Glycated Hemoglobin and Outcomes in Patients with Advanced Diabetic Chronic Kidney Disease

Diabetes is the major risk factor for end-stage renal disease (ESRD) worldwide. In advanced chronic kidney disease (CKD), less is known about the predictive value of HbA1c. We enrolled 2401 diabetic patients with stage 3–4 and stage 5 CKD, who were classified into 4 groups according to their baseline HbA1c values (<6%, 6%–7%, 7%–9%, and >9%). During the median follow-up of 3 years, 895 patients developed ESRD, and 530 died. In linear regression analysis, higher HbA1c correlated with higher eGFR in patients with stage 5 CKD but not in stage 3–4 CKD. In Cox regression analysis, a trend toward worse clinical outcomes existed when the HbA1c level exceeded 6% in stage 3–4 CKD, but the significance was only observed for >9%. The hazard ratios (HRs) for ESRD, all-cause mortality and combined CV events with mortality in the group of HbA1c >9% were 1.6 (95% CI, 1.07 to 2.38), 1.52 (95% CI, 0.97 to 2.38) and 1.46 (95% CI, 1.02 to 2.09), respectively. This study demonstrates that the higher HbA1c level is associated higher risks for clinical outcomes in diabetic patients with stage 3–4 CKD but not in stage 5 CKD.

the risks of ESRD and mortality in the advanced stages of diabetic CKD and tested whether different CKD stages affected these relationships.

Methods
Participants and Measurements. This was an observational study that enrolled patients with CKD who were treated as part of the integrated or traditional care program of 2 affiliated hospitals of Kaohsiung Medical University in Southern Taiwan. The study was conducted from November 11, 2002 to May 31, 2009, with follow-up until May 31, 2010. We excluded patients who had a record of acute kidney injury, defined as a more than 50% decrease in the eGFR within 3 months, or had received chronic renal replacement therapy (RRT) before their first visit. The cohort comprised 4824 patients, and we selected 2401 patients with stage 3-5 CKD and type 2 DM as defined by the World Health Organization for this study 10 . CKD stages were defined as follows: stage 3, eGFR of 30 to 59 mL/min/1.73 m2; stage 4, eGFR of 15 to 29 mL/min/1.73 m2; and stage 5, eGFR less than 15 mL/ min/1.73 m2 based on staging criteria from the National Kidney Foundation Kidney Disease Outcomes Quality Initiative (KDOQI) 11 . All participants were followed at clinic visits periodically for routine biochemical blood exams and evaluation of CKD complications. The institutional review board of Kaohsiung Medical University Hospital approved the study protocol, and informed consent was obtained from all participants. The methods were carried out in accordance with the Declaration of Helsinki ethical principles for medical research.
Participant demographic information was gathered upon their first visit, and their medical histories were obtained using a chart review. Their baseline biochemical data and comorbidities were analyzed. The eGFR of the participants was calculated using the simplified modification of diet in renal disease (MDRD) study equation: eGFR mL s-1 [1.73m2]-1 = 186 × serum creatinine − 1.154 × age − 0.203 × 0.742 (if female) × 1.212 (if black). In Taiwan, the MDRD formula was applied in the Taiwan National Database to evaluate CKD prevalence and dialysis initiation 12,13 . Therefore, we chose MDRD formula over CKD-EPI (Epidemiology Collaboration) as our study equation. The HbA1c value was measured as clinically indicated by the hospital laboratory using automated cation-exchange high-performance liquid chromatography. There was no substantial change to the HgbA1c measurement methodology during the study.
The patients were classified into 4 groups according to the following thresholds, which were selected according to guidelines and clinical trials, depending on their first HbA1c measurement: < 6%, 6%-7%, 7%-9%, and > 9%. The participants were diagnosed with hypertension if their office blood pressure was > 140/90 mmHg or if they took any antihypertensive medications. Cardiovascular (CV) diseases were defined as clinically diagnosed myocardial infarction, heart failure, ischemic heart disease, and cerebrovascular disease. The first measurements of laboratory data, namely the urine protein-to-creatinine ratio (UPCR), albumin, hemoglobin, blood glucose, HbA1c, total calcium, phosphate, uric acid, total cholesterol, triglyceride, and C-reactive protein (CRP), were used as baseline variables.
