Diabetologia

, Volume 55, Issue 8, pp 2128–2131

HbA1c variability and the development of microalbuminuria in type 2 diabetes: Tsukuba Kawai Diabetes Registry 2

  • A. Sugawara
  • K. Kawai
  • S. Motohashi
  • K. Saito
  • S. Kodama
  • Y. Yachi
  • R. Hirasawa
  • H. Shimano
  • K. Yamazaki
  • H. Sone
Short Communication

DOI: 10.1007/s00125-012-2572-7

Cite this article as:
Sugawara, A., Kawai, K., Motohashi, S. et al. Diabetologia (2012) 55: 2128. doi:10.1007/s00125-012-2572-7

Abstract

Aims/hypothesis

The aim of this study was to examine the association between HbA1c variability and the development of microalbuminuria as defined by an albumin/creatinine ratio ≥3.4 mg/mmol (≥30 mg/g) in at least two of three consecutive urine samples in Japanese patients with type 2 diabetes.

Methods

HbA1c level was measured in 812 serially registered normoalbuminuric adults aged 21–79 years with type 2 diabetes. After registration, a 1-year period to establish baseline values for mean HbA1c and HbA1c variability (measured as the intrapersonal SD of serially collected HbA1c) was decided upon. The association between HbA1c variability and the development of microalbuminuria was determined by Cox regression analysis after adjustment for other risk factors for microalbuminuria.

Results

Microalbuminuria occurred in 193 patients during the observation period of (mean ± SD) 4.3 ± 2.7 years. Even after adjustment for mean HbA1c, HbA1c variability was a significant predictor of microalbuminuria independently of the mean HbA1c; the HR for every 1% (95% CI) increase in mean HbA1c was 1.22 (1.06, 1.40) (p = 0.005), and that for HbA1c variability was 1.35 (1.05, 1.72) (p = 0.019). The effects of these two variables were quite similar when 1 SD was used; the HR for every 1 SD increase (95% CI) in HbA1c was 1.23 (1.07, 1.43) (p = 0.005), and that for HbA1c variability was 1.20 (1.03, 1.39) (p = 0.019).

Conclusions/interpretation

HbA1c variability affects the development of microalbuminuria independently of mean HbA1c in type 2 diabetes. Further studies should be performed to evaluate the influence of HbA1c variability on other complications and in individuals of other ethnicities with type 2 diabetes.

Keywords

Haemoglobin A1c Microalbuminuria Prospective 

Abbreviation

ACR

Urinary albumin/creatinine ratio

Introduction

The risk of vascular complications increases exponentially as the mean HbA1c level increases [1]. Furthermore, it has been suggested that HbA1c variability, which is expressed as the intrapersonal SD of serially measured HbA1c, is also an independent risk factor for the development of diabetic complications in individuals with type 1 diabetes [2, 3]. However, it is unknown whether HbA1c variability influences the development of microvascular complications in those with type 2 diabetes since risk factors for diabetic complications have been reported to vary depending on the type of diabetes [4]. Sone and colleagues reported that risk factors for diabetic neuropathy were different between patients with type 1 and type 2 diabetes [4]. Moreover, we are not aware of previous studies undertaking a qualitative comparison between mean HbA1c and HbA1c variability using 1 SD increments, which allows for a direct comparison of the effects of these two variables, except for the study of Waden et al. [3] that compared the effect of these two variables on categorised data using Kaplan–Meier analysis.

Therefore, we have prospectively investigated the following in Japanese patients with type 2 diabetes: (1) whether HbA1c variability is associated with the development of microalbuminuria; and (2) whether mean HbA1c or HbA1c variability has the stronger effect. In addition, in this study, we selected a 1-year period beginning immediately after enrollment in the registry during which we calculated the mean HbA1c and HbA1c variability to acquire baseline information that would provide highly accurate information for Cox regression analysis.

