| Background/AimsGlucocorticoids remain a valuable and necessary component of treatment for many diseases including chronic kidney disease (CKD). However, glucocorticoids have been shown to impair glucose metabolism, increase the risk of new-onset diabetes (steroid-induced diabetes), and thereby increase the potential for cardiovascular morbidity and mortality. Steroid-induced abnormal glucose metabolism (AGM), including diabetes and pre-diabetes (impaired glucose tolerance, IGT and impaired fasting plasma glucose, IFG), has been reported in 1.5-47% of patients depending on differences in the patient population, therapeutic protocols, and the definition of diagnosis. Identification of individuals who are at high risk for developing AGM is of great importance as it may help to recognize those for whom preventive approaches might be considered. However, although steroid-induced AGM has long been noted, the risk factors for AGM and their clinical impact have not been well characterized or quantified. Few published studies have focused on the predictive factors related to steroid-induced AGM, probably due to the use of retrospective or cross-sectional design in the majority of these studies. Oral administration of glucocorticoids, as an immunosuppressive therapy, is commonly used in the treatment of non-diabetic CKD. In the present study, we conducted a prospective, single-center, observational study in 303 non-diabetic CKD patients who did not have AGM (IFG, IGT or diabetes) before initiation of glucocorticoid therapy. The primary objective was to test the hypothesis that fasting glucose levels before the initiation of glucocorticoid therapy may help to identify individuals who are at an increased risk of steroid-induced AGM in patients with non-diabetic CKD.Methods1. DesignThis was a prospective cohort study conducted between September 2008 and September 2013 in Nanfang Hospital, an affiliated hospital of Southern Medical University. The study was approved by the Institutional Review Board of the National Clinical Research Center for Kidney Disease. All of the study subjects provided written informed consent.2. PatientsEligible subjects were non-diabetic CKD patients who were treated with glucocorticoids and (1) aged 18-70; (2) with fasting plasma glucose (FPG) levels<5.6 mmol/l (100 mg/dl); and (3) 2-hour plasma glucose levels in oral glucose tolerance test (OGTT)<7.8 mmol/l (140 mg/dl). Exclusion criteria included (1) pre-existing AGM (diabetes, IGT or IFG) or using hypoglycemic drugs; (2) use of drugs that might cause hyperglycemia such as calcineurin inhibitors, hydrochlorothiazide, and protease inhibitors; (3) body-mass index (BMI)≥ 28 kg/m2 or metabolic syndrome defined as International Diabetes Federation Consensus Worldwide,2005; (4) severe hepatopathy or cirrhosis. Patients who had a history of glucocorticoid use were also excluded from the study.3. ProceduresAll eligible patients were treated with an oral administration of glucocorticoid according to the recommendations for their given disease. Data collection and OGTT were performed before glucocorticoid therapy (baseline) and every 3 months during glucocorticoid treatment at the out-patient visits. In patients who developed AGM (diabetes, IGT or IFG) during follow-up, a repeat test was conducted within 2 weeks to confirm the diagnosis. In patients with established AGM, the glucocorticoid doses were reduced and withdrawn within 3 months. Diet control and hypoglycemic drugs were given as necessary. All patients (including those who completed and had early withdrawal of the therapy) were continuously followed up for 12 months after glucocorticoid withdrawal.4. Laboratory MethodOral Glucose Tolerance Test (OGTT)After at least 12 h of overnight fasting, a standard oral glucose test using 75 g glucose was conducted according to the guidelines. Venous blood samples were drawn at 0-hour and 2-hours after the glucose load and collected in a vacuum tube containing sodium fluoride for measurement of plasma glucose concentrations. All tests were performed between 8 and 11 o’clock in the morning. Patients were asked to fast and postpone their morning medication until the test was completed. Plasma glucose concentrations were measured from fresh samples (within 2 h) by the hexokinase enzymatic method using an automated analyzer (Olympus Au 480, Japan).Other Biochemical ParametersSerum total cholesterol, triglycerides, high-density lipoprotein-cholesterol, and creatinine levels were assessed enzymatically with commercially available reagents using an automated analyzer (Olympus Au 480, Japan). Serum albumin levels were determined by the bromo-cresol green method. Serum cystatin-C levels were measured by the particle-enhanced turbidimetric immunoassay method. Urinary protein excretion was measured with the Biuret method using 24-hour urine samples. Estimated GFR (eGFR) was determined by CKD-EPI equation based on serum creatinine and cystatin C.All biochemical measurements were performed in the Clinical Laboratory of Nanfang Hospital, which is authorized to perform the tests according to the international standard ISO-15189.5. Outcomes DefinitionThe primary outcome was development of AGM (diabetes, IGT, or IFG) during glucocorticoid therapy. AGM was diagnosedaccording to the criteria published by the American Diabetes