Abstract
Abstract. The issue of whether recombinant human erythropoietin (rhEPO) increases thrombosis of arteriovenous (AV) fistulae used for hemodialysis remains unclear. Thrombosis often occurs at stenotic segments of fistulae where there is marked intimal hyperplasia and extracellular matrix accumulation. Increased expression of transforming growth factor-β1 (TGF-β1) has been shown to be involved in the development of atherosclerotic lesions by promoting intimal hyperplasia and extracellular matrix accumulation. To clarify the role of rhEPO in the development of stenosis of AV fistulae, this study examined expression of the erythropoietin receptor (EPO-R), TGF-β1, plasminogen activator inhibitor type 1 (PAI-1), cellular fibronectin containing an extra domain A (EDA+), and TGF-β1 mRNA, and assessed in situ rhEPO binding in tissue specimens from seven cutaneous veins and eight patent and seven stenosed portions of AV fistulae of patients undergoing dialysis. Prominent intimal hyperplasia was evident in the stenosed segments. Significant elevation in expression of EPO-R and TGF-β1 was noted in patent AV fistulae compared to the cutaneous veins. Significant enhancement of EPO-R and TGF-β expression was detected in the stenotic fistulae. Fibronectin EDA+ and PAI-1 expression was increased in intimal hyperplasia compared to patent fistulae and cutaneous veins. Elevated EPO-R expression was further confirmed by in situ binding of biotin-labeled rhEPO in stenosed tissue specimens. It is hypothesized that increased rhEPO binding due to elevated EPO-R expression contributes to the development of AV fistula stenosis caused by intimal hyperplasia and extracellular matrix accumulation in response to increased TGF-β1 expression in patients receiving hemodialysis.
Recombinant human erythropoietin (rhEPO) is currently widely used for treating anemia in hemodialysis patients and is clearly associated with improved quality of life (1,2,3). Various side effects, however, such as hypertension and thrombotic complications, have been reported with the use of rhEPO in hemodialysis patients (3,4,5). The association of increased incidence of thrombotic events in vascular access sites with rhEPO is controversial: Vascular access thrombosis rates are considered by some to be the same for rhEPO-treated and -untreated groups (5,6), but others have reported an increase in thrombosis with the use of rhEPO (7,8).
Thrombosis of vascular access often occurs at stenosed segments (9,10,11). As in the case of arteriosclerotic lesions associated with cerebrovascular disorders or myocardial infarction, stenotic lesions of arteriovenous (AV) fistulae used for hemodialysis exhibit intimal hyperplasia, marked cell proliferation, extracellular matrix (ECM) accumulation, and angiogenesis (12,13,14,15). Iseki et al. (16) reported that rhEPO increased risk of cardiovascular diseases in chronic dialysis patients. Thus, rhEPO may accelerate development of stenotic vascular lesions in hemodialysis patients, but this has yet to be confirmed.
Transforming growth factor-β1 (TGF-β1) is thought to induce intimal and medial hyperplasia and ECM accumulation in arteries (17,18,19). rhEPO has been shown to enhance the expression of TGF-β1 and platelet-derived growth factor (PDGF) mRNA in cultured vascular smooth muscle cells (20) and to increase plasminogen activator inhibitor type 1 (PAI-1) in cultured endothelial cells (21), events that may be linked with delayed turnover and increased net accumulation of ECM and thrombosis formation. rhEPO also increases the proliferation of vascular endothelium and aortic vascular smooth muscle cells (22,23).
Based on these findings, rhEPO seems to be one of several growth factors contributing to vascular stenosis due to intimal hyperplasia and ECM accumulation in hemodialysis patients. In this study, we hypothesized that rhEPO can bind to the receptor for EPO (EPO-R) in vascular access vessels with consequent promotion of intimal thickening and eventual stenosis. We examined the expression of EPO-R, TGF-β1, PAI-1, cellular fibronectin extra domain A (fibronectin containing EDA+), and in situ binding of rhEPO in stenotic AV fistulae in hemodialysis patients. The data were compared with those for patent parts of AV fistulae in the vicinity of stenotic lesions and also data from cutaneous veins obtained at the time of operation for stenotic AV fistula repair.
