- Email: [email protected]
Reduced erythropoietin responsiveness to anemia in diabetic patients before advanced diabetic nephropathy
Diabetes Research and Clinical Practice 46 (1999) 223 – 229 www.elsevier.com/locate/diabres
Reduced erythropoietin responsiveness to anemia in diabetic patients before advanced diabetic nephropathy Yong Seok Yun a, Hyun Chul Lee a,*, Nae Chun Yoo b, Young Duk Song a, Sung Kil Lim a, Kyung Rae Kim a, Jee Sook Hahn b, Kap Bum Huh a a
Department of Endocrinology and Metabolism, Diabetes Center, Yonsei Uni6ersity College of Medicine, Seodaemun-Gu, Shinchon-Dong 134, Seoul 120 -752, South Korea b Department of Hematology and Oncology, Se6erance Hospital, Yonsei Uni6ersity College of Medicine, Seoul, South Korea Received 2 December 1998; received in revised form 16 July 1999; accepted 29 July 1999
Abstract We often encounter diabetic patients with anemia in whom the causes of anemia were not clearly identified despite differential hematologic studies. We therefore studied the clinical and biochemical characteristics of diabetic patients with anemia of uncertain cause and measured erythropoietin (Epo) concentrations in 35 diabetic subjects without significant diabetic renal disease. Among 62 medical records of diabetic patients with anemia, showing no evidence of advanced diabetic nephropathy (creatinine clearance ] 30 mg/kg/1.73 m2), the causes of the anemia were not able to be identified in 28 cases (45.2%). In addition, we enrolled 35 diabetic patients with uncertain causes of anemia in order to evaluate the serum Epo responsiveness to anemia, and compared levels to a group of non-diabetic subjects also with anemia. The serum Epo concentrations of diabetic patients (17.6 98.1 mIU/ml) were significantly lower than those of non-diabetic patients with similar degree of decrease in hemoglobin concentrations (144.9 9 108.0 mIU/ml, PB 0.001). The hemoglobin concentrations of diabetic patients correlated with creatinine clearance (r= 0.34, P =0.03), serum creatinine (r= − 0.49, P=0.003) and albumin excretion rate (r= − 0.44, P=0.009), but showed no relation to age, duration of diabetes, glycated hemoglobin, presence of retinopathy or neuropathy. We concluded that reduced Epo responsiveness to anemia could explain the anemia present in diabetic patient but without advanced diabetic nephropathy. This may reflect early renal interstitial damage. © 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Anemia; Diabetes mellitus; Erythropoietin
1. Introduction
* Corresponding author. Tel.: +82-2-361-5425; fax: +822-393-6884. E-mail address: [email protected] (H.C. Lee)
In diabetes mellitus, several changes can be found in the hematologic system, such as hypercoagulability, a decrease of red blood cell deformability or impairment of phagocytosis and
0168-8227/99/$ - see front matter © 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 8 - 8 2 2 7 ( 9 9 ) 0 0 0 9 7 - 2
224
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229
leukocyte chemotaxis [1]. Anemia is also a frequently noted disorder of the red blood cell system in diabetes. The causes of anemia in diabetes include advanced renal damage and uremia by progression of diabetic nephropathy or a coexisting condition not associated with diabetes such as iron deficiency, chronic inflammatory disease, malignancy or chronic infection. However, we have often encountered cases with anemia, the causes of which were not clearly explained despite routine hematologic studies. We speculated that there could be an anemia associated with diabetes [2]. Recently, a few reports have suggested that a relative deficiency of erythropoietin (Epo) may develop in diabetes without any evidence of renal damage and that the administration of synthetic Epo could be useful in treating this type of anemia [3]. Therefore, we have studied the clinical and biochemical characteristics of diabetic patients with anemia of uncertain cause and measured Epo concentrations, comparing these to non-diabetic subjects with similar degree of anemia.
2. Materials and methods
2.1. Two patient groups were studied We retrospectively reviewed the medical records of 192 non-uremic non-insulin-dependent diabetes mellitus (NIDDM) patients with anemia who had been followed up by the Diabetes Center at Yonsei University College of Medicine from January 1991 to December 1995. Among them, 62 cases (32%) had allowed determination of the cause of the anemia. Cases with a history of liver disease, malignancy or any features of advanced nephropathy such as creatinine clearance less than 30 ml/min/1.73 m2, were excluded from this analysis. WHO criteria for anemia were used: hemoglobin (Hb) 5 13 g/dl in males and 5 12 g/dl in females. In order to evaluate Epo concentrations in those with anemia of uncertain cause, we recruited further 37 diabetic patients meeting this clinical diagnosis. Twenty-eight were enrolled from the database of patients involved in prior
studies and nine were selected from patients admitted to diabetic clinics from January 1996 to December 1996. Two patients recruited from earlier studies were excluded due to progression of diabetic nephropathy. Therefore, 35 patients were included in the Epo analysis. All subjects underwent formal hematologic studies again to evaluate the causes of anemia. Anyone with a medical history of hypertension, advanced nephropathy, liver disease, malignancy, chronic infection or inflammatory disorder as well as drug use such as angiotensin-converting enzyme inhibitors, sex hormones or iron were excluded from the study. Twenty subjects with anemia but with normal glucose tolerance were enrolled as a control group. In control subjects, the causes of anemia were iron deficiency and bone marrow failure due to chronic leukemia or aplasia. The hemoglobin concentrations (mean 9 SD) of the control group (9.29 0.8 g/dl) were similar to the diabetic group (9.89 1.2 g/dl).
