The importance of HbA1c and glucose variability in ... - Springer Link

Acta Diabetol (2012) 49 (Suppl 1):S153–S160 DOI 10.1007/s00592-012-0391-4

ORIGINAL ARTICLE

The importance of HbA1c and glucose variability in patients with type 1 and type 2 diabetes: outcome of continuous glucose monitoring (CGM) Giovanni Sartore • Nino Cristiano Chilelli • Silvia Burlina • Paola Di Stefano • Francesco Piarulli • Domenico Fedele • Andrea Mosca • Annunziata Lapolla

Received: 22 December 2011 / Accepted: 14 March 2012 / Published online: 1 April 2012 Springer-Verlag 2012

Abstract Glucose variability has recently been investigated in diabetic patients in several studies, but most of them considered only a few variability indicators and did not systematically correlate them with patients’ HbA1c levels and other important characteristics. In thus study, the correlations between HbA1c levels and metabolic control (average glucose, AG), glucose variability (SD, CONGA, MAGE, MODD, BG ROC), hyperglycemia (HBGI), hypoglycemia (LBGI) and postprandial (AUC PP) indices were investigated in patients with type 1 and type 2 diabetes. The study involved 68 patients divided into 3 groups as follows: 35 patients had type 1 diabetes (group 1); 17 had type 2 diabetes and were taking multiple daily injections (MDI) of insulin (group 2); and 16 patients had type 2 diabetes treated with OHA and/or basal insulin (group 3). The indicators were obtained over at least 48 h using a continuous glucose monitoring (CGM) system. HbA1c levels were measured at the baseline and after CGM. HbA1c correlated significantly with AG (r = 0.74), AUC PP (r = 0.69) and HBGI (r = 0.74), but only in type 1 diabetic patients. Patients with longstanding disease and type 1 diabetes had a greater

Communicated by Antonio Secchi. G. Sartore N. C. Chilelli (&) S. Burlina F. Piarulli D. Fedele A. Lapolla Department of Medical and Surgical Sciences, University of Padova, Via Giustiniani n 2, 35128 Padua, Italy e-mail: [email protected] P. D. Stefano Medtronic Italia S.p.A., Rome, Italy A. Mosca Department of Biomedical Sciences and Technologies, University of Milano, Milan, Italy

glucose variability, irrespective of their HbA1c levels. Insulin therapy with MDI correlated strongly with HbA1c, but not with glucose variability. HbA1c levels identify states of sustained hyperglycemia and seem to be unaffected by hypoglycemic episodes or short-lived glucose spikes, consequently revealing shortcomings as a ‘‘gold standard’’ indicator of metabolic control. Glucose variability indicators describe the glucose profile of type 1 diabetic patients and identify any worsening glycemic control (typical of longstanding diabetes) more accurately than HbA1c tests. Keywords Glucose variability Continuous glucose monitoring HbA1c Standard deviation Hyperglycemia

Introduction The large, randomized DCCT-EDIC and UKPDS trials on subjects with diabetes mellitus type 1 (DM1) and type 2 (DM2) have shown that daily blood glucose profiles can differ quite considerably between patients with comparable HbA1c and mean blood glucose levels [1, 2]. In particular, HbA1c alone is unable to provide information on day-today changes in glucose levels (in terms of their amplitude and frequency) [3]. Numerous reports in the literature have indicated that oscillating glucose levels, associated with an increase in free radicals and endothelial dysfunction, could contribute to the pathogenesis of diabetic complications, irrespective of a patient’s HbA1c levels [4]. Ceriello et al. [5] also recently emphasized the role of acute hyperglycemic spikes in the physiopathology of micro- and macrovascular complications of DM: These authors showed that a reduction in hyperglycemic excursions coincides with a reduction in some markers of oxidative stress.

