.04, 1.44) 0.02 Cluster 5: Low level with increasing trend10044194693.9889.261.20 (1.07, 1.35) 0.002 1.19 (1.06, 1.33) 0.003 1.19 (1.06, 1.33) 0.003 † Number of incident cases/person-years*1000; HR: hazard ratio; CI: confidence interval; BMI: body mass index; WC: waist circumference; WHR: waist-to-hip ratio; ABSI: A body shape index; BRI: body roundness index. The bold text represents the statistically significant results (p<0.05).Multivariate model 1 adjusted for age and sexMultivariate model 2 adjusted for age, sex, smoking, alcohol drinking, exercising, duration of diabetes, types of diabetes treatment, fasting plasma glucose and HbA1cMultivariate model 3 adjusted for age, sex, smoking, alcohol drinking, exercising, duration of diabetes, types of diabetes treatment, fasting plasma glucose, HbA1c, hypertension, hyperlipidemia, stroke, coronary artery disease, angina, severe hypoglycemia, peripheral neuropathy, nephropathy, diabetes retinopathy, amputation, DKA, HHNK, hypertension medications, hyperlipidemia medications, cardiovascular medications, total cholesterol, triglyceride, high density lipoprotein-cholesterol, and low density lipoprotein-cholesterol Hazard ratios of chronic kidney disease (CKD) incidence for body index trajectories in patients with type 2 diabetes (n = 8,445) † Number of incident cases/person-years*1000; HR: hazard ratio; CI: confidence interval; BMI: body mass index; WC: waist circumference; WHR: waist-to-hip ratio; ABSI: A body shape index; BRI: body roundness index. The bold text represents the statistically significant results (p<0.05). Multivariate model 1 adjusted for age and sex Multivariate model 2 adjusted for age, sex, smoking, alcohol drinking, exercising, duration of diabetes, types of diabetes treatment, fasting plasma glucose and HbA1c Multivariate model 3 adjusted for age, sex, smoking, alcohol drinking, exercising, duration of diabetes, types of diabetes treatment, fasting plasma glucose, HbA1c, hypertension, hyperlipidemia, stroke, coronary artery disease, angina, severe hypoglycemia, peripheral neuropathy, nephropathy, diabetes retinopathy, amputation, DKA, HHNK, hypertension medications, hyperlipidemia medications, cardiovascular medications, total cholesterol, triglyceride, high density lipoprotein-cholesterol, and low density lipoprotein-cholesterol The association between body shape trajectory patterns and mortality is provided in Table 3. Compared with individuals in the BMI cluster 3 (constant high-level), those in cluster 5 (low level with increasing trend) had an increased risk of mortality, with multivariable adjusted HRs of 1.93 (95% CI, 1.42–2.61). Compared with individuals in cluster 2 (middle level with increasing trend or constant middle-level), increased mortality risks were found in cluster 1 (constant low–level) for WC (1.21, 1.04–1.42), WHtR (1.30, 1.08–1.57), and BRI (1.23, 1.03–1.47); for cluster 3 (constant high–level) for ABSI (1.26, 1.05–1.50), WHtR (1.28, 1.05–1.55), and BRI (1.23, 1.01–1.48). No significant association was observed between all-cause mortality and WHR in either unadjusted or adjusted analyses. Multivariate-adjusted cumulative incidence of CKD and all-cause mortality for each body shape index trajectory were shown in Supplemental Fig. 3. Table 3Hazard ratios of mortality for body index trajectories in patients with type 2 diabetes (n = 8,445)Variables n casePerson-yearsIncidence rate †Multivariate model 1Multivariate model 2Multivariate model 3HR (95% CI)p valueHR (95% CI)p valueHR (95% CI)p valueBMI TrajectoriesCluster 1: Constant low-level490653039852.0313.301.10 (0.89, 1.37)0.381.12 (0.90, 1.40)0.301.13 (0.90, 1.42)0.28Cluster 2: Middle level with increasing trend908746831.2210.831.12 (0.83, 1.52)0.451.06 (0.78, 1.43)0.731.07 (0.79, 1.46)0.65Cluster 3: Constant high-level15469611442.558.391.001.001.00Cluster 4: High level with decreasing trend564684433.5915.341.28 (0.94, 1.75)0.121.30 (0.95, 1.77)0.101.30 (0.95, 1.78)0.11Cluster 5: Low level with increasing trend521883950.9422.272.10 (1.57, 2.81) < 0.001 1.93 (1.44, 2.59) < 0.001 1.93 (1.42, 2.61) < 0.001 