2.2. E6aluation of anemia The following laboratory studies were performed in all patients from both groups: (1) complete blood count using Sysmex SE 9000 automatic counter; (2) iron studies of serum iron, total iron-binding capacity (TIBC) by colorimetry and serum ferritin by nephelometry (Array protein system, Beckman, USA); (3) vitamin B12 and folate concentrations by dual RIA kit (OrthoClinical Diagnostics, Amersham, UK) utilizing a competitive radioassay technique (normal range: vitamin B12: 180710 pg/ml, folate: ]2.2 ng/m); (4) reticulocyte count by automated flow cytometry (Sysmex R-3000, Japan) and supravital staining; (5) stool examination to exclude gastrointestinal bleeding. Coombs test, serum haptoglobulin, peripheral blood smear or bone marrow studies were also performed, if indicated, to exclude hemolytic anemia or aplastic anemia. Transferrin saturation was calculated from serum iron and TIBC. Renal function was determined by creatinine clearance using a 24-h urine collection. Creatinine concentrations were measured by Jaffee reaction (Hitachi 747 automatic analyzer, Japan) and 24-h
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229
urinary albumin excretion was measured by nephelometry (Boehring nephelometer analyser II, Germany). Retinopathy and nephropathy were Table 1 Clinical characteristics of diabetic subjects with anemia in 62 medical records Age (years) Gender (female) Body mass index (kg/m2) Duration of diabetes (years) Glycated hemoglobin (%) Hemoglobin (g/dl) Hematocrit (%)
57.59 11.2 27 (43.5%) 30.09 3.0 10.2 97.4 13.3 94.3 10.791.4 31.8 9 4.4
Red blood cell indices Mean corpuscular volume (mm3) Mean corpuscular hemoglobin (pg/cell) Mean corpuscular hemoglobin concentration (g/dl) Absolute percentage of reticulocytes (%) Serum creatinine concentrations (mg/dl) Creatinine clearance (ml/min/1.73 m2) Microvasular complication (%) Retinopathy Neuropathy Nephropathya a
91.89 9.1 30.9 93.6 33.693.1 1.1 91.5 0.9 9 0.3 74.89 27.0 25/62 (40.3%) 37/62 (59.7%) 25/62 (40.3%)
Nephropathy: 24-h urine albumin more than 30 mg.
Table 2 The causes of anemia in 62 diabetic patientsa Male (n =20)
Female (n= 42)
Total (n= 62)
Causes not identified
9 (14.5)
19 (30.7)
28 (45.2)
Causes identified Iron deficiency Chronic disorder Folate deficiency Vitamin B12 deficiency Hemolytic anemia Aplastic anemia
11 (17.7)
23 (37.1)
34 (54.8)
2 (3.2)
13 (21.0)
15 (24.2)
4 (6.5)
9 (14.5)
13 (21.0)
2 (3.2)
1 (1.6)
3 (4.8)
1 (1.6)
0 (0.0)
1 (1.6)
1 (1.6)
0 (0.0)
1 (1.6)
1 (1.6)
0 (0.0)
1 (1.6)
a
Data are presented as numbers (% in parentheses).
225
evaluated by physical examination, fundus photography and nerve conduction velocity. Cases, in whom cause was not identified despite the above studies, were categorized as having anemia of uncertain cause.
2.3. Serum erythropoietin measurement Serum Epo concentrations were measured by radio-immunoassay with an Epo-Tranc™ RIA kit (23200, INCSTAR Corporation Stillwater, MN, USA). Serum samples with a minimum of 400 ml were collected and stored at −20°C in a non-selfdefrosting freezer. A polyclonal goat anti-human Epo antibody directed against recombinant human Epo (rhEpo) was employed as a primary antibody. Samples were tested in duplicate.
2.4. Statistical analysis SPSS/PC + software was used for statistical analysis. Results were expressed as mean9 S.D. Spearman’s rank correlation analysis was done to access the association between hemoglobin concentrations and selected variables. Difference in serum Epo concentrations between diabetes and controls were compared using the Mann–Whitney test. The level of significance was PB 0.05.
3. Results The clinical characteristics of the 62 patients with anemia are shown in Table 1. All patients included in this study were Korean. The causes of anemia in 28 cases (45.2%) were not clearly identified despite differential hematologic studies. Iron deficiency and chronic disorders were the most common findings of identified causes (Table 2). Of those with chronic disorders, there were ten cases with tuberculosis, two with rheumatoid arthritis and one with systemic lupus erythematosus. The clinical characteristics of the newly recruited patients with anemia of uncertain cause are shown in Table 3. A mild decrease of hemoglobin concentrations with a reduced absolute percentage of reticulocytes was shown. Red blood cell indices were normocytic in 30 subjects
226
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229
Table 3 The clinical characteristics of patients with anemia of uncertain causes and non-diabetic control group Diabetic group (n= 35) Sex (female) (%) 23 (65.7) Age (years) 57.4913.4 Hemoglobin (g/dl) 9.891.1 Transferrin 0.2190.11 saturationa Serum ferritin (ng/ml) 198.59 122.9 Erythrocyte indices Mean red cell 87.39 5.5 volume (fl) Mean red cell 29.09 2.0 hemoglobin (pg) Mean red cell 33.491.6 concentration (%) Absolute percentage 0.51 90.4 of reticulocytes (%) Serum erythropoietin 17.69 8.1 (mIU/ml) Serum creatinine 1.090.1 (mg/dl) Creatinine 58.8923.0 clearance (ml/min/1.73 m2) Duration of diabetes 13.49 7.6 (years) Glycated Hb (%) 12.79 2.6 Microvascular complication (%) Retinopathy 28 (80.0) Neuropathy 29 (82.9) Nephropathyb 28 (80.0)
Control group (n = 20) 4 (20)e 39.89 11.7e 9.290.8 0.50 9 0.21e 176.1 9 31.8
tence of retinopathy or nephropathy (Table 4). All patients had more than two of the stigmata of microvascular complications. The Epo concentrations of diabetic patients were 17.69 8.1 mIU/ml (range: 9.0–40.1 mIU/ ml), which were within or slightly above the normal range (9.1–30.8 mIU/ml). However, the Epo concentrations of non-diabetic subjects, also with anemia were 144.9 9 108.0 mIU/ml, were significantly higher (PB 0.001) compared to the subjects with diabetes (Fig. 1).
