Introduction

Diabetes mellitus (DM) is currently considered a modern global epidemic with a continuously increasing burden. Registry data show that the incidence of type 2 diabetes mellitus (T2DM) has more than doubled in the last 30 years, and time projections of diabetes burden over the next decades depict further rising trends [1, 2].

Diabetic nephropathy (DN) comprises one of the major microvascular complications of DM [3] and is characterized by albuminuria (>300 mg/24 h), a progressive decline of the glomerular filtration rate (GFR) by 2–20 mL/min per year, arterial hypertension and increased cardiovascular morbidity and mortality [4–6]. The overall prevalence of microalbuminuria and macroalbuminuria is nearly 30–35 % in both types of DM. Diabetic nephropathy in patients with type 1 DM (T1DM) rarely develops earlier than 10 years after diagnosis, whereas approximately 3 % of patients with newly diagnosed type 2 DM (T2DM) already have overt nephropathy [7–9]. Diabetic nephropathy is the most common cause of CKD [10–12], and diabetic patients in some countries account for 40–50 % of patients receiving dialysis [13].

Anemia is one of the most significant consequences of CKD, and it is mainly attributed to insufficient erythropoietin (EPO) production. The control of red blood cell (RBC) mass is based on a classic negative feedback loop mediated by changes in the production of the hormone EPO, which is mainly produced in the kidney and regulates the production of erythrocytes by interaction with specific EPO receptors (EPOR) on bone marrow erythroid progenitors. Anemia can manifest itself early in the course of CKD, and its prevalence and severity go in parallel with the progression of kidney disease. The National Kidney Foundation’s clinical practice guidelines define anemia as a hemoglobin (Hb) concentration lower than 13.5 g/dL for adult men and less than 12.0 g/dL for adult women [14]. However, anemia develops earlier in the course of CKD among patients with DM [15–17] and the severity of anemia tends to be more marked in these patients compared to nondiabetic subjects, regardless of the stage of CKD, while this difference is even most prominent among CKD stage 3 patients [18–21]. Several studies have demonstrated the importance of early detection of anemia in patients with T1DM and T2DM [22–24].

In this review, we focus on the “less known” complex interacting mechanisms which are involved in the pathophysiology of anemia associated with DN.

The role of the kidneys in hematopoiesis

The kidneys play a key role in hematopoiesis since the physiologic regulator of RBC production, the glycoprotein hormone EPO, is produced and released within the kidney, in response to impaired oxygen delivery from the circulating erythrocytes. EPO is a member of the family of class 1 cytokines and is produced by peritubular interstitial cells, which were subsequently identified as peritubular fibroblasts located within the renal cortex [25, 26]. Subsequently, EPO is excreted in the peritubular capillaries and enters the systemic circulation through the renal vein [26]. Although a small quantity of EPO is produced in the liver, the spleen, the brain, the lungs and the testes, its production by these organs cannot substitute for the renal production and cannot meet the increased needs in patients with CKD [27]. The normal plasma EPO concentration ranges from 6 to 32 IU/L [27].

Erythropoietin exerts its action by binding to the EPOR which is present on erythroid progenitors in the bone marrow, from the colony-forming units erythroid stage (CFU-E) to late basophilic erythroblasts. Without EPO present, these cells are rapidly lost via programmed cell death [28, 29]. Thus, EPO is an essential factor determining survival for erythroid progenitors, beginning at the CFU-E stage all the way to basophilic erythroblasts. The production and release of EPO are regulated by a feedback mechanism, which is related to hypoxia, as the fundamental stimulus for EPO production is the availability of O2 for tissue metabolic needs [30]. Renal cells responsible for the production of EPO possess specific hypoxia-sensing regulatory mechanisms, which are mostly based on transcription factors induced by hypoxia [31, 32] called hypoxia-inducible factors (HIFs).

Pathogenesis of anemia in diabetic nephropathy

The ability of the kidneys to produce EPO is not impaired in renal disease, and the absolute value of EPO can be normal or even high; however, EPO levels will be inappropriately low relative to the degree of anemia, resulting in a functional EPO deficiency. An interplay of pathophysiological mechanisms associated with EPO production and action together with factors unrelated to EPO determines the pathogenesis of anemia in diabetic nephropathy (Fig. 1) [33–35].

