Glucose Metabolism Indices and the Development of Chronic Kidney Disease: A Cohort Study of Middle-Aged and Elderly Chinese Persons

Objective Chronic kidney disease (CKD) has become a major global health issue, and abnormalities of glucose metabolism are a risk factor responsible for development of CKD. We aimed to investigate associations between glucose metabolism indices and CKD in a Chinese population and determine which index is superior for predicting incident CKD. Methods We performed a community-based population on 5232 subjects aged ≥40 years without baseline CKD. CKD was defined as an estimated glomerular filtration rate (eGFR) < 60 mL/min/1.73 m2 or urinary albumin-to-creatinine ratio (UACR) ≥30 mg/g. We examined the associations of glucose metabolism indices, including fasting plasma glucose (FPG), 2-hour (2 h) oral glucose tolerance test (OGTT), hemoglobin A1c (HbA1c), fasting insulin level, homeostasis model assessment of insulin resistance (HOMA-IR), and HOMA-β and the development of CKD. Results With an average follow-up of 3.6 years, 6.4% of the subjects developed CKD. Pearson's correlation analysis revealed that FPG, HbA1c, fasting insulin, and HOMA-IR were all significantly correlated with UACR and eGFR. The association persisted in multivariate linear regression analysis adjusted for age and sex. Compared with other glucose indices, HOMA-IR exhibited the strongest associations with CKD in COX multivariate regression analysis (HR = 1.17, 95% CI: 1.04–1.31). Conclusion HOMA-IR is superior to other routine indices of glucose metabolism for predicting the development of CKD in middle-aged Chinese persons. Screening with HOMA-IR may help prevent the development of CKD in the general population.

increased all-cause mortality [6]. CKD and associated morbidities are important drivers of increased health care costs [7]. Importantly, in many patients, CKD is not diagnosed until it is in a later stage [8]. Tus, it is important to identify factors that may predict the development of CKD so that early interventions may be given to prevent or delay its development.
Te association of glucose metabolism and the development of CKD have been extensively investigated over the past decade. Diabetes is the leading cause of CKD, and up to one-third of adults with newly diagnosed diabetes already have CKD [9]. Diabetes is thought to be responsible for almost 40% of new cases of CKD [10]. Prediabetes also increases the risk of developing CKD [11,12]. Diabetes and prediabetes diagnosed according to an elevated fasting plasma glucose (FPG) and/or 2-hour oral glucose tolerance test (OGTT) and/or elevated hemoglobins (HbA1c) based on the World Health Organization (WHO) [13] or American Diabetes Association (ADA) [14] criteria are closely related with the development of CKD. Several studies [11,15,16] have shown positive associations between CKD and FPG, OGTT, and HbA1c, as well as fasting insulin level, homeostasis model assessment of insulin resistance (HOMA-IR), and HOMA-β. A wide body of evidence supports that abnormalities of glucose metabolism are related with the development if CKD. However, associations have mainly been identifed through cross-sectional studies; longitudinal data associating indices of glucose metabolisms and CKD are limited. In addition, few studies have examined if any single index of glucose metabolism is superior for predicting the development of CKD.
Tus, the purpose of this study was to use longitudinal data to investigate associations between glucose metabolism indices and CKD in a Chinese population and determine which index is superior for predicting CKD.

