Hemoglobin Concentrations Are Closely Linked to ...

Original Paper Kidney Blood Press Res 2008;31:313–321 DOI: 10.1159/000155230

Received: March 27, 2008 Accepted: July 27, 2008 Published online: September 15, 2008

Hemoglobin Concentrations Are Closely Linked to Renal Function in Patients with Type 1 or 2 Diabetes Mellitus Gunter Wolf Nicolle Müller Wilgard Hunger-Battefeld Christof Kloos Ulrich A. Müller Klinik für Innere Medizin III, Klinikum der Friedrich-Schiller-Universität, Jena, Germany

Key Words Diabetes mellitus, type 1 and 2 ⴢ Anemia ⴢ Diabetic nephropathy ⴢ Metabolic control

Abstract Background/Aims: It has been reported that anemia is more common in patients with diabetes mellitus, and that it occurs early in the disease process. Methods: In this study, we evaluated hemoglobin (Hb) values of patients with diabetes type 1 or 2 from a large collective receiving care from a tertiary center. A total of 751 patients with type 1 diabetes and 3,306 patients with type 2 were studied. Correlations were calculated for Hb with the following parameters: metabolic control (HbA1c and blood glucose), renal function [estimated glomerular filtration rate (eGFR), serum creatinine, albuminuria, proteinuria], blood leukocytes, duration of diabetes and use of ACE inhibitors/AT1-receptor antagonists. Results: 17% of patients with type 1 diabetes and 14% of those with type 2 had anemia [defined as an Hb !8.5 mmol/l (!13.68 g/dl) in men and !7.5 mmol/l (!12.07 g/dl) in women). There was a close positive correlation between Hb and eGFR, and a negative correlation with albuminuria and proteinuria. These close associations were also confirmed with linear regression analysis. A significant negative correlation was observed between serum creatinine levels and Hb. There was no negative correlation between actual Hb and mean HbA1c in the individual follow-up periods. No correlation was found

© 2008 S. Karger AG, Basel 1420–4096/08/0315–0313$24.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/kbr

between blood glucose (morning and postprandial blood glucose) and Hb. Blood leukocyte numbers, as a parameter of systemic inflammation, were not correlated with Hb. The use of ACE inhibitors/AT1-receptor antagonists had no adverse effect on Hb in our study cohort. Conclusion: No negative effects of metabolic control on Hb could be demonstrated in this study. Copyright © 2008 S. Karger AG, Basel

Introduction

Anemia is increasingly recognized as a complication of diabetes mellitus [1]. In the general population, but particularly in diabetic patients, anemia is an independent risk factor for cardiovascular mortality [2, 3]. Cardiac hypertrophy induced by chronic hypoxia in the presence of anemia is 1 major mechanism of this increased cardiovascular mortality [4]. In addition, anemia causes morbidity and adversely affects health by symptoms such as dyspnea, reduced capacity to exercise, fatigue and lack of energy. Importantly, anemia has also been linked to faster progression of diabetic nephropathy [5, 6]. Epidemiological studies suggest that anemia in patients with diabetes often remains unrecognized and undiagnosed [7, 8]. Many patients with diabetes suffer from diabetic nephropathy, and the reduced hemoglobin (Hb) values found in diabetics may be simply the consequence of reProf. Gunter Wolf University of Jena, Department of Medicine Erlanger Allee 101 DE–07740 Jena (Germany) Tel. +49 3641 932 4301, Fax +49 3641 932 4302, E-Mail [email protected]

Table 1. Characterization of patients with type 1 diabetes

Table 2. Characterization of patients with type 2 diabetes

Patients, n Male Female Age, years Duration of diabetes, years Follow-up, years BMI Adjusted HbA1c, % Blood pressure (systolic/diastolic) mm Hg Pulse, beats/min Albuminuria, mg/l Proteinuria, mg/l eGFR (MDRD formula), ml/min/1.73 m2 Insulin injections/day, n Insulin dose/day, U Received ACE inhibitors or AT1-receptor antagonists, %

Patients, n Male Female Age, years Duration of diabetes, years Follow-up, years BMI Adjusted HbA1c, % Blood pressure (systolic/diastolic) mm Hg Pulse, beats/min Albuminuria, mg/l Proteinuria, mg/l eGFR (MDRD formula), ml/min/1.73 m2 On insulin therapy, n (%) Insulin injections/day, n Insulin dose/day, U Received ACE inhibitors or AT1-receptor antagonists, %