Fig. 1
figure 1

Suggested pathophysiological mechanisms and causes of anemia in CKD and DN. A number of mechanisms (white-colored boxes) are common among all CKD patients independently of the primary renal disease, and they usually appear with the progression of CKD, while other “less known” mechanisms (gray-colored boxes) are mainly EPO dependent and initiated at the early, preclinical stages of DN

Full size image

Erythropoietin-dependent mechanisms

Glomerular hyperfiltration

The preclinical normoalbuminuric phase of DN is characterized by kidney hypertrophy together with increased renal blood flow and a simultaneous increase in the GFR which augments by 20–40 % above normal levels [36]. The hemodynamic abnormalities in DN have been attributed to hyperglycemia as well as other contributory factors including insulin-like growth factor I, atrial natriuretic peptide, sex hormones, intracellular sorbitol, early glycation products, which cause dilation of the afferent arteriole, constriction of the efferent arteriole and suppression of the tubuloglomerular feedback, thus resulting in increased glomerular pressure and hyperfiltration [3, 37, 38]. Hyperfiltration is observed in 25–75 % of patients with T1DM and 5–40 % of patients with T2DM [36, 39]. Enhanced renal blood flow results in increased oxygen supply in the renal tissue and subsequently removal of the hypoxia stimulus for the production of EPO. Thus, augmented oxygenation of the renal parenchyma, especially at the specific areas of EPO synthesis, results in premature reduction in the renal EPO production [30].

Proteinuria

Severe proteinuria causes loss of the endogenously produced EPO in patients with impaired function of the glomerular barrier, as occurs in DN, since its molecular weight is smaller than albumin. Indeed, patients with DN who present with severe proteinuria appear to have a significant reduction in EPO concentration in the plasma due to increased EPO urinary losses [30, 39–42]. It should be noted that anemia in the setting of proteinuria is not related to the loss of renal function [24]. Finally, both the fractional and the total urinary excretion of EPO are not elevated in patients without significant proteinuria (nonnephrotic type) [43], while hypoalbuminemia has been shown to be significantly associated with severe anemia in patients with DN [44].

Chronic inflammation

In DM, hyperglycemia, arterial hypertension and dyslipidemia stimulate the proinflammatory activity of the endothelium via various mechanisms which involve increased oxidative stress, dysregulation of the nitric oxide synthase (NOS), production of advanced glycosylated end products (AGEs) and the activation of converting factor NF-kappaB [3, 45]. The activated endothelial cells express proinflammatory cytokines such as interleukin-1 (IL-1), tumor necrosis factor α (TNFα) and adhesion molecules such as ICAM-1 and VCAM-1 [30, 46, 47]. It has been suggested that the inflammatory mediators suppress the hypoxia-induced production of EPO [48]; however, there is no solid evidence available to support this hypothesis [49]. A cross-sectional survey in T2DM patients did not reveal any correlation between C-reactive protein (CRP) and EPO levels [50], whereas another study managed to show an inverse significant correlation between CRP and hemoglobin concentration (Hb) in T2DM patients [51]. However, chronic systemic inflammation together with microvascular disease causes a reduction in the activity of EPO on its target cells [52, 53]. The resistance to EPO action involves inhibition of the maturation of erythrocyte precursors, reduction in EPOR expression, inhibition of EPO binding to its receptor by competitors and inhibition of activation of signaling pathways [54]. In addition, EPO can be nonenzymatic glycosylated, similar to apolipoproteins thus causing decreased levels of active circulating EPO [55]. Finally, the degree of hyperglycemia severity appears to modify the expression of EPOR through receptor glycosylation, whereas a significant inverse correlation between serum EPO and the fraction of glycosylated Hb has been found in diabetic patients [13].

Microvascular damage

The early microvascular damage, observed in patients with DM, which is independent of the progressive reduction in the GFR, also contributes to the anemia of DM [30]. Thickening of the glomerular basement membrane together with simultaneous thickening of the tubular basement membrane is reported to be among the first pathological alterations of nephron architecture in patients with DM even in absence of established albuminuria/proteinuria [56]. The changes of the tubular basement membrane disturb the equilibrium between the renal tubules, the peritubular fibroblasts and the endothelium, thus further disrupting the renal excretion of EPO [30]. Inomata et al. [35] reported that interstitial fibrosis also contributes to the anemia of DN. In a recent study, Mise et al. [57] investigated the impact of tubulointerstitial lesions on anemia in diabetic patients with biopsy-proven DN and demonstrated that higher interstitial fibrosis and tubular atrophy scores were more strongly associated with reduced Hb values, whereas glomerular and vascular lesion scores did not seem to correlate significantly with the presence of anemia. However, it remains to be further clarified whether the process of tubular atrophy and interstitial fibrosis is more severe in DN than in other kidney diseases with underlying severe interstitial lesions.