Participants.
Te study population was from the Risk Evaluation of cAncers in Chinese diabeTic Individuals: a IONgitudinal (REACTION) study, which was a multicenter prospective observational study with the aim of evaluating chronic diseases in the Chinese population. Detailed information about the study design and protocols has been published previously [17]. A total of 9916 subjects from a community in Guangzhou, China, signed the informed consent and were included in the baseline survey, from June to November 2011. According to the ADA diagnostic criteria, the baseline diagnosis for the DM group was 1186 cases. All baseline examinations were performed in 2011, and follow-up examinations were carried out from 2014 to 2016. Of the 9,166 individuals included in the baseline survey, 2,917 were lost to follow-up. Tus, a total of 6,999 participants were included in the fnal dataset (follow-up success rate � 71%). With a mean of 3.6 years followup, 125 subjects died and 995 subjects completed the questionnaire by telephone interview. Te subjects who failed to provide baseline information of albumin-tocreatinine ratio (ACR) (n � 78), creatinine (n � 8), FPG (n � 7), PPG (n � 35), OGTT 2 h glucose (n � 35), and HbA1c (n � 19) were excluded from the analyses. Te individuals who failed to provide follow-up information of ACR (n � 50), creatinine (n � 3), OGTT 2 h glucose (n � 23), and HbA1c (n � 5) were excluded from the analyses. For the current study, subjects with baseline CKD were excluded (n � 378). Tus, a total of 5,273 individuals were included in the current analyses, which included 1319 people in the DM group and 3954 people in the nondiabetic group. Detailed patient selection for this study is shown in Figure 1.
Te study protocol was approved by the Institutional Review Board of Sun Yat-sen Memorial Hospital, Sun Yatsen University (2014 [2]). Te study was performed in accordance with the principles of the Helsinki Declaration, and all participants provided written informed consent.

Clinical and Biochemical Measurement.
A standardized questionnaire was used to collect baseline data and was administered by trained interviewers during a face-to-face interview. Information collected included lifestyle factors, sociodemographic characteristics, family history, and the type of diabetes. Current smoking and drinking status were divided into 3 groups: never, ever (the cessation of smoking and drinking for more than half a year), and current (smoking or drinking regularly in the recent half year). Te frequency and duration of physical activity were obtained using the International Physical Activity Questionnaire (IRAQ), and the level of physical activity was evaluated by calculating the metabolic equivalent hours per week (METh/week). Te types of diabetes were divided into 3 groups: type 1 diabetes, type 2 diabetes, other types of diabetes, and the rest. In our study questionnaire, we asked about the types of diabetes and recorded the following responses: 10 cases reported having type 1 diabetes, 241 cases reported having type 2 diabetes, 38 cases reported having other types of diabetes, and the rest were missing data. Notably, although the proportion of type 2 diabetes among diabetes varies by region and year, it generally ranges from 90% to 95% [18][19][20]. Terefore, in this study, we believe that most of the newly diagnosed diabetic patients in our community-based study were type 2 diabetic patients, especially considering that our study population was aged ≥40 years.
Anthropometrical examinations were conducted by trained staf using standard protocols. Body weight and height were measured with subjects wearing light indoor clothing without shoes and recorded to the nearest 1.0 kg and 0.1 cm, respectively. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared (kg/m 2 ), and BMI was used to defne obesity. Obesity was defned as a BMI ≥ 28.0 kg/m 2 , and overweight when 24.0 kg/m 2 ≤ BMI < 28.0 kg/m 2 . Waist circumference (WC) was measured at the umbilical level to the nearest 0.1 cm with subjects in the standing position using a nonelastic measuring tape. Central obesity was defned as a WC ≥ 90 cm in men and ≥80 cm in women. Blood pressure was obtained with the subject seated 3 consecutive times at 5 minutes intervals using an automated electronic device (OMRON, Omron Company, Dalian, China). Te average of the 3 measurements was used in the analysis. Hypertension was defned as a systolic blood pressure (SBP) ≥ 140 mmHg or a diastolic blood pressure (DBP) ≥ 90 mmHg or the subject reporting that they were receiving regular antihypertensive treatment.

