The prevalence of diabetes mellitus in the Chinese population is rapidly increasing, causing China to be the country with the largest population of individuals with diabetes in the world. Type 2 diabetes mellitus (T2DM) is the most common type of diabetes, and its prevalence was reported to be 0.67% in 1980, 11.6% in 2020 and 10.9% according to latest research (1). Approximately 46% of adults in China are obese or overweight, and the prevalence of T2DM in overweight and obese populations is 12.8% and 18.5%, respectively (2). Obesity and T2DM have common triggers in China, such as rapid urbanization and lifestyle changes (3, 4). Obesity can lead to impaired glucose tolerance and insulin resistance (5, 6). Recently, it has been reported that the influence of body weight on T2DM is more obvious among Asian people (7). A nationwide prospective study suggested that insulin resistance was significantly correlated with T2DM, more so than with β-cell dysfunction, in Chinese adults, and that this association pattern was more prominent among those with obesity (8). Body mass index (BMI) is the preferred factor for evaluating adiposity. A strong positive association between BMI and an elevated risk of T2DM has been widely reported by previous studies (9–12). However, the research mostly focused on current BMI or BMI at baseline. Life-time maximum body mass index (BMImax) is an indicator of BMI history and has been reported to be a screening tool for predicting the function of β-cells at the time of diagnosis and reflecting the period when obesity more obviously affected blood glucose (13). Although the associations between BMImax and hypertension (14), fatty liver (15), chronic kidney disease (16) and cardiovascular disease (17) has been reported, less is known about the effect of BMImax on the onset of T2DM.

We hypothesized that BMImax, as an index independent of any intervention, such as medication and lifestyle management, might have a marked association with T2DM on the progression of diabetes. To further understand the effect of obesity on T2DM, we conducted an epidemiological study in a Chinese population that was followed up for 8 years.

Our study of Chinese individuals demonstrated that obesity defined by BMImax was positively related to the risk of progression to diabetes, especially for females. BMImax, could be regarded as an independent predictor in the screening and prevention of diabetes in high-risk populations. To our knowledge, this is the first study to explore the association between the historical BMImax and the risk of diabetes in a Chinese population.

The relationship between obesity and diabetes has been reported in several previous studies. Most of the research focused on body weight or body weight change at a certain time point during the research period. Few studies have evaluated the effect of highest lifetime BMI on the progression of diabetes. In 2014, Yoshizawa examined current BMI and BMI history as screening indicators for diabetes. In that study, individuals in the highest tertile of BMImax (≥26.09 kg/m² in men and ≥24.27 kg/m² in women) had a significantly higher risk for diabetes than those in the lowest tertile (18). Recently, a study from Japan investigated the clinical significance of BMImax before the onset of type 2 diabetes for predicting pancreatic beta cell function. They found that BMImax was an independent factor indicating beta cell function and could predict insulin secretion capacity at diagnosis. However, it did not affect the rate of decrease in islet secretion function after a diagnosis of diabetes (13). Several fundamental studies have also explored the close link between obesity and T2DM. In 2014, Mallafre reported that signaling of glucose-dependent insulinotropic peptide (GIP) and its receptor (GIPR) played key roles in the pathogenesis of obesity and its associated insulin resistance. GIP has important metabolic actions other than its physiological glucose-dependent incretin effect. Particularly in human adipose tissue, GIP might act as an insulin sensitizer and an anti-inflammatory factor. Insulin-resistant states related to obesity might be associated with diminished GIP sensitivity, and GIPR expression in human adipose tissue might determine systemic insulin sensitivity (19). Recently, a study provided compelling evidence showing that an inverse correlation between increasing body mass index and decreasing insulin receptor expression in visceral adipose tissue. Visceral adipose tissue-specific downregulation of the insulin receptor was found to be an early event in obesity-related adipose cell dysfunction, which increases systemic insulin resistance in obese humans (20). As insulin resistance and beta cell dysfunction were found to be the main pathogenic mechanisms of T2DM, both of these effects caused by the highest lifetime BMI might influence the progression of diabetes.

Our study suggested that the association between BMImax and diabetes is sex-specific, in accordance with previous research. A study published in PLOS Genetics revealed sex differences in the effect of obesity traits on the risk of type 2 diabetes—a higher BMI led to a higher risk of type 2 diabetes in women than in men. In addition, Logue reported that the mean BMI closest to the date of diagnosis of type 2 diabetes was 31.83 kg/m2 in men and 33.69 kg/m2 in women, so the inverse relationship between BMI and type 2 diabetes was significantly steeper in women than in men (21). A study published in Diabetologia noted the effect of historical weight on glucose metabolism in women. This study was designed to explore the association of the duration of obesity and cumulative exposure with disease. The results indicated that age at onset of obesity was negatively associated with type 2 diabetes in a cohort of women aged 18-23 years at baseline, and the duration of obesity was positively related to type 2 diabetes (22). In contrast, according to the report of Yutaka, the effect of BMI on diabetes in men and women might be related to genetic susceptibility. The authors analyzed the correlations of single-nucleotide polymorphisms (SNPs) of the FTO gene, current BMI, BMImax and type 2 diabetes, and found that SNPs within FTO influence the development of type 2 diabetes by acting on BMImax. However, the result was not obvious in women (23). The researchers attributed the sex-specific effect to the differences in fat distribution and/or physical activity between men and women, which might influence the effects of FTO variants on obesity/adiposity. However, that study was a cross-sectional study with a small sample. The differences in the metabolic effects of obesity between different sexes might be related to estrogen. The ventromedial hypothalamus plays chief roles in regulating energy and glucose homeostasis and is sexually dimorphic. Research has reported the sex-specific role of GLUR5 in regulating the firing rate, intrinsic excitability, and excitatory and inhibitory transmission in ventromedial hypothalamic (VMH) SF1 neurons in female mice, facilitating glycemic control and lipid metabolism. Furthermore, the study also found that intact mGluR5 function in SF1 neurons is required for estrogen to achieve its positive effects on VMH neuronal activity and glucose homeostasis (22).