ABSI TrajectoriesCluster 1: Constant low-level277320422945.948.890.95 (0.78, 1.16)0.620.94 (0.77, 1.14)0.520.95 (0.78, 1.15)0.59Cluster 2: Middle level with increasing trend224720817277.7212.041.001.001.00Cluster 3: Constant high-level246633217527.5118.941.29 (1.08, 1.54) 0.004 1.24 (1.04, 1.48) 0.02 1.26 (1.05, 1.50) 0.01 Cluster 4: High level with decreasing trend409673535.2918.951.13 (0.85, 1.49)0.401.07 (0.81, 1.42)0.631.08 (0.81, 1.43)0.61Cluster 5: Low level with increasing trend550455223.868.610.81 (0.58, 1.11)0.190.82 (0.59, 1.13)0.230.82 (0.59, 1.14)0.23WC TrajectoriesCluster 1: Constant low-level277631722561.0114.051.25 (1.08, 1.46) 0.004 1.23 (1.05, 1.43) 0.009 1.21 (1.04, 1.42) 0.01 Cluster 2: Constant middle-level388737430108.4512.421.001.001.00Cluster 3: Constant high-level1192847998.3210.501.05 (0.83, 1.34)0.671.00 (0.78, 1.26)0.970.98 (0.77, 1.25)0.90Cluster 4: High level with decreasing trend141241586.5715.130.90 (0.59, 1.37)0.620.88 (0.58, 1.34)0.550.83 (0.54, 1.26)0.37Cluster 5: Low level with increasing trend449574255.9813.390.99 (0.74, 1.31)0.920.99 (0.75, 1.31)0.940.96 (0.72, 1.27)0.76WHR TrajectoriesCluster 1: Constant low-level164213113446.049.740.89 (0.72, 1.09)0.260.92 (0.75, 1.13)0.430.93 (0.76, 1.15)0.52Cluster 2: Constant middle-level386237930114.7112.591.001.001.00Cluster 3: Constant high-level231525916932.3615.301.06 (0.90, 1.25)0.471.01 (0.86, 1.19)0.901.02 (0.86, 1.20)0.83Cluster 4: High level with decreasing trend182331956.1916.870.99 (0.69, 1.42)0.950.96 (0.67, 1.38)0.840.95 (0.66, 1.37)0.80Cluster 5: Low level with increasing trend444544061.0313.300.90 (0.68, 1.20)0.480.93 (0.69, 1.23)0.600.91 (0.68, 1.22)0.54WHtR TrajectoriesCluster 1: Constant low-level209322917215.8613.301.31 (1.09, 1.58) 0.004 1.32 (1.09, 1.59) 0.004 1.30 (1.08, 1.57) 0.006 Cluster 2: Constant middle-level247722119347.711.421.001.001.00Cluster 3: Constant high-level225921515541.3613.831.29 (1.07, 1.56) 0.008 1.29 (1.07, 1.56) 0.008 1.28 (1.05, 1.55) 0.01 Cluster 4: High level with decreasing trend462624643.6913.350.87 (0.65, 1.15)0.320.91 (0.68, 1.21)0.500.89 (0.67, 1.19)0.44Cluster 5: Low level with increasing trend11541299761.7213.211.04 (0.84, 1.30)0.711.08 (0.87, 1.34)0.501.10 (0.88, 1.37)0.40BRI TrajectoriesCluster 1: Constant low-level203722416982.8213.191.28 (1.08, 1.53) 0.005 1.25 (1.05, 1.50) 0.01 1.23 (1.03, 1.47) 0.02 Cluster 2: Constant middle-level316329224719.6811.811.001.001.00Cluster 3: Constant high-level189917512805.2713.671.28 (1.06, 1.55) 0.01 1.26 (1.05, 1.53) 0.02 1.23 (1.01, 1.48) 0.04 Cluster 4: High level with decreasing trend342503405.0514.680.87 (0.65, 1.18)0.380.88 (0.65, 1.20)0.430.84 (0.62, 1.14)0.26Cluster 5: Low level with increasing trend10041158597.5113.381.02 (0.82, 1.26)0.871.03 (0.83, 1.28)0.820.99 (0.79, 1.23)0.89† Number of incident cases/person-years*1000; HR: hazard ratio; CI: confidence interval; BMI: body mass index; WC: waist circumference; WHR: waist-to-hip ratio; ABSI: A body shape index; BRI: body roundness index. The bold text represents the statistically significant results (p<0.05).Multivariate model 1 adjusted for age and sexMultivariate model 2 adjusted for age, sex, smoking, alcohol drinking, exercising, duration of diabetes, types of diabetes treatment, fasting plasma glucose and HbA1cMultivariate model 3 adjusted for age, sex, smoking, alcohol drinking, exercising, duration of diabetes, types of diabetes treatment, fasting plasma glucose, HbA1c, hypertension, hyperlipidemia, stroke, coronary artery disease, angina, severe hypoglycemia, peripheral neuropathy, nephropathy, diabetes retinopathy, amputation, DKA, HHNK, hypertension medications, hyperlipidemia medications, cardiovascular medications, total cholesterol, triglyceride, high density lipoprotein-cholesterol, and low density lipoprotein-cholesterol Hazard ratios of mortality for body index trajectories in patients with type 2 diabetes (n = 8,445) † Number of incident cases/person-years*1000; HR: hazard ratio; CI: confidence interval; BMI: body mass index; WC: waist circumference; WHR: waist-to-hip ratio; ABSI: A body shape