Association of HbA1C Variability and Renal Progression in Patients with Type 2 Diabetes with Chronic Kidney Disease Stages 3–4

Little is known about the predictive value of glycosylated hemoglobin (HbA1C) variability in patients with advanced chronic kidney disease (CKD). The aim of this study was to investigate whether HbA1C variability is associated with progression to end-stage renal disease in diabetic patients with stages 3–5 CKD, and whether different stages of CKD affect these associations. Three hundred and eighty-eight patients with diabetes and stages 3–5 CKD were enrolled in this longitudinal study. Intra-individual HbA1C variability was defined as the standard deviation (SD) of HbA1C, and the renal endpoint was defined as commencing dialysis. The results indicated that, during a median follow-up period of 3.5 years, 108 patients started dialysis. Adjusted Cox analysis showed an association between the highest tertile of HbA1C SD (tertile 3 vs. tertile 1) and a lower risk of the renal endpoint (hazard ratio = 0.175; 95% confidence interval = 0.059–0.518; p = 0.002) in the patients with an HbA1C level ≥ 7% and stages 3–4 CKD, but not in stage 5 CKD. Further subgroup analysis showed that the highest two tertiles of HbA1C SD were associated with a lower risk of the renal endpoint in the group with a decreasing trend of HbA1C. Our results demonstrated that greater HbA1C variability and a decreasing trend of HbA1C, which may be related to intensive diabetes control, was associated with a lower risk of progression to dialysis in the patients with stages 3–4 CKD and poor glycemic control (HbA1c ≥ 7%).