In addition to BMImax, our results suggested that fasting plasma glucose, 2 h postload glucose and HbA_1c could predict the risk of diabetes. However, the correlation among fasting plasma glucose, HbA_1c and diabetes was much stronger than that among 2 h postload glucose. In line with our results, two cohort studies proposed the predictive value of fasting glucose and HbA_1c. Yoriko suggested that an HbA_1c of 5.7%-6.4% could be regarded as an indicator of diabetes, with a value similar to that of fasting glucose. The two tests used together could efficiently target people who are most likely to develop diabetes (24). Heianza’s research reported that the combination of an HbA_1c of 5.7%-6.4% and a fasting glucose of 5.6-6.9 mmol/l had the best performance in reducing the likelihood of missing future cases of diabetes. An HbA_1c of 6.0%-6.4% and a fasting glucose of 6.1-6.9 mmol/l could be used to screen for diabetes among prediabetic individuals (24). As glucose tolerance tests are not generally conducted during routine medical examinations, fasting glucose combined with HbA_1c, which could reflect chronic hyperglycemia and better evaluate the potential risk of impaired glucose regulation.

Nomograms provided a statistical predictive model that includes simple scores of different clinical indicators in a visual and quantitative way and have been widely used in the evaluation of the risk and possibility of specific diseases or clinical events. Considering that factors associated with diabetes vary by sex, we made separate predictive models for males and females. The variables included in the nomograms were accessible in clinic and could provide a reference for clinicians to assess the risk of diabetes for different patients, facilitating individualized clinical management. In our study, HbA_1c, followed by fasting glucose, was the most effective predictor among both men and women. BMImax and postload glucose ranked next in effectiveness. Consistent with a recently published study in a Chinese population, HbA_1c could improve the performance of nomograms for predicting the 5-year incidence of type 2 diabetes (25). Although we innovatively incorporated BMImax into the predictive model, the efficacy of the nomogram assessed by the ROC curve was moderate, which might be related to the following factors. The effect of obesity on the prevalence of diabetes might be exerted in several ways, for example, fluctuations in weight, highest weight in adulthood, or even the duration. The predictive model developed in this study explored only the effect of historical highest BMI on the risk of developing diabetes. If we could have included more obesity-related factors, such as those mentioned above, a more comprehensive analysis of the effect of body weight on glucose metabolism might have been possible. In addition, the 92 patients who developed diabetes in this study did not include patients with reduced glucose tolerance and impaired fasting glucose regulation, thus failing to mention the effect of maximum body weight on prediabetes.

Our study focused on the correlation between BMImax and diabetes. Several limitations should be considered in this study. First, the definitive types of diabetes were not clear due to unavailable clinical evidence, including diabetes-associated autoantibodies. We hypothesized that our conclusions could be extended to individuals with type 2 diabetes. Second, lifetime BMImax before the onset of diabetes was self-reported and inevitably had a certain recall bias. Further studies should be conducted to confirm these findings with actual measurements of body weight and height. To preliminarily investigate whether recall bias might affect the findings of our study, we divided the subjects into two groups according to the time of BMImax. One group of subjects reached maximum weight at baseline. Subjects in the second group reached their maximum weight before the onset of the study, and thus BMImax was obtained by self-report. We found a correlation between BMImax and diabetes in both groups.

In conclusion, obesity defined by lifetime BMImax has an obvious association with the risk of diabetes in Chinese individuals over 40 years old, especially women. Given the ongoing increases in the prevalence of diabetes and obesity, a better understanding of the effect of obesity is certainly needed to improve the prevention and management of diabetes and diabetes-related complications.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding authors.

The studies involving human participants were reviewed and approved by Clinical Research Ethics Committee of the people’s Liberation Army General Hospital. The patients/participants provided their written informed consent to participate in this study.

XJ, AW and ZL designed the study. XJ, AW, LY, YC, YW collected and analysed the data. JB, JD, YM and DZ contributed samples for this study and provided intellectual input. XJ drafted and wrote the manuscript. AW, DZ and ZL performed critical revisions to the manuscript. All authors contributed to the article and approved the submitted version.

This work was supported by a grant from the National Key Research and Development Program of China (2018YFC1314100).

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.