123

S154

Finally, glucose variability has been seen as a plausible candidate for predicting severe hypoglycemia because this condition is preceded by blood glucose disruptions [6]; several studies have reported that fewer hypoglycemic episodes coincide with a lower glucose variability [7]. Analyzing glucose profiles obtained with continuous interstitial CGM systems in diabetic patients enrolled in the A1c Derived Average Glucose (ADAG) study demonstrated a close relationship between HbA1c and mean plasma glucose in the months preceding the test [8]. On the other hand, it has been demonstrated that patients with comparable mean blood glucose or HbA1c levels may have quite different daily blood glucose profiles [9], so other measures have been suggested for the long-term monitoring of diabetes mellitus. Glycated albumin has been suggested, for instance, as a useful alternative marker of glycemic control [10]: It has some properties that are better than HbA1c [11], but several studies have demonstrated that oxidation states influence glycated albumin levels, and this detracts from its validity [12, 13]. The ability of CGM systems to discriminate between chronic exposure to hyperglycemia and acute hypoor hyperglycemic spikes has led to the development of new glucose variability indicators [14], but none of them have so far come to be considered as the ‘‘gold standard’’ for assessing glucose variability in diabetic individuals. A recent trial shed light on the relationship between HbA1c levels and several glucose variability indicators [15], but there is still debate in the literature on this topic, some studies supporting this association [16, 17], while others refute it [18, 19]. The reasons for these divergences of opinion could be that too few variability indicators were considered, and over-simple statistical models were used. More in-depth investigations into this association would help us to understand whether HbA1c levels alone suffice to describe the glucose profile of diabetics or whether other variables should be taken into account. Hence, this study examined whether HbA1c reflects averages and slows glucose fluctuations, or whether it might also be influenced by any rapid glucose changes, such as hypo- and hyperglycemic spikes. All available indicators were considered, that is, average metabolic control (average glucose, AG), postprandial glycemia (area under curve, postprandial, AUC PP), hyperglycemia (high blood glucose index, HBGI), hypoglycemia (low blood glucose index, LBGI), and glucose variability (standard deviation, SD), the mean amplitude of glucose excursion (MAGE), the continuous overlapping net action (CONGA), the mean of daily differences (MODD), and the standard deviation of the blood glucose rate of change (SD-BG ROC) obtained during the A1c-Derived Average Glucose (ADAG) study. Our ultimate aim was to establish which type of glucose variability best determines HbA1c levels, using all the most common and the latest CGM indicators.

123

Acta Diabetol (2012) 49 (Suppl 1):S153–S160

The influence of the study population’s clinical characteristics on the indicators considered was also analyzed, focusing particularly on establishing which types of patient are characterized by a greater glucose variability.

Materials and methods Data were obtained on 68 consecutive type 1 and type 2 diabetic patients (37 women and 31 men, aged 19–69 years) attending the Diabetology and Dietetics Service (ULSS 16) in Padua, Italy. These patients had been recruited for the ADAG study, which has been reported in detail elsewhere [8]. The patients had to have a stable glycemic control, as demonstrated by two HbA1c readings differing by no more than 1 % during the 6 months prior to the start of the study. At the baseline, patients underwent venous blood sampling to establish their HbA1c levels. For the ADAG study, continuous interstitial glucose monitoring (Medtronic Minimed, Northridge, CA) was performed 4 times, at 4-week intervals, over a 16-week study period. The monitoring periods lasted at least 48 h, and the glucose levels were measured every 5 min. HbA1c levels were tested monthly in blood samples every time a new CGM session started, and four different DCCT-aligned assays were performed, that is, a high-performance liquid chromatography assay, two immunoassays, and one affinity assay (all approved by the National Glycohemoglobin Study Program), and the mean HbA1c level was adopted [8]. For the statistical analysis, the HbA1c levels measured at the end of the first CGM session were considered to avoid any ‘‘trial effect’’ of the ADAG study and to capture the period characterized by the greatest glucose variability. Patients were divided into three groups as follows: group 1 consisted of 35 patients (17 men and 18 women) with DM 1, aged 39 ± 11 (mean ± standard deviation, SD); group 2 included 17 patients (6 men and 11 women) with DM 2, being treated with multiple daily injections (MDI) of insulin, aged 58 ± 9 (mean ± SD); and group 3 comprised 16 patients (10 men and 6 women) with DM 2 treated with dietary restrictions alone or oral hypoglycemic agents (OHA), or a basal insulin regimen, aged 58 ± 9 (mean ± SD). Table 1 summarizes the demographic and clinical characteristics of the patients analyzed in the 3 groups. The glucose variability indicators were compared in the different groups vis-a`-vis certain demographic and clinical characteristics, considering age ([46 vs.\46 years old), type of diabetes (group 2 vs. group 1), type of treatment (group 3 vs. group 2), and duration of diabetes ([14 vs.\14 years). The study was approved by the local human studies committees, and informed consent was obtained from all participants.