Materials and Methods
Specimens
Samples were obtained from 16 patients whose characteristics are listed in Table 1. Overall study results are given in Table 2. The 16 patients were all chronic hemodialysis patients with the exception of one, who was entering the end-stage renal disease program. Twelve of 16 patients had almost occluded or severely narrowed AV fistula segments. Of these 12 patients, occluded or severely narrowed AV fistula segments were obtained from seven hemodialysis patients undergoing surgical AV fistula repair. All seven patients had been treated with intravenous rhEPO. All AV fistulae were of the Brescica-Cimino type. Stenosis was diagnosed on the basis of insufficient blood flow at hemodialysis and was sometimes detected by angiography. Patent portions adjacent to the stenotic fistulae were obtained from eight patients at the time of fistula reconstruction. In four cases, only patent segments adjacent to the stenotic portions were obtained at the time of operation. In addition, cutaneous veins were obtained from seven patients at fistula construction and reconstruction. Clinical characteristics of the three patient groups were not statistically different with respect to age, duration of hemodialysis, duration of EPO treatment, and mean dose of weekly intravenous rhEPO (Table 1). Informed consent was obtained from all patients before inclusion in the study. Specimens were fixed overnight in 4% paraformaldehyde and paraffin-embedded.
Patient characteristicsa
Overall study results
Histochemistry
Serial sections (3 μm thick) of paraffin-embedded tissues were mounted on slides coated with aminopropyl-triethoxysilane (Matsunami Co., Osaka, Japan). The deparaffinized sections were stained with hematoxylin-eosin and elastic van Gieson.
Immunohistochemistry
Serial sections were also used for immunohistochemical analysis. The primary antibodies used were rabbit anti-mouse EPO-R (Upstate Biotechnology, Lake Placid, NY), which detects the human EPO receptor, and rabbit anti-human TGF-β1 (Santa Cruz Biotechnology, Santa Cruz, CA). Goat anti-human PAI-1 (Bio-pool, Umea, Sweden) was used to detect the proteinase inhibitor. Extracellular matrices were examined using anti-fibronectin EDA+ (a gift from Dr. L. Zardi, Istituto Nazionale per la Ricerca sul Cancro, Genova, Italy). Anti-human CD68 (Dako, Carpinteria, CA) and anti-α-smooth muscle actin (Dako) were used to identify macrophages and smooth muscle cells, respectively. Sections incubated with nonimmune sera instead of the primary antibodies were used as controls.
Staining was performed using the standard avidin-biotin-peroxidase complex method. Deparaffinized sections were treated with 3% H2O2 for 30 min. To block nonspecific binding of primary antibodies, sections were incubated with 3% normal donkey serum in phosphate-buffered saline (PBS) for 30 min at room temperature, followed by incubation with primary antibody for 1 h at 37°C in a humidified chamber. After three washes in PBS, biotinylated donkey anti-rabbit, -mouse, or -goat IgG (Vector Laboratories, Burlingame, CA), diluted 1:500 in PBS, was applied, and incubation was continued for another 30 min at room temperature. The sections were then rinsed in PBS and incubated with avidin-biotin-peroxidase complex (ABC Elite, Vector Laboratories) for 30 min. Finally, color was developed by treating sections with the chromogenic component True-Blue (Kirkegaard and Perry Laboratories, Gaithersburg, MD) or chromogenic component containing 0.05% 3,3′-diaminobenzidine (Sigma Chemical Co., St. Louis, MO). Tissue sections were lightly counterstained with Contrast-Red (Kirkegaard and Perry Laboratories), dehydrated with ethanol, and mounted with coverslips.
