Metabolic-Related Index to Predict Post Transplantation Diabetes After Renal Transplantation


 Background Early identification of post‑transplantation diabetes mellitus (PTDM) can be helpful to attenuate the rapid development of diabetic complications. This article aims to explore the beneficial values of metabolic-related markers, including TyG, TyG-BMI, TG/HDL-C and non-HDL-C/HDL-C, for predicting PTDM.Material and methods The data of 191 renal transplant recipients in our center were collected retrospectively. The association between the metabolic-related markers and the risk of PTDM was examined by the area under the curve and logistic regression analyses. Results During 6 months follow-up, 12.04% of RT recipients developed PTDM, and significantly higher values of TyG-BMI, TyG and non-HDL-C/HDL-C was found in patients with PTDM than in nondiabetic patients, especially among the recipients taking FK506. The incidence of PTDM increased along with the values of TyG or TyG-BMI. After adjusting for multiple potential factors, recipients with the highest trisector of TyG or TyG-BMI still had a higher risk of PTDM morbidity. Conclusions TyG, TyG-BMI, TG/HDL-C and non-HDL-C/HDL-C can predict PTDM in renal transplant recipients, and TyG-BMI was the best alternative marker among the four markers. TyG and TyG-BMI can be used as cost-effective and complementary monitors to identify individuals at high risk of PTDM.


Introduction
Renal transplantation (RT) is the best renal replacement therapy for end stage renal disease, but post-transplantation diabetes mellitus (PTDM), one of common complications of RT, can adversely affects both short-term and long-term outcomes of RT recipients [1]. The incidence of PTDM ranges between 4% and 39%, and most PTDM is diagnosed in the rst-year posttransplant [2,3]. PTDM can lead to adverse diabetes-related outcomes at a faster rate and early development of PTDM is associated with up to a 3-fold greater risk of cardiac events than nondiabetic recipients [2] . So early identi cation and treatment of PTDM is pivotal to attenuate the rapid development of diabetic complications. Insulin resistance (IR) and defective insulin secretion [4] are two main pathogeneses of PTDM. IR has been reported to be associated with an increased risk of progressive diabetic nephropathy [5]. The frequency of IR was found to be as high as 30% in stable RT patients, which is nearly three times the incidence in the general population. IR can be recognized as an early and strong predictor of PTDM even in the absence of hyperglycemia. Thus, the detection of IR can be helpful to predict the onset of PTDM.
However, the methods of estimating IR remain a challenge in clinical practice. The hyperinsulinemic-euglycemic clamp test is an acknowledged gold standard method for measuring IR [6], but complex operations and high costs restrict its clinical applicability. The homeostasis model assessment of insulin resistance index (HOMA-IR) is a commendable alternative indicator of IR, but the measurement of insulin required for HOMA-IR has been unstandardized and has poor repeatability. Recently, various metabolic risk factors such as the triglyceride-glucose index (TyG), TyG-body mass index (TyG-BMI), triglyceride/ high-density lipoprotein cholesterol ratio (TG/HDL-C) and nonhigh-density lipoprotein cholesterol/ high-density lipoprotein cholesterol ratio (non-HDL-C/HDL-C) have been suggested as simple and economical surrogate markers of IR. In both healthy subjects and abnormal glucose metabolism patients, TyG, TyG-BMI, TG/HDL-C and non-HDL-C/HDL-C had a high sensitivity and speci city for recognizing IR in previous studies [7,8]. However, the association between metabolicrelated markers and PTDM has not reached a consensus until now, so the aim of this article was to explore the bene cial values of metabolic-related markers, including TyG, TyG-BMI, TG/HDL-C and non-HDL-C/HDL-C, for predicting PTDM pretransplantation and earlystage posttransplantation.