S155

Table 1 Comparisons between groups’ demographic and clinical characteristics

Statistical analysis

Parameters

Group 1

Group 2

Group 3

p value

Age (years)

39 ± 11*

58 ± 9

58 ± 9

\0.001

Gender (M/F)

17/18

6/11

10/6

n.s.

Descriptive statistics are reported as means and standard deviations. Normality of distribution was tested by calculating skew and kurtosis values. Continuous variables were compared using Student’s t test or the Wilcoxon test (for two groups), or analysis of variance with Bonferroni’s correction (for three groups), as appropriate. Correlations were analyzed for the three groups of patients to investigate the relationship between HbA1c and the glucose variability indicators, and p values were corrected with the Bonferroni method. Pearson’s correlation coefficient was computed separately for the three analyses. A univariate analysis was conducted using linear regression to assess independent predictors among demographic and clinical characteristics. On the basis of the outcome of the univariate analysis, a multivariate regression analysis was performed to assess independent predictors of HbA1c and glucose variability indicators and to adjust for age, gender and hypertension. Statistical significance was assumed for p \ 0.05. The statistical data analysis was conducted with Stata/SE 11.0 software for Windows (StataCorp LP, Texas, USA).

Duration of diabetes (years)

16.25 ± 8.97

16.9 ± 8.03

10.9 ± 8.73

n.s.

BMI (kg/m2)

25.1 ± 2.6*

31.0 ± 5.7

27.0 ± 4.6

\0.001

HbA1c (%)

8.4 ± 1.6

8.4 ± 1.5à

7.0 ± 1.0

\0.05

All parameters are expressed as mean ± standard deviation. * group 1 versus group 2; group 1 versus group 3; à group 2 versus group 3

Indicators of glucose variability and metabolic control All indicators of metabolic control were calculated from the glucose monitoring data, disregarding the first 2 h of the first monitoring session, which served as the initial instrument calibration period. Average glucose (AG) was considered as an indicator of metabolic control [20] and the postprandial area under the curve (AUC PP) to establish postprandial exposure to hyperglycemia. Glucose variability was evaluated by measuring the SD, MAGE, CONGA, MODD, HBGI, LBGI and SD-BG ROC. The HBGI and LBGI were also considered for the purpose of assessing the hyperglycemic/hypoglycemic risk. The AUC PP was the postprandial AUC calculated from the preprandial glucose levels to the highest peak over a 2-h period after a meal. The MAGE was calculated as the arithmetic mean of the differences between consecutive glycemic peaks and nadirs, only considering changes in glycemic values of more than 1 SD [21]. The CONGA was the SD of the glycemic differences recorded between a specific point on the CGM profile and a point n hours previously (where n = 1,2,3,4,…) [22]. The CONGA was analyzed at 2 h, both during the day (CONGA 2,_day) and at night (CONGA 2,_night), comparing the variability during fasting and in the postabsorptive period. The MODD was calculated as the mean of the absolute differences between glycemic gaps observed during the same time interval on 2 consecutive days [23], as an expression of the ‘‘between-day’’ glucose variability. Finally, the LBGI, HBGI and SD-BG ROC were calculated using complex formulas specifically designed to derive these indicators from CGM data [24]. HBGI and LBGI parameters account for the frequency and amplitude of hyperglycemic and hypoglycemic events, respectively, enabling an assessment of the risk of patients encountering adverse glycemic event [25].