Outcomes. The primary outcomes of this study were ESRD, CV events, and all-cause mortality. ESRD was defined as the initiation of hemodialysis, peritoneal dialysis, or renal transplantation. The development of ESRD was ascertained using catastrophic illness cards issued by the Bureau of National Health Insurance. CV events were confirmed by examining records of hospitalization with the responsible diagnosis for acute coronary syndrome (International Classification of Diseases, Ninth Revision, Clinical Modification: 410.x-412.x), acute cerebrovascular disease (430.x-438.x), and congestive heart failure (428.x) or death from the aforementioned causes, but only in patients with CV event occurrence after the index date. All-cause mortality was determined using death certificates and the National Death Index. Hypoglycemia was defined as the presence of typical symptoms and signs of hypoglycemia requiring medical assistance. Statistical Analysis. Descriptive statistics were expressed as counts and percentages for the categorical data, and means with standard deviations or medians with interquartile ranges were determined for continuous variables with approximately normal distributions. Differences in baseline characteristics between groups were analyzed using ANOVA tests for continuous variables and chi-squared tests for categorical variables. For skewed distributions of some continuous variables, we applied logarithmic transformation to make data conform more closely to the normal distribution (cholesterol and CRP). We used multivariate linear regression analysis to investigate possible individual related variables, with the HbA1c level as the dependent variable. To determine the relative associations between the baseline HbA1c level and clinical outcomes, a Cox multivariate regression model was employed and adjusted for age, sex, the eGFR, the log-transformed UPCR, CV disease, mean blood pressure, hemoglobin, albumin, log-transformed cholesterol, log-transformed CRP, phosphorus, and the body mass index. Covariates were selected on the basis of their significance in statistics or on the basis of clinical relevance 14 . Furthermore, we performed subsequent subgroup analysis to observe the relationship of HbA1c with clinical outcomes among these demographic, clinical, and laboratory categories. A P value of < 0.05 was considered statistically significant. All analyses were conducted using R 2.15.2 software (R Foundation for Statistical Computing, Vienna, Austria) and the Statistical Package for Social Sciences for Windows Version 18.0 (SPSS Inc., Chicago, IL).
Over the approximately 3-year median follow-up period, 312 (20.0%) and 583 (69.1%) cases of RRT were identified among the patients with stages 3-4 and stage 5 CKD, respectively. The patients with higher HbA1c      Multivariate Linear Regression for HbA1c. Table 2 shows the results of multivariate linear regression for HbA1c level. In all participants, age, the eGFR, the log UPCR, albumin, hemoglobin, log cholesterol, and log CRP were associated with the baseline HbA1c level. In the subgroup analysis, higher HbA1c level correlated with higher eGFR in patients with stage 5 CKD (95% confidence interval [95% CI], 0.031 to 0.108, P < 0.001) but not in patients with stage 3-4 CKD (95% CI, −0.005 to 0.012, P = 0.451) ( Table 2 and Fig. 1). By contrast, a significant correlation existed between a higher HbA1c level and a poorer UPCR in patients with stage 3-4 CKD, but the correlation was nonsignificant in patients with stage 5 CKD (Table 2 and Fig. 1).

HbA1c and Clinical Outcome Associations.
In patients with stage 3-4 CKD (  (Table 3), no relationship between the HbA1c levels and clinical outcomes was observed. In addition, a HbA1c level of 7%-9% was associated with a higher likelihood of RRT, with a HR of 1.28 (95% CI, 1.01 to 1.63, P = 0.04), whereas the risk of RRT was lower for a HbA1c level > 9%, with an HR of 1.01 (95% CI, 0.70 to 1.45, P = 0.97), compared with a level < 6%. Similarly, higher all-cause mortality was observed for a HbA1c level of 7%-9%, but this rate was lower for a HbA1c level > 9%. Figure 2 presents subgroup analysis results regarding the adjusted risks of RRT. Both the eGFR and hemoglobin affected the association between the HbA1c level and RRT (both P for interaction = 0.001); a HbA1c level > 9% was associated with a higher risk (HR, 1.44; 95% CI, 0.93 to 2.25) of RRT in patients with hemoglobin > 10 mg/dl whereas HbA1c > 9% was not associated with a higher risk (HR, 0.85; 95% CI, 0.59 to 1.22) in patients with hemoglobin < 10 mg/dl.

Discussion
In the diabetic patients with stage 3-4 CKD, our study identified that the baseline HbA1c > 9% is correlated with higher risks for multiple relevant outcomes, including RRT and combined CV events with all-cause mortality. By contrast, the relationship between HbA1c and clinical outcomes was not significant for those with stage 5 CKD. Accordingly, these findings demonstrate that the predictive value of HbA1c level is stronger at earlier CKD stages.