Introduction
Diabetes mellitus (DM) is the leading cause of chronic kidney disease (CKD) worldwide including Taiwan, where it has been reported to account for approximately 45% of all cases of end-stage renal disease (ESRD) in patients undergoing dialysis. Glucose control has been reported to be an important factor in controlling diabetic nephropathy, and time-averaged mean levels of glycemia, as assessed by glycosylated hemoglobin (HbA 1C ) level, is the gold standard of treatment to control glycemia and reduce the complications associated with diabetes [1,2]. Current guidelines suggest a target level of The median follow-up period was 3.5 (0.5-9.3) years. Of the 388 patients, 108 developed ESRD during the follow-up period, including 105 with hemodialysis and three with peritoneal dialysis. The patients in tertile 3 (vs. tertile 1) were associated with a lower risk of the renal endpoint in the unadjusted model (hazard ratio (HR), 0.493; 95% confidence interval (CI), 0.305-0.796; p = 0.004). Figure 1 illustrates the Kaplan-Meier curves for dialysis-free survival (log-rank p = 0.012) for all of the patients subdivided according to tertiles of HbA 1C SD. The patients in tertile 3 had a better dialysis-free survival than those in tertile 1. Abbreviations. SD, standard deviation; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; ACEI, angiotensin converting enzyme inhibitor; ARB, angiotensin II receptor blocker.
The median follow-up period was 3.5 (0.5-9.3) years. Of the 388 patients, 108 developed ESRD during the follow-up period, including 105 with hemodialysis and three with peritoneal dialysis. The patients in tertile 3 (vs. tertile 1) were associated with a lower risk of the renal endpoint in the unadjusted model (hazard ratio (HR), 0.493; 95% confidence interval (CI), 0.305-0.796; p = 0.004). Figure 1 illustrates the Kaplan-Meier curves for dialysis-free survival (log-rank p = 0.012) for all of the patients subdivided according to tertiles of HbA1C SD. The patients in tertile 3 had a better dialysis-free survival than those in tertile 1.  Table 2 shows the HRs of HbA1C SD tertiles for progression to dialysis using the univariate Cox proportional hazards model among different subgroups. The subjects were further divided into two  Table 2 shows the HRs of HbA 1C SD tertiles for progression to dialysis using the univariate Cox proportional hazards model among different subgroups. The subjects were further divided into two groups based on a mean HbA 1C value of 7%. In the patients with an HbA 1C level ≥7%, those in tertile 2 (vs. tertile 1; HR, 0.369; 95% CI, 0.167-00.814; p = 0.014) and tertile 3 (vs. tertile 1; HR, 0.307; 95% CI, 0.143-0.662; p = 0.003) were associated with a lower risk of the renal endpoint in the unadjusted model. However, in the patients with an HbA 1C level <7%, HbA 1C SD tertiles were not associated with progression to dialysis. Figure 2 illustrates the Kaplan-Meier curves for dialysis-free survival among the subjects with (A) an HbA 1C level ≥7% (log-rank p = 0.004) and (B) an HbA 1C level <7% (log-rank p = 0.902).   The patients with an HbA1C level ≥ 7% were further divided into two groups based on the stage of CKD. In the patients with an HbA1C level ≥ 7% and CKD stages 3-4, those in tertile 3 (vs. tertile 1; HR, 0.329; 95% CI, 0.122-0.887; p = 0.028) were associated with a lower risk of the renal endpoint in the unadjusted model. However, this relationship was not observed in the patients with an HbA1C level ≥ 7% and CKD stage 5. Figure 3 illustrates the Kaplan-Meier curves for dialysis-free survival among the subjects with (A) an HbA1C level ≥ 7% and CKD stages 3-4 (log-rank p = 0.045) and (B) an HbA1C level ≥ 7% and CKD stage 5 (log-rank p = 0.808). The patients with an HbA 1C level ≥ 7% were further divided into two groups based on the stage of CKD. In the patients with an HbA 1C level ≥ 7% and CKD stages 3-4, those in tertile 3 (vs. tertile 1; HR, 0.329; 95% CI, 0.122-0.887; p = 0.028) were associated with a lower risk of the renal endpoint in the unadjusted model. However, this relationship was not observed in the patients with an HbA 1C level ≥ 7% and CKD stage 5. Figure 3 illustrates the Kaplan-Meier curves for dialysis-free survival among the subjects with (A) an HbA 1C level ≥ 7% and CKD stages 3-4 (log-rank p = 0.045) and (B) an HbA 1C level ≥ 7% and CKD stage 5 (log-rank p = 0.808).  