Results All 68 patients completed the study. The characteristics of the study population are given in Table 1: There was a statistically significant difference between the groups in terms of age (group 1 vs. group 2, group 1 vs. group 3; p \ 0.001), basal HbA1c levels (group 1 vs. group 3, group 2 vs. group 3; p \ 0.05) and BMI (group 1 vs. group 2, p \ 0.05). Simple regression analysis The analysis focused on the association between the HbA1c levels obtained at the end of the first CGM session and indicators of metabolic control, hyperglycemia and glucose variability. The correlations between the HbA1c levels and the indicators of metabolic control for the three groups are summarized in Table 2. Significant correlations were only found between HbA1c and AG_w (r = 0.74), AUC PP (r = 0.69) and HBGI (r = 0.74) in type 1 diabetic patients (group 1), while none of the other correlations reached statistical significance. This was probably due to our needing to use Bonferroni’s correction (significance considered for p value < 0.0017) because of the large number of indicators considered. A correlation coming closer to significance was seen between HbA1c and AG (which expresses the mean blood

123

S156


0.7369*

0.7087

0.7069

0.0000

0.0021

0.0022

Pearson’s correlation coefficient

0.6857*

0.3205

0.5271

p value

0.0000

0.2097

0.0359


0.3907

0.3527

0.6273

p value

0.0203

0.1802

0.0093

glucose derived from the 48 h monitoring period) in groups 2 and 3 (p values = 0.0021 and 0.0022, respectively). Then, some of the characteristics of the study population were correlated with HbA1c levels, metabolic control indicators, hyperglycemia and glucose variability. Figure 1a shows that HbA1c levels were significantly lower in patients treated with OHA or a basal insulin regimen than in those on MDI of insulin (group 3 vs. group 2; p \ 0.01), whereas HbA1c levels were unaffected by the type of diabetes (group 2 vs. group 1) or a more or less longstanding disease ([14 years vs. \14 years). Much the same results emerged when the AUC PP was considered. On the other hand, SD_w (Fig. 1b) was significantly lower in patients with a shorter-lived disease (p \ 0.01) and with type 2 diabetes (p \ 0.01). Similar results emerged for MAGE and CONGA2.

0.2734

0.4331

0.3193

Multiple regression analysis

0.1121

0.0938

0.2280


0.4796

-0.0320

0.2823

p value

0.0036

0.9064

0.2895


0.3783

0.4460

0.6972

p value

0.0250

0.0834

0.0270

-0.0307

0.0910

0.0012

0.8610

0.7374

0.9965

-0.1890

-0.5273

-0.2163

0.2769

0.0358

0.4211


0.7411*

0.6886

0.6511

p value

0.0000

0.0032

0.0063


0.1045

-0.5695

0.4251

p value

0.5501

0.0213

0.1008

On the basis of the above results, a multiple regression analysis was performed, considering HbA1c and the glucose variability indicators (MAGE, SD_w, CONGA 2) as dependent variables, while the type of therapy, the duration of the disease and the type of diabetes were considered as explanatory variables (Table 3). All comparisons were corrected for age, gender and hypertension. The analysis confirmed the influence of the type of therapy on HbA1c levels: DM 2 patients treated with OHA or basal insulin had lower HbA1c levels than patients on MDI of insulin (p \ 0.05). The other characteristics did not reach statistical significance as predictors of HbA1c levels. Multiple regression nonetheless showed that glucose variability was influenced by the type of diabetes and its duration: Type 2 diabetic patients on MDI of insulin had lower SD_w and CONGA 2 than those with DM 1, although the two groups had the same HbA1c levels. Patients with longstanding diabetes (be it DM1 or DM2) also had higher levels of SD_w and CONGA 2 than those with a shorter history of diabetes (p \ 0.05). Our model, taking the type of therapy, duration of disease and type of diabetes into account as independent variables, explained 18.8 % of the HbA1c levels, while it explained 33.8 % of the MAGE, 44.0 % of the SD_w and 37.3 % of the CONGA 2 variability.