Methods

The study population comprised 1,713 type 2 diabetic individuals aged 21–79 years who had been consecutively registered on the Tsukuba Kawai Diabetes Registry database [5] from 2000 to 2007 beginning at their first visit to the Kawai Clinic. This clinic is a typical outpatient diabetes referral centre located in a suburb of Tokyo. Participants who had fewer than two measurements of their urinary albumin/creatinine ratio (ACR; n = 284), dropped out during the 1-year period after enrollment (n = 29), were not normoalbuminuric (n = 562), had a high serum creatinine level (>130 μmol/l; n = 2), or had a history of cancer (n = 24) or cardiovascular disease (n = 1), were excluded. Subsequently, 812 normoalbuminuric patients (558 males, 254 females) were eligible for the current analysis.

During the 1-year period after enrollment in the registry, we accumulated data on the mean HbA1c and HbA1c variability for baseline information. Participants were defined as normoalbuminuric if their ACR was <3.4 mg/mmol for their first and second urine samples, and were considered microalbuminuric if the ACR was ≥3.4 mg/mmol for at least two of three consecutive urine samples. During the follow-up period, ACR was examined every 6 months by turbidimetric immunoassay (Microalbumin-HA test; Wako Pure Chemicals, Osaka, Japan) and HbA1c was examined using HPLC (HLC-723; TOSOH, Tokyo, Japan). The value for HbA1c was estimated as a National Glycohemoglobin Standardization Program equivalent value calculated using the formula [6]: HbA1c (%) = HbA1c (Japan Diabetes Society) (%) + 0.4.

At baseline, serum lipid levels (determined by a direct enzyme method, BM-80-60; Nihon Denshi, Tokyo, Japan), blood pressure (measured by a doctor with the patient seated) and smoking status (never/ever) were also assessed.

The observation period began at the patient’s first clinic visit after the 1-year period in which data were collected on mean HbA1c and HbA1c variability and lasted up to the date of development of microalbuminuria or to the last ACR measurement. HbA1c variability was calculated as the intrapersonal SD for HbA1c for each patient during the 1-year period during which the baseline was established. The difference in number of HbA1c assessments between patients was adjusted according to the formula [2]: \( SD/\surd \left[ {n/\left( {n - 1} \right)} \right] \). To compare the direct impact of mean HbA1c and HbA1c variability, we changed the increment from 1% to 1 SD (mean ± SD being 0.0 ± 1.0).

The study protocol was consistent with the Japanese government’s Ethical Guidelines Regarding Epidemiological Studies in accordance with the Declaration of Helsinki and was reviewed by the institutional review board. All participants gave their informed consent. We used χ2 tests to compare proportions and t tests to compare the variables between normoalbuminuria and microalbuminuria. Cox regression analysis was used to determine whether HbA1c variability was an independent predictor of microalbuminuria. All statistical analyses were performed using SPSS (version 17.0 for Windows; SPSS, Chicago, IL, USA).

Results

During the 1-year period in which baseline values were established, the median (range) number of HbA1c measurements per patient was 11.0 (5–12) and the mean (SD) follow-up period was 4.3 (2.7) years. At baseline, the mean age of the participants was 54.9 (10.4) years. Of these participants, 727 used glucose-lowering agents (625, oral hypoglycaemic agents; 102, insulin), and 259 patients used antihypertensive agents. The proportion of patients who used ACE inhibitors/angiotensin II receptor blockers did not differ significantly between those with normoalbuminuria and those with microalbuminuria (7.5% vs 12.2%; p = 0.15). Table 1 shows the baseline characteristics of study participants who did or did not subsequently develop microalbuminuria. Participants who developed microalbuminuria (n = 193) were significantly younger, had had diabetes for a longer period, and had a higher mean HbA1c and a higher ACR than participants who remained normoalbuminuric. HbA1c variability was not significant but was greater in participants with microalbuminuria than in those with normoalbuminuria.
Table 1