Glycemic control has been clarified in previous studies as being related to microvascular complications. Chronic hyperglycemia promotes advanced glycation end-product formation, which can increase growth factor production and consequently contribute to extracellular protein deposition, mesangial expansion, gradual glomerular scelrosis, thereby reducing GFR 15 . In a 3-year cohort study, diabetic participants with preserved renal function whose baseline HbA1c exceeded 6% exhibited accelerated eGFR decline 16 . Recently, a meta-analysis of 7 randomized controlled trials (RCTs) of intensive glycemic therapy, defined by lower HbA1c, versus the standard regimen for type 2 DM, reported a significant reduction in microalbuminuria and macroalbuminuria occurrences, but the benefits were inconclusive concerning the effect on clinical renal outcomes, defined by the doubling of the SCr level or ESRD 17 . Little evidence is available regarding the relationship between HbA1c levels and clinical outcomes in patients with advanced CKD. Se Won Oh et al. enrolled a 5-year cohort of 799 patients with DM and an eGFR < 60 ml/ min/1.73m 2 and reported that patients with a baseline HbA1c of < 6.5% had reduced a risk for ESRD by comparing those with a HbA1c of > 6.5% 18 . In people requiring chronic hemodialysis, Oomichi T et al. found poor glycemic control is an independent predictor of survival from an observational study in which 114 diabetic CKD patients were enrolled 19 . However, others reported that the CKD stages could influence the association between HbA1c and renal outcomes. A population-based cohort study on patients with DM and stage 3-4 CKD revealed that a baseline HbA1c higher than 7% was strongly associated with an increased risk of ESRD. Moreover, the magnitude of increased risk with higher HbA1c levels seemed attenuate in patients with stage 4 CKD compared with patients with stage 3 CKD 8 . Considering the rate of eGFR decline, a recent cohort study in Taiwan demonstrated that for patients with higher preceding HbA1c levels, the negative effects on eGFR deterioration were stronger at stage 3-4 CKD than stage 1-2 or stage 5, but the outcomes of ESRD were not reported 20 . Our results were consistent with these data. We observed a HbA1c > 9%, comparing with a HbA1c < 6%, was associated with an increased risk for ESRD in the stage 3-4 CKD group. Conversely, a corresponding trend was not observed in patients with stage 5 CKD. Our study is the first to recruit a large-scale sample of patients with stage 5 CKD and,

HR (95% CI) for RRT
The prognostic role of HbA1c in patients with stage 5 CKD was unclear because impaired glucose metabolism in advanced CKD, and the HbA1c level may be altered by anemia, or erythropoiesis-stimulating agent use. First, it is well-known that a marked reduction in insulin clearance can occur until the GFR falls to less than 15-20 ml/ min 21 . Agarwal et al. demonstrated that glycemic control, as assessed by random blood glucose, improved in patients with late-stage CKD 22 . Our data found that a lower HbA1c level was correlated with a lower eGFR in patients with stage 5 CKD but not in those with stage 3-4 CKD, which could be partly explained by hyperinsulinemia affecting the HbA1c level in stage 5 CKD. Second, glycated hemoglobin formation is reduced in patients with CKD because the fragile red blood cell (RBC) has shortened lifespan by 30%-70%, and carbamylated hemoglobin molecules in the uremic environment become resistant to glycosylation 23,24 . Administering erythropoietin stimulating agents (ESAs) to patients with anemia also augments, in peripheral blood, the proportion of young RBCs, which have a lower rate of glycosylation than do old RBCs, thereby altering glycosylated hemoglobin formation 25 . The results of some studies support this notion. Agarwal et al. reported that, among 128 patients with DM and stage 1-5 CKD, a decline in HbA1c was correlated with CKD stages, but this relationship disappeared after adjustment for hemoglobin 22 . In addition, Freedman et al. confirmed, in diabetic patients with stage 3-4 CKD, an inverse correlation between the eGFR and the glucose/HbA1c ratio, which indicated that HbA1c could be falsely low in lower eGFR 26 . Accordingly, HbA1c levels appear to be falsely low in subjects with DM and advanced CKD [27][28][29] . Our data again confirm the positive correlation between HbA1c and Hb in stage 3-5 CKD, but the positive correlation between HbA1c and eGFR only exists in stage 5 CKD. HbA1c level may not accurately indicate glycemic control during the deterioration of kidney function, and based on our study it is less prognostic in stage 5 CKD.
