Study design and population
This study is based on our own established Ageing and Body Composition of Diabetes (ABCD) cohort, a single-center, retrospective cohort specifically designed to explore the impact of body composition on adverse outcomes in patients with T2DM. The study utilized electronic medical records (EMRs) from the Department of Endocrinology at the First Affiliated Hospital of Chongqing Medical University, covering the period from January 2015 to August 2019. The EMR system includes demographic information, clinical diagnoses, laboratory test results, medication histories, and imaging findings. All aspects of the study, including data collection, follow-up, data cleaning, statistical analysis, and interpretation of results, were independently completed by the authors, with rigorous quality control measures in place. This ensures the completeness and authenticity of the research data.
The study included individuals aged 50 years or older who met the diagnostic criteria for T2DM established by the World Health Organization in 1999. Initially, a total of 5,971 patients were enrolled in the study, as illustrated in Fig. S1. The exclusion criteria were as follows: (1) not undergoing whole-body dual-energy X-ray absorptiometry (DXA); (2) having preexisting medical conditions or medication usage that could influence body composition, such as malignant tumors, other endocrine diseases (e.g., dysfunction from thyroid, parathyroid, pituitary or adrenal glands), autoimmune diseases (e.g., rheumatoid arthritis, systemic lupus erythematosus), organ failure (moderate to severe heart failure, chronic lung disease and liver insufficiency, chronic kidney disease with an estimated glomerular filtration rate (eGFR) ≤ 45 ml/min/1.73 m2), gastrointestinal disorders (inflammatory bowel disease, malabsorption), neuromuscular disorders (e.g., muscular dystrophy, myasthenia), history of bariatric surgery, depression, other psychiatric disorders, or use of steroids and psychotropic medications or bisphosphonate; (3) having acute medical disorders such as severe infections, acute diabetes complications, pregnancy, trauma, surgery, or other stressful situations; (4) having factors that could affect the interpretation of the DXA results, such as edema, amputation or body part loss, or the presence of artificial joints or implants; (5) declining to participate in the study; (6) having incomplete clinical data; and (7) being lost to follow-up. Ultimately, 1,818 eligible participants were included in the current analysis. The analyses were conducted for the total population and separately for males and females.
Body composition assessment
Total and appendicular lean mass were estimated using DXA with an identical Hologic scanner (Discovery A, S/N 87352, Hologic, Bedford, USA), and all measurements were taken at baseline. Quality control of the DXA scans in this study is described in detail in the Supplementary material, Methods. We considered 8 different scaling factors to adjust Total Lean Mass (TLM) and ALM: height2, weight, BMI, BSA, BFM, PBF, AFM, and VFM.
BMI was calculated by dividing weight (in kg) by the square of height (in m2). The BSA was calculated using the Mosteller formula19: BSA (in m2) = [weight (in kg) × height (in cm)/3600]½. ALM or AFM was determined by summing the lean or adipose tissue mass of all four extremities. The appendicular muscle-to-fat ratio (MFR) was defined as the ratio of ALM to AFM. Both TLM/BMI and ALM/BMI are expressed in units of g/BMI.
Measurements of covariates and biomarkers
Our research team members underwent unified training to ensure consistency and accuracy in baseline data collection, which formed the foundation for all subsequent analyses and evaluations.These data included key demographic characteristics such as sex, age, current smoking or drinking status, and duration of diabetes. Standardized procedures were used for the measurement of blood pressure (both systolic and diastolic) and anthropometric variables such as height and weight. Laboratory evaluation focused on metabolic traits, including fasting plasma glucose, hemoglobin A1c (HbA1c), and lipid profile (low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), total cholesterol, and triglycerides). Additionally, renal metabolic panel assessments, which included serum uric acid, urinary albumin-to-creatinine ratio (uACR) and eGFR, along with nutritional and inflammatory markers like albumin, prealbumin, hemoglobin, and high-sensitivity C-reactive protein(hsCRP) levels, were conducted. This study also examined clinical characteristics, including diabetes-related complications (such as retinopathy, nephropathy, peripheral neuropathy, carotid atherosclerosis, and peripheral artery disease), common comorbidities (hypertension, coronary heart disease, stroke, and prior CVDs or fractures), and corresponding drug usage. The severity of diabetic complications was evaluated using the Diabetes Complications Severity Index (DCSI).