Autonomic neuropathy

Autonomic neuropathy is one of the culprits considered to play a significant role in the pathogenesis of anemia in DN, although the mechanism has not been completely elucidated. Several studies have demonstrated a positive correlation between polyneuropathy and anemia in patients with DM [58]. These findings are further supported by the fact that patients suffering from multiple system atrophy or pure autonomic neuropathy have EPO deficiency as well [33, 59–61]. Additionally, experimental studies have failed to produce EPO in response to hypoxia after visceral denervation, as occurs in patients with DM [62]. Interestingly, diabetic patients who display signs of autonomic neuropathy such as orthostatic hypotension have lower EPO levels compared to diabetic patients without autonomic neuropathy despite equal Hb concentration levels [63]. On the other hand, an increase in the production of EPO is observed after renal transplantation in the setting of graft denervation, which may cause erythrocytosis [64]. Moreover, in a recent study investigating the response of EPO to anemia in T2DM patients without advanced renal failure, anemic patients had a longer duration of T2DM and a higher cardiovascular autonomic neuropathy score [65]. Although serum EPO level was weakly correlated with hemoglobin values, multiple linear regression analysis revealed that autonomic neuropathy score was independently related to Hb or EPO level, thus suggesting that autonomic neuropathy is associated with a blunted EPO response to anemia in this group of patients [65].

The renin–angiotensin–aldosterone system

The renin–angiotensin–aldosterone system (RAAS), apart from its key role in blood pressure control and cardiovascular homeostasis, is considered to contribute to erythropoiesis as well. Specifically, angiotensin II stimulates erythropoiesis via an increase in the production of EPO and by acting like a growth factor in the bone marrow, especially on erythrocyte precursors [66, 67]. Moreover, it has been suggested that a functional unit of the juxtaglomerular apparatus in which angiotensin II acts is modulated by hematocrit values, thus named the critmeter [68]. Angiotensin II causes increased oxygen demands in the area adjacent to the proximal tubule by stimulating several processes in the area such as sodium reabsorption or ammoniagenesis [68].

Hyporeninemic hypoaldosteronism is very common among patients with T1DM, and it is characterized by low EPO and Hb levels in these patients [69, 70]. Reduced plasma renin activity (PRA) is common among patients with DM and arterial hypertension, but not in diabetic patients who display normal blood pressure levels and have no other complications of DM [71]. In this group of patients, reduced PRA goes in parallel with the occurrence of the other complications of DM such as arterial hypertension, retinopathy and nephropathy [72]. The most likely mechanisms responsible for reduced PRA and consequently suppression of the RAAS system are hyalinosis of the efferent arteriole, diminished levels of circulating catecholamines, nonactive prorenin and expansion of the intravascular volume [69, 72]. Suppression of the RAAS system leads to inhibition of the beneficial role of angiotensin II in erythropoiesis and anemia occurring in patients with mild CKD [70].

The two most widely used groups of antihypertensive drugs in patients with DN, which act on the RAAS system, angiotensin-converting enzyme inhibitors (ACEIs) and angiotensin receptor blockers (ARBs), might also worsen anemia in these patients [73]. RAAS stimulation causes afferent arteriolar dilation and efferent arteriolar constriction in order to preserve steady levels of the GFR, thus causing reduced renal blood flow and oxygen supply in the peritubular capillaries as well as the renal medulla and subsequently increased EPO production [30]. Moreover, there is a direct effect of angiotensin II in the bone marrow [66]. Consequently, inhibition of the RAAS system reverses all the above mechanisms and results in anemia. Additionally, these medications increase the concentration of the tetrapeptide N-acetyl-seryl-aspartyllysyl-proline, which is a natural inhibitor of erythropoiesis [73]. However, Hayashi et al. [74] did not demonstrate any correlation between Hb levels and the use of ACEI, whereas according to a more recent report, use of ARBs has better correlation with the severity of anemia in diabetic patients with CKD compared to ACEI [75].