index; BRI: body roundness index. The bold text represents the statistically significant results (p<0.05). Multivariate model 1 adjusted for age and sex Multivariate model 2 adjusted for age, sex, smoking, alcohol drinking, exercising, duration of diabetes, types of diabetes treatment, fasting plasma glucose and HbA1c Multivariate model 3 adjusted for age, sex, smoking, alcohol drinking, exercising, duration of diabetes, types of diabetes treatment, fasting plasma glucose, HbA1c, hypertension, hyperlipidemia, stroke, coronary artery disease, angina, severe hypoglycemia, peripheral neuropathy, nephropathy, diabetes retinopathy, amputation, DKA, HHNK, hypertension medications, hyperlipidemia medications, cardiovascular medications, total cholesterol, triglyceride, high density lipoprotein-cholesterol, and low density lipoprotein-cholesterol Significant interactions were observed between sex and WHR (P = 0.024), WHtR (P = 0.031), and BRI (P = 0.021) for CKD incidence, as well as between sex and WC (P = 0.027) for all-cause mortality. Stratified analyses by sex revealed that WHtR and BRI trajectories were associated with CKD incidence in men, and WC trajectories were associated with mortality risk in men. However, no such associations were observed in women (Fig. 1). These findings suggested that changes in body shape may have a greater impact on renal function in men than in women. No significant interactions were found between diabetes duration and any body shape trajectory for CKD incidence or mortality, so stratified analyses for diabetes duration were not reported. Adjusted hazard ratios of chronic kidney disease (CKD) incidence and mortality by body shape trajectories, stratified by sex *:p < 0.05; **:p < 0.01; ***:p < 0.0001 Discussion This study identified five distinct longitudinal trajectory patterns for body shape indices (BMI, ABSI, WC, WHR, WHtR, and BRI) in T2DM patients: constant low, stable/increasing middle, constant high, decreasing from high, and increasing from low. Importantly, specific trajectory patterns were associated with differential risks of CKD and mortality. An increased risk of incident CKD was observed in both sexes for individuals following low-baseline increasing BMI and WC trajectories, a middle-baseline increasing BMI trajectory, or consistently low ABSI and WC trajectories. Among men, non-constant middle-level trajectories for WHtR and BRI (specifically, consistently low WHtR and high-decreasing BRI) were also associated with higher CKD risk. Elevated mortality risk in both sexes was linked to low-baseline increasing BMI, consistently high ABSI, WHtR, and BRI; and consistently low WHtR and BRI. Consistently low WC increased mortality risk solely in men. Research on how changes in BMI or WC over time relate to CKD risk in the general population has yield inconsistent results. While some large cohort studies link significant BMI increases [22] or shifts in obesity status [23–25] to altered CKD risk, others find no such predictive association [26, 27]. Crucially, studies looking at the long-term effects of body shape changes on CKD in individuals with T2DM are rare. A recent study found no significant link between one-year BMI changes and CKD risk in T2DM patients, possibly due to its short follow-up [10]. This suggests that the impact of body composition changes might differ significantly between the general population and those with T2DM. Our study directly addresses this critical gap by tracking the long-term evolution of various body shape indices in a large T2DM cohort. This offers needed clarity on these complex relationships and providing new insights into high-risk trajectories—like “low baseline-increasing BMI/WC” for both CKD and mortality, or paradoxical “persistently low” patterns (e.g., low ABSI/WC for CKD, low WHtR/BRI for mortality)—thereby enabling earlier identification of vulnerable patients. Regarding specific indices, although baseline WHR is reportedly a predictor of diabetic renal disease or the severity of diabetic nephropathy in T2DM populations [28], research on the association between