Table 2 Correlations between HbA1c and indicators of metabolic control derived from CGM data in groups 1, 2 and 3 Indicators

HbA1c group 1

HbA1c group 2

HbA1c group 3

AG_w Pearson’s correlation coefficient p value AUC PP

SD_w

CONGA2_day Pearson’s correlation coefficient p value CONGA2_night

MAGE

MODD Pearson’s correlation coefficient p value LBGI Pearson’s correlation coefficient p value HBGI

BG ROC

AG_w average glucose ‘‘within day’’; AG_b average glucose ‘‘between days’’; AUC area under curve; SD standard deviation; CONGA2 continuous overlapping net glycemic action (at 2-h intervals); MAGE mean amplitude of glycemic excursion; MODD mean of daily differences; LBGI low blood glucose index; HBGI high blood glucose index; BG ROC blood glucose rate of change * Statistical significance was considered for p value £ 0.0017, as appropriate for Bonferroni’s correction

123

Discussion Testing HbA1c is currently considered the ‘‘gold standard’’ for assessing metabolic control in diabetics, but it has been


S157

Fig. 1 Comparisons between HbA1c (a) and SD_w (b) in different classes of patients (divided by type of therapy, duration of diabetes, type of diabetes). Statistical correlation performed with Student’s t test; statistical significance was assumed for p \ 0.05

demonstrated that patients with comparable mean blood glucose or HbA1c levels may have quite different daily blood glucose profiles, with significant differences in the number and extent of fluctuations in their blood glucose levels [9]. Our results suggest that HbA1c is influenced by hyperglycemic spikes (in terms of AUC PP and HBGI), particularly in type 1 diabetic patients, confirming the importance of hyperglycemia in causing high HbA1c levels

[26]. A close correlation was also found between AG and HbA1c in type 2 diabetic patients, although its statistically significance was lost because of the model adopted, which sets the threshold for significance at a p value of B0.0017. Borg et al. [15] showed a weak correlation between variability indicators and HbA1c, while they found a close correlation between the latter and hyperglycemia indicators. Our results are consistent with the latter finding, but not with the former, because the correlation between

123

S158


Table 3 Influence of clinical characteristics on HbA1c and glucose variability indicators Type of therapy (OHA/basal insulin vs. MDI)

Duration of disease ([14 vs. \14 years)

Type of diabetes (DM1 vs. DM2)

R2 (%)

HbA1c (%)

-1.29*

0.46

0.33

18.8

(-2.43; -0.16)

(-0.31; 1.23)

(-0.86; 1.53)

MAGE (mg/dl)

- 0.26

0.18

0.23

(-0.61; 0.08)

(-0.05; 0.41)

(-0.13; 0.59)

-9.11

10.62*

19.06*

(-22.12; 3.89)

(1.79; 19.44)

(5.34; 32.80)

-4.9 (-21.67; 11.87)

13.73 * (2.35; 25.11)

18.68 * (0.98; 36.38)

SD_w (mg/dl) CONGA 2 (mg/dl)