751 371 380 43816 (8.8–94) 17812 (0.1–68) 5.485.7 (0–19) 25.984.4 (13.1–48.9) 8.381.8 (4.7–19.04) 131819/76810 (90–220/33–124) 77.2813.3 (48–122) 1068376 (0–3,614) 2208603 (0–5,427) 94.8833.2 (6.6–233.7) 3.981.8 (1–8) 47.9821.9 (4–152) 42.9

Data are means 8 standard deviations with ranges in parentheses, unless otherwise stated.

nal anemia due to erythropoietin deficiency [9]. However, smaller studies have provided evidence that anemia in patients with diabetic nephropathy occurs earlier than in patients with other renal diseases and it may even be independent of renal function [10–12]. Other investigations have found no direct evidence of anemia in diabetic patients independent of renal function [13, 14]. The present retrospective cohort study investigates the potential factors influencing Hb concentration in a large cohort of patients with type 1 or 2 diabetes who were treated at a single center. We found a close association of Hb with the decline of renal function but no evidence for an influence of metabolic control on the development of anemia.

Patients and Methods The main objective of this study was to identify potential factors associated with the development of anemia in patients with diabetes type 1 or 2. This retrospective cohort study was performed on a collective of patients with diabetes type 1 or 2 from a tertiary university hospital which had a large diabetes clinic. The hospital stores patient data in an electronic database (EMILTM). Data from 1989–2006 were analyzed. A total of 749 patients with type 1 (59% with 2 or more visits, mean follow-up 7.8 years) and 3,306 patients with type 2 diabetes (59% with 2 or more visits, mean follow-up 4.4 years) were investigated. Very few dialysis patients were part of this ambulatory collective and no patient received erythropoietin treatment. Albuminuria was deter-

314

Kidney Blood Press Res 2008;31:313–321

3,306 1,599 1,707 65812 (18.8–95.1) 1289 (0.1–66) 3.184.3 (0–35.7) 30.686.0 (15.9–80.8) 8.381.9 (2.8–20.7) 144822/79812 (76–260/39–140) 76813.9 (44–130) 1898628 (0–9,138) 3478861 (0–12,828) 84.1836.3 (3.3–757.4) 1,958 (59.2) 2.8581.1 (1–8) 59.4842.7 (2–380) 78.6

Data are means 8 standard deviations with ranges in parentheses, unless otherwise stated.

mined by nephelometry of the first morning urine samples and is expressed in mg/l according to the recommendations of the German Diabetes Society [15]. Estimated glomerular filtration rate (eGFR) was calculated according to the modified equation 7 MDRD formula [16]. Subgroup analyses were performed by grouping the study cohort according to the duration of diabetes (! 2 years, 2–5 years, 5–10 years and 110 years). The majority of the patients had no urinary sediment findings suggesting glomerulonephritis (e.g. hematuria, acanthocytes), but not all patients had received a recent urinary sediment analysis. HbA1c levels were analyzed by high-performance liquid chromatography. HbA1c values were adjusted to Diabetes Control and Complications Trail (DCCT) standards with an evaluated standardization procedure. HbA1c is registered as an absolute and relative value by dividing the absolute value by the mean of the local healthy subjects. DCCT-adjusted absolute HbA1c is then calculated by multiplying relative HbA1c values by 5.05% [17]. For each patient the mean HbA1c of all visits was used for calculation. Hb and blood leukocytes were measured with the Hitachi Multianalyzer 747 (Roche, Mannheim, Germany). Anemia was defined as an Hb !8.5 mmol/l (!13.68 g/dl) in men and !7.5 mmol/l (!12.07 g/dl) in women. Morning and postprandial blood glucose levels were determined with the glucose oxidase method. The complete drug therapy was classified by the Anatomical Therapeutic Chemical Classification (ATC) from 2003. Therefore, the analysis of antihypertensive drug treatment comprises only patients who last visited after January 1, 2003. Statistical Analyses For descriptive statistical analysis, means, standard deviations, and absolute and relative frequencies were calculated. Normal distribution was confirmed with the Kolmogorov-Smirnov-

Wolf/Müller/Hunger-Battefeld/Kloos/ Müller

Table 3. Correlations of type 1 diabetes (all patients)