Table 3 shows the HR estimates for progression to dialysis with multivariate adjustments in the patients with an HbA1C level ≥ 7% and CKD stage 3-4. The patients in tertile 3 (vs. tertile 1) were associated with a lower risk of the renal endpoint in the adjusted model (HR, 0.243; 95% CI, 0.086-0.688; p = 0.008) after adjusting for age, sex, hypertension, coronary artery disease and cerebrovascular disease. This relationship remained significant after further adjustments for mean HbA1C, triglycerides, total cholesterol, baseline eGFR, calcium-phosphorous product, uric acid, and the use of angiotensin-converting-enzyme inhibitors (ACEIs) and/or angiotensin II receptor blockers (ARBs) (HR, 0.175; 95% CI, 0.059-0.518; p = 0.002).  Table 3 shows the HR estimates for progression to dialysis with multivariate adjustments in the patients with an HbA 1C level ≥ 7% and CKD stage 3-4. The patients in tertile 3 (vs. tertile 1) were associated with a lower risk of the renal endpoint in the adjusted model (HR, 0.243; 95% CI, 0.086-0.688; p = 0.008) after adjusting for age, sex, hypertension, coronary artery disease and cerebrovascular disease. This relationship remained significant after further adjustments for mean HbA 1C , triglycerides, total cholesterol, baseline eGFR, calcium-phosphorous product, uric acid, and the use of angiotensin-converting-enzyme inhibitors (ACEIs) and/or angiotensin II receptor blockers (ARBs) (HR, 0.175; 95% CI, 0.059-0.518; p = 0.002). The patients with an HbA 1C level ≥ 7% and CKD stages 3-4 were further divided into two groups based on the trend of HbA 1C level ( Table 2). In the patients with an HbA 1C level ≥ 7%, CKD stages 3-4 and a decreasing HbA 1C trend, those in tertile 2 (vs. tertile 1; HR, 0.261; 95% CI, 0.069-0.996; p = 0.049) and tertile 3 (vs. tertile 1; HR, 0.245; 95% CI, 0.069-0.869; p = 0.029) were associated with a lower risk of the renal endpoint in the unadjusted model. However, in the patients with an HbA 1C level ≥ 7%, CKD stages 3-4 and an increasing HbA 1C trend, HbA 1C SD tertiles were not associated with progression to dialysis. Figure 4 illustrates the Kaplan-Meier curves for dialysis-free survival among the subjects with (A) an HbA 1C level ≥ 7%, CKD stages 3-4 and a decreasing HbA 1C trend (log-rank p = 0.050) and (B) an HbA 1C level ≥ 7%, CKD stages 3-4, and an increasing HbA 1C trend (log-rank p = 0.324).  The patients with an HbA1C level ≥ 7% and CKD stages 3-4 were further divided into two groups based on the trend of HbA1C level ( Table 2). In the patients with an HbA1C level ≥ 7%, CKD stages 3-4 and a decreasing HbA1C trend, those in tertile 2 (vs. tertile 1; HR, 0.261; 95% CI, 0.069-0.996; p = 0.049) and tertile 3 (vs. tertile 1; HR, 0.245; 95% CI, 0.069-0.869; p = 0.029) were associated with a lower risk of the renal endpoint in the unadjusted model. However, in the patients with an HbA1C level ≥ 7%, CKD stages 3-4 and an increasing HbA1C trend, HbA1C SD tertiles were not associated with progression to dialysis. Figure 4 illustrates the Kaplan-Meier curves for dialysis-free survival among the subjects with (A) an HbA1C level ≥ 7%, CKD stages 3-4 and a decreasing HbA1C trend (log-rank p = 0.050) and (B) an HbA1C level ≥ 7%, CKD stages 3-4, and an increasing HbA1C trend (log-rank p = 0.324).