The eGFR was calculated using the formula developed by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI). The DCSI, a 13-point scale developed by Young BA et al.20, better predicts the risk of adverse outcomes in diabetes patients based on automated diagnostic, pharmacy, and laboratory data.
Outcome ascertainment
Patients in the study were diligently tracked and monitored using various methods, including in-person interviews, meticulous review of medical records, and thorough telephone interviews. Additionally, unless an event occurred prior to the follow-up cutoff, all participants were contacted via telephone, ensuring thorough data collection and patient monitoring. The follow-up data focuses solely on new health events after the initial visit, ensuring it is distinct from baseline data. Medical records were reviewed on an irregular basis, being checked as needed. Follow-up began at the time of patient recruitment and continued until the first occurrence of any primary endpoint (CVD, fragility fracture, or all-cause mortality) or until June 30, 2022, whichever came first. Furthermore, in the sub-analysis, the secondary endpoints were considered as components of the composite primary endpoint, a practice based on previous studies21. This approach not only comprehensively evaluates the mechanical and metabolic functions of muscle mass but also increases statistical power by expanding the number of cases.
The determination of mortality status relied on the electronic medical record system of the First Affiliated Hospital of Chongqing Medical University, death certificates from specialist physicians, and follow-up phone calls with family members for confirmation. During interviews with the study participants, we primarily sought information on their general health status, with particular emphasis on whether they had experienced any CVDs or fractures as specified in the manuscript. CVDs encompass a range of conditions, including fatal cardiovascular/cerebrovascular events, nonfatal myocardial infarction or unstable angina, procedures involving the revascularization of coronary or peripheral arteries, nonfatal ischemic or hemorrhagic stroke, and transient ischemic attack. Additionally, we inquired about the occurrence of acute events accompanied by relevant symptoms and the availability of diagnostic evidence from specialist physicians or imaging reports. Furthermore, we asked about the treatments received and the methods employed. Notably, we excluded cases where CVDs were solely diagnosed based on imaging without symptomatic evidence. We also excluded pathological fractures, traumatic fractures, or fractures in specific anatomical regions (e.g., skull, hand, finger, foot, toe).
Statistical analysis
Continuous quantitative data are presented as the means ± standard deviations (SDs) or medians with interquartile ranges (IQRs) in accordance with the results of the Shapiro‒Wilk normality test. Categorical data are summarized as numbers and frequencies (%). Comparative analyses between the two groups were conducted using independent Student’s t tests, nonparametric tests, or chi-square tests, as appropriate. To assess trends in population characteristics across the tertiles, the linear trend χ2 test was used for categorical data, and either ANOVA or the Jonckheere-Terpstra test was used for quantitative data. Spearman correlation analysis was performed to examine the relationships between lean mass indices and specified scaling factors.
Cox proportional hazards models were utilized to calculate HRs and 95% CIs for the primary endpoints based on putative sarcopenia variables. Tests using Schoenfeld’s residuals indicated no evidence of violation of the proportional hazards assumption. A single Cox model was employed, adjusting for age, smoking, alcohol consumption, and the severity of diabetes-related complications (including diabetes duration, HbA1c, and DCSI score). In the analysis of secondary endpoints, adjustments were made for other well-established CV risk factors, such as prior CVDs, LDL-C, and hypertension, to account for CVD endpoints. Similarly, additional adjustments were made for prior fractures and factors related to body composition, such as T scores ≤-2.5 for the lumbar spine, femoral neck, or total hip, in relation to fragility fracture endpoints. Receiver-operating characteristic (ROC) curve analysis was used to compare the predictive performance of ALM/weight and ALM/BMI in predicting the primary endpoints in patients with T2DM. The area under the ROC curve (AUC) was compared using the nonparametric Z test. A Venn diagram was also generated to evaluate the quantitative differences, both overlapping and nonoverlapping, between ALM/weight and ALM/BMI in relation to determining the primary endpoint.
A two-tailed p value less than 0.05 was considered to indicate statistical significance. The statistical analysis was conducted using IBM SPSS version 26.0 (SPSS Inc., Chicago, IL, USA), MedCalc® Statistical Software version 22.002 (MedCalc Software Ltd, Ostend, Belgium), and R (version 4.2.2; R Foundation).
Ethical approval
The study was conducted in accordance with the principles of the Declaration of Helsinki and received approval from the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University (No. K2023-402). All procedures were conducted in compliance with local legislation and institutional protocols. Due to the retrospective nature of the study, the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University waived the need of obtaining informed consent.
link