Baseline characteristics of those who did or did not subsequently develop microalbuminuria

Variable

Normoalbuminuria

Microalbuminuria

p value

n

619

193

 

Age (years)

54.3 ± 10.2

56.8 ± 10.9

0.003

Sex (% male)

69

67.9

0.772

Duration of diabetes (years)

5.8 ± 6.2

7.8 ± 8.1

0.002

Mean HbA1c (%)

7.0 ± 1.0

7.4 ± 1.2

0.001

Mean HbA1c (mmol/mol)

53.4 ± 11.1

56.8 ± 12.8

0.001

HbA1c variability (%)

0.79 ± 0.60

0.88 ± 0.62

0.110

Systolic blood pressure (mmHg)

125 ± 11

128 ± 10

0.001

Body mass index (kg/m2)

24.7 ± 3.6

25.4 ± 4.0

0.031

Total cholesterol (mmol/l)

5.3 ± 1.0

5.1 ± 0.8

0.011

HDL-cholesterol (mmol/l)

1.4 ± 0.4

1.4 ± 0.4

0.248

ACR (mg/mmol)

1.1 ± 0.5

1.7 ± 0.6

<0.001

Ever smoker (%)

60.6

67.4

0.090

Retinopathy (%)

19.4

28.5

0.007

Neuropathy (%)

6.5

9.3

0.117

Data are expressed as numbers, means ± SD, or percentages

The value for HbA1c was estimated as a National Glycohemoglobin Standardization Program equivalent value calculated using the formula [6]: HbA1c (%) = HbA1c (Japan Diabetes Society) (%) + 0.4

The results of Cox regression analysis showed that, in addition to known predictors of nephropathy, both mean HbA1c and HbA1c variability were significant and independent risk factors for microalbuminuria (Table 2, model 1). By using a 1 SD increment instead of a 1% increment, we showed that the HRs of these two variables were similar (Table 2, model 2).
Table 2

Cox regression models for the development of microalbuminuria

Variable

HR (95% CI)

p value

Age (years)

1.02 (1.00, 1.04)

0.013

Male sex (%)

0.70 (0.47, 1.06)

0.089

Duration of diabetes (years)

1.03 (1.01, 1.05)

0.005

Systolic blood pressure (mmHg)

1.01 (1.00, 1.03)

0.090

Body mass index (kg/m2)

1.05 (1.01, 1.10)

0.016

Total cholesterol (mmol/l)

0.87 (0.73, 1.03)

0.097

HDL-cholesterol (mmol/l)

0.90 (0.59, 1.40)

0.650

Ever smoker (%)

1.68 (1.13, 2.49)

0.010

Model 1: 1% increment

 Mean HbA1c (%)

1.22 (1.06, 1.40)

0.004

 HbA1c variability (%)

1.35 (1.05–1.72)

0.019

Model 2: 1 SD increment

 Mean HbA1c

1.23 (1.07, 1.43)

0.005

 HbA1c variability

1.20 (1.03, 1.39)

0.019

Model 1: effect of 1% increments in mean HbA1c and HbA1c variability (intrapersonal SD of HbA1c)

Model 2: comparison of direct effect of mean HbA1c and HbA1c variability using 1 SD increments

The observation period was the time variable in the Cox proportional hazards model. All models were adjusted by various known predictors of nephropathy in addition to those related to HbA1c (upper section of the table). Mean HbA1c and HbA1c variability were calculated during the 1-year period before start of the study, and other variables were obtained at the start of the study

HRs of variables except mean HbA1c and HbA1c variability were determined by multivariate analysis from model 1. The result did not change in model 2

Discussion

We clarified that HbA1c variability was independently associated with the development of microalbuminuria in our Japanese patients with type 2 diabetes even after adjustment for known predictors of nephropathy [5]. To our knowledge, this is the only prospective study to examine this topic in individuals with type 2 diabetes except for one Korean study that determined the progression of carotid artery intima–media thickness [7] and compared the direct impact of HbA1c variability and mean HbA1c. This association was in accordance with results of studies of type 1 diabetes [2, 3, 8], especially with regard to findings by Waden and colleagues in which adjustments were made for blood pressure and smoking status [3]. We found that HbA1c variability had an effect on the development of microalbuminuria similar to that of mean HbA1c, since the HRs of these two variables were similar, a finding that was also in accordance with results of Waden et al. [3] shown by Kaplan–Meier survival curves.

