Introduction

Type 2 diabetes is a growing disease worldwide, imposing a severe burden on society along with its complications [1–4]. In Iran, the incidence rate of type 2 diabetes in the urban population over 20 years is more than 1 % annually [5]. The incidence rate of pre-diabetes is significantly higher and estimated to be more than 4 % each year, indicating a very high prevalence of diabetes soon [6].

Pre-diabetes is associated with a high risk of micro and macrovascular complications [7–10]. As a result, early detection of people susceptible to pre-diabetes, lifestyle modifications, and effective medications is necessary to prevent develo** pre-diabetes and its complications in these people [4, 11]. 30-year results of the Da Qing Diabetes Prevention Outcome Study showed that lifestyle intervention in people with pre-diabetes could reduce 40% development of diabetes and delay its onset by 3·96 years [12]. Insulin resistance and dysfunction of pancreatic beta cells are important factors in the pathophysiology of diabetes and pre-diabetes, contributing to different degrees among pre-diabetic subgroups according to different races and ethnicities [13–27]; thus, the effects of HOMA-B changes on the incidence of pre-diabetes can be affected by the time interval until the desired outcome in the future.

Also, in a study by Derakhshan et al., beta-cell dysfunction or low HOMA-B (HOMA-B < 25 percentile) was associated with increased risk of iIFG in a multivariate model [H.R. = 1.37(1.03–1.81) in men, and HR = 1.36(1.02–1.80) in women] which is in line with our findings. This study further showed that increased HOMA-IR was associated with increased HR in all four groups, which is in line with the results of our study, too [28]. A 2009 prospective study in Denmark to determine the association between insulin sensitivity and insulin secretion with the development of normoglycemic individuals to IFG or IGT reported that liver insulin resistance and subsequent dysfunction of beta cells in the secretion of insulin are the main factors in the progression of normal blood glucose to iIFG during 5 years, while the reduction of whole-body sensitivity and consequently insulin secretion dysfunction was reported as the major factors leading to iIGT [21]. It should be noted that, in this study, OGTT was performed, HOMA-S and ISI (Insulin Sensitivity Index) were used to assess insulin sensitivity, and EPIR (Early Phase Insulin Release) and DI (Disposition Index) were used to measure insulin secretion [29].

Too many studies investigated the association of HOMA-IR and HOMA-B with IFG and IGT. However, most of these were cross-sectional studies and evaluated this association concurrently with the occurrence of phenotypes, making it impossible to compare their results with the findings of this study. In this study, increasing HOMA-B per SD resulted in a reduced incidence of iIFG and an increased incidence of iIGT, CGI, and DM; Thus, AUC was not an appropriate criterion for predicting the development of pre-diabetes. On the other hand, HOMA-IR does not produce an acceptable AUC and cut-off point, so HOMA-IR is not a good index for predicting pre-diabetes. None of the ROC diagrams of HOMA-IR and HOMA-B, which were prepared separately for iIFG and iIGT, resulted in appropriate AUCs. In a study by Dr. Ghassemi et al., the HOMA-IR cut-off point for diagnosis of diabetes was 2.17 (50% sensitivity and 76.7% specificity) in women and 1.85% (75.9% sensitivity and 58.3% specificity) in men [19]. Compared with the results of Dr. Ghassemi’s study for diabetes, the cut-off point of our study seems to be inappropriate for predicting the incidence of pre-diabetes because the HOMA-IR cut-off point for pre-diabetes is expected to be lower than that for diabetes. Also, in a cross-sectional study by Baek et al., the reported cut-off point of HOMA-IR for detecting dysglycemia based on FBS and HbA1C (equal to 1.6) was lower than that obtained in our study [30]. Given that the HOMA-IR formula only includes fasting glucose and insulin, mainly related to insulin resistance in the liver, it seems complicated to obtain a single HOMA-IR cut-off point for all pre-diabetic subjects.

Since insulin levels and glucose concentrations are the only measurements needed to calculate HOMA-IR and HOMA-B, they have become the most widely used surrogate indicators, providing valuable insights into insulin resistance, b-cell function, and glucose metabolism [26, 31, 32]. However, in our findings, these were not good criteria for predicting the incidence of diabetes/pre-diabetes, emphasizing that HOMA should not be considered exclusively in the framework of clinical practice. Using other clinical features of subjects or indices such as triglyceride-glucose index [33], which has been shown that are important in predicting metabolic syndrome alongside HOMA-B/HOMA-IR, could help to decide more precisely for individuals concerning the prediction of pre-diabetes/diabetes in the future. Nonetheless, significant differences in insulin levels have been reported between different populations and ethnicities [34], and further research may be needed to re-evaluate our findings in other populations.

This study has some limitations and benefits. The main limitation of this study is the inability to use more precise methods to measure insulin resistance and its secretion (such as the clamp technique). Other limitations include the lack of information on HbA1C and missing participants during the study. Although the technique used in this study for measuring insulin (Electrochemical Luminescence) is currently one of the most precise methods, since there is no reference method for measuring insulin and the results of insulin measurement by different methods can vary up to 2 times, the method of insulin measurement should be considered while using the results [35].

The advantages of this study are as follows: It is the only study that evaluated the HOMA-IR cut-off point to predict the incidence of pre-diabetes. This study was performed in a high sample size and is the only study that prospectively investigates the association between changes in HOMA-B and HOMA-IR with the incidence of diabetes and pre-diabetes subtypes. In this study, the data analysis was performed using a multinomial logistic regression method, suitable for investigating the concurrent relationship between HOMA-B and HOMA-IR with several outcomes (iIFG, iIGT, CGI, and DM). This method seems to be more appropriate for investigating this relationship than the COX method used in the previous study.

In conclusion, this study showed that increased HOMA-IR in normoglycemic individuals is associated with increased incidence of pre-diabetes in both iIFG and iIGT subtypes, while increased HOMA-B leads to a higher risk of develo** iIGT and lower risk of iIFG. There was no significant difference between the incidence of iIGT and iIFG with increased HOMA-IR. Based on the results of this study, it seems that cut-off points of HOMA-IR and HOMA-B are not suitable criteria for predicting the incidence of pre-diabetes.