Role of renal tubular epithelial cells

The main functions of proximal renal tubule cells include reabsorption of filtered substances such as glucose, sodium or proteins, excretion of metabolic products as well as synthesis and release of growth factors in the peritubular capillaries, such as transforming growth factor-β (TGF-β) and connective tissue growth factor (CTGF) [76, 77]. During the early stages of DN, renal tubular epithelial cell function is suppressed because of both morphological changes like cell hypertrophy and functional changes involving increased metabolic demands and oxygen consumption [76, 78]. As a result, in the setting of glucosuria and increased supply of glucose in the proximal tubule, there is enhanced reabsorption of sodium through the sodium–glucose cotransporter (SGLT1) [79]. The increased sodium reabsorption in the proximal tubule leads to diminished sodium delivery in the macula densa, stimulation of the RAAS system and augmentation of aldosterone-mediated sodium reabsorption, thus causing expansion of the intravascular volume and increased ultrafiltration [80]. The end result of this energy-consuming process is reduced oxygen supply to the renal parenchyma and tissue hypoxia which stimulate EPO production. Indeed, EPO production is positively correlated with sodium reabsorption in patients with DN [68], whereas inhibition of sodium reabsorption with acetazolamide results in decreased EPO concentrations in normal subjects [81]. However, considering the fact that anemia occurs early during the course of disease in patients with DN, reduced Hb levels and increased sodium reabsorption are probably opposite stimuli for the production of EPO in diabetic nephrons [30].

As already mentioned, in CKD, the EPO plasma concentration is within the normal range, despite low Hb levels [82]. In addition, reduced fractional reabsorption of sodium diminishes oxygen consumption in the epithelial renal tubular cells [83], thus resulting in increased tissue oxygenation of the renal parenchyma [84] and suppression of EPO production and excretion [85].

Finally, recent evidence regarding EPO production by the nephron suggests that EPO is produced by cortical nephrons mainly in the intercalated cells and not in the peritubular cells in normal conditions and by mainly peritubular cells in hypoxia, thus highlighting additional roles of the renal tubular cells in the pathogenesis of anemia [86]. Thus, ongoing research is expected to shed light on the role of renal tubular epithelium on the anemia of DN.

Erythropoietin-independent mechanisms

Renal blood cells disorders

Patients with DM demonstrate significant hyperglycemia-induced metabolic and functional disorders of the RBC such as cell membrane lipid abnormalities [87], suppression of RBC filterability [88] and deformability [89] and alteration of RBC adhesion properties [90].

Impaired activity of the Na+/K+-ATPase pump which leads to RBC swelling [88] together with protein oxidation and accumulation of AGEs on the RBC cell membrane are the main mechanisms responsible for erythrocyte dysfunction [91, 92], thus leading to shortened RBC lifespan [93, 94]. In addition, hyperglycemia promotes the expression of aminophospholipids such as phosphatidylserine on the surface of RBC, which results in their recognition and trapping by the reticuloendothelial system [92].

Medications

There is strong evidence that some of the medications frequently prescribed to diabetic patients may worsen anemia, which is referred to as iatrogenic anemia. Apart from the antihypertensives ACEIs and ARBs mentioned earlier, other culprits are thiazolidinediones and fibrates.

The hypoglycemic agents, thiazolidinediones, decrease glucose plasma levels, improve the action of insulin in the peripheral tissues and augment pancreatic insulin secretion [95]. Fluid retention and weight gain are among the most common side effects of these agents, thus resulting in increased plasma volume and peripheral edema [95]. The main mechanisms responsible for the edema formation are increased sodium reabsorption by the distal nephron, vascular dilation due to improved insulin sensitivity, increased activity of the autonomic nervous system and increased endothelial permeability [96]. Increased plasma volume causes hemodilution, thus reducing plasma Hb values [30]. In addition, large doses of thiazolidinediones may directly suppress the bone marrow [97]. Results from a recent study showed that the occurrence of anemia after use of thiazolidinediones is not only dose dependent, but it also worsens with simultaneous administration of ACEI or ARBs [98].

Accumulating experimental evidence investigates the possible therapeutic potential of fibrates in DN [99]. One the other hand, earlier studies have indicated that fibrates are related to anemia in patients with DN [100]. Their mode of action involves certain nuclear receptors (perixosome proliferator-activated receptors); however, their exact mechanism of action has not been completely elucidated [30, 101]. Fibrates reduce the oxygen affinity of Hb and cause oxygen release in the tissues, thus removing the stimulating role of tissue hypoxia for EPO synthesis [102–104].

Conclusions

Diabetic nephropathy is the most common cause of CKD. Anemia in patients with DN occurs earlier and is more severe than anemia in other CKD patients. On the other hand, anemia in CKD is an important predictor of quality of life and contributes to increased cardiovascular morbidity and mortality in this group of patients. Further research is required to elucidate the pathogenesis of anemia in DN as existing evidence indicates that several, diabetes-related complex and interacting mechanisms are implicated.