Hb Hb HbA1c eGFR Albuminuria Proteinuria Leukocytes Duration of diabetes

– –0.013 0.353** –0.195** –0.154* –0.102** –0.133**

HbA1c

eGFR

Albuminuria

–0.013 – 0.015 –0.011 –0.062 0.068 –0.232**

0.353** 0.015 – –0.282** –0.220** –0.107** –0.371**

–0.195** 0.011 –0.282** – 0.964** 0.076 0.094*

Proteinuria –0.154* –0.062 –0.220** 0.964** – 0.153* 0.129

Leukocytes –0.102** 0.068 –0.107** 0.076 0.153* – 0.025

Duration of diabetes –0.133** –0.232** –0.71** 0.094* 0.129 0.025 –

* p < 0.05, ** p < 0.01, Pearson’s correlation coefficient, n = 751.

Characteristics of the Study Cohort Characteristics of the patients included in the study are shown in tables 1 and 2. As expected, patients with type 2 diabetes were older and had a greater body mass index than type 1 patients. Blood pressure was also higher in patients with type 2 diabetes. However, metabolic controls (HbA1c), eGFR and proteinuria were not significantly different between type 1 and 2 patients. 59.2% of the type 2 diabetics received insulin therapy.

Hemoglobin, mmol/l (g/dl)

Results

12 (19.3) 10 (16.1) 8 (12.8) 6 (9.6) 4 (6.4) 2 (3.2) 0

10

10

a 18 (28.9)


Lilliefors test. Statistical analysis was performed using the t test. Correlations were calculated with the Pearson test. p ! 0.05 was considered significant. Linear regression analysis was performed using the method of least squares. Statistical analysis was performed with SPSS 14.0 (SPSS, Chicago, Ill., USA) and SAS 8.02 (SAS Institute Inc., Cary, N.C., USA).

16 (25.7) 14 (22.5) 12 (19.3) 10 (16.1) 8 (12.8) 6 (9.6)

Factors Influencing Anemia in Patients with Type 1 Diabetes A total of 128 type 1 patients (17%) had anemia. 58 patients were female (15.3% of all female patients) and 70 were male (18.9% of all male patients). The mean Hb in all patients with type 1 was 8.63 8 1.04 mmol/l (13.89 8 1.67 g/dl). Since the duration of diabetes may significantly influence the development of anemia, patients were grouped according to the duration of the disease (duration !2 years, 2–5 years, 5–10 years and 110 years). The mean Hb values of patients grouped according to the duration of their disease are shown in figure 1a. The correlations between Hb, HbA1c, eGFR, albuminuria, proteinuria, duration of diabetes and leukocytes for all type 1 diabetics, irrespective of diabetes duration, are

shown in table 3. There was a significant positive correlation between Hb and eGFR, and a negative correlation between Hb, albuminuria, proteinuria, leukocytes and the duration of diabetes. Linear regression analysis revealed a highly significant correlation between eGFR

Anemia and Nephropathy


4 (6.4) 2 (3.2) 0

b

Fig. 1. Hb values in type 1 (a) and type 2 (b) diabetic patients

grouped according to the duration of their disease.

315

12 (19.3)


10 (16.1)

8 (12.8)

6 (9.6)

4 (6.4)

2 (3.2)

0 0

50

a

100 150 eGFR (ml/min/1.73 m2)

200

250


20 (32.2)

Fig. 2. Linear regression analysis for eGFR and Hb in type 1 (a) and type 2 (b) dia-

betic patients. There was a highly significant correlation between eGFR and Hb in both groups (type 1: Hb = 0.011 ! eGFR + 7.574, R2 = 0.145, p = 0.001; type 2: Hb = 0.012 ! eGFR + 7.56, R2 = 0.132, p = 0.001).