93.1 9 6.1d 30.1 91.5c 34.7 91.7d 2.23 90.73e
Table 4 Relationships between hemoglobin concentrations and clinical factors in patients with anemia of uncertain cause Anemia of uncertain cause (n = 35)
144.9 9108.0e 0.8 9 0.2d 103.5920.5
e
– –
– – –
Age (years) Duration of diabetes (years) Glycated hemoglobin (%) Serum creatinine (mg/dl) Creatinine clearance (ml/min/1.73 m2) Albumin excretion rate (mg/24 h urine)
Correlation coefficient (r)
P-value
0.01 0.20 0.22 −0.49 0.34
0.95 0.26 0.21 0.003 0.03
−0.44
0.009
a
Transferrin saturation: [serum-Fe (mg/dl)]/[total iron binding capacity (mg/dl)]. b Nephropathy: 24-h urine albumin more than 30 mg. c PB0.05. d PB0.01. e PB0.001.
(85.7%) and microcytic in the remaining five patients. Transferrin saturation ratios of diabetic patients were lower than controls, but serum ferritin concentrations were similar between two groups. Hemoglobin concentrations correlated with creatinine clearance (r =0.34, P = 0.03), serum creatinine (r = −0.49, P =0.003) and albumin excretion rate (r = −0.44, P = 0.009). However, no relationship was found with age, duration of diabetes, glycated hemoglobin, exis-
Fig. 1. Serum erythropoietin and hemoglobin concentrations of diabetic patients with anemia of uncertain cause ( ) and non-diabetic patients with a similar degree of anemia ().
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229
4. Discussion Causes of anemia are variable, ranging from relatively benign conditions such as dietary iron deficiency to serious chronic inflammatory disease or malignancy. Careful examination is essential to identify the underlying cause. Though sufficient investigations were performed in only 62 of 192 patients identified through medical records, we were able to verify a considerable number of cases in which the causes of anemia were not clearly determined despite hematologic studies. The common features were normocytic red blood cell indices with a relatively low absolute percentage of reticulocytes [4], accompanied by diabetic microvascular complications. These findings led us to speculate that impairment of erythropoiesis might be responsible. As a preliminary study, we examined the relationship between hemoglobin concentrations and clinical features of non-uremic diabetic patients in 192 medical records. The hemoglobin concentrations were associated with duration of diabetes, creatinine clearance and albumin excretion rate (data not shown). These 192 cases, although retrospectively reviewed, not randomly selected and heterogeneous in causes of anemia, suggested some relationship between anemia and renal damage. In general, renal anemia is not suspected when creatinine clearance exceeds 30 ml/min/1.73 m2, except in the setting of acute tubular necrosis, crescentic glomerular nephritis or tubulointerstitial disease [5]. The main anatomical focus of the lesion of diabetic nephropathy has been the glomerulus. Therefore, the attenuation of renal Epo responsiveness associated with renal interstitial damage has not been considered as a cause of anemia of diabetic patients prior to progression into advanced diabetic nephropathy [6]. However, in this study, of 35 anemic patients with well preserved excretory renal function, hemoglobin concentrations were related to the biochemical parameters of renal damage, albumin excretion rate, creatinine clearance and serum creatinine concentrations. Serum concentrations of Epo are regulated by the degree of oxygen demand. Values vary from 10 to 12 mIU/ml in non-anemic states to 100 –
227
1000 mIU/ml in severe anemia [7]. We have shown that the serum Epo concentrations in diabetic patients with anemia of uncertain cause were significantly lower than those of non-diabetic subjects with anemia. Values were mainly within the normal range and failed to show the expected increase in response to the anemia. These results suggest that relatively low Epo responsiveness may be present even at a very early stage of diabetic nephropathy and could be one of the reasons for anemia of uncertain cause in diabetic patients. The prevalence of renal anemia in diabetic patients without advanced diabetic nephropathy has probably been underestimated. Recently, reports have indicated that focal or general tubulointerstitial lesions may be frequently found in diabetic patients with microalbuminuria and normal or near normal renal function [8]. These findings support our suggestion that tubulointerstitial damage and reduced renal Epo responsiveness could be possible before advanced diabetic nephropathy. Inomato et al. also reported that serum Epo concentrations could provide clues of the presence of renal tubulointerstitial lesions and a potential for acceleration of renal damage [9]. Therefore strict attention should be given to renal function in diabetic patients with anemia of uncertain cause. However, our results also raise question about serum Epo responsiveness. In anemia associated with end-stage renal disease, serum Epo concentrations are usually above the normal range reflecting partial compensation for anemia [10]. However, in this study, values were almost within the normal range and failed to show the compensatory response. These findings suggest that a blunted Epo response could be related, not only to renal parenchymal damage, but also to impairment of regulation of Epo responsiveness in diabetes mellitus. Kojima et al. proposed the possibility that a reduced Epo response could appear as a manifestation of diabetic autonomic neuropathy, describing two cases with autonomic neuropathy, but without evidence of renal damage [3]. Peripheral neuropathy was documented in 29 of the 35 patients in this study, although autonomic nerve function was not directly tested. However, patients with severe diabetic neuropa-
