The prevalence of type 2 diabetes mellitus (T2DM) is a growing global health issue (1, 2). Presently, about 1 in 11 adults have diabetes mellitus (DM), 90% of whom develop T2DM (3). Asia is a significant region of the fast spreading T2DM global epidemic. During the past few decades, the prevalence of T2DM has increased significantly in China (4).

T2DM is known as a complex and heterogeneous disease, for which obesity is one of the strongest risk factors (5). Increased (mostly visceral) adipose tissue mass-related insulin resistance (IR) has been recognized as a critical component that may contribute to concurrent increases in the prevalence of T2DM (6). IR is present in many metabolic disorders, such as T2DM and metabolic syndrome (7, 8). However, due to its complicated process, lengthy laboratory test, and high cost, the method for the IR measurement is not widely used in clinical trials or sizable population-based studies (9). Excessive abdominal fat accumulation has been linked to glucose and lipid metabolic problems (10). In particular, increased visceral adipose tissue (VAT) has been linked to IR (11). In diabetic studies, the Chinese visceral adiposity index (CVAI) and lipid accumulation product (LAP) have been employed as trustworthy indicators of visceral adiposity (12–14). Triglyceride-glucose (TyG) index and the triglyceride-to-HDL-cholesterol (TG/HDL-C) ratio have also emerged as straightforward, affordable, repeatable, and reliable surrogates for IR measurement (15).

Previous studies show that high TyG index is a risk factor for incident DM and cardiovascular disease (16, 17). However, a drawback of those studies is that they only examine data from one moment in time and do not examine how the index changes over time. Given that each individual has unique underlying conditions, assessing dynamic changes may be more beneficial than measuring data at a single time point (18). We calculate the difference in index value at the end of follow-up (2015) minus that at baseline (2011) for analysis, denoted as -D. This research aims to assess the applicability of longitudinal changes in CVAI, LAP, TyG, and TG/HDL-C index in predicting T2DM in middle-aged and elderly Chinese population in a large community-based prospective cohort study.

In this prospective cohort study, we analyzed the baseline and follow-up data of 4,596 Chinese participants and found that longitudinal changes in the four surrogate IR indexes were significantly associated with the risk of T2DM. TyG-D showed the highest predictive accuracy of all the indexes.

Although relevant studies have been conducted previously, they have some limitations. Li et al. found the association between the TyG and new-onset diabetes was positive and linear, with an AUC of 0.597 (0.559–0.636) (16). Compared to Li's study, our research found that the AUC of the TyG-D was 0.66 (0.63–0.69), which was higher than the TyG index alone in the risk of incident diabetes. It was possible that TyG-D had higher predictive performance than TyG index alone for the risk of T2DM. Zhang et al. reported that the risk of T2DM increased with elevated TyG index in normal-weighted Chinese and suggested that TyG might be an important indicator in identifying population at high risk of T2DM. However, the applicability and utility of TyG index for predicting T2DM in normal-weighted people should be further confirmed in the entire population (24).

In Chinese adults, IR showed a stronger correlation with T2DM risk compared to β-cell dysfunction (25). In clinical practice, in addition to BMI, measurement of WC might be useful in identifying and managing overweight/obese population at high cardiometabolic risk (26). Anthropometric parameters (WC, waist-to-hip ratio, BMI), surrogate measurements of IR (fasting plasma glucose, insulin, fasting insulin-glucose product), fasting lipids, SBP and DBP were also important in predicting T2DM risk (27). Although BMI was a major risk factor of T2DM-related mortality, many people with normal BMI were referred to as “metabolically obese but normal weight (MONW),” which was characterized by a variety of metabolic risk factors and would significantly increased T2DM risk in many ethnic groups.

TyG, a combination of fasting blood glucose and TG, was considered a surrogate marker of IR. Previous studies on middle-aged and elderly adults in China suggested that TyG could identify metabolic syndrome (Mets) (28, 29). There was a significant correlation between the TyG index and incidence of T2DM in individuals with obesity (including visceral fat obesity), fatty liver and normal BMI, which was not a measure of body fat distribution (30). In Japanese with normal glycemic levels, the longitudinal cohort research showed a positive correlation between baseline TyG-BMI and the risk of T2DM, which was noticeably higher in young people, women, non-hypertensive people, and non-drinkers (31). Previous studies suggested that TyG-waist-to-height ratio and TyG-BMI were both clinically effective markers for predicting diabetes in individuals with normal and impaired fasting glucose levels, respectively (16). Numerous studies on the TyG index in individuals of normal weight demonstrated the significance of the TyG index (23, 32).

For female, TyG, VAI and LAP indexes were considered better predictors for T2DM than conventional anthropometric and laboratory measures (27). Based on BMI, WC, TG, and HDL-C, VAI was a gender-specific indicator of visceral adiposity. The risk of new-onset T2DM was significantly correlated with baseline VAI and its transitions. Early prevention was required to reduce the risk of T2DM in Chinese with high VAI levels (33). LAP index, a phenotype of obesity produced by WC and fasting TG, was independently linked to T2DM in hypertensive female. Elevated LAP levels and patterns of low-to-high and maintained-high LAP transitions were significant risk factors for T2DM in women (34). Prevention was needed to combat T2DM at an early dyslipidemic stage. Previous studies suggested that TG/HDL-C was a better marker for assessing IR and DM in elderly Chinese when compared with other common lipid measurements (35, 36). For African American women, the TG/HDL-C ratio was a predictor of β-cell function (37), but it was not a predictor of IR (38). Although the TyG presented significantly better predictive performance than CVAI, LAP and TG/HDL-C in this study, it still required further research in other populations.

The data of this study were obtained from a cohort, so there was a strong causal inference. Compared to the traditional HOMA-IR detection method, surrogate IR indexes were cheaper, faster, and helpful to earlier prevention of T2DM. Researchers could fully utilize both the investigated data and laboratory data when considering the longitudinal changes of IR to reflect the dynamic changes of the human body. Calculating the difference of indexes could reduce the absolute difference between the data and avoid the influence of individual extreme values. This study provided a potential direction for predicting the risk of T2DM development. There were some limitations. First, the follow-up period of this study was not long enough. Second, as the majority of the study's participants were middle-aged and elderly individuals, further researches on other age groups would be necessary. Previous studies suggested that there was a slight difference in DM incidence between women and men. More researches should be encouraged to explore specific T2DM predictors in both women and men. Finally, we were unable to include some other potential confounders such as physical activity, diet and genes in our analysis, which were not contained in the database.

Our study found that TyG-D index was a stronger predictor in the clinical setting for identifying individuals with T2DM in middle-aged and elderly Chinese compared with other indexes. Among the four analyzed indexes, TyG-D presented the largest AUC value [0.66 (0.63–0.69)] and highest sensitivity.

Publicly available datasets were analyzed in this study. This data can be found at: China Health and Retirement Longitudinal Survey (2020). Available online at: http://charls.pku.edu.cn/pages/data/111/zh-cn.html.

YH and LP designed the study. LL, HP, JL, HW, and GX contributed to data acquisition. MW and LP performed the statistical analysis. JH and YH contributed to the discussion. LP and YG drafted the manuscript. JH edited the manuscript. All authors read and approved the final manuscript.

This work was supported by the Research Program of Deyang Municipal Science and Technology Department (2022SCZ089 and 2022SCZ130).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.