Tissue sections were examined by light microscopy at ×200 magnification equipped with an ocular grid. EPO-R- and TGF-β1-positive cells were counted separately in the intima, media, and adventitia. In each of the vessel layers, four randomly selected fields were counted and the average values of the four fields were obtained (14). Analysis was performed by two independent investigators who were blinded to the clinical status of the patient.
In Situ Hybridization
Sections of paraffin-embedded tissues (3 mm thick) were subjected to in situ hybridization to detect the expression of TGF-β1 mRNA. The plasmid pmTGF-β1-A, expressing murine TGF-β1 cDNA, was kindly provided by Dr. H. L. Moses (Vanderbilt University, Nashville, TN) (24). Antisense and sense cRNA riboprobes were generated after digestion with restriction enzymes HindIII and EcoRI (Takara Shuzo, Otsu, Japan), respectively. Riboprobe was labeled with digoxigenin using a DIG RNA labeling kit (Boehringer Mannheim Biochemica, Mannheim, Germany). Tissue sections were rehydrated, digested with proteinase K (10 μg/ml; Wako Chemical, Osaka, Japan) for 15 min at 37°C, post-fixed in 4% paraformaldehyde, treated with 0.2N HCl, and acetylated in 0.1 M triethanolamine and 0.25% acetic anhydride. A hybridization solution containing 50% deionized formamide, 0.3 M NaCl, 10 mM Tris HCl, pH 8.0, 5 mM ethylenediaminetetra-acetic acid, 10 mM NaH2PO4, 0.2 mg/ml salmon sperm DNA (Sigma), 0.1 mg/ml yeast transfer RNA (Takara), 1 × Denhardt's solution, and 10% dextran sulfate was added to the probe. Hybridization was performed with 10 ng/μl of the digoxigenin-labeled riboprobe added to each slide and incubated for 18 h at 42°C, as described previously (25).
In Situ Binding Analysis
Staining for biotin-EPO was performed using purified rhEPO (a kind gift from Chugai Pharmaceuticals, Tokyo, Japan) labeled with biotin, as described previously (26).
Serial sections were also used for in situ binding studies according to the method of Gabius et al. (27). Deparaffinized sections were treated with 3% H2O2 for 30 min. To block nonspecific binding of rhEPO, sections were incubated with 3% normal donkey serum in PBS for 30 min at room temperature. Sections were then incubated with biotin-EPO. Subsequently, sections were rinsed in PBS and incubated with avidin-biotin-peroxidase complex (ABC Elite, Vector Laboratories) for 30 min. Finally, color was developed by treating sections with a peroxidase substrate, True-Blue (Kirkegaard and Perry Laboratories). Tissue sections were lightly counterstained with Contrast-Red (Kirkegaard and Perry Laboratories), dehydrated by ethanol, and mounted with coverslips.
Statistical Analyses
Values are expressed as mean ± SD. The average value was compared among groups by one-way ANOVA. Correlations between variables were assessed by linear regression analysis. Significance was defined as P < 0.05.
Results
Histology
Cutaneous veins used as normal controls showed no intimal thickening (Figure 1A). In contrast, marked intimal hyperplasia was evident in all stenosed fistulae (Figure 1B), some of which were almost occluded (Figure 1C). The intimal hyperplasia was irregular with marked cellular proliferation and was associated with angiogenesis in some stenosed areas. In contrast, the vascular media in stenosed lesions was uniformly thickened. Compared with stenosed fistulae, patent portions of the fistulae near stenotic lesions showed only mild intimal hyperplasia.
Histologic changes in sections. (A) Cutaneous veins were thin and showed no intimal thickening as detected by elastic van Gieson staining. Arrows show thin intima (patient 4). (B) Prominent intimal hyperplasia was evident in stenosed fistulae. Note that the magnification of this photograph is similar to that of the cutaneous vein shown in A. Arrows show intimal hyperplasia (patient 1). (C) An almost occluded fistula showing irregular intimal hyperplasia and marked cell proliferation. Arrows show areas of angiogenesis (patient 10). Magnification, ×48.