WHR changes over time and CKD incidence in individuals with T2DM is limited. A study in the Korean general population observed that transitioning to a non-obese WHR trajectory lowers CKD risk [23]. In contrast, the present study did not find a significant association between WHR trajectory and CKD incidence in individuals with T2DM. Several factors may explain these differing results between the general population and individuals with T2DM, including comorbidities, medication use, and the dominant influence of T2DM-specific pathophysiology. Known mechanisms in T2DM including hyperglycemia, insulin resistance, and inflammation directly impact kidney health [29] and can potentially overshadow or mask the influence of WHR changes, which may otherwise be a more prominent risk indicator in individuals without these severe metabolic perturbations. Furthermore, the interplay of comorbidities and complex medication regimens in T2DM could also modulate the expression of WHR’s impact on CKD progression. Clarifying these complex interactions through further dedicated research in T2DM is warranted to understand the precise contribution of WHR dynamics. Contrary to the expectation that lower central obesity is protective [30], this study found a paradoxical association when persistently low ABSI and WC trajectories (in both sexes) and persistently low WHtR and BRI trajectories (in men) were linked to an increased CKD incidence among individuals with T2DM. This phenomenon may primarily be explained by two intertwined mechanisms. Firstly, it could indicate “normal-weight central obesity,” where individuals exhibit increased visceral fat accumulation despite having seemingly normal or low overall anthropometric measures. Supporting this, Velludo et al. (2010) observed poor correlation between WC changes and CT-measured visceral adipose tissue (VAT) changes in non-dialysis CKD patients, suggesting WC may be an imperfect VAT surrogate [31]. This altered body composition, potentially involving higher relative VAT, can promote renal dysfunction via metabolic disturbances and inflammation [32]. Secondly, persistently low body shape indices can indicate underlying issues like nutritional deficiencies or sarcopenia, which are also known risk factors for poor kidney health [33]. Thus, these ‘low’ trajectories may identify a particularly vulnerable phenotype in T2DM, where outwardly favorable anthropometrics belie an underlying detrimental body composition or advanced disease state. We observed sex-specific associations, where abdominal indices (WHtR and BRI) trajectories predicted CKD risk in men but not women, may be correlated with fundamental differences in body-fat distribution and metabolism. Men tend to accumulate more visceral fat abdominally, whereas women typically store more subcutaneous fat in gluteofemoral regions [34]. Visceral adipose tissue is metabolically active and known to release pro-inflammatory cytokines that can contribute to systemic low-grade inflammation, potentially damaging renal structures and impairing function over time [35]. This mechanism can explain why abdominal fat-related indices showed a clearer link to CKD risk specifically in men. While our findings highlight this important sex dimorphism, further investigation is needed to elucidate the precise pathophysiological underpinnings of such sex differences. We observed association of consistently high ABSI, WHtR, and BRI with increased mortality risk aligns with previous studies highlighting the detrimental effects of central adiposity on overall health and longevity [36]. This consistent finding underscores the critical role of visceral fat accumulation in T2DM mortality. Proposed mechanisms include increased systemic inflammation, insulin resistance, and direct adverse cardiovascular effects often associated with visceral-fat accumulation [37]. Conversely, the higher mortality risk associated with consistently low WC, WHtR, and BRI may, in part, be attributable to reverse causality, whereby underlying chronic illness leads to weight loss, muscle wasting, or reduced anthropometric