Subjects
We retrospectively collected the data of RT recipients who received renal transplantation in our center between January

Measurements
The study subjects were followed monthly from RT surgery to at least 6 months posttransplant. Medical history information including age at RT, primary nephropathy, donor source, weight, height, and use of immunosuppressants was reviewed. Smoking history was de ned as at least one cigarette per day for at least one year, and drinking history was de ned as at least 1 drink per day for at least one year [9]. Weight was measured when patients wore light clothing to the nearest 0.1kg, and height was measured without shoes to the nearest 0.1cm. Body mass index (BMI) was calculated as weight (kg) divided by height (m) squared. Blood pressure (BP) was measured twice on the right arm while the subjects were in a seated position after at least 10 minutes of rest, and the average of the BP readings was calculated as the nal result. Blood samples were obtained from the antecubital vein after fasting for at least 8 hours using BD vacutainer tubes containing EDTA and analyzed in our central laboratory under strict quality control. FBG, TG, TC, LDL-C, high-density lipoprotein cholesterol (HDL-C), serum creatinine (SCr), albumin (ALB), alanine aminotransferase (ALT), glutamine aminotransferase (AST), gamma-glutamyl transferase (γ-GT) and hemoglobin (HB) were tested using a Beckman Coulter chemistry analyzer AU5800 (U.S.A.) and Sysmex XE2100 hematology analyzer (Japan  [11]. In addition, TG/HDL-C ratio and non-HDL-C/HDL-C ratio were calculated. Variables were collected at three time points: pretransplant, 1-month posttransplant and 3-month posttransplant. De nition of PTDM and obesity PTDM was diagnosed according to the criteria of the American Diabetes Association (ADA) as FBG ≥ 126 mg/dl (7.0 mmol/l) or 2-h blood glucose (2h BG) ≥200 mg/dl (11.1 mmol/l) during a 75g oral glucose tolerance test(OGTT) or glycosylated hemoglobin ≥6.5% (48 mmol/mol) or a random blood glucose ≥200 mg/dl (11.1 mmol/l) in a patient with classic symptoms of hyperglycemia or hyperglycemic crisis. In the absence of unequivocal hyperglycemia, the diagnosis requires two abnormal test results from the same sample or in two separate test samples [12]. Subjects with BMI ≥25 kg/m 2 were diagnosed as overweight.

Statistical analyses
The continuous variables were presented as the mean value ± standard deviation (SD), and categorical variables were presented as frequencies and percentages. The normality of the parameters was veri ed using Kolmogorov-Smirnov tests. The means of continuous variables between groups were assessed by a one-way analysis of variance (ANOVA) with replicate measures, and a post hoc analysis of the least signi cant difference (LSD) was used when the means of the continuous variables were normally distributed; if not, the Tamhane's T2 test was used. The Student's t-test was used for two independent continuous variables, and the Chi-square test was used for categorical variables. According to trisections of each studied marker, the non-DM and PTDM subjects were divided into four groups, and the incidence of PTDM was compared between the four groups. The diagnostic accuracy of each studied marker in predicting PTDM was estimated using the AUC from the ROC curve, and an AUC >0.7 was considered a good predictive value. Youden's index was calculated as (speci city+ sensitivity-1) and then was used to determine the optimal cutoffs of the studied markers for predicting PTDM. A logistic regression model after adjustment for multiple confounders was built to calculate the odds ratios (ORs) and 95% con dence intervals (CIs) of the studied markers. Model