228
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229
thy frequently also exhibit some degree of diabetic nephropathy. A relative deficiency of Epo to tissue oxygen demand can also be found in other types of anemia in which renal damage is absent, for example the anemia of chronic disorders (ACD) [11,12]. The mechanism in this condition (ACD) is inhibition of Epo production and Epo action by certain cytokines such as tumor necrosis factor or interleukin 1 associated with chronic inflammation, infection or malignancy [13]. It is not known whether similar cytokine-mediated effects could also occur in diabetes mellitus. It also needs to be considered that age is associated with blunted Epo responsiveness, usually in those exceeding 70 years of age [14,15]. Some of the subjects in our study were within this age group. Though transferrin saturation ratios of diabetic patients were lower than controls, serum ferritin concentrations were similar between two groups. Therefore, iron profile of anemia of uncertain causes is not suitable for iron deficiency anemia. Besides renal Epo responsiveness, the Epo action on erythropoiesis could be altered by the metabolic environment associated with diabetes. Hyperglycemia influences the action of Epo in diabetic patients. Ritchey et al. demonstrated that improved diabetic control by continuous insulin infusion enhanced the action of Epo in the bone marrow, which augmented the proliferation of erythroid stem cells [16]. Thus it is considered that synthetic Epo replacement therapy should be allied with measures to improve glycemic control. Epo concentrations of non-diabetic patients in this study showed a bimodal pattern without any relation to hemoglobin concentrations (Fig. 1). Some patients show a relatively smaller response to anemia than others. We can not exclude the possibility that diet composition could have affected the Epo responsiveness in our subjects. In Korea, a large proportion of diabetic patients are vegetarian, and consume a lower amount of protein than Western patients [17], which could curtail Epo responsiveness. In summary, we have investigated a number of cases of anemia of uncertain cause, among patients with diabetes before advanced diabetic nephropathy. Though the reason for this anemia
remains unknown, it could be closely related to early change in renal function with inappropriately low levels of Epo.
References [1] L.J. Robert, M.P. Charles, Hematologic alterations in diabetes mellitus, Am. J. Med. 70 (1981) 339 – 352. [2] C.M. Peterson, R.L. Jones, R.J. Koenig, E.T. Melvin, M.L. Lehrman, Reversible hematologic sequelae of diabetes mellitus, Ann. Intern. Med. 86 (1977) 425 – 429. [3] K. Kojima, Y. Totsuka, Anemia due to reduced serum erythropoietin concentration in non-uremic diabetic patients, Diabetes Res. Clin. Pract. 27 (1995) 229 – 233. [4] E.W. Friedman, Reticulocyte counts: how to use them, what they mean, Diagn. Med. 7 (1984) 29 – 35. [5] A.J. Erslew, J. Wilson, J. Caro, Erythropoietin titers in anemic, non-uremic patients, J. Lab. Clin. Med. 109 (1987) 429 – 433. [6] P.H. Maxwell, M.K. Osmond, C.W. Pugh, A. Heryet, L.G. Nicholls, G.C. Tan, B.G. Doe, D.J.P. Ferguson, M.H. Johnson, P.J. Ratcliffe, Identification of the renal erythropoietin-producing cells using transgenic mice, Kidney. Int. 44 (1993) 1149 – 1162. [7] S.J. Schwster, J.H. Wilson, A.J. Ersler, J. Coro, Physiologic regulation and tissue localization of renal erythropoietin messenger RNA, Blood 70 (1987) 316 – 318. [8] P. Fioretto, M. Mauer, E. Brocco, M. Velussi, F. Frigato, B. Muollo, M. Sambataro, C. Abaterusso, B. Baggio, G. Crepaldi, R. Nosadini, Patterns of renal injury in NIDDM patients with microalbuminuria, Diabetologia 39 (1996) 1569 – 1576. [9] S. Inomata, M. Itoh, H. Imai, T. Sato, Serum levels of erythropoietin as a novel marker reflecting the severity of diabetic nephropathy, Nephron 75 (1996) 426 – 430. [10] W.R. Heinz, A. Claussner, P.M. Erbes, E.H. Scheuermann, W. Schoeppe, K.M. Koch, Serum erythropoietin concentration in chronic renal failure: relationship to degree of anemia and excretory renal fuction, Blood 54 (1979) 877 – 884. [11] G.E. Cartwright, The anemia of chronic disorders, Semin. Hematol. 3 (1996) 3351 – 3375. [12] H.K. Boyd, T.R.J. Lappin, Erythropoietin deficiency in the anemia of chronic disorder, Eur. J. Hematol. 46 (1991) 198 – 201. [13] J. Camacho, F. Arnalich, A.F. Zamorano, J.J. Vazauez, Serum erythropoietin levels in the anemia of chronic disorders, J. Intern. Med. 229 (1991) 49 – 54. [14] J. Nafziger, K. Pailla, L. Luciani, J.P. Andreux, O.S. Jean, N. Casaderall, Decreased erythropoietin response to iron deficiency anemia in the elderly, Am. J. Hematol. 43 (1993) 172 – 176. [15] J.S. Power, B.K. Scanford, J.C. Collins, K. Meurer, A. Failinger, T. Buchholz, M. Blank, J. Spivak, M. Hochberg, A. Baer, M. Cotes, E. Goidwasser, Erythropoietin
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229 response to anemia as a function of age, J. Am. Geriatr. Soc. 39 (1991) 30 – 32. [16] A.K. Ritchey, W.V. Tamborlane, J. Gertner, Improved diabetic control enhances erythroid stem cell proliferation
.
229
in vitro, J. Clin. Endocrinol. Metab. 60 (1984) 1257 – 1260. [17] Kap Bum Huh, Malnutrition related diabetes mellitus. J. Korean Med. Assoc. 30 (1987) 774 – 750.