measures [38–40], rather than low body measures being a direct cause of increased mortality. This possibility aligns with the concept of a U-shaped relationship between body shape indices and survival in T2DM. These results collectively indicate that extremes of body-fat distribution over time (both consistently high and consistently low) may be detrimental for longevity in individuals with T2DM. This emphasizes the need for a nuanced understanding of body composition beyond simple weight status. Clinically, this suggests that healthcare providers should consider not only a patient’s current weight but also their historical body shape patterns and fat distribution, particularly central obesity, when assessing long-term health risks and developing interventions for patients with T2DM. The differential risks associated with these trajectories can inform tailored interventions: for instance, aggressive weight management for individuals exhibiting ‘increasing’ or ‘consistently high’ trajectories, whereas muscle preservation, nutritional support, and investigation for underlying frailty might be prioritized in ‘consistently low’ groups. This empowers clinicians with a more dynamic toolset for assessing cardiometabolic and renal risk, thereby enabling targeted and timely interventions. Recent evidence further supports this approach: a randomized controlled trial demonstrated that incorporating anthropometric measures such as waist circumference or BMI into sonographic diagnostic criteria refined case selection for primary and secondary prevention, thereby improving risk stratification in metabolic dysfunction-associated fatty liver disease [41]. These findings underscore the promise of integrated diagnostic and therapeutic strategies for optimizing patient outcomes. A key strength of the current work was its longitudinal design in individuals with T2DM, which allowed us to examine changes in body shape indices over time and their association with CKD and mortality. However, several limitations should be considered when interpreting these results. First, owing to the observational nature of the study, our findings represent associations rather than proven cause-and-effect relationships. While our longitudinal design allows for the exploration of temporal sequences, it lacks the experimental rigor required to establish causality conclusively. Second, our findings may have limited generalizability to the broader population of individuals with T2DM, particularly those managed in primary care or community settings. As our study participants were drawn exclusively from a single tertiary hospital in Taiwan, they may represent a patient cohort with more complex or severe diabetes, distinct comorbidity profiles, or different healthcare-seeking behaviors compared to the general T2DM population. This specificity of our cohort could influence the observed body shape trajectory patterns and their associations with CKD and mortality. This may limit the external validity of our findings, highlighting the need for validation in other populations and settings, including those from primary care settings and different geographical regions within and outside Taiwan. Third, potential biases from missing data and loss to follow-up warrant consideration. Despite our efforts to mitigate this, individuals with missing anthropometric data or who were lost to follow-up may differ systematically from those who completed the study. This could potentially introduce selection bias, affecting the representativeness of our findings. Fourth, the restricted specificity regarding cause-specific mortality, despite the availability of ICD-9 or ICD-10 diagnostic codes for deaths. While these codes allow for the investigation of more granular causes, the subsequent reduction in the number of deaths for highly specific etiologies precluded the generation of robust or statistically reliable estimates (cerebrovascular disease deaths: 35, 4.1% of all deaths; renal-related deaths: 22, 2.54% of all deaths). Consequently, this constrains our capacity to offer highly nuanced insights into