Immunohistochemical Staining for EPO-R
In cutaneous veins, positive immunoreactivity for EPO-R was noted in the thin intima, media, and adventitia (Figure 2A), but strong staining was evident in the adventitia of capillaries. Positive immunostaining was evident in the cytoplasmic and perinuclear areas of cells. Compared with cutaneous veins, immunoreactivity for EPO-R was significantly increased in areas of intimal hyperplasia in stenosed fistulae (Figure 2B). EPO-R-positive cells were positively stained for actin but negative for CD68 (data not shown). As shown in Table 3, the number of EPO-R-positive cells was significantly higher, even in patent portions of AV fistulae, compared to cutaneous veins, and was even higher in stenotic lesions. Strong staining for EPO-R was especially marked in areas of intimal hyperplastic lesions exhibiting angiogenesis (Figure 2C). The number of EPO-R-positive cells did not correlate significantly with total amount of EPO treatment (r = 0.165, P = 0.487).
Immunostaining for the erythropoietin receptor (EPO-R). (A) In cutaneous veins, positive staining for EPO-R was observed in the thin intima, media, and adventitia. Arrows show thin intima (patient 4). (B) Staining for EPO-R was significantly increased in hyperplastic intima of the stenotic fistula (patient 1). (C) Strong EPO-R staining was observed in angiogenesis of the intimal hyperplastic lesions (patient 10). Magnification, ×120.
Immunopositivity for EPO-R and TGF-β1a
Staining for TGF-β, PAI-1, and Fibronectin
The staining pattern of TGF-β1 was similar to that of EPO-R. In cutaneous veins, positive staining for TGF-β1 was detected in the thin intima, media, and adventitia (Figure 3A). Positive immunostaining for TGF-β1 was evident in the cytoplasmic and perinuclear areas of cells. Staining for TGF-β1 was significantly increased in areas of intimal hyperplasia in stenosed fistulae (Figure 3B). As shown in Table 3, cells positively stained for TGF-β1 were significantly more numerous in stenotic compared to patent portions of fistulae and cutaneous veins. In patent parts of fistulae, TGF-β1 immunostaining was also increased compared with cutaneous veins. TGF-β expression was significantly correlated with that of EPO-R (r = 0.928, P < 0.05) (Figure 4). Immunostaining for PAI-1 was also increased in areas of intimal hyperplasia of stenosed fistulae and the staining was essentially the same as that of TGF-β1 (Figure 3D). Fibronectin EDA+ staining was also similar to that of PAI-1 (Figure 3E). Sections treated with nonimmune rabbit serum showed no immunostaining.
Immunostaining of transforming growth factor-β1 (TGF-β1), plasminogen activator inhibitor type 1 (PAI-1), and cellular fibronectin extra domain A (EDA+) and in situ hybridization of TGF-β1. (A) Positive staining for TGF-β1 was observed in the thin intima, media, and adventitia of cutaneous veins (patient 4). (B) TGF-β1 staining was markedly increased in hyperplastic intima of stenotic fistulae (patient 1). (C) Weak expression of TGF-β1 mRNA in cutaneous veins (patient 4). (D) Increased expression of TGF-β1 mRNA in stenotic fistula (patient 1). Increased immunoreactivity for PAI-1 (E) (patient 1) and fibronectin EDA+ (F) (patient 1). Magnification, ×120.
Correlation between numbers of EPO-R- and TGF-β-positive cells. The number of EPO-R-positive cells in the intima significantly correlated with the number of TGF-β-positive cells in the intima in all samples (r = 0.928, P < 0.05).