Pretransplant demographic characteristics and laboratory data
The pretransplant demographic characteristics and laboratory data of the study subjects are summarized in Table 1. Of the 191 subjects, the mean age was 43.12±11.87 years with 67.54% (129/191) male. Patients with PTDM or DM were older and had higher 2h-BG, higher BMI and higher HDL-C than nondiabetic patients. Patients with PTDM had the highest proportion of tacrolimus (FK506) usage and patients with DM had the highest proportion of overweight. BP, smoking and drinking history, TC, TG, LDL-C, FBG, SCr, eGFR, ALB, ALT, AST, γ-GT and HB were compared between the three groups but no signi cant differences were found. The predictive accuracy of the studied markers for PTDM In the overall subjects, four markers including TyG-BMI, TyG, TG/HDL-C and non-HDL-C/HDL-C were compared among the NC group, PTDM group and DM group at pretransplant, 1-month posttransplant and 3-month posttransplant ( Table 2). A higher TyG-BMI was found in the PTDM group and DM group than in the NC group from pretransplant to 3-month posttransplant (P< 0.01). TyG and non-HDL-C/HDL-C were markedly higher in the PTDM group and DM group at pretransplant and 3-month posttransplant (P <0.05), but not at 1-month posttransplant, while no signi cant difference was found in the value of non-HDL-C/HDL-C among the three groups at all three timepoints. In view of the identi ed correlation between FK506 usage and PTDM, a subgroup analysis according to the type of calcineurin inhibitor (CNI) was conducted. Among the patients taking FK506, a higher TyG-BMI was a rmed in the PTDM group than in the NC group (P <0.01); nonetheless, no similar result was observed in the DM group. However, TyG, TG/HDL-C and non-HDL-  posttransplant, but not at 1-month posttransplant. In the FK506 subgroup, positive relationships were also found between TG/HDL-C and PTDM onset, because from the lowest TG/HDL-C trisection quartile to the highest, the incidence of PTDM increased from 0 to 26.83% at pretransplant and from 2.50% to 22.73% at 3-month posttransplant. A similar association was found between non-HDL-C/HDL-C and PTDM onset in the FK506 subgroup at pretransplant and 1-month posttransplant. In contrast, in the CsA subgroup, no association was veri ed between the studied markers and the incidence of PTDM.

Correlation analysis of the markers and PTDM
Subjects with non-DM and PTDM were the study objects of correlation analysis in our research. Direct receiver-operating characteristic (ROC) regression was implemented to further determine the discriminative power of the four markers to predict PTDM (Table 3) 10.83) in the highest trisector of TG/HDL-C, and 6.95 (95%CI: 4.37-11.04) in the highest trisector of non-HDL-C/HDL-C. The relationships between the studied markers and PTDM were still de nite at 1-month and 3-month posttransplant regardless of whether the interference factors had been adjusted. After transplantation, the largest ORs of TyG-BMI and non-HDL-C/HDL-C were found at 1-month posttransplant, while the largest ORs of TyG and TG/HDL-C were found at 3-month posttransplant. At the early-stage posttransplant, the four studied markers had all been approved as independent predictors for PTDM.

