Fasting Glucose Levels in Predicting 1-Year All-Cause Mortality in Patients Who Do Not Have Diabetes and Are on Maintenance Hemodialysis

Characteristics of the Study Population

A total of 693 MHD patients (329 men and 364 women) with mean MHD duration of 5.7 ± 0.2 yr were enrolled for analysis. Table 1 lists the baseline clinical characteristics, including age, gender, body mass index (BMI), and biologic and hematologic data. Mean patient age was 55.5 ± 0.5 yr (range 14 to 92 yr). The numbers of patients with renal diseases were as follows: Diabetes 189 (27.3%), chronic glomerulonephritis 226 (32.6%), and hypertensive-related nephropathy 121 (17.5%). Table 1 also lists patient characteristics for the three subgroups. The group with diabetes (n = 189) had lower values of Kt/V and intact parathyroid hormone (iPTH), shorter duration of HD, higher prevalence of CVD, and higher BMI and white blood cell (WBC) counts than the other groups. Meanwhile, the normal FGL group (n = 423) were younger; had higher levels of serum albumin, creatinine, and normalized protein catabolism rate (nPCR); and had lower values of triglycerides, high-sensitivity C-reactive protein (hsCRP), and cardiothoracic ratio (CTR) than the other groups. Furthermore, the IFG group (n = 81) had higher levels of triglycerides, hsCRP, and ferritin and lower levels of serum albumin, creatinine, and nPCR than the normal FGL group. The groups did not differ in terms of gender; smoking status; presence of hypertension or viral hepatitis; values of Ca × P; and use of fistula, biocompatible membrane dialyzers, statins, and/or aspirin (data not shown). The IFG and diabetes group patients had significantly higher percentages of malnutrition (χ² = 24.55, P < 0.0001) and inflammatory status (χ² = 9.32, P = 0.0095) than the normal FGL group patients (Table 2).

Significant Correlations among FGL, Baseline Data, Nutrition, and Inflammation Markers in Patients Who Did Not Have Diabetes and Were on MHD

The Spearman correlation analysis demonstrated that FGL was significantly and positively correlated with age and inflammatory markers, such as hsCRP and ferritin levels, but negatively correlated with nutritional parameters, such as albumin, creatinine, and hemoglobin levels in nondiabetic MHD patients (Table 3).

Significant Relations between Serum Albumin and FGL in Patients Who Did Not Have Diabetes and Were on MHD

The variables that were considered as potential covariates of serum albumin^7,9–11 were age, gender, BMI, HD years, smoking status, CVD, viral hepatitis B and C, use of aspirin or statins, use of antihypertension drugs, using fistula for vascular access, biocompatible membrane dialyzers, Ca × P, iPTH, Kt/V_urea (Daugirdas), and log hsCRP. Simple linear regression analysis indicated that age, gender, smoking, BMI, previous CVD, use of a fistula, log hsCRP, CTR, and FGL were associated with serum albumin in the study patients (Table 4). After adjustment for these significant variables, stepwise multiple linear regression analysis revealed a significant inverse correlation between serum albumin levels and FGL (Table 4). A 1-mg/dl increase in FGL was associated with a 0.003-g/dl decrease in serum albumin level (P = 0.0026).

Significant Relationship between Log hsCRP and FGL in Patients Who Did Not Have Diabetes and Were on MHD

The variables^7,9–11 that were considered as potential covariates of inflammation were age, gender, BMI, HD years, smoking status, CVD, viral hepatitis B and C, aspirin or statins drugs, antihypertension drugs, using a fistula for vascular access, biocompatible membrane dialyzers, Ca × P, iPTH, and Kt/V_urea (Daugirdas). Simple linear regression analysis indicated that age, smoking, BMI, previous CVD, Kt/V_urea (Daugirdas), CTR, and FGL were associated with log hsCRP in these patients (Table 5). After adjustment for these significant variables, stepwise multiple linear regression analysis identified a significant correlation between log hsCRP levels and FGL (P = 0.0004; Table 5).

Cox Regression Multivariate Analysis for 1-Yr Mortality in MHD Patients

At the end of the 12-mo observation period, the following data were obtained: 23 (12.2%) patients in the diabetic group, eight (9.9%) patients in the IFG group, and 10 (2.4%) patients in the normal FGL group died during the 1-yr follow-up. None of IFG group with sugar <110 died. A total of 606 patients completed the 1-yr follow-up. During this period, 34 patients were transferred to other outpatient hemodialysis units, 12 patients received kidney transplantation, and 41 patients died. Among these patients, 35 (72.9%) died of CVD, five (12.2%) died of infection, and one died of hepatoma. Cox multivariate regression analysis revealed that—after adjustment for other significant related factors—presence of abnormal CTR (>50%) and FGL were significant risk factors for 1-yr mortality of patients who did not have diabetes and were on MHD (Table 6). Similarly, instead of FGL, presence of IFG is a significant risk factor (relative risk 3.798, 95% confidence interval 1.168 to 12.344; P = 0.0265) for 1-yr mortality in these patients after adjustment of related significant factors. The Kaplan-Meier survival analysis for all MHD patients showed that diabetic and nondiabetic IFG group patients had higher mortality than those in the normal FGL group (Log rank test, χ² = 11.48, P = 0.0007; Figure 1). At the study end, four (4.9%) of 81 patients with IFG had developed diabetes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

The Kaplan-Meier survival analysis demonstrated that diabetic and nondiabetic IFG group patients had a higher 1-yr mortality rate than that of the nondiabetic normal FGL group (Log rank test, χ² = 11.48, P = 0.0007).