In Situ Hybridization
Cells expressing TGF-β1 mRNA as demonstrated by in situ hybridization were identical in most respects to those that stained positively with anti-TGF-β1 antibody. In cutaneous veins, TGF-β1 mRNA was observed in the thin intima, media, and adventitia (Figure 3C). Remarkably strong expression of TGF-β1 mRNA was evident in areas of intimal hyperplasia in stenotic fistulae (Figure 3D). Marked expression of TGF-β1 mRNA was noted even in patent portions of fistulae, and this was still less than in stenotic fistulae, but much higher than in cutaneous veins.
In Situ Binding Study
EPO-R expression was determined by in situ binding of biotin-EPO in serial tissue specimens. Binding was detected in cutaneous veins and fistulae. Staining of biotin-EPO bound to tissue was essentially the same as that of immunostaining for EPO-R using anti-EPO-R antibody. Greater binding of biotin-EPO was detected in areas of intimal hyperplasia (Figure 5A). Excess nonlabeled rhEPO prevented for the most part the binding of biotin-EPO (Figure 5B).
In situ binding of biotin-EPO. Increased binding of biotin-EPO was demonstrated in hyperplastic intima in stenosed sections (A) (patient 10). Addition of excess nonlabeled rhEPO almost blocked the binding of biotin-EPO (B). Magnification, ×120.
Discussion
The pathogenesis of fistula stenosis has been reviewed previously (9,10,11). Marked intimal hyperplasia of stenotic fistulae may occur through various mechanisms such as turbulence at sites of venous valves, compliance mismatch, and vessel damage due to previous venipuncture. Upstream or local release of PDGF or other growth factors and shear-induced intimal injury and repair specifically may mediate the proliferative response.
Whether access failures increase subsequent to rhEPO administration as treatment for renal anemia remains to be determined (5,6,7,8). To date, studies of the pathogenesis of fistula failures related to rhEPO focus on thrombosis formation, in that rhEPO increases the hematocrit and may increase blood viscosity and cause changes in coagulation systems (28,29,30). rhEPO improves platelet function in hemodialysis patients (31) and increases PAI-1 in cultured endothelial cells (20), possibly contributing to fistula thrombosis. Although thrombosis often occurs at sites of stenosis (9,10,11), little investigation of the pathogenic significance of rhEPO in vascular stenosis in hemodialysis patients has been undertaken. rhEPO has been shown to enhance the proliferation of cultured vascular smooth muscle and vascular endothelial cells (21,22). Intimal hyperplasia results from the proliferation of these same cell types, and migration of medial smooth muscle cells into intima is characteristic of vascular stenotic lesions. We therefore hypothesized that rhEPO supplementation for treatment of renal anemia may in some way be related to vascular stenosis in hemodialysis patients.
In this study, we examined the expression of EPO-R in stenosed and patent parts of AV fistulae and cutaneous veins obtained from patients receiving regular dialysis. We detected proliferation of actin-positive cells and increased expression of TGF-β1, PAI-1, and fibronectin EDA+ in the hyperplastic intima of stenosed fistulae compared to patent fistulae and cutaneous veins. We also observed increased expression of EPO-R in actin-positive cells in hyperplastic intima of stenosed fistulae. Furthermore, EPO-R-positive cells exhibited in situ binding of biotin-EPO.
Although the expression of EPO-R in systemic organs outside bone marrow has not yet been completely elucidated, endothelial and smooth muscle cells of systemic vascular beds are reported to express EPO-R (21,22). We confirmed the presence of EPO-R in three layers of cutaneous veins of hemodialysis patients. Because hypoxia is a strong stimulus for EPO production (32) and hypoxia might stimulate EPO-R expression in tissues, enhanced expression of EPO-R in the vessels may simply be the result of anemia in dialysis patients. In addition, the relative abundance of EPO-R and TGF-β1 in intimal lesions was directly proportional to the number of cells. It was not possible to discern whether upregulation of the measured proteins was the cause of the observed proliferation or a consequence of a greater number of cells capable of expressing them. As stated in the results, since the number of EPO-R-positive cells did not correlate with the total or weekly amount or duration of rhEPO administered, we did not observe a direct correlation between use of rhEPO and stenosis. However, binding studies did indicate that binding of rhEPO increased markedly in the area of intimal hyperplasia. It is possible that exogenous rhEPO could bind to vascular cells with increased receptor expression and contribute to increased cell proliferation and matrix synthesis in stenotic vascular lesions. Moreover, it has been reported that endogenous EPO correlates with BP in essential hypertension (33). Because elevated BP is hemodynamically explained by increased total peripheral resistance, these data support the results of previous experimental studies describing the proliferative and vasoconstrictive effects of EPO.