33.8 44.0 37.3

Data are b coefficients from multiple linear regression models (95 % CI) and express the mean increase or decrease in indicators between the groups. R2 is the proportion of the indicator explained by the clinical characteristics considered in the model. Analysis corrected for age, gender and hypertension. * p \ 0.05

the variability indicators and HbA1c was no longer significant when the variability indicators and the mean and postprandial glucose were plotted together. This was particularly evident for Kovatchev’s BG ROC and LBGI, previously shown to predict glycemic variability and hypoglycemic risk [25]. HbA1c levels therefore reflect averages and sustained hyperglycemic fluctuations, but are insensitive to short and rapid glucose swings occurring during the day. Our results are also consistent with evidence emerging at the end of the DCCT study [1] of HbA1c levels being of limited value as an indicator of hypoglycemic events [27]. HbA1c should therefore not be considered alone as the ‘‘gold standard’’ for maintaining an optimal glycemic control in diabetic patients; A patient’s glucose profile needs to be considered more thoroughly. There is still debate on how several important characteristics of diabetic patients (e.g., type of diabetes, type of therapy and duration of disease) affect their glucose variability. Our results showed that indicators of metabolic control or hyperglycemia, glucose variability and HbA1c levels were all higher in patients on MDI of insulin. On the other hand, glucose variability indicators—and MAGE, SD_w and CONGA 2 in particular—were significantly higher in DM 1 patients and in cases of longstanding disease (both type 1 and type 2 DM). SD, in particular, is a major indicator of glucose variability and consequently a fundamental parameter for optimal diabetes management, as previously reported [20]. Our results partly reflect some well-established evidence in the literature of a greater glucose variability in DM 1 than in DM 2, but our findings also contrast with a recent report from Greven et al. [17], who reported finding no relationship between disease duration and glucose variability (despite a study design similar to ours and patients with comparable characteristics), while they did find a correlation between the latter and the duration of insulin therapy.

123

In our multiple regression analysis to identify the characteristics most strongly correlated with glucose profile, all comparisons were corrected for age, gender and hypertension, and our results confirmed that a longer history of diabetes is associated with a more pronounced glucose variability, in both DM1 and DM2. This higher blood glucose variability in longstanding diabetes may be partially responsible for the observed increase in the episodes of severe hypoglycemia, as the disease progresses [28]. Monnier et al. [29] showed a higher glucose variability in patients on MDI of insulin. In our study, when the type of therapy was considered alone (see Fig. 1), it was found to correlate with glucose variability, but the significance of this correlation was lost in the multivariate model, whereas the duration of disease and type of diabetes showed a marked correlation with glucose variability indicators. This might be partly due to the variety of insulin treatments used in our patients on MDI regimens, deliberately designed to tailor the therapy to individual patients. It may be that a more widespread use of new (rapid and basal) insulin analogs can improve glucose variability and contain glucose swings, even if they cannot improve HbA1c levels [30], but this consideration needs to be supported by further studies. Much evidence has led to the hypothesis that the pathophysiology of DM-related complications relating to glycation and the excessive generation of oxidative stress is attributable to all three major blood glucose imbalances, that is, hyperglycemia (during fasting and after meals) and both positive and negative acute glucose fluctuations (glucose variability) [31]. Judging from our results, HbA1c levels cannot represent any specific impairment in glucose homeostasis—a job that glucose variability indicators could do better. Although, for the time being, there is not enough evidence to support the targeting of glucose variability


irrespective of mean glucose and/or HbA1c levels [9], our results highlight the importance of aiming for customized patient monitoring. Limitations of our study include the small number of patients involved, the extensive characterization of the study population, and the variety of treatments being taken by patients on MDI-based insulin therapy. In conclusion, this study supports the use of tools such as CGM and glucose variability indicators in clinical practice—as recently suggested by other authors [32]—in addition to measuring HbA1c, which remains the reference parameter for the time being. Given the known problems of the cost/benefit ratio of using CGM [14], our results identify certain classes of patients who would benefit most from these devices, that is, DM1 patients and those with a longer history of diabetes. In these subgroups of patients, our results have shown that SD (which reflects glucose variability) describes glucose fluctuations more accurately than single HbA1c assessments. Although the specific role of glucose variability in the management of diabetes remains an important open question [33], it may be that the risk of chronic diabetic complications can be reduced by focusing clinical measures not only on reducing HbA1c levels, but also on a better control of a patients’ glucose profiles. Conflict of interest interests to disclose.