Although the mechanisms of the association between HbA1c variability and the development of vascular complications are unclear from this observational study, increasing oxidative stress is speculated to be one explanation since unstable glucose fluctuation induces greater oxidant production [9]. Monnier and colleagues showed a strong relationship between oxidative stress and daily glucose variability [9]. However, it is yet to be clarified whether there is a correlation between short-term glucose variability and HbA1c variability. Future study is expected to clarify the pathway.

The strength of the current study is that we used a 1-year period to calculate the baseline data of mean HbA1c and HbA1c variability for use in Cox regression analysis. Another strength is that we measured HbA1c much more frequently than in previous studies [2, 3, 7, 8], which could contribute to the reliability of the data. Such a testing frequency is routine in clinical settings in Japan. In addition, because the use of monthly measurements of HbA1c might cause partial interdependence, we performed a reanalysis using 3-monthly measurements of HbA1c and found that results were no different (HbA1c variability for 1% increase: HR [95% CI] = 1.33 [1.02, 1.72]; p = 0.036).

In contrast, the fact that the study participants were from a single clinic in Japan is one of the limitations. However, the characteristics of our study participants are similar to those reported in a previous large-scale study in Japan [10]. Another limitation is that the hyperglycaemic treatment modalities (i.e. oral agents and insulin) or use of an ACE inhibitor was not explored in the context of HbA1c variability. However, the results of the present study are very much in line with those of previous studies of type 1 diabetes [2, 3, 7, 8]. In addition, changes in HbA1c during the entire observation period before the development of microalbuminuria were not taken into account in this study. However, this is also a limitation of Cox regression analysis since this method is a way to predict future risk using present data, and further analysis is needed to clarify it.

In conclusion, HbA1c variability might be a risk factor for the development of microalbuminuria in patients with type 2 diabetes, and the impact of the effect might be similar to that of the mean HbA1c. Further studies are recommended to evaluate complications other than nephropathy and complications in other ethnic groups with type 2 diabetes.

Acknowledgements

The authors thank F. Satomi at the University of Tsukuba for her excellent secretarial work. The authors are also grateful to the patients who took part in the Tsukuba Kawai Diabetes Registry.

Funding

A. Sugawara and H. Sone are recipients of a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science. This work was also funded by the Ministry of Health, Labor and Welfare, Japan.

Duality of interest

The authors declare that there is no duality of interest associated with this manuscript.

Contribution statement

AS designed and conducted the study, analysed the results and wrote the manuscript. KK, SM and KY contributed to the acquisition of data, review of the data, and discussion of the data, and reviewed and edited the manuscript. KS, SK, YY, RH, HSh and HSo designed and conducted the study, analysed the results, and critically reviewed and edited the manuscript. All the authors gave final approval of the version to be published.

Copyright information

© Springer-Verlag 2012

Authors and Affiliations

  • A. Sugawara
    • 1
    • 2
  • K. Kawai
    • 2
  • S. Motohashi
    • 2
  • K. Saito
    • 1
    • 2
  • S. Kodama
    • 1
  • Y. Yachi
    • 1
  • R. Hirasawa
    • 1
  • H. Shimano
    • 1
  • K. Yamazaki
    • 2
  • H. Sone
    • 1
  1. 1.Department of Internal Medicine, Institute of Clinical MedicineUniversity of TsukubaMitoJapan
  2. 2.Kawai ClinicIbarakiJapan

Personalised recommendations