the precise mechanisms underlying the observed mortality associations. Fifth, the estimation of abdominal fat relied on anthropometric indices (WC, WHR, WHtR, ABSI, and BRI), which represent indirect measures. The precise correlation between these indicators and true adiposity warrants further investigation. Despite this limitation, these measurements offer the advantage of being simple, easily implemented methods suitable for clinical application. Finally, because body weight and anthropometric measurements can be influenced by an individual’s hydration and fluid status. The fluctuations due to factors such as medication use (e.g., diuretics), glycemic control, or other clinical conditions (e.g., heart failure, kidney disease) could introduce day-to-day or visit-to-visit variability into the anthropometric indices used. This potential source of measurement error could affect the precision of our trajectory analyses. Conclusion An increased risk of CKD and/or mortality among individuals with T2DM was associated with distinct longitudinal trajectories of body shape indices, specifically patterns involving weight gain from low or middle baselines, sustained high central adiposity (via ABSI, WHtR, and BRI), or even persistently low values for certain indices (ABSI, WC, WHtR, and BRI). Notably, the associations between trajectories of abdominal adiposity indices (WHtR and BRI) and CKD risk were evident primarily in men, suggesting sex-specific differences in how body shape dynamics impact renal function. These findings underscore the importance of longitudinal monitoring of body shape indices, beyond static measurements, in T2DM management. Furthermore, these identify adverse body shape trajectories as potential targets for interventions aimed at improving long-term clinical outcomes in this high-risk population. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Abbreviations Diabetes mellitus Chronic kidney disease Type 2 DM End-stage renal disease Body mass index Waist circumference Waist to hip ratio Waist-to-height ratio A body shape index Body roundness index Diabetes care management program International Classification Disease, Ninth Revision, Clinical Modification Estimated glomerular filtration rate Glycated hemoglobin Systolic blood pressure Diastolic blood pressure Total cholesterol Low density lipoprotein cholesterol High density lipoprotein cholesterol Fasting plasma glucose Triglycerides Urine albumin-to-creatinine ratio Chronic Kidney Disease Epidemiology Collaboration Hazard ratios Confidence interval Visceral adipose tissue Author contributions C.I L., T.C.L., and C.C.L. developed the study design, obtained funding and drafted manuscript. C.I L. and S.Y.Y acquired data, carried out the statistical analysis, analyzed and interpreted data, and critically revised the manuscript. C.S.L., C.H.L., T.C.L., and C.C.L. contributed to discussion, and reviewed and edited this manuscript. All authors read and approved the final manuscript. T.C.L. and C.C.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Funding This study was supported primarily by the Ministry of Science and Technology, Taiwan (MOST 109-2314-B-039-031-MY2, MOST 110-2314-B-039-021- & MOST 111-2314-B-039-018-), National Science and Technology Council, Taiwan (NSTC 112-2314-B-039-042-, NSTC 113-2314-B-039-042- & NSTC 114-2314-B-039-027) and China Medical University Hospital, Taiwan (DMR-114-096). Data availability The datasets generated and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This study was approved by the Ethical Review Board of China Medical University Hospital (CMUH113-REC1-198). Informed consent of the study participants was not required because the dataset used in this study consists of de-identified secondary data released for research purposes. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Cheng-Chieh Lin, Email: 002229@tool.caaumed.org.tw. Tsai-Chung Li, Email: tcli@mail.cmu.edu.tw. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Data Availability Statement The datasets generated and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.