The authors have no competing financial

References 1. DCCT Research Group (1993) The effect of intensive treatment of diabetes on development and progression of long-term complications in insulin dependent diabetes mellitus. N Eng J Med 329:977–986 2. UK Prospective Diabetes Study group (1998) Intensive bloodglucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patient with type 2 diabetes. Lancet 352:837–853 3. Dailey G (2007) Assessing glycemic control with self-monitoring of blood glucose and hemoglobin A(1c) measurements. Mayo Clin Proc 82(2):229–235 4. Ceriello A, Ihnat MA (2010) ‘Glycaemic variability’: a new therapeutic challenge in diabetes and the critical care setting. Diabet Med 27(8):862–867 5. Ceriello A (2005) Postprandial hyperglycemia and diabetes complications. Is it time to treat? Diabetes 54:1–7 6. Kovatchev BP, Cox DJ, Farhy LS, Straume M, Gonder-Frederick L, Clarke WL (2000) Episodes of severe hypoglycemia in type 1 diabetes are preceded and followed within 48 hours by measurable disturbances in blood glucose. J Clin Endocrinol Metab 85:4287–4429 7. Kudva YC, Basu A, Jenkins GD, Pons GM, Vogelsang DA, Rizza RA, Smith SA, Isley WL (2007) Glycemic variation and hypoglycemia in patients with well-controlled type 1 diabetes on a multiple daily insulin injection program with use of glargine and ultralente as basal insulin. Endocr Pract 13:244–250

S159 8. Nathan DM, Kuenen J, Borg R, Zheng H, Schoenfeld D, Heine RJ, for the A1c-Derived Average Glucose (ADAG) Study Group (2008) Translating the A1C assay into estimated average glucose values. Diabetes Care 31(8):1473–1478 9. Siegelaar SE, Holleman F, Hoekstra JBL, DeVries JH (2010) Glucose variability, does it matter? Endocr Rev 31:171–182 10. Lee EY, Lee BW, Kim D, Lee YH, Kim KJ, Kang ES, Cha BS, Lee EJ, Lee HC (2011) Glycated albumin is a useful glycation index for monitoring fluctuating and poorly controlled type 2 diabetic patients. Acta Diabetol 48:167–172 11. Koga M, Saito H, Mukai M, Matsumoto S, Kasayama S (2010) Influence of iron metabolism indices on glycated haemoglobin but not glycated albumin levels in premenopausal women. Acta Diabetol 47:65–69 12. Koga M, Murai J, Saito H, Mukai M, Kasayama S (2010) Serum glycated albumin, but not glycated haemoglobin, is low in relation to glycemia in hyperuricemic men. Acta Diabetol 47:173–177 13. Koga M, Saito H, Mukai M, Otsuki M, Kasayama S (2009) Serum glycated albumin levels are influenced by smoking status, independent of plasma glucose levels. Acta Diabetol 46:141–144 14. Klonoff DC (2005) Continuous glucose monitoring: roadmap for 21st century diabetes therapy. Diabetes Care 28:1231–1239 15. Borg R, Kuenen JC, Carstensen B, Zheng H, Nathan D, Heine RJ, Nerup J, Borch-Johnsen K, Witte DR, the ADAG Study Group (2010) Associations between features of glucose exposure and HbA1c. The A1c-Derived Average Glucose (ADAG) study. Diabetes 59:1585–1590 16. Pickup JC, Kidd J, Burmiston S, Yemane N (2006) Determinants of glycaemic control in type 1 diabetes during intensified therapy with multiple daily insulin injections or continuous subcutaneous insulin infusion: importance of blood glucose variability. Diabetes Metab Res Rev 22:232–237 17. Greven WL, Beulens JW, Biesma DH, Faiz S, de Valk HW (2010) Glycemic variability in inadequately controlled type 1 diabetes and type 2 diabetes on intensive insulin therapy: a cross-sectional, observational study. Diabetes Technol Ther 12:695–699 18. Monnier L, Mas E, Ginet C, Michel F, Villon L, Cristol JP, Colette C (2006) Activation of oxidative stress by acute glucose fluctuations compared with sustained chronic hyperglycemia in patients with type 2 diabetes. JAMA 295(14):1681–1687 19. Kohnert KD, Augstein P, Heinke P, Zander E, Peterson K, Freyse EJ, Salzsieder E (2007) Chronic hyperglycemia but not glucose variability determines HbA1c levels in well-controlled patients with type 2 diabetes. Diabetes Res Clin Pract 77:420–426 20. Rodbard D (2009) New and improved methods to characterize glycemic variability using continuous glucose monitoring. Diabetes Technol Ther 11:551–565 21. Service FJ, Molnar GD, Rosevear JW, Ackerman E, Gatewood LC, Taylor WF (1970) Mean amplitude of glycemic excursions, a measure of diabetic instability. Diabetes 19:644–656 22. McDonnell CM, Donath SM, Vidmar SI, Werther GA, Cameron FJ (2005) A novel approach to continuous glucose analysis utilizing glycemic variation. Diabetes Technol Ther 7(2):253–263 23. Molnar GD, Taylor WF, Ho MM (1972) Day-to-day variation of continuously monitored glycaemia: a further measure of diabetic instability. Diabetologia 8(5):342–348 24. McCall AL, Cox DJ, Crean J, Gloster M, Kovatchev BP (2006) A novel analytical method for assessing glucose variability: using CGMS in type 1 diabetes mellitus. Diabetes Technol Ther 8:644–653 25. Kovatchev BP, Cox DJ, Gonder-Frederick LA, Clarke W (2006) Evaluation of a new measure of blood glucose variability in diabetes. Diabetes Care 29:2433–2438 26. Kovatchev BP, Cox DJ, Gonder-Frederick LA, Clarke WL (2002) Methods for quantifying and monitoring blood glucose profiles exemplified by an examination of blood glucose patterns in