Reduced erythropoietin responsiveness to anemia in diabetic patients before advanced diabetic nephropathy Yong Seok Yun a, Hyun Chul Lee a,*, Nae Chun Yoo b, Young Duk Song a, Sung Kil Lim a, Kyung Rae Kim a, Jee Sook Hahn b, Kap Bum Huh a a
Department of Endocrinology and Metabolism, Diabetes Center, Yonsei Uni6ersity College of Medicine, Seodaemun-Gu, Shinchon-Dong 134, Seoul 120 -752, South Korea b Department of Hematology and Oncology, Se6erance Hospital, Yonsei Uni6ersity College of Medicine, Seoul, South Korea Received 2 December 1998; received in revised form 16 July 1999; accepted 29 July 1999
Abstract We often encounter diabetic patients with anemia in whom the causes of anemia were not clearly identified despite differential hematologic studies. We therefore studied the clinical and biochemical characteristics of diabetic patients with anemia of uncertain cause and measured erythropoietin (Epo) concentrations in 35 diabetic subjects without significant diabetic renal disease. Among 62 medical records of diabetic patients with anemia, showing no evidence of advanced diabetic nephropathy (creatinine clearance ] 30 mg/kg/1.73 m2), the causes of the anemia were not able to be identified in 28 cases (45.2%). In addition, we enrolled 35 diabetic patients with uncertain causes of anemia in order to evaluate the serum Epo responsiveness to anemia, and compared levels to a group of non-diabetic subjects also with anemia. The serum Epo concentrations of diabetic patients (17.6 98.1 mIU/ml) were significantly lower than those of non-diabetic patients with similar degree of decrease in hemoglobin concentrations (144.9 9 108.0 mIU/ml, PB 0.001). The hemoglobin concentrations of diabetic patients correlated with creatinine clearance (r= 0.34, P =0.03), serum creatinine (r= − 0.49, P=0.003) and albumin excretion rate (r= − 0.44, P=0.009), but showed no relation to age, duration of diabetes, glycated hemoglobin, presence of retinopathy or neuropathy. We concluded that reduced Epo responsiveness to anemia could explain the anemia present in diabetic patient but without advanced diabetic nephropathy. This may reflect early renal interstitial damage. © 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Anemia; Diabetes mellitus; Erythropoietin
1. Introduction
* Corresponding author. Tel.: +82-2-361-5425; fax: +822-393-6884. E-mail address: [email protected] (H.C. Lee)
In diabetes mellitus, several changes can be found in the hematologic system, such as hypercoagulability, a decrease of red blood cell deformability or impairment of phagocytosis and
0168-8227/99/$ - see front matter © 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 8 - 8 2 2 7 ( 9 9 ) 0 0 0 9 7 - 2
224
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229
leukocyte chemotaxis [1]. Anemia is also a frequently noted disorder of the red blood cell system in diabetes. The causes of anemia in diabetes include advanced renal damage and uremia by progression of diabetic nephropathy or a coexisting condition not associated with diabetes such as iron deficiency, chronic inflammatory disease, malignancy or chronic infection. However, we have often encountered cases with anemia, the causes of which were not clearly explained despite routine hematologic studies. We speculated that there could be an anemia associated with diabetes [2]. Recently, a few reports have suggested that a relative deficiency of erythropoietin (Epo) may develop in diabetes without any evidence of renal damage and that the administration of synthetic Epo could be useful in treating this type of anemia [3]. Therefore, we have studied the clinical and biochemical characteristics of diabetic patients with anemia of uncertain cause and measured Epo concentrations, comparing these to non-diabetic subjects with similar degree of anemia.
2. Materials and methods
2.1. Two patient groups were studied We retrospectively reviewed the medical records of 192 non-uremic non-insulin-dependent diabetes mellitus (NIDDM) patients with anemia who had been followed up by the Diabetes Center at Yonsei University College of Medicine from January 1991 to December 1995. Among them, 62 cases (32%) had allowed determination of the cause of the anemia. Cases with a history of liver disease, malignancy or any features of advanced nephropathy such as creatinine clearance less than 30 ml/min/1.73 m2, were excluded from this analysis. WHO criteria for anemia were used: hemoglobin (Hb) 5 13 g/dl in males and 5 12 g/dl in females. In order to evaluate Epo concentrations in those with anemia of uncertain cause, we recruited further 37 diabetic patients meeting this clinical diagnosis. Twenty-eight were enrolled from the database of patients involved in prior
studies and nine were selected from patients admitted to diabetic clinics from January 1996 to December 1996. Two patients recruited from earlier studies were excluded due to progression of diabetic nephropathy. Therefore, 35 patients were included in the Epo analysis. All subjects underwent formal hematologic studies again to evaluate the causes of anemia. Anyone with a medical history of hypertension, advanced nephropathy, liver disease, malignancy, chronic infection or inflammatory disorder as well as drug use such as angiotensin-converting enzyme inhibitors, sex hormones or iron were excluded from the study. Twenty subjects with anemia but with normal glucose tolerance were enrolled as a control group. In control subjects, the causes of anemia were iron deficiency and bone marrow failure due to chronic leukemia or aplasia. The hemoglobin concentrations (mean 9 SD) of the control group (9.29 0.8 g/dl) were similar to the diabetic group (9.89 1.2 g/dl).