Measuring HbA1c levels earlier might increase their prognostic value. The term "legacy effect" has been used in some RCTs to describe the ongoing benefits of better glycemic control even after intervention ceases. A proposed mechanism for this is that fewer advanced glycation end products confer long-term protection. In the Epidemiology of Diabetes Interventions and Complications trial, participants who were originally assigned to an intensive control group continued exhibiting benefits in sustainable reductions of macroalbuminuria and renal function impairment during the 8-year post-trial period, even though the average HbA1c levels of the intensive group and conventional groups lost their difference (8.0% and 8.2%) [30][31][32] . This phenomenon implies that HbA1c measurement in late CKD stages may be too late to determine their long term glycemic control. Our observational study identified no increase in the risks of ESRD and the composite endpoint of CV events and all-cause mortality in stage 5 CKD, even when the baseline HbA1c level was > 9%. In other words, the HbAlc level is more useful in stage 3-4 CKD than it is in stage 5 CKD in predicting clinical prognosis, either because of multiple factors influencing HbA1c production in stage 5 CKD or the possible legacy effect.
Regarding patients with CKD, attempting to control HbA1c levels as low as possible is controversial if we take safety into account because doing so increases the risk of hypoglycemia. This is on account of prolonged half-life of antidiabetic drugs and reduced renal insulin clearance, degradation of insulin in peripheral tissues, glycogen stores, and renal gluconeogenesis 33 . Furthermore, the CV mechanism of hypoglycemia exists in sympathoadrenal stimulation and the inflammation reaction, which may bring about endothelial dysfunction and coagulation abnormalities, eventually causing QT prolongation, cardiac arrhythmia, and eventual CV events 34 . Several RCTs have attempted to lower HbA1c levels aggressively in diabetic patients with preserved kidney function, which illustrated the inconsistent findings regarding CV benefits and mortality, and the rates of hypoglycemia were greater in the intensive therapy groups 4,5,35 . In addition, one meta-analysis of 5 RCTs involving patients with type 2 DM also suggested that intensive glucose therapy reduced nonfatal myocardial infarction and coronary heart disease but not all-cause mortality 7 . These data told that the adverse sequelae of hypoglycemia might partly cause the inconsistent CV and mortality outcomes. Moreover, a large-scale UK observational study reported a general U-shaped association of the mean HbA1c level with all-cause mortality and CV events, with the HbA1c threshold at approximately 7.5% and higher or lower levels related to increased risks 36 . Nevertheless, our cohort did not demonstrate a U-shape association, which seems not in concordance with prior studies. This discrepancy could be interpreted by the low incidence of hypoglycemia in our patients, who were provided with behavioral instruction to prevent hypoglycemia. Furthermore, patients with low HbA1c level in an observational study do not parallel to those in a clinical trial receiving strict glycemic control. These patients with low HbA1c level in our study group were not forced to receive intensive intervention and may have less severe diabetes. Thus, hypoglycemia and CV disadvantages did not emerge.
There are several limitations to this study. First, we relied on the baseline HbA1c rather than the mean HbA1c level. HbA1c level at one point could not reflect the actual glucose control during the follow-up period. However, because of the gradual decline of HbA1c as CKD stage progresses and the possible legacy effect, we considered baseline HbA1c could serve a reasonable indicator when CKD stage was classified at the same time. Second, we did not measure serial blood glucose levels which may represent actual glucose control better than the HbA1c values, and thus we could not analyze the relationship between the CKD stages with variation of blood glucose. Third, it is our limitation to use immunoassays in the hospital as HbA1c testing methodology since immunological methods for the detection of HbA1c is more reliable in uremic environment. Fourth, we did not have data on medications that the patients used to control their DM, such as oral hypoglycemic agents and insulin. In addition, we did not collect the information on the ESAs doses or other medications that may alter RBC production; some studies have found that the ESA dose is inversely related to HbA1c level 37 . Therefore, we could not adjust for these potential confounders. Fourth, this was a cohort study that cannot evaluate the clinical effects of intensive intervention on glucose control in patients with advanced CKD.
In conclusion, an HbA1c > 9% predicts an increased risk for ESRD, and composite outcome of CV event and mortality in patients with stage 3-4 CKD. By contrast, in stage 5 CKD, predictive value of HbA1c level is weaker.