123

S160 patients with type 1 and type 2 diabetes. Diabetes Technol Ther 4:295–303 27. Cox DJ, Kovatchev BP, Julian DM, Gonder-Frederick LA, Polonsky WH, Schlundt DG, Clarke WL (1994) Frequency of severe hypoglycemia in insulin-dependent diabetes mellitus can be predicted from self-monitoring blood glucose data. J Clin Endocrinol Metab 79:1659–1662 28. UK Hypoglycemia Study Group (2007) Risk of hypoglycaemia in types 1 and 2 diabetes: effects of treatment modalities and their duration. Diabetologia 50:1140–1147 29. Monnier L, Colette C, Mas E, Michel F, Cristol JP, Boegner C, Owens DR (2010) Regulation of oxidative stress by glycaemic control: evidence for an independent inhibitory effect of insulin therapy. Diabetologia 53:562–571

123

Acta Diabetol (2012) 49 (Suppl 1):S153–S160 30. Rossetti P, Porcellati F, Fanelli CG, Perriello G, Torlone E, Bolli GB (2008) Superiority of insulin analogues versus human insulin in the treatment of diabetes mellitus. Arch Physiol Biochem 114(1):3–10 31. Monnier L, Colette C, Owens DR (2008) Glycemic variability: the third component of the dysglycemia in diabetes. Is it important? How to measure it? J Diabetes Sci Technol 2:1094–1100 32. Nardacci EA, Bode BW, Hirsch IB (2010) Individualizing care for the many: the evolving role of professional continuous glucose monitoring systems in clinical practice. Diabetes Educ 36:4S–19S 33. Kilpatrick ES, Rigby AS, Atkin SL (2010) Glucose variability and diabetes complication risk: we need to know the answer. Diabet Med 27:868–871