Association of TyG-related indices with incident DM among NAFLD patients: a retrospective study

Data source

The current investigation represents a secondary analysis using data from the DATADRYAD database (accessible at http://www.datadryad.org/). The research utilized data shared by Takuro Okamura et al. with the Dryad database [13]. The terms of use for this database permit researchers from various backgrounds to employ the data for additional analysis, thereby supporting a range of research hypotheses and enhancing the applicability of the database. The database includes comprehensive data on the participants, including their age, gender, waist circumference (WC), body weight, body mass index (BMI), systolic blood pressure (SBP), diastolic blood pressure (DBP), alanine aminotransferase (ALT) levels, fasting plasma glucose (FPG), total cholesterol (TC), aspartate transaminase (AST) levels, gamma-glutamyl transferase (GGT) levels, high-density lipoprotein cholesterol (HDL-C) concentration, smoking habits, physical activity, presence of fatty liver, alcohol consumption, Triglyceride (TG) levels, Hemoglobin A1c (HbA1c) measurements, obesity classification, visceral adiposity, patterns of ethanol consumption, DM status, and the length of follow-up observations.

Study population

The data collected from a research program conducted at a medical institution in Gifu, Japan, formed the basis of the NAFLD in the Gifu Area, Longitudinal Analysis (NAGALA) database [13]. From 2004 to 2015, participants in the NAGALA study underwent physical assessments, with about 60% receiving one or two examinations annually. The research question was updated, leading to the establishment of the following exclusion criteria: (1) Individuals previously diagnosed with DM, using diabetic medications, or presenting with fasting plasma glucose levels of 7.0 mmol/L or higher during the initial assessment. (2) Participants missing essential demographic information such as age and gender, crucial blood tests, or anthropometric measurements. (3) Those not diagnosed with NAFLD at the start of the study. (4) Subjects taking drugs at the time of the baseline examination. The diagnosis of NAFLD was confirmed through the identification of fatty liver via ultrasound, while ruling out pharmaceuticals, viral infections, or ethanol as contributing factors. The definitive identification was based on four distinct echographic features, each assigned a value ranging from 0 to 6, as follows: hepatic luminosity (graded 0–4), renal-hepatic acoustic differentiation (valued 0–4), unclear vascular delineation (marked 0–1), and reduction in ultrasound penetration (rated 0–2). A total score of 2 or higher indicated a diagnosis of NAFLD. Informed consent for the use of data was obtained from all subjects, and the NAGALA study was approved by the Murakami Memorial Hospital Research Ethics Committee. Since this investigation served as a secondary retrospective cohort analysis, separate ethical approval was not necessary. The entire research process adhered to the principles outlined in the Declaration of Helsinki.

Data collection and measurements

As per previous references, standard self-administered questionnaires captured the subjects’ comprehensive medical backgrounds and lifestyle aspects. The BMI was calculated by dividing an individual’s weight in kilograms by the square of their height in meters. Based on these calculations, participants were categorized into two groups, as follows: non-obese individuals, defined as those with a BMI less than 25 kg/m², and obese individuals, defined as those with a BMI of 25 kg/m² or greater [14]. Participants were categorized as non-smokers, ex-smokers, or active smokers based on their smoking habits at the study’s onset. Exercise habit was defined as participating in any form of physical activity regularly, at least once a week [15]. Blood samples were collected in the morning following an overnight fast of at least 8 h and analyzed using an automated analyzer with standardized protocols. High blood pressure (HBP) classification followed the guidelines established by the Japanese Society of Hypertension for Hypertension Management (JSH 2019), which identifies HBP as an office blood pressure with a systolic blood pressure (SBP) of 140 mmHg or higher and/or a diastolic blood pressure (DBP) of 90 mmHg or higher [16].

TyG index and modified indices

The concentrations of FPG and triglycerides were analyzed and determined using standard methods. Anthropometric measurements, including height, weight, and WC, which were utilized for final data analysis, were recorded through standardized, self-completed questionnaires. The WHtR was calculated as WC divided by height. The TyG index was computed using the logarithmic formula: TyG index = Ln [(triglycerides in mg/dL) × (glucose in mg/dL) / 2]. Modified versions of the TyG index were derived by multiplying the TyG by BMI for TyG-BMI, by WC for TyG-WC, and by WHtR for TyG-WHtR, respectively.

Assessment of new-onset DM events

Incident DM was defined as the primary outcome measure in this study. The definition of incident DM referred to a situation where a primary care physician documented a patient’s DM diagnosis for the first time. Baseline prevalent diabetes was characterized by DM cases reported by the primary care physician, the patient themselves, or the date of first use of antidiabetic drugs [17]. Furthermore, the incidence of DM was assessed during follow-up examinations, allowing for the identification of new-onset cases. The evaluation of diabetes as an outcome measure involved rigorous adherence to recognized diagnostic criteria and robust surveillance to capture relevant clinical events accurately.

Statistical analysis

Variables were categorized as either continuous or categorical. Continuous variables were then assessed for normal distribution. Variables that met the normality criteria were reported as means ± standard deviations (SD) and compared between groups using the Student’s t-test. Conversely, continuous variables that did not exhibit a normal distribution were presented as medians ± interquartile ranges (IQRs) and analyzed for group differences using the Wilcoxon rank-sum test. Categorical variables were summarized as percentages and analyzed using the chi-square test. To estimate the odds ratios (ORs) and corresponding 95% confidence intervals (CI) for the association between DM and the TyG index, as well as its modified indices, both univariate and multivariate logistic regression models were utilized. In these models, the continuous variables TyG and modified TyG indices were categorized into three groups using tertiles and subsequently transformed into categorical variables for logistic regression analysis. We utilized restricted cubic spline function analysis to examine the linear relationship between the TyG index and its modified indices with incident DM. This methodological approach enabled a detailed exploration of potential non-linear associations, thereby enhancing our understanding of the relationship between these indices and the prevalence of diabetes mellitus within our study cohort. Receiver operating characteristic (ROC) curves were analyzed using logistic regression models. The focus was on comparing the areas underneath the related ROC curves, along with their respective 95% CIs, for the TyG and the modified TyG indices’ ROC area under the curve (AUC). Considering potential variations and interactions, subgroup analyses were conducted based on age, sex, hypertension (HBP), and exercise habits, with the ORs in the subgroup analysis being adjusted. We assessed the possibility of unaccounted confounding factors influencing the association between TyG-related indices and the development of incident DM by calculating E-values [18]. The statistical analysis was conducted using R software, specifically version 4.1.0, provided by the R Foundation for Statistical Computing. A two-tailed P-value less than 0.05 was considered to indicate a statistically significant difference.

Comments (0)

No login
gif