15 (24.1)

10 (16.1)

5 (8.0)

0 0

b

and Hb (fig. 2a). A significant correlation was also seen between serum creatinine and Hb in the whole collective of type 1 patients as well as in those patients with an eGFR 160 ml/min/1.74 m2 (all patients’ serum creatinine and Hb: r = –0.253, p ! 0.01; patients with an eGFR 160 ml/min/1.73 m2: r = –0.153, p ! 0.01). There was no significant correlation between Hb and HbA1c. In addition, no significant correlation was found between morning and postprandial blood glucose and Hb from the last visit (morning glucose and Hb: r = –0.008, not signifi316


50

100

150

200

250

300

eGFR (ml/min/1.73 m2)

cant; postprandial glucose and Hb: r = –0.037, not significant). When the patients were grouped according to the duration of diabetes, Hb was no longer correlated with eGFR, albuminuria or proteinuria in those patients with diabetes of less than 5 years’ duration (data not shown). The significant positive correlation between Hb and eGFR and the significant negative correlation between Hb and the duration of diabetes appeared in patients with 110 years of disease (table 4). A significant negative correlation between Hb and leukocyte numbers Wolf/Müller/Hunger-Battefeld/Kloos/ Müller

Table 4. Correlations of type 1 diabetes (duration of diabetes >10 years)


– 0.012 0.436** –0.219** –0.155 –0.132** –0.115*

HbA1c

eGFR

Albuminuria

0.012 – –0.061 0.075 –0.061 0.075 –0.102*

0.436* –0.061 – –0.306** –0.225* –0.111* –0.308*

–0.219* 0.075 –0.306** – 0.968* 0.098* –0.015

Proteinuria –0.155 –0.061 –0.225* 0.968** – 0.193* 0.018

Leukocytes –0.132** 0.075 –0.111* 0.098* 0.193* – 0.026

Duration of diabetes –0.115* –0.102* –0.308** –0.015 0.018 0.026 –

* p < 0.05, ** p < 0.01, Pearson’s correlation coefficient, n = 496.

Table 5. Correlations of type 2 diabetes (all patients)


– 0.048** 0.266** –0.091** –0.112** –0.009 –0.128**

HbA1c

eGFR

Albuminuria

0.048** – 0.008 0.015 –0.001 –0.001 0.007

0.266** 0.008 – –0.203** –0.196** –0.086** –0.309**

–0.091** 0.015 –0.203** – 0.827** 0.068** 0.112**

Proteinuria –0.112** –0.001 –0.196** 0.827** – 0.119** 0.063

Leukocytes –0.048** –0.001 –0.086** 0.068** 0.119** – –0.008

Duration of diabetes –0.138** 0.007 –0.309** 0.112** 0.063 –0.008 –

* p < 0.05, ** p < 0.01, Pearson’s correlation coefficient, n = 3,306.

was found in patients with type 1 diabetes for 110 years (table 4). There was no significant difference in Hb or leukocytes between the use of ACE inhibitors/AT1-receptor blocker and no use (Hb without ACE inhibitors/AT1-receptor blocker: 8.74 8 0.94 mmol/l (14.07 8 1.51 g/dl); with ACE inhibitors/AT1-receptor blocker: 8.60 8 1.17 mmol/l (13.84 8 1.88 g/dl), not significant; leukocytes without ACE inhibitors/AT1-receptor blocker: 7.13 8 2.13 GPT/l; with ACE inhibitors/AT1-receptor blocker: 7.29 8 2.19 GPT/l, not significant). Factors Influencing Anemia in Patients with Type 2 Diabetes Of all the patients with type 2 diabetes, 468 had anemia (14%). 174 patients with anemia were female (10.1% of all female patients) and 294 were male (18.2% of all male patients). The mean Hb of all anemic patients with type 2 diabetes was 9.00 8 3.06 mmol/l (14.49 8 4.92

g/dl). Figure 1a demonstrates the mean Hb values of patients with type 2 diabetes, grouped according to the disease duration. The correlation analysis of all patients with type 2 diabetes is shown in table 5. A significant positive correlation was found between Hb and eGFR as well as Hb and HbA1c. As shown in figure 2b, linear regression analysis revealed a highly significant association between eGFR and Hb. In addition, serum creatinine levels were significantly correlated with Hb in the whole collective of type 2 patients and also in those patients with an eGFR 160 ml/min/1.74 m2 (all patients’ serum creatinine and Hb: r = –0.206, p ! 0.01; patients with an eGFR 160 ml/min/ 1.73 m2: r = –0.055, p ! 0.01). A significant negative correlation existed between Hb, albuminuria, proteinuria and duration of diabetes and leukocytes (table 5). In contrast to the findings in type 1 diabetics, these correlations in patients with type 2 diabetes were already present even when the duration of disease was !2 years (table 6). These

Anemia and Nephropathy


317

Table 6. Correlations of type 2 diabetes (duration of diabetes