Statistical Analyses.
Baseline characteristics of study participants were expressed as mean ± standard deviation for continuous variables with a normal distribution, or median and interquartile range (IQR) for continuous variables with a skewed distribution. Categorical variables were summarized as count and percentage. UACR, FPG, 2 h OGTT, HbA1c, fasting insulin, HOMA-IR, HOMA-β, TG, ALT, AST, c-GGT, and MET-h/week were logarithmically transformed prior to the analysis due to skewed distributions. Characteristics between groups were compared using one-way ANOVA. Comparisons between categorical variables were performed with the χ 2 test. Correlations between the indices of glucose metabolism (FPG, 2 h OGTT, HbA1c, fasting insulin, HOMA-IR, and HOMA-β) with UACR and eGFR were examined with Pearson's correlation analysis and multivariate linear regression. Cox proportional hazard analyses were used to calculate incidence of CKD, and the results were expressed as hazard ratio (HR) and 95% confdence interval (CI). Model 1 was unadjusted; Model 2 was adjusted for age, sex, and BMI; Model 3 was further adjusted for current smoking status, current drinking status, physical activity level, SBP, c-GGT, and LDL-C, which are wellestablished metabolic factors that may infuence glucose metabolism status and CKD progression according to clinical evidence and previous literature. Using continuous variables for logistic regression analysis would answer how much the risk of CKD increases per unit increase in HOMA-IR, but in clinical practice, the signifcance of HOMA-IR increasing by one unit is negligible. Hence, grouping and comparing by HOMA-IR levels is more clinically relevant. Moreover, quartile categorization evenly divides the data into 4 groups according to sample size, which may minimize the issue of low testing efciency due to large sample size diferences among groups. Finally, we calculated the bin width by dividing the specifcation tolerance or range (USL-LSL or max-min value) by the number of bins (Bin width � (max-min)/sqrt (n)). Te relations of indices of glucose metabolism with CKD were also examined in subgroups stratifed by age (≥58 or <58 years), sex (male or female), degree of obesity categorized by BMI (normal, overweight, or obese), central obesity (yes or no), diabetes (yes or no), and hypertension (yes or no). Interactions were tested by including strata factors, the quartile of glucose metabolism index, and the respective interaction terms (strata factors multiplied by quantiles of glucose metabolism index) simultaneously in the models. All statistical analyses were performed using SAS software version 9.3 (SAS Institute Inc., Cary, NC, USA). All statistical tests were 2-sided, and values of P < 0.05 were considered statistically signifcant.

Clinical Characteristics of the Study Population.
Te 5,273 subjects had a mean age at baseline of 58.7 ± 7.2 years, and after a mean follow-up interval of 3.6 ± 0.7 years, 335 (6.4%) had developed new CKD. Te clinical and biochemical characteristics of the subjects at baseline are summarized in Table 1. Compared to those who did not develop CKD, subjects that did had increased FPG, 2 h OGTT, HbA1c, fasting insulin, and HOMA-IR (all P < 0.001). However, HOMA-β was not diferent between the 2 groups. Table 2, Pearson's correlation analysis revealed that baseline FPG, 2 h OGTT, HbA1c, fasting insulin, and HOMA-IR were signifcantly correlated with follow-up UACR and eGFR. No signifcant relation between HOMA-β and UCAR was observed in the correlation analysis. After adjusting for sex and age, a multivariate linear regression analysis showed that the correlations between the baseline glucose indices and UACR persisted (all P < 0.001), except for HOMA-β. HOMA-IR was most strongly correlated with UACR (β = 0.13 and P < 0.001) and eGFR (β = −0.05 and P � 0.0001). Meanwhile, the linear correlation between glucose indices and UACR or eGFR in diabetic and nondiabetic populations separately is shown in Supplementary Tables 1 and 2. 3.3. Associations between Baseline Glucose Indices and Risk of CKD. Figure 2 shows the incidence of CKD in quartiles of the diferent baseline glucose indices. Te incidence of CKD tended to increase with increasing FPG, 2 h OGTT, HbA1c, fasting insulin, and HOMA-IR (all P for trend <0.001). No diferences of CKD were found in HOMA-β quartiles (P � 0.734). HOMA-IR exhibited the strongest correlations with increased risk of CKD in all of the Cox regression analysis models, indicating that the association between HOMA-IR and CKD was robust and independent (Table 3). Te HR for the risk of CKD in the frst (lowest) quartile of HOMA-IR was 1.00 (reference); for the second quartile, the HR � 1.15 (95% CI: 0.79-1.69); for the third quartile, the HR � 1.22 (95% CI: 0.84-1.78); and for the fourth quartile (highest), the HR � 1.61 (95% CI: 1.10-2.34).

Subgroup Analyses of HOMA-IR with Risk of CKD.
As shown in Figure 3, multivariate analyses of subgroups indicated that the association of baseline HOMA-IR with the development of CKD was diferent in diferent subgroups. Signifcant diference of such relationship was detected in subjects with age <58 years, women subjects, those BMI in normal range, those with diabetes, and those without hypertension.