2.2. E6aluation of anemia The following laboratory studies were performed in all patients from both groups: (1) complete blood count using Sysmex SE 9000 automatic counter; (2) iron studies of serum iron, total iron-binding capacity (TIBC) by colorimetry and serum ferritin by nephelometry (Array protein system, Beckman, USA); (3) vitamin B12 and folate concentrations by dual RIA kit (OrthoClinical Diagnostics, Amersham, UK) utilizing a competitive radioassay technique (normal range: vitamin B12: 180710 pg/ml, folate: ]2.2 ng/m); (4) reticulocyte count by automated flow cytometry (Sysmex R-3000, Japan) and supravital staining; (5) stool examination to exclude gastrointestinal bleeding. Coombs test, serum haptoglobulin, peripheral blood smear or bone marrow studies were also performed, if indicated, to exclude hemolytic anemia or aplastic anemia. Transferrin saturation was calculated from serum iron and TIBC. Renal function was determined by creatinine clearance using a 24-h urine collection. Creatinine concentrations were measured by Jaffee reaction (Hitachi 747 automatic analyzer, Japan) and 24-h
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229
urinary albumin excretion was measured by nephelometry (Boehring nephelometer analyser II, Germany). Retinopathy and nephropathy were Table 1 Clinical characteristics of diabetic subjects with anemia in 62 medical records Age (years) Gender (female) Body mass index (kg/m2) Duration of diabetes (years) Glycated hemoglobin (%) Hemoglobin (g/dl) Hematocrit (%)
57.59 11.2 27 (43.5%) 30.09 3.0 10.2 97.4 13.3 94.3 10.791.4 31.8 9 4.4
Red blood cell indices Mean corpuscular volume (mm3) Mean corpuscular hemoglobin (pg/cell) Mean corpuscular hemoglobin concentration (g/dl) Absolute percentage of reticulocytes (%) Serum creatinine concentrations (mg/dl) Creatinine clearance (ml/min/1.73 m2) Microvasular complication (%) Retinopathy Neuropathy Nephropathya a
91.89 9.1 30.9 93.6 33.693.1 1.1 91.5 0.9 9 0.3 74.89 27.0 25/62 (40.3%) 37/62 (59.7%) 25/62 (40.3%)
Nephropathy: 24-h urine albumin more than 30 mg.
Table 2 The causes of anemia in 62 diabetic patientsa Male (n =20)
Female (n= 42)
Total (n= 62)
Causes not identified
9 (14.5)
19 (30.7)
28 (45.2)
Causes identified Iron deficiency Chronic disorder Folate deficiency Vitamin B12 deficiency Hemolytic anemia Aplastic anemia
11 (17.7)
23 (37.1)
34 (54.8)
2 (3.2)
13 (21.0)
15 (24.2)
4 (6.5)
9 (14.5)
13 (21.0)
2 (3.2)
1 (1.6)
3 (4.8)
1 (1.6)
0 (0.0)
1 (1.6)
1 (1.6)
0 (0.0)
1 (1.6)
1 (1.6)
0 (0.0)
1 (1.6)
a
Data are presented as numbers (% in parentheses).
225
evaluated by physical examination, fundus photography and nerve conduction velocity. Cases, in whom cause was not identified despite the above studies, were categorized as having anemia of uncertain cause.
2.3. Serum erythropoietin measurement Serum Epo concentrations were measured by radio-immunoassay with an Epo-Tranc™ RIA kit (23200, INCSTAR Corporation Stillwater, MN, USA). Serum samples with a minimum of 400 ml were collected and stored at −20°C in a non-selfdefrosting freezer. A polyclonal goat anti-human Epo antibody directed against recombinant human Epo (rhEpo) was employed as a primary antibody. Samples were tested in duplicate.
2.4. Statistical analysis SPSS/PC + software was used for statistical analysis. Results were expressed as mean9 S.D. Spearman’s rank correlation analysis was done to access the association between hemoglobin concentrations and selected variables. Difference in serum Epo concentrations between diabetes and controls were compared using the Mann–Whitney test. The level of significance was PB 0.05.
3. Results The clinical characteristics of the 62 patients with anemia are shown in Table 1. All patients included in this study were Korean. The causes of anemia in 28 cases (45.2%) were not clearly identified despite differential hematologic studies. Iron deficiency and chronic disorders were the most common findings of identified causes (Table 2). Of those with chronic disorders, there were ten cases with tuberculosis, two with rheumatoid arthritis and one with systemic lupus erythematosus. The clinical characteristics of the newly recruited patients with anemia of uncertain cause are shown in Table 3. A mild decrease of hemoglobin concentrations with a reduced absolute percentage of reticulocytes was shown. Red blood cell indices were normocytic in 30 subjects
226
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229
Table 3 The clinical characteristics of patients with anemia of uncertain causes and non-diabetic control group Diabetic group (n= 35) Sex (female) (%) 23 (65.7) Age (years) 57.4913.4 Hemoglobin (g/dl) 9.891.1 Transferrin 0.2190.11 saturationa Serum ferritin (ng/ml) 198.59 122.9 Erythrocyte indices Mean red cell 87.39 5.5 volume (fl) Mean red cell 29.09 2.0 hemoglobin (pg) Mean red cell 33.491.6 concentration (%) Absolute percentage 0.51 90.4 of reticulocytes (%) Serum erythropoietin 17.69 8.1 (mIU/ml) Serum creatinine 1.090.1 (mg/dl) Creatinine 58.8923.0 clearance (ml/min/1.73 m2) Duration of diabetes 13.49 7.6 (years) Glycated Hb (%) 12.79 2.6 Microvascular complication (%) Retinopathy 28 (80.0) Neuropathy 29 (82.9) Nephropathyb 28 (80.0)
Control group (n = 20) 4 (20)e 39.89 11.7e 9.290.8 0.50 9 0.21e 176.1 9 31.8
tence of retinopathy or nephropathy (Table 4). All patients had more than two of the stigmata of microvascular complications. The Epo concentrations of diabetic patients were 17.69 8.1 mIU/ml (range: 9.0–40.1 mIU/ ml), which were within or slightly above the normal range (9.1–30.8 mIU/ml). However, the Epo concentrations of non-diabetic subjects, also with anemia were 144.9 9 108.0 mIU/ml, were significantly higher (PB 0.001) compared to the subjects with diabetes (Fig. 1).