Discussion
PTDM is a noteworthy complication of renal transplantation. In our study we observed that 12.04% of RT recipients developed PTDM, and signi cantly higher values of TyG-BMI, TyG and non-HDL-C/HDL-C was found in patients with PTDM than in nondiabetic patients, especially among the recipients taking FK506 at pretransplant. The incidence of PTDM increased along with the values of TyG or TyG-BMI. After adjusting for multiple potential factors, recipients with the highest quartile of TyG or TyG-BMI still had a higher risk of PTDM morbidity than those with the lowest quartile, suggesting the potential for TyG and TyG-BMI to serve as independent risk indicators of PTDM in clinical practice.
IR may induce glomerular and vascular hemodynamic changes because compensatory hyperinsulinemia can lead to dysfunctional tubuloglomerular feedback caused by an increase in sodium absorption at the proximal tubules and in the loop of Henle, stimulating mesangial cell proliferation and mesangial matrix production [13]. Posttransplant IR has been associated with subclinical atheromasia and chronic subclinical in ammation. Therefore, IR after RT remains prevalent and important, and is an early and very strong predictor of graft function and survival, CV disease and metabolic complications, including PTDM.
Recently, TyG-related indices were introduced as easily measurable, economical and applicable surrogate markers of IR instead of the hyperinsulinemic-euglycemic clamp test and HOMA-IR index [7,14]. The TyG proposed by Guerrero-Romero has shown high sensitivity and speci city in the measurement of IR, metabolic abnormalities, T2DM onset and T2DM-related complication risk [7,[15][16][17]; moreover, it was revealed to be a more e cient marker for the early identi cation of IR than lipid ratios, VAI and LAP in a cross-sectional study [18]. In the diabetic patients with poor glycemic control, TyG increased signi cantly and was signi cantly correlated with HbA1c [19]. The results of several population-based studies have demonstrated that TyG was positively associated with fat distribution, subclinical atherosclerosis, coronary artery disease, prehypertension, hyperuricemia and nonalcoholic fatty liver disease [20][21][22].
The application of TyG-related indices was based on the fact that both lipotoxicity and glucotoxicity play crucial roles in IR modulation.
Accumulating evidence suggests close relationships between IR, obesity and hyperlipoidemia in both diabetic and nondiabetic subjects [11]. One integrative physiological research proposed that hypertriglyceridemia had a negative effect on IR, because it might cause fatty acid accumulation in non-adipose tissues such as the liver, muscle and heart, resulting in ectopic lipid deposition with lipotoxicity [23]. Elevated TG, LDL-C and low HDL-C are probably detected secondary to IR and hyperinsulinism, but before the onset of prediabetes [24]. For RT recipients, preceding studies proposed that the presence of obesity and hyperlipoidemia increases the risks of PTDM and long-term patient and graft loss [25].  [27], regardless of the sex [26]. We obtained similar results, rea rming that TyG-BMI had a larger AUC than TyG, TG/HDL-C and non-HLD-C/HDL-C from pretransplant to 3-month posttransplant, regardless of the type of CNI. The highest TyG-BMI AUC of 0.775 was found in recipients taking FK506 with the optimal cutoff point of 209.84 at pretransplant. A cross-sectional study in Taiwanese individuals also a rmed that TyG-BMI was the most favorable surrogate marker of IR, and TyG-BMI had the strongest association with HOMA-IR [26] . In nondiabetic Chinese subjects, TyG-BMI was a better indicator for detecting IR in both the normal glucose and all glucose categories subjects [28]. In Colombian men, TyG-BMI also had higher OR and AUC values in prediabetes patients than in the nondiabetic population [29]. It is noteworthy that in the current investigation, the values of TyG increased early after transplantation among the recipients taking FK506, but the values of TyG-BMI decreased. Malnutrition and acceleration of metabolism maybe the reasons for these changes, which were caused by surgical operations and the usage of prednisone respectively. The nutritional assessment of patients veri ed inadequate body composition, with increased fat and reduced lean body mass early after renal transplantation [30]. Therefore, it is not recommended to predict PTDM using the dynamic changes in TyG-BMI from pretransplant to early posttransplant.
Lipid parameters such as TG/HDL-C and non-HDL-C/HDL-C are additional markers of IR that have been suggested by some researchers. Hyperinsulinemia was positively associated with the serum TG concentration [31] and inversely associated with the serum HDL-C concentration; moreover, the concentrations of TG and HDL-C might be in uenced by glycemic control in patients with diabetes [32].