Discussion
In this study, we evaluated the relations between diferent indices of glucose metabolism and the development of CKD in a large population of middle-aged Chinese individuals from the REACTION study. Te results showed that 3 indices of glucose metabolism, 2 h OGTT, HbA1c, and HOMA-IR, were signifcantly associated the development of CKD, independent of potential confounding risk factors. Of the 3 indices, HOMA-IR exhibited the best predictive ability.
To the best of our knowledge, this is the frst and largest population-based cohort study to examine the best index of glucose metabolism for predicting the development of CKD. Because the only efective treatment for ESRD is transplantation, controlling risk factors for the development of CKD and screening methods to determine persons at greater risk of developing CKD are important for decreasing the number of patients developing ESRD. Te fndings of the present study may assist in identifying persons who are at risk of developing CKD and who may beneft from early interventions. CKD is becoming a global health problem, and global death rates from kidney diseases have risen by 83% since 1990 [24]. Glucose metabolism has been shown to be an important factor in the development of CKD [12,25,26]. Moreover, we performed a subgroup analysis by diabetes status in Figure 3, where we explored the association between HOMA-IR and CKD in diabetic and nondiabetic populations separately. Tis study indicated that in the diabetic population, after adjusting for confounding factors, the risk of CKD increased signifcantly with increasing quartiles of HOMA-IR (OR: 1.37 (1.07-1.74)), while in the nondiabetic population, after adjusting for confounding factors, there was also an increasing trend of CKD risk with increasing HOMA-IR levels, OR = 1.03, but it did not reach statistical signifcance (94% CI: 0.89-1.20). Tus, our fnding that an elevated 2 h OGTT and an elevated HbA1c level are independent risk factors for development of CKD is    Data are odds ratios (95% confdence interval). Participants without CKD at follow-up are defned as 0 and with CKD as 1. # All variables were calculated for 1-quartile change of glucose indices. * AUC (95% CI), area under the receiver operating characteristic curve (AUC), and the corresponding 95% confdence intervals (CIs). Model 1 is unadjusted. Model 2 is adjusted for age, sex, and BMI. Model 3 is adjusted for age, sex, BMI, current smoking status, current drinking status, physical activity level, SBP, c-GGT, and LDL-C. consistent with that of prior studies [26,27]. Gabir et al. [27] studied 5023 Pima Indian adults, and with a follow-up period of 10 years showed that a 2 h OGTT can predict the development of CKD. Gwang et al. [26] studied 7728 subjects with a median follow-up of 8.7 years; 871 (11.3%) developed CKD and HbA1c was an independent predictor for the development of CKD. However, there have been conficting reports of the association of CKD development and glucose metabolism. In a study in Germany, Schöttker et al. [28] followed 3,538 participants during 8 years and found that prediabetes might not contribute to the development of CKD and that preventive eforts such as regular exercise might reduce the risk of developing CKD [29]. An animal study using rats showed that physical training increased insulin sensitivity by enhancing muscle glucose uptake and glucose utilization via glycolysis [30]. A study of hemodialysis patients showed that moderate physical training, using plasma insulin level for patients reduces by 40% [31]. In addition, some crosssectional studies showed that neither glucose tolerance nor insulin secretion was associated with CKD. Hanssen et al. [32] studied 1796 persons with normal glucose metabolism, 478 with prediabetes and 669 with type 2 DM, and reported no association of CKD with 2 h OGTT or HbA1c. However, the follow-up was relatively short, and the results interpreted as short or intermediate follow-up periods might not capture the associations of 2 h OGTT or HbA1c with CKD. It is possible that there might be geographic or race variability with respect to the association of 2 h OGTT and HbA1c and the development of CKD, as the study has shown that both are predictors of insulin resistance, which is strongly associated with the development of CKD [33]. Clinical signifcance of each indicator is expected to be studied in the future and further explored with long-term and multistage longitudinal measures to better defne the relationship.
Te results of this study showed that HOMA-IR was the strongest predictor of the development of CKD in middleaged and elderly Chinese persons. Te HOMA-IR refects a pathological state in which target tissues fail to respond normally to the biological efects of insulin and is generally considered as an important infuential factor for development of CKD. In this study, we investigated the linear correlation between glucose indices and UACR or eGFR in diabetic and nondiabetic populations separately (Supplementary Tables 1 and 2). Besides, we acknowledge that the correlation coefcients between 0.11 and 0.14 represent a relatively low correlation, which may be due to the large sample size of our study, and the wide distribution range of glucose indices and eGFR data, which are not as homogeneous as animal and cell experiments, results in larger correlation coefcients in the linear correlation analysis. Te purpose of our correlation analysis was to serve as an exploratory supplementary material, that is, to analyze the same batch of data using multiple statistical methods and to see whether the results of the analysis are consistent, thus increasing the credibility of the article's conclusions. Although the correlation coefcients between 0.11 and 0.14 are relatively low, they all show signifcant statistical signifcance, indicating the necessity of further exploring the  relationship between HOMA-IR and eGFR. Meanwhile, we selected FPG, OGTT 2 h glucose, HbA1c, fasting insulin, HOMA-IR, and HOMA-β as indicators that are commonly used in clinical practice to assess glucose metabolism and insulin resistance. We aimed to explore the association of these glucose metabolism indicators with CKD and compare their predictive value. Terefore, we believe that our analysis of these six indicators is relevant and informative. Meanwhile, this study may