Research design and methods
Study design and subjects
From January 2018 to April 2020, a total of 251 patients with T2DM from the Department of Endocrinology, Zhongshan Hospital, Fudan University (Xiamen Branch) (Xiamen, China) had been recruited into the present ongoing cohort. Patients were diagnosed as diabetes based on American Diabetes Association 2018 criteria: (1) a self-reported history of diabetes previously diagnosed by healthcare professionals; (2) fasting plasma glucose (FGP) ≥126 mg/dL (7.0 mmol/L); (3) 2-hour plasma glucose (2-hour PG, oral glucose tolerance test) ≥200 mg/dL (11.1 mmol/L) or (4) glycosylated hemoglobin A1c (HbA1c) ≥6.5% (48 mmol/mol).24 T2DM was identified for diabetes cases with the age of 20 years or older who are overweight or obese and/or have a family history of diabetes. Finally, 18 patients were excluded due to incomplete data and 233 patients were left for the present analyses. This study was designed as a cross-sectional analysis of this T2DM patient cohort.
Measurements
Face-to-face interview was conducted for each patient to collect sociodemographic status, lifestyle habits, present and previous history of health and medications, including histories of diabetic complications and treatment. Subjects underwent weight and height measurements by using a calibrated scale after removing shoes and heavy clothes. Body mass index (BMI) was calculated as the weight in kilograms divided by the square of the height in meters. Arterial blood pressure was measured with OMRON electronic sphygmomanometer after sitting for at least 15 min. Three readings were taken at 5 min intervals and the mean of them was recorded.
After a 12-hour overnight fasting, blood samples were collected to measure FPG, HbA1c, liver function, renal function and lipid profiles. All biochemical measurements were tested in the clinical laboratory of the Zhongshan Hospital, Fudan University (Xiamen Branch). Serum creatinine, uric acid (UA), triglyceride (TG), total cholesterol (TC) and high-density lipoprotein cholesterol (HDL-C), aspartate aminotransferase (AST), alanine aminotransferase (ALT) were determined on an analyzer (Roche Elecsys Insulin Test, Roche Diagnostics). Low-density lipoprotein cholesterol (LDL-C) was calculated by Friedewald’s formula: LDL-C=(TC−HDL-C)−TG/5.25 FPG concentration was measured by the hexokinase method and HbA1c by the Bio-Rad Variant Hemoglobin A1c assay.
Liver ultrasonography and definition of non-alcoholic fatty liver disease
Hepatic ultrasonography scanning was performed by an experienced radiologist who was blinded to the patients’ health status using GE LOGIQ P5 scanner (GE Healthcare, Milwaukee, USA) with a 4 MHz probe. Hepatic steatosis was diagnosed on the basis of characteristic sonographic features, including hepatorenal echo contrast, liver parenchymal brightness, deep beam attenuation and vessel blurring.26 The definition of NAFLD was based on hepatic ultrasonography diagnosis of hepatic steatosis without excessive alcohol consumption, viral or autoimmune liver disease.
Fibrosis-4 (FIB-4) score was calculated for each subject based on the formula: FIB-4=age ((year)×AST (U/L))/((PLT (109/L))×(ALT (U/L))1/2), and a cut-off of >3.25 was used to define advanced hepatic fibrosis.27
Definition of cancers and BMI categories
All types of cancers were identified by checking the patients’ medical records after they recalled histories of any kind of cancer which were diagnosed by professional health workers previously or after admission to the hospital. Subjects were classified by WHO guidelines for the Asian Pacific population into five BMI categories: underweight (<18.5 kg/m2), normal weight (18.5–22.9 kg/m2), overweight (23.0–24.9 kg/m2), obesity I (25.0–29.9 kg/m2) and obesity II (≥30.0 kg/m2).28 29 Since there were only 7 (3.0%) patients with underweight and 17 (7.3%) patients with obesity II, three BMI categories were used in the present study, including underweight or normal weight (non-obese or lean, <23.0 kg/m2), overweight (23.0–24.9 kg/m2) and obesity (25.0 kg/m2 or over).
Statistical analyses
Data were presented as the mean±SD for continuous variables or number and percentage for categorical variables. Skewness and kurtosis tests for continuous variables were conducted and found all followed approximation of normal distributions. Differences between subjects categorized by NAFLD (vs non-NAFLD) and cancer (yes vs no) were analyzed using one-way analysis of variance for continuous variables and χ2 test for categorical variables. Bar graphs showing prevalence rates of all cancers were made stratified by BMI categories and NAFLD.
Multivariable logistic regression models were used to calculate the adjusted ORs and 95% CIs for all cancers with adjustment for potential confounders (including age, sex, ever smoking and drinking habits, systolic and diastolic blood pressure, TG, TC, HDL-C and LDL-C, HbA1c, serum UA, oral hypoglycemic medications and insulin use, BMI and NAFLD). Interaction tests between BMI and NAFLD were conducted. Furthermore, multivariable logistic regression analyses of NAFLD (yes vs no) stratified by BMI categories as well as multivariable logistic regression analyses of BMI categories (underweight or normal weight as the reference) stratified by NAFLD for all cancers were conducted separately with adjustment for the same potential confounding variables. All p values were two-sided and p<0.05 was considered statistically significant. All statistical analyses were performed using Stata V.14.0 (StataCorp, College Station, Texas, USA).