93.1 9 6.1d 30.1 91.5c 34.7 91.7d 2.23 90.73e
Table 4 Relationships between hemoglobin concentrations and clinical factors in patients with anemia of uncertain cause Anemia of uncertain cause (n = 35)
144.9 9108.0e 0.8 9 0.2d 103.5920.5
e
– –
– – –
Age (years) Duration of diabetes (years) Glycated hemoglobin (%) Serum creatinine (mg/dl) Creatinine clearance (ml/min/1.73 m2) Albumin excretion rate (mg/24 h urine)
Correlation coefficient (r)
P-value
0.01 0.20 0.22 −0.49 0.34
0.95 0.26 0.21 0.003 0.03
−0.44
0.009
a
Transferrin saturation: [serum-Fe (mg/dl)]/[total iron binding capacity (mg/dl)]. b Nephropathy: 24-h urine albumin more than 30 mg. c PB0.05. d PB0.01. e PB0.001.
(85.7%) and microcytic in the remaining five patients. Transferrin saturation ratios of diabetic patients were lower than controls, but serum ferritin concentrations were similar between two groups. Hemoglobin concentrations correlated with creatinine clearance (r =0.34, P = 0.03), serum creatinine (r = −0.49, P =0.003) and albumin excretion rate (r = −0.44, P = 0.009). However, no relationship was found with age, duration of diabetes, glycated hemoglobin, exis-
Fig. 1. Serum erythropoietin and hemoglobin concentrations of diabetic patients with anemia of uncertain cause ( ) and non-diabetic patients with a similar degree of anemia ().
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229
4. Discussion Causes of anemia are variable, ranging from relatively benign conditions such as dietary iron deficiency to serious chronic inflammatory disease or malignancy. Careful examination is essential to identify the underlying cause. Though sufficient investigations were performed in only 62 of 192 patients identified through medical records, we were able to verify a considerable number of cases in which the causes of anemia were not clearly determined despite hematologic studies. The common features were normocytic red blood cell indices with a relatively low absolute percentage of reticulocytes [4], accompanied by diabetic microvascular complications. These findings led us to speculate that impairment of erythropoiesis might be responsible. As a preliminary study, we examined the relationship between hemoglobin concentrations and clinical features of non-uremic diabetic patients in 192 medical records. The hemoglobin concentrations were associated with duration of diabetes, creatinine clearance and albumin excretion rate (data not shown). These 192 cases, although retrospectively reviewed, not randomly selected and heterogeneous in causes of anemia, suggested some relationship between anemia and renal damage. In general, renal anemia is not suspected when creatinine clearance exceeds 30 ml/min/1.73 m2, except in the setting of acute tubular necrosis, crescentic glomerular nephritis or tubulointerstitial disease [5]. The main anatomical focus of the lesion of diabetic nephropathy has been the glomerulus. Therefore, the attenuation of renal Epo responsiveness associated with renal interstitial damage has not been considered as a cause of anemia of diabetic patients prior to progression into advanced diabetic nephropathy [6]. However, in this study, of 35 anemic patients with well preserved excretory renal function, hemoglobin concentrations were related to the biochemical parameters of renal damage, albumin excretion rate, creatinine clearance and serum creatinine concentrations. Serum concentrations of Epo are regulated by the degree of oxygen demand. Values vary from 10 to 12 mIU/ml in non-anemic states to 100 –
227
1000 mIU/ml in severe anemia [7]. We have shown that the serum Epo concentrations in diabetic patients with anemia of uncertain cause were significantly lower than those of non-diabetic subjects with anemia. Values were mainly within the normal range and failed to show the expected increase in response to the anemia. These results suggest that relatively low Epo responsiveness may be present even at a very early stage of diabetic nephropathy and could be one of the reasons for anemia of uncertain cause in diabetic patients. The prevalence of renal anemia in diabetic patients without advanced diabetic nephropathy has probably been underestimated. Recently, reports have indicated that focal or general tubulointerstitial lesions may be frequently found in diabetic patients with microalbuminuria and normal or near normal renal function [8]. These findings support our suggestion that tubulointerstitial damage and reduced renal Epo responsiveness could be possible before advanced diabetic nephropathy. Inomato et al. also reported that serum Epo concentrations could provide clues of the presence of renal tubulointerstitial lesions and a potential for acceleration of renal damage [9]. Therefore strict attention should be given to renal function in diabetic patients with anemia of uncertain cause. However, our results also raise question about serum Epo responsiveness. In anemia associated with end-stage renal disease, serum Epo concentrations are usually above the normal range reflecting partial compensation for anemia [10]. However, in this study, values were almost within the normal range and failed to show the compensatory response. These findings suggest that a blunted Epo response could be related, not only to renal parenchymal damage, but also to impairment of regulation of Epo responsiveness in diabetes mellitus. Kojima et al. proposed the possibility that a reduced Epo response could appear as a manifestation of diabetic autonomic neuropathy, describing two cases with autonomic neuropathy, but without evidence of renal damage [3]. Peripheral neuropathy was documented in 29 of the 35 patients in this study, although autonomic nerve function was not directly tested. However, patients with severe diabetic neuropa-