Patients

All study patients came from three HD centers: Chang Gung Memorial Hospitals (CGMH) in Taipei, Lin-Kou, and Toayuan. All MHD patients were enrolled, after exclusion of those with malignancies and obvious infectious diseases, as well as those who were hospitalized or underwent surgery or renal transplantation during the 3 mo before the study. Patients who were currently using FG-elevating drugs, such as postmenopausal replacement hormones, steroids, thyroid hormone; who could not tolerate fasting time; or who had been receiving regular HD for <6 mo were also excluded. This longitudinal observational study enrolled a total of 693 patients. Most HD patients had been undergoing 4 h of HD three times per week. HD for these patients used single-use hollow-fiber dialyzers equipped with modified cellulose-based polyamide or polysulfone membranes. The dialysate used was a standard ionic composition, and a bicarbonate-based buffer was used in all cases. Presence of diabetes and CVD, including cerebrovascular disease, coronary arterial disease, congestive heart failure, and peripheral vascular disease, were recorded. Smoking behavior and use of drugs that effect inflammatory properties, such as statins and aspirin, were also recorded. This clinical study followed the Declaration of Helsinki and was approved by the Medical Ethics Committee of CGMH, Taipei.

Study Groups

Patients who met the inclusion criteria were classified into three groups according to their FGL, which were checked at least three times at 1-mo intervals. All values had to be within the classification of the IFG and normal FGL ranges, based on the 1997 and 2003 definitions.¹² The three patient classification groups were as follows: The nondiabetic with normal FGL group (fasting glucose <105 mg/dl; n = 423), the nondiabetic with IFG group (fasting glucose <126 and ≥105 mg/dl; n = 81), and the diabetic group (twice subsequent fasting glucose ≥126 mg/dl or diabetes was diagnosed by a physician in the history; n = 189). All patients were followed up for 1 yr, and mortality during this period was the primary end point.

Laboratory, Nutritional, and Inflammatory Parameters

Laboratory data for each patient were obtained within a few days of the clinical examination during stable outpatient HD sessions to minimize any effects of acute events. Blood samples were drawn from the arterial end of the vascular access immediately before initiation of a 2-d interval HD, centrifuged, and stored at −70°C until they were used in assays.

Serum albumin, creatinine levels, transferrin saturation, nPCR, triglycerides, and total cholesterol were assayed and recorded as nutritional markers. Moreover, WBC and hemoglobin were measured as the hematologic indicators. This study used both serum ferritin and hsCRP as inflammation markers. Serum hsCRP concentrations were measured using immunonephelometry (Nanopia CRP; Daiichi, Tokyo. Japan). The lowest detection limit was <0.15 mg/L. All other markers were measured using standard laboratory approaches with an automatic analyzer. nPCR in HD patients was calculated via validated equations and normalized to actual body weight.²⁵ Dialysis clearance of urea was expressed as Kt/V_urea, using the method described by Daugirdas²⁶ in HD patients. Serum calcium, phosphate, and iPTH were also detected. Corrected calcium = serum calcium (mg/dl) + [0.8(4.0 − serum albumin [g/dl])].

Definition of Malnutrition and Inflammation

To determine whether FGL reflects the inflammation and/or malnutrition status of MHD patients, this work investigated serum albumin and hsCRP levels in different population subgroups on the basis of the absence or presence of malnutrition and inflammation. The presence of inflammation in MHD patients was defined as an hsCRP level >3 mg/dl, a level that is correlated with raised cardiovascular risk in the general population.²⁷ This work defined albumin levels <3.6 g/dl (<36 g/L) as a state of malnutrition, levels that represent the 10th percentile of the Third National Health and Nutrition Examination Survey.^28,29

Statistical Analyses

Unless otherwise stated, continuous variables are expressed as means ± SEM, and categorical variables are expressed as numbers or percentages for each item. Comparisons among the three study groups of patients were analyzed via one-way ANOVA using Bonferroni test for post hoc analysis. The Kruskal-Wallis test and Mann-Whitney U test were used to detect significant differences among non-normally distributed variables, and logarithmic conversion was conducted for hsCRP levels. The χ² test was used for categorical variables, including gender, smoking status, CVD, and the presence of malnutrition and inflammation. Moreover, Spearman correlation analysis was conducted to assess the relations between variables. To identify factors that were associated with inflammation and nutrition, this investigation used multiple linear regression models with backward stepwise procedures for biochemical and demographic variables,^7,9–11 such as age, gender, BMI, HD years, smoking status, CVD, viral hepatitis B and C, use of aspirin or statins, use of antihypertension drugs, using fistula for vascular access, biocompatible membrane dialyzers, Ca × P, iPTH, and Kt/V_urea (Daugirdas). The Cox proportional hazards model was applied to determine the significance of variables for predicting 1-yr mortality (the primary end point). This model included all of the variables that were identified in the literature^7,9–11 as related to HD mortality. The level of significance was set at P < 0.05. All statistical calculations were performed using StatView 2.0 for Windows (SAS Institute, Cary, NC).