Discussion
In the present study, we investigated the association between HbA1C variability and renal outcomes in diabetic patients with stages 3-5 CKD over a follow-up period of 3.5 years. The results showed that the diabetic patients with CKD stages 3-4 and an HbA1C level ≥ 7% in the top HbA1C SD tertile had a decreased risk of progression to dialysis. In contrast, this relationship between HbA1C variability and renal outcome was not significant for those with an HbA1C level < 7% or CKD stage 5.
The most important finding of this study is that greater HbA1C variability was associated with a decreased risk of progression to dialysis in diabetic patients with CKD stages 3-4 and an HbA1C level ≥ 7%. Luk et al. [13] investigated the associations between HbA1C variability and incident CKD and cardiovascular disease in 8439 patients with type 2 diabetic with preserved renal function, and over

Discussion
In the present study, we investigated the association between HbA 1C variability and renal outcomes in diabetic patients with stages 3-5 CKD over a follow-up period of 3.5 years. The results showed that the diabetic patients with CKD stages 3-4 and an HbA 1C level ≥ 7% in the top HbA 1C SD tertile had a decreased risk of progression to dialysis. In contrast, this relationship between HbA 1C variability and renal outcome was not significant for those with an HbA 1C level < 7% or CKD stage 5.
The most important finding of this study is that greater HbA 1C variability was associated with a decreased risk of progression to dialysis in diabetic patients with CKD stages 3-4 and an HbA 1C level ≥ 7%. Luk et al. [13] investigated the associations between HbA 1C variability and incident CKD and cardiovascular disease in 8439 patients with type 2 diabetic with preserved renal function, and over a follow-up period of 7.2 years found that a high SD of HbA 1C was associated with incident CKD and cardiovascular disease, independent of mean HbA 1C [13]. Wang et al. [16] also investigated the role of glucose variability, expressed as fluctuations between fasting and 2-h postload glucose in patients with type 2 DM and an HbA 1C level ≥ 7%, and found that high short-term glucose variability was associated with decreased eGFR and increased risk of CKD in the patients with poor glycemic control [16]. Furthermore, Yang et al. [14] investigated the relationship between coefficient of variation (CV) of HbA 1C and progression to ESRD in patients with type 2 DM during a follow-up period of 8.2 years, and found that high HbA 1C -CV predicted the development of ESRD. In contrast to these studies [13,14,16], we found that greater HbA 1C variability was associated with better, not worse, real outcomes in the diabetic patients with CKD stages 3-4 and an HbA 1C level ≥ 7%. To further investigate this inconsistency, we performed subgroup analysis according to HbA 1C trend, and found that in the patients with a decreasing HbA 1C trend, those with greater HbA 1C variability were associated with better real outcomes, but not in those with an increasing HbA 1C trend. We hypothesize that patients with a high glucose level may change their hypoglycemic medications or receive more intensive diabetes control, and that this could lead to a decrease in glucose level and higher variation in fasting plasma glucose level. The higher rate of insulin use in patients with tertile 3 may partially support our hypothesis. The United Kingdom Prospective Diabetes Study (UKPDS) trial of patients with type 2 DM and preserved kidney function demonstrated that intensive glycemic control targeting an HbA 1C level of < 6-6.5% reduced the development and progression of diabetic nephropathy [17]. Possible mechanisms which may explain the impact of high glucose and renal toxicity include increases in glomerular permeability, circulating levels of inflammatory cytokines, mesangial lipid accumulation, mesangial and tubulointerstitial cell matrix production, expression of fibrinogenesis markers, endothelial dysfunction, and the generation of free radicals that induce diabetic complications [18][19][20][21][22]. Our findings suggest that greater HbA 1C variability may be associated with aggressive glucose control, and that this had a positive impact on the renal outcomes of the diabetic patients with stages 3-4 CKD and poor glycemic control (HbA 1C ≥ 7%). Our findings would remind physicians the importance of intensive glucose control on renal function progression in CKD patients.
Another important finding of our study is that, in contrast to the patients with CKD stages 3-4, the significance of HbA 1C variability and renal outcomes was not observed in those with CKD stage 5. The prognostic role of HbA 1C in patients with CKD stage 5 is unclear because of impaired glucose metabolism in patients with advanced CKD, and because HbA 1C level can be altered by anemia or the use of erythropoiesis-stimulating agents. A marked reduction in insulin clearance is known to occur until the eGFR falls to < 15-20 mL/min [23]. In addition, the formation of HbA 1C is known to be lower in patients with CKD due to a decrease of 30-70% in the lifespan of red blood cells (RBCs), and resistance of carbamylated hemoglobin molecules to glycosylation in a uremic environment [24]. In addition, administering erythropoiesis-stimulating agents to patients with anemia has been shown to augment, the proportion of young RBCs in peripheral blood, and these young RBCs have been shown to have a lower rate of glycosylation than old RBCs, thereby altering the formation of HbA 1C [25]. Several studies have reported that in diabetic patients with CKD stages 3-4, a higher HbA 1C level (>9%) appeared to be associated with poorer clinical outcomes, but that this was not observed in patients with CKD stage 5 [26,27]. Aggressive glycemic control appears to be beneficial for early diabetic nephropathy, however data supporting intensive glycemic control in patients with advanced CKD (including ESRD) are lacking [28]. Challenges in the management of such patients include therapeutic criteria, monitoring difficulties, and the complexity of management. In summary, HbA 1C variability seems to be more useful in predicting clinical outcomes in patients with CKD stages 3-4 than in those with CKD stage 5, possibly because of multiple factors influencing the production of HbA 1C or other factors influencing the progression to dialysis in patients with CKD stage 5.
There are several limitations to this study. First, because this was an observational study, the number and frequency of HbA 1C measurements varied between individual patients. To minimize the influence of the number and frequency of HbA 1C measurements on the results, we excluded patients with fewer than three HbA 1C measurements during the follow-up period, and those who were followed for less than six months. However, the lack of uniformity in the number and frequency of HbA 1C measurements remains an important limitation of this analysis. In addition, the use of immunoassays in the hospital to test for HbA 1C was a limitation, since immunological methods for the detection of HbA 1C are more reliable in a uremic environment. Finally, the limited number of study patients severely reduced the power of the study.