We found that expression of TGF-β1 significantly correlated with that of EPO-R. TGF-β1 is a multipotent growth factor that strongly stimulates production of ECM (17). Several studies have shown that TGF-β1 is an important mediator of intimal hyperplasia in vascular diseases (18,19). Increased fibronectin expression with TGF-β was compatible with previous findings reported in atherosclerotic arteries (18,19). Although the precise pathogenic role of EPO in the expression of TGF-β1 in stenotic AV fistulae has not been elucidated, a recent study reported the rhEPO-induced increase in TGF-β1 and PDGF mRNA expression in cultured vascular smooth muscle cells via angiotensin II induction (20). This finding could explain the significant correlation we observed between expression of EPO-R and TGF-β1.
We also found increased expression of PAI-1 in stenosed fistulae. PAI-1 is a primary regulator of plasmin-mediated proteolytic cascades, which contribute to both fibrinolysis and ECM turnover (34). PAI-1 exerts a negative influence on fibrinolysis and matrix degradation, and could contribute to local accumulation of ECM and fibrin deposits. Increased PAI-1 expression was noted in the thickened intima at the base of human atherosclerotic plaques (35,36).
Angiogenesis has been detected previously in intimal hyperplasia of AV fistulae (12). We observed strong staining for EPO-R in areas of angiogenesis. Because rhEPO has been reported to increase angiogenesis in cultured aortic rings (37), rhEPO may also be a stimulator of angiogenesis in the hyperplastic intima.
The present results suggest that rhEPO might be linked with progression of atherosclerosis, because intimal hyperplasia is a prominent feature of atherosclerosis. rhEPO-induced increase in the risk of cardiovascular diseases in chronic dialysis patients has previously been suggested (16). It may therefore be useful to investigate the expression of EPO-R in stenosed arteries.
Based on the findings above, we assume that rhEPO may be one of the growth factors for intimal hyperplasia in hemodialysis fistulae. This is the first study to show increased expression of EPO-R and colocalization of EPO-R and TGF-β in stenotic AV fistulae. Because there was no significant correlation between number of EPO-R- and TGF-β1-positive cells and total amount of rhEPO administered, our results do not indicate that EPO is the main cause of fistula stenosis. However, we assume that EPO may be associated with stenosis due to the increased expression of EPO-R in stenotic fistula. The mechanism for increased EPO-R expression in hyperplastic intima in fistulae remains to be clarified, as does the question of whether the dose of rhEPO commonly used for renal anemia in hemodialysis patients is sufficient to promote intimal hyperplasia.
An ideal research design would include stenotic fistulae from patients not receiving rhEPO as a control group to prove that EPO is a vascular growth-promoting factor in fistulae. However, we were unable to include such a group in this study. Additional longitudinal studies are necessary to examine the pathogenic significance of exogenous administration of rhEPO on the development of vascular stenosis in hemodialysis patients.
Acknowledgments
We thank Dr. F. G. Issa (Word-Medex, Sydney, Australia) for the careful reading and editing of the manuscript, Dr. L. Zardi (Istituto Nazionale per la Ricerca sul Cancro, Genova, Italy) for providing fibronectin EDA+ antibody, and Dr. H. L. Moses (Vanderbilt University, Nashville, TN) for providing plasmid containing TGF-β1.
Footnotes
American Society of Nephrology
- © 2000 American Society of Nephrology