228
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229
thy frequently also exhibit some degree of diabetic nephropathy. A relative deficiency of Epo to tissue oxygen demand can also be found in other types of anemia in which renal damage is absent, for example the anemia of chronic disorders (ACD) [11,12]. The mechanism in this condition (ACD) is inhibition of Epo production and Epo action by certain cytokines such as tumor necrosis factor or interleukin 1 associated with chronic inflammation, infection or malignancy [13]. It is not known whether similar cytokine-mediated effects could also occur in diabetes mellitus. It also needs to be considered that age is associated with blunted Epo responsiveness, usually in those exceeding 70 years of age [14,15]. Some of the subjects in our study were within this age group. Though transferrin saturation ratios of diabetic patients were lower than controls, serum ferritin concentrations were similar between two groups. Therefore, iron profile of anemia of uncertain causes is not suitable for iron deficiency anemia. Besides renal Epo responsiveness, the Epo action on erythropoiesis could be altered by the metabolic environment associated with diabetes. Hyperglycemia influences the action of Epo in diabetic patients. Ritchey et al. demonstrated that improved diabetic control by continuous insulin infusion enhanced the action of Epo in the bone marrow, which augmented the proliferation of erythroid stem cells [16]. Thus it is considered that synthetic Epo replacement therapy should be allied with measures to improve glycemic control. Epo concentrations of non-diabetic patients in this study showed a bimodal pattern without any relation to hemoglobin concentrations (Fig. 1). Some patients show a relatively smaller response to anemia than others. We can not exclude the possibility that diet composition could have affected the Epo responsiveness in our subjects. In Korea, a large proportion of diabetic patients are vegetarian, and consume a lower amount of protein than Western patients [17], which could curtail Epo responsiveness. In summary, we have investigated a number of cases of anemia of uncertain cause, among patients with diabetes before advanced diabetic nephropathy. Though the reason for this anemia
remains unknown, it could be closely related to early change in renal function with inappropriately low levels of Epo.
References [1] L.J. Robert, M.P. Charles, Hematologic alterations in diabetes mellitus, Am. J. Med. 70 (1981) 339 – 352. [2] C.M. Peterson, R.L. Jones, R.J. Koenig, E.T. Melvin, M.L. Lehrman, Reversible hematologic sequelae of diabetes mellitus, Ann. Intern. Med. 86 (1977) 425 – 429. [3] K. Kojima, Y. Totsuka, Anemia due to reduced serum erythropoietin concentration in non-uremic diabetic patients, Diabetes Res. Clin. Pract. 27 (1995) 229 – 233. [4] E.W. Friedman, Reticulocyte counts: how to use them, what they mean, Diagn. Med. 7 (1984) 29 – 35. [5] A.J. Erslew, J. Wilson, J. Caro, Erythropoietin titers in anemic, non-uremic patients, J. Lab. Clin. Med. 109 (1987) 429 – 433. [6] P.H. Maxwell, M.K. Osmond, C.W. Pugh, A. Heryet, L.G. Nicholls, G.C. Tan, B.G. Doe, D.J.P. Ferguson, M.H. Johnson, P.J. Ratcliffe, Identification of the renal erythropoietin-producing cells using transgenic mice, Kidney. Int. 44 (1993) 1149 – 1162. [7] S.J. Schwster, J.H. Wilson, A.J. Ersler, J. Coro, Physiologic regulation and tissue localization of renal erythropoietin messenger RNA, Blood 70 (1987) 316 – 318. [8] P. Fioretto, M. Mauer, E. Brocco, M. Velussi, F. Frigato, B. Muollo, M. Sambataro, C. Abaterusso, B. Baggio, G. Crepaldi, R. Nosadini, Patterns of renal injury in NIDDM patients with microalbuminuria, Diabetologia 39 (1996) 1569 – 1576. [9] S. Inomata, M. Itoh, H. Imai, T. Sato, Serum levels of erythropoietin as a novel marker reflecting the severity of diabetic nephropathy, Nephron 75 (1996) 426 – 430. [10] W.R. Heinz, A. Claussner, P.M. Erbes, E.H. Scheuermann, W. Schoeppe, K.M. Koch, Serum erythropoietin concentration in chronic renal failure: relationship to degree of anemia and excretory renal fuction, Blood 54 (1979) 877 – 884. [11] G.E. Cartwright, The anemia of chronic disorders, Semin. Hematol. 3 (1996) 3351 – 3375. [12] H.K. Boyd, T.R.J. Lappin, Erythropoietin deficiency in the anemia of chronic disorder, Eur. J. Hematol. 46 (1991) 198 – 201. [13] J. Camacho, F. Arnalich, A.F. Zamorano, J.J. Vazauez, Serum erythropoietin levels in the anemia of chronic disorders, J. Intern. Med. 229 (1991) 49 – 54. [14] J. Nafziger, K. Pailla, L. Luciani, J.P. Andreux, O.S. Jean, N. Casaderall, Decreased erythropoietin response to iron deficiency anemia in the elderly, Am. J. Hematol. 43 (1993) 172 – 176. [15] J.S. Power, B.K. Scanford, J.C. Collins, K. Meurer, A. Failinger, T. Buchholz, M. Blank, J. Spivak, M. Hochberg, A. Baer, M. Cotes, E. Goidwasser, Erythropoietin
Y.S. Yun et al. / Diabetes Research and Clinical Practice 46 (1999) 223–229 response to anemia as a function of age, J. Am. Geriatr. Soc. 39 (1991) 30 – 32. [16] A.K. Ritchey, W.V. Tamborlane, J. Gertner, Improved diabetic control enhances erythroid stem cell proliferation
.
229
in vitro, J. Clin. Endocrinol. Metab. 60 (1984) 1257 – 1260. [17] Kap Bum Huh, Malnutrition related diabetes mellitus. J. Korean Med. Assoc. 30 (1987) 774 – 750.