Study Patients and Design
The study was conducted at a regional hospital in southern Taiwan. We consecutively recruited patients with type 2 DM and evidence of kidney damage lasting for more than three months with stages 3-5 CKD according to the National Kidney Foundation-Kidney Disease Outcomes Quality Initiative (K/DOQI) guidelines [29] from January 2007 to September 2015. The patients were classified as having CKD stages 3, 4, and 5, based on an estimated glomerular filtration rate (eGFR) (mL/min/1.73 m 2 ) of 30-59, 15-29, and <15, respectively. All of the patients were regularly followed up at our Outpatient Department of Internal Medicine. The exclusion criteria were patients with fewer than three HbA 1C measurements during the follow-up period, patients who died, and those who commenced dialysis within six months after enrollment. Finally, 388 patients (mean age 65.7 ± 10.9 years, 234 males) were included in this study. The study protocol was approved by our Institutional Review Board of Kaohsiung Medical University Hospital (KMUHIRB-E (I)-20160032), and all enrolled patients provided written informed consent.

Collection of Demographic, Medical, and Laboratory Data
Demographic data including age and sex and medical data on comorbidities were obtained from medical records or patient interviews. DM was defined as a fasting blood glucose level of >126 mg/dL or the use of hypoglycemic agents to control the level of blood glucose. Hypertension was defined as systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg or the use of anti-hypertensive medications. Coronary artery disease was defined as evidence of old myocardial infarction, coronary artery disease on angiography, and a history of typical angina with a positive stress test, coronary artery bypass surgery, or angioplasty. Fasting blood samples were obtained from each patient within 1 month of enrollment, and laboratory analyses were performed using an autoanalyzer (Roche Diagnostics GmbH, D-68298 Mannheim COBAS Integra 400, Mannheim, Germany). The compensated Jaffé method was used to measure levels of serum creatinine using a Roche/Integra 400 Analyzer (Roche Diagnostics, Mannheim, Germany) and isotope-dilution mass spectrometry [30]. The four-variable Modification of Diet in Renal Disease equation was used to calculate eGFR [31]. HbA 1C was measured using automated cation-exchange high-performance liquid chromatography. Data on the prescriptions of angiotensin converting enzyme inhibitors (ACEIs) and angiotensin II receptor blockers (ARBs) during the study period were obtained from the patients' medical records.

Serial HbA 1C Measurements
HbA 1C measurements were recorded for each patient from the date of enrollment until the development of the renal endpoint (see the Definition of Renal Endpoint section below) or the censor date (April 2016), whichever occurred first. The mean and standard deviation (SD) of HbA 1C were calculated for each patient, with the SD being considered an index of HbA 1C variability. The rate of change of HbA 1C was assessed using the HbA 1C slope, defined as the regression coefficient between HbA 1C and time. At least three HbA 1C measurements after enrollment were required to estimate the HbA 1C slope. An HbA 1C slope ≥ 0 was considered to be an increasing trend, whereas an HbA 1C slope < 0 was considered to be a decreasing trend.

Definition of Renal Endpoint
The renal endpoint was defined as commencing dialysis. In the patients who reached the endpoint, data on renal function were censored when they commenced renal replacement therapy. All other patients were followed until April 2016. The regulations of the National Health Insurance Bureau of Taiwan were used to confirm the patients who started dialysis therapy, according to laboratory data, nutritional status and symptoms and signs of uremia.

Statistical Analysis
SPSS version 15.0 for Windows (SPSS Inc., Chicago, IL, USA) was used for all statistical analyses. Data are expressed as percentages or mean ± SD. Multiple comparisons among groups were performed using one-way analysis of variance (ANOVA) followed by post hoc tests with Bonferroni correction. The Kaplan-Meier method was used to plot survival curves for the renal endpoint. A Cox proportional hazards model was used to analyze the time to renal endpoint and the risk factors. A p-value of <0.05 was considered to indicate a significant difference.

Conclusions
In conclusion, greater HbA 1C variability with a decreasing HbA 1C trend, which may be related to intensive diabetes control, was associated with a decreased risk of progression to dialysis in patients with stages 3-4 CKD and poor glycemic control (HbA 1C ≥ 7%), but this association was not found in the patients with CKD stage 5. Our results support the potential role of aggressive glycemic control on clinical outcomes and highlight its importance in diabetic patient with stages 3-4 CKD.

Conflicts of Interest:
The authors declare no conflict of interest.