provide new ideas and evidence for exploring the mechanism of glucose metabolism and CKD. A study using NHANES data showed that individuals with the highest insulin levels had a 2.65 times greater risk for the development of CKD (95% CI: 1.25-5.62) [34]. Ma et al. [35] studied 3,237 middle-aged and elderly Chinese persons with a 3-year follow-up and showed that an elevated HOMA-IR was associated with accelerated progression of CKD. Huh et al. [36] studied 6,065 Korean persons without CKD at baseline, and over a follow-up period of 10 years showed that insulin resistance was an independent risk factor for development of CKD. Interestingly, the study by Feng et al. also found that the Metabolism Score for Visceral Fat (METS-VF) was a new indicator for predicting CKD risk, which integrated HOMA-IR, waist-to-height ratio, visceral adiposity index, and body mass index [37]. METS-VF can comprehensively evaluate the impact of abdominal adipose tissue on whole-body metabolism, and abdominal adipose tissue is one of the important sources of insulin resistance. Meanwhile, a cross-sectional study conducted by Lin et al. in Southeast China found that HOMA-IR had gender-specifc and age-specifc associations with albuminuria and renal function impairment, which suggested that HOMA-IR may have diferent efects on renal function depending on sex hormones and aging [38]. Terefore, diferent HOMA-IR thresholds may be needed to determine insulin resistance and CKD risk in diferent gender and age groups. In addition, a prospective study of 73 nondiabetic subjects with CKD showed that of incident CKD in Korean population. However, a prospective study of 73 nondiabetic subjects with CKD showed that HOMA-IR was not signifcantly diferent in patients with or without renal endpoints [39]. And the fndings of the study might not extend to our study since it was conducted from one region of Turkey with a small sample (n � 73).
Tere are some theories as to why insulin resistance increases the risk of developing CKD. First, normally, insulin binds to the insulin receptor can activate insulin receptor substrate-1 (IRS-1), which can phosphorylated phosphatidylinositol 3-kinase (PI3-K). Under insulin resistance conditions, impaired PI3-K lead to reductions in bioavailable nitric oxide (NO) directly resulting in the development of endothelial dysfunction and CKD [40]. Second, insulin resistance promotes CKD at the molecular level by infammation through endoplasmic reticulum (ER) stress and is involved in the pathophysiology of chronic kidney injury with tubulointerstitial damage [41]. Moreover, insulin resistance can increase the levels of infammatory cytokines, which can lead to basement membrane thickening, glomerular mesangial expansion, and the loss of slit pore diaphragm integrity, ultimately leading to glomerulosclerosis and tubule-interstitial injury [42]. Tird, insulin resistance may cause overproduction of LDL-C and contribute to hypertriglyceridemia, which can result in renal diseases [43]. Triglyceride-rich apolipoprotein B-containing lipoproteins promote the progression of renal insufciency [44]. Lastly, insulin resistance promotes CKD by worsening renal hemodynamics through mechanisms such as activation of the sympathetic nervous system, sodium retention, decreased Na + , K + -ATPase activity, and increased GFR [45,46].
Te causes of insulin resistance are complex and multifactorial and involve genetic factors such as postreceptor signaling defects, an unhealthy lifestyle that includes a lack of physical activity and poor diet which can lead to obesity, obesity, medications, aging, metabolic acidosis, oxidative stress, infammation, vitamin D defciency, uremic toxicity, and anemia, as shown by previous human and animal studies. Other comorbid conditions that are strongly associated with insulin resistance are hypertension, diabetes, and hyperlipidemia.
Several limitations of this study need to be addressed. First, by including only middle-aged and elderly Chinese subjects, the results might not apply to diferent races or a population of younger individuals. Second, our study population was predominantly female; this was partially because we invited person ≥40 years to participate and females are predominant in this age range in China, and women were more likely to participate than men in all surveys [47,48]. Tird, as with any observational study, all confounding factors that may contribute to the development of CKD may not have been included in the models. Fourth, the "gold-standard" for documenting insulin resistance is the euglycemic clamp test. However, the euglycemic clamp test is time-consuming and requires trained personal so it is rarely used in large epidemiological studies. HOMA-IR is a common method used to assess insulin resistance in large epidemiological studies, and it is relatively well-correlated with the euglycemic clamp technique (r � 0.88) [49]. Fifth, we defned the CKD based on the frst measurements of eGFR and UACR; however, the gold standard is two measurement results. Tis approach may have reduced the accuracy of our results; however, the results of 1 measurement correlate well with those of 2 measurements and use of 1 measurement is common in large epidemiological studies [50]. Besides, the follow-up rate in this study of 71% was relatively low; however, large epidemiological investigations rare studies can achieve follow-up rate of ≥85% [51,52]. It is worth noting that the follow-up rate of another REACTION study with a defned 3-year follow-up in Shandong Province, China, from 2012 to 2015 was 77.8%, which was similar to that of our study [35]. Finally, the follow-up time of 3.6 years on average was relatively short to determine the long-term risk of CKD progression based on HOMA-IR or other glucose metabolism indicators. Terefore, our fndings need to be validated by studies with longer follow-up periods that can capture more advanced renal outcomes or declining eGFR as markers of progressive CKD. However, our results still suggested that HOMA-IR may be useful to detect early signs of CKD deterioration, such as increased ACR or reduced eGFR, which may have clinical implications for early prevention and management of CKD. Meanwhile, our study provides new ideas and evidence for the subsequent research on the mechanism of the association between glucose metabolism and CKD.