Furthermore, low HDL-C concentrations may exacerbate abnormal glucose homeostasis [33]. For RT recipients in our study, although the associations between TG/HDL-C, non-HDL-C/HDL-C and the incidence of PTDM were signi cant even after adjusting for various factors, the signi cant relationships were not maintained when compared according to the trisectors of the values. Obvious predictive abilities were particularly found for TG/HDL-C at pretransplant and non-HDL-C/HDL-C at 3-month posttransplant only among recipients taking FK506. Among recipients taking FK506, high AUCs were obtained at pretransplant and 3-month posttransplant both for TG/HDL-C and non-HDL-C/HDL-C, while among recipients taking CsA, a high AUC was obtained only at 1-month posttransplant for non-HDL-C/HDL-C. Our results were partly in concordance with previous ndings that TG/HDL-C is a practical approach for identifying individuals with diabetes [34]. In the Bypass Angioplasty Revascularization Investigation 2 Diabetes trial, TG/HDL-C was proposed to be a useful marker in individuals who achieved better glycemic control [35], and another study found that TG/HDL-C can be an effective screening tool to predict success with dose reductions of antidiabetic medications in obese patients who successfully lose weight [36]. TG/HDL-C can also be a pretransplant index of IR to predict both PTDM and glucose intolerance after transplantation [37,38]. Among all lipid parameters, it was reported by several articles that TG/HDL-C was more e cient than any other lipid measures or ratios in terms of the strength of the association with IR [26]. We found that a TG/HDL-C value above 2.2 was a better marker of PTDM in the FK506 subgroup, while McLaughlin et al. [39] found that a TG/HDL-C value above 3.0 mg/dL was a better marker of IR, indicating that TG/HDL-C can be a more sensitive marker in RT recipients than in the general population. Non-HDL-C/HDL-C has also been suggested as a better predictor of CVD than LDL-C, non-HDL-C, apolipoprotein B or apolipoprotein A1 [40], however, its predictive value for PTDM has not been studied extensively, so our data can be a valuable reference in this regard in future research.
We further conducted strati ed analyses to investigate whether the type of CNIs affected the association between the studied markers and PTDM. High values of TyG, TyG-BMI, TG/HDL-C and non-HDL-C/HDL-C were signi cantly associated with PTDM only in the FK506 subgroup, but not in the CsA subgroup. The AUCs of the markers above were all higher in the FK506 subgroup than in the CsA subgroup, and among the four markers, only TyG-BMI had a statistically signi cant predictive value for PTDM in the CsA subgroup. This nding was consistent with previous studies, among which Esteban et al. observed that the combination of FK506 plus pretransplant hypertriglyceridemia was a risk factor for PTDM, highlighting the importance of FK506 in the pathogenesis of PTDM [37]. Nevertheless, there was no obvious difference in the values of the four markers between the FK506 subgroup and CsA subgroup, indicating that the etiology of PTDM involved many elements and not just the type of CNI.
Our study still had several limitations that need to be considered. First, it was a retrospective study with a relatively small sample size, so confounding factors cannot be controlled well. There are few similar studies focusing on RT recipients, which may limit possible comparisons. Second, the HOMA-IR index was not calculated, so direct comparisons between the indices and the HOMA-IR index were missing in this study. Additionally, we did not analyze WC or WHtR. Measurement accuracy was a main limitation of WC or WHtR application [41], and interobserver variabilities of WC may range from 86% to 99% [42]. The interobserver variabilities of circumferences observed for WC were higher than those of BMI [43]. Moreover, WC alone cannot fully represent visceral adipose tissues due to its poor differentiation between subcutaneous and visceral adipose tissues [44]. That is, the advantages of obesity indices remain controversial, and further additional research is required. Third, our study cohort mostly comprised the South Zhejiang population in China which is not nationally representative and lacks geographical diversity. Therefore, further prospective and exhaustive studies are necessary to con rm the results we found.

Conclusions
In conclusion, predicted insulin resistance and its related metabolic risk factors play an essential role in the development of PTDM. TyG, TyG-BMI, TG/HDL-C and non-HDL-C/HDL-C were signi cantly associated with a higher incidence of PTDM in RT recipients, and TyG-BMI was the best alternative marker among the four markers for early identi cation of PTDM. In routine management of renal transplantation, TyG and TyG-BMI can be used as cost-effective and complementary monitors to identify individuals at high risk of PTDM.

Declarations Data Availability
All data generated or analyzed during this study are included in this published article and its supplementary information le.
Author contributions BC.C. participated in study design and coordination, made contributions to acquisition of data and helped to write the manuscript. XJ.N. participated in design of the study, performed the statistical analysis and interpretation and wrote the manuscript. YL.L., TT.H. and Y.Z. participated in the analysis and the acquisition of data and discussion. All authors contributed to discussion and critically reviewed and revised the manuscript. BC.C. is the guarantor of this work and takes responsibility for the integrity of the data and the accuracy of the data analysis.