Association expert committee. Diabetes was defined as 2-hour plasma glucose≥ 11.lmmol/l (200mg/dl) during an OGTT or FPG> 7.0mmol/l (126 mg/dl). IGT was defined as a 2-hour plasma glucose level of 7.8mmol/l (140mg/dl) to 11.0mmol/L (199mg/dl) during an OGTT. IFG was defined as FPG> 5.6mmol/l (100 mg/dl) but < 6.9mmol/l (125 mg/dl).The secondary outcome was the recovery rates of steroid-induced AGM for 12 months after steroid withdrawal. Recovery was defined as hyperglycemia, IFG, or IGT restored to normal without the requirement of any hypoglycemic drug.6. Statistical AnalysisIn the univariate analysis, we used a two-sample t-test or Mann-Whitney rank sum test to compare the continuous variables, and the Pearson X2 test to compare the categorical variables. A multivariable Cox regression model was used to determine the risk factors for AGM. In the univariate analysis, we included all of known or possible risk factors, such as age, hypertension, family history of diabetes, BMI, FPG, serum lipids, and eGFR.Variables with a p< 0.05 in the univariate analysis were then entered into the stepwise multivariate model. We also conducted a Cox proportional-hazards analysis during follow-up to estimate the HR and 95% CI for development of AGM or diabetes across FPG quartiles or △FPG tertiles. Interaction terms were analyzed by modeling the quartile as continuous variables. We evaluated the joint risk attributed to FPG levels (categorized according to the quartiles) with either age or triglyceride levels. To test the predictive sensitivity and specificity of baseline FPG or the combination of baseline FPG and AFPG at different cutoff values, a receiver-operating characteristic (ROC) curve was generated and the area under the curve (AUC) was analyzed for steroid-induced AGM or diabetes. Optimal cutoffs were determined by selecting the data point that minimized the geometric distance from 100% sensitivity and 100% specificity on the ROC curve. Kaplan-Meier plots were built to assess the cumulative probability of AGM or diabetes according to the category of baseline FPG or △FPG at month 3. All of the analyses were conducted using the statistical package SPSS software, version 18.0. All tests were two-tailed and p< 0.05 was considered significant.ResultsNon-diabetic patients, totaling to 379, were screened for potential participation. A total of 69 were excluded according to the exclusion criteria and 7 were lost during follow-up.During 593 person-years, there were 107 incident cases of steroid-induced AGM (18/100 person-year), including 55 (51.4%) diabetes and 52 (48.6%) pre-diabetes. Compared to those who did not develop AGM, patients with steroid-induced AGM were older, more frequent to have a family history of diabetes, and with higher levels of FPG, BMI, and serum triglycerides, and lower levels of eGFR.In a multivariate model, each millimole increase per liter in FPG enhanced the risk of AGM by 4.6-fold (hazard ratio 4.58,95% confidence interval,2.67-7.83). Other risk factors included triglycerides, age, and changes of 3 months’ follow-up in FPG (△FPG). After adjusting other risk factors, a progressively increased risk of AGM or DM was observed in patients with FPG levels≧4.8mmol/L, as compared with those whose levels were(?) 4.3mmol/L (P for trend<0.001). The adjusted risk of steroid-induced AGM or diabetes significantly increased across tertiles of △FPG at month 3. A △FPG level of ≥ 0.3 mmol/1 (5.4 mg/dl) was associated with a 2.8-fold increased risk for AGM and a 10.7-fold for steroid-induced diabetes even after multivariate adjustment, as compared with those whose levels were<0 mmol/L (P for trend<0.001).When we cross-classified age with baseline FPG levels, we found that older patients (≧ 35years) with baseline FPG in the higher normal range (5.2～5.5mmol/L [93～99 mg/dl]) had an HR of 16.0 (95%CI 7.1～36.1) for the development of steroid-induced AGM, as compared with the reference group. Patients with baseline FPG levels in the higher normal range (5.2～5.5mmol/L [93-99mg/dl]) and serum triglyceride levels≧ 1.7mmol/L had an HR of 8.7 (95%CI,3.6～21.3) for the development of steroid-induced AGM as compared to those with FPG levels< 4.8mmol/L (86mg/dl) and serum triglyceride levels< 1.7mmol/L,For predicting steroid-induced diabetes, we generated an ROC, for which the AUC of baseline FPG for all participants was 0.81. An optimal cutoff of 4.8 mmol/1 (86mg/dl) yielded both good sensitivity (0.82) and specificity (0.70). Furthermore, a greater increase in FPG level (≧ 0.3mmol/L) during the first 3 months of glucocorticoids treatment was associated with an increased risk for future diabetes. For predicting steroid-induced diabetes, the area under the receiver-operating characteristic curve was 0.90 for the combination of FPG and changes in FPG levels at month 3. A baseline FPG≧ the cutoff (4.8mmol/L) was associated with a significantly increased probability of steroid-induced diabetes (HR,7.8,95%CI, 3.1～8.9). An increase in the FPG≧0.3mmol/L (5.4 mg/dl) at month 3 was also associated with an increased cumulative probability of developing diabetes (HR,2.8, 95%CI,1.5～5.0).ConclusionHigher-normal FPG and a greater increase in FPG levels during glucocorticoid treatment may help to identify non-diabetic CKD patients at increased risk of steroid-induced AGM or diabetes. |
PDF Full Text Request