Conclusion
Our study demonstrated that HOMA-IR is superior to other glucose metabolism indices in predicting the development of CKD in a middle-aged and elderly Chinese population. HOMA-IR may be a useful screening method to determine persons at risk for the development of CKD and who may beneft from early interventions.

Data Availability
Main document data and additional unpublished data from the study are available by sending email to youlli@mail.sysu.edu.cn with proper purposes.

Additional Points
Patient and Public Involvement. Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Strengths and Limitations of Tis Study. (i) Tis study was performed in a large population-based cohort of 5232 Chinese subjects. (ii) Finding of this study based on cohort study may be applied to the majority of CKD patients, and HOMA-IR has a better prediction index for development of CKD. (iii) By including only middle-aged and elderly Chinese subjects, the results might not apply to diferent races or a population of younger individuals. (iv) Our study population was predominantly female; this was partially because we invited person ≥40 years old to participate and females are predominant in this age range in China.

Ethical Approval
Te study protocol was approved by the Institutional Review Board of the Sun Yat-sen Memorial Hospital afliated to Sun Yat-sen University and was in accordance with the principle of the Declaration of Helsinki II.

Consent
Written informed consent was obtained from the participants.

Disclosure
A preprint has previously been published [53]. Lili You, Xiaosi Hong, and Qiling Feng are co-frst authors.

Conflicts of Interest
Te authors declare that they have no conficts of interest.