- Top of page
Cardiovascular disease (CVD) is the main cause of mortality in renal transplant recipients (RTR). Classical factors only partly explain the excess risk. We hypothesized that high EPO—a marker for inflammation, angiogenesis and hypoxia—is associated with CVD in RTR. A total of 568 RTR (51±12 years; 45% female; creatinine clearance (CrCl) 57±20 mL/min/1.73 m2) were included at median 6 [IQR 3–11] years after transplantation. Subjects on exogenous EPO and ferritin-depleted subjects were excluded. Median EPO level was 17.3 [IQR 11.9–24.2] IU/L. Gender-stratified tertiles of age-corrected EPO were positively associated with waist circumference (but not BMI), CVD history, time since transplantation, diuretics, azathioprine, CRP, mean corpuscular volume and triglyceride levels, and inversely with CrCl, RAAS-inhibition, cyclosporine, hemoglobin, total- and HDL-cholesterol. During follow-up for 7 [6–7] years, 121 RTR (21%) died, 64 of cardiovascular (CV) causes. Higher EPO (per 10 IU/L) was associated with total (HR1.16 [1.04–1.29], p = 0.01) and CV mortality (HR1.22 [1.06–1.40], p = 0.005), independent of age, gender, hemoglobin, inflammation, renal function and Framingham risk factors. Thus, EPO and mortality are linked in RTR, independent of potential confounders. This suggests that yet other mechanisms are involved. Dissecting determinants of EPO in RTR may improve understanding of mechanisms behind excess CV risk in this population.
- Top of page
Cardiovascular disease is the most important cause of mortality among renal patients, in patients with chronic kidney disease (CKD) in native kidneys as well as in renal transplant recipients (RTR). RTR are at an increased risk for cardiovascular disease of roughly six times when compared to the general population (1). Half of the mortality in this population is attributable to cardiovascular causes (2). Identifying patients at risk has gained a lot of attention in recent decennia, yet long-term survival has not improved markedly.
One of the possible targets for improving cardiovascular risk in RTR was thought to be anemia, as it is a predictor of mortality (3) and potentially accessible to intervention, yet large interventional trials with exogenous EPO in native CKD patients, with (4) and without diabetes (5), did not show an improvement of cardiovascular morbidity or mortality after anemia correction. Rather, an increased incidence of cardiovascular events was observed in nondiabetic patients with CKD (5). Likewise, an increased incidence of cerebrovascular accidents was observed in diabetic patients with CKD receiving high doses of exogenous EPO (4).
The above studies shifted attention to EPO as a factor in cardiovascular risk. In line with this assumption, studies in heart failure and native diabetic CKD showed that endogenous EPO levels have prognostic impact for cardiovascular events (6,7). Whether endogenous EPO levels are associated with cardiovascular events in RTR is so far unknown. Considering the elevated risk for cardiovascular disease, and the specific abnormalities of EPO regulation in RTR, this would require specific investigation. Therefore, in the current study we studied the association of endogenous EPO levels with cardiovascular and all-cause mortality in a large single-center cohort of RTR.
- Top of page
Of all study subjects, 55% were male. Median EPO levels were 17.3 [IQR 11.9–24.2] IU/L.
Age-adjusted baseline characteristics according to gender-stratified tertiles of EPO are given in Table 1. From this table, it becomes apparent that EPO levels are associated with waist circumference (but not BMI), prior history of CVD, use of azathioprine, use of diuretics, high-sensitive CRP (hsCRP) levels, mean corpuscular volume (MCV), triglycerides and proteinuria and inversely with hemoglobin (Hb), ferritin levels, use of renin–angiotensin–aldosterone System (RAAS) inhibiting medication, use of cyclosporine, total-, and HDL-cholesterol and renal function.
Table 1. Baseline table for gender-stratified tertiles of age-adjusted EPO levels (N = 568)
| || ||Tertiles of EPO||p-Value|
|Demographics|| ||N = 189||N = 190||N = 189|| |
| Male||n (%)||108 (55%)||109 (55%)||108 (55%)||NA|
| Age||(years)||47 ± 12||52 ± 12||54 ± 10||<0.001|
| EPO||(U/L)||10.2 [9.5–10.8]||17.4 [16.9–18.0]||35.1 [34.6–35.6]||<0.001|
| Hemoglobin||(g/dL)||14.1 ± 1.5||13.9 ± 1.5||13.6 ± 1.6||0.008|
| Anemia||n (%)||32 (17%)||25 (13%)||41 (23%)||0.002|
| MCV||(fl)||90 [89–90]||91 [91–92]||92 [91–92]||0.01|
| Ferritin||(μg/L)||157 [144–170]||158 [146–171]||121 [111–131]||0.07|
| CAD||n (%)||9 (5%)||20 (10%)||18 (9%)||0.05|
| CVA / TIA||n (%)||7 (4%)||8 (4%)||19 (10%)||0.04|
| Time on transplant||(years)||4.9 [2.3–9.6]||6.4 [3.4–11.6]||6.7 [3.0–13.7]||0.001|
| Time on dialysis||(months)||27 [16–45]||27 [12–48]||29 [14–51]||0.89|
|Immunosuppression|| || || || ||<0.001|
| Cyclosporine||n (%)||143 (73%)||117 (60%)||115 (59%)|| |
| Tacrolimus||n (%)||28 (14%)||30 (15%)||24 (12%)|| |
|Proliferation inhibitor|| || || || ||0.003|
| Azathioprine||n (%)||34 (18%)||65 (33%)||93 (48%)|| |
| Mycophenolic acid||n (%)||96 (49%)||80 (41%)||63 (32%)|| |
| Length||(m)||1.72 ± 0.10||1.72 ± 0.09||1.72 ± 0.09||0.97|
| BMI||(kg /m2)||25.9 ± 4.3||25.9 ± 4.2||26.3 ± 4.3||0.57|
| Waist||(cm)||95.2 ± 13.1||96.3 ± 12.9||99.6 ± 13.1||0.006|
|Smoking|| || || || ||0.32|
| Current smoking||n (%)||47 (24%)||33 (17%)||50 (25%)|| |
| Previous smoking||n (%)||82 (42%)||84 (42%)||83 (42%)|| |
| SBP||(mmHg)||153 ± 22||152 ± 22||155 ± 22||0.42|
| DBP||(mmHg)||90 ± 10||91 ± 10||90 ± 10||0.64|
| Number of AHT||(n)||2 [1–3]||2 [1–3]||2 [1–3]||0.12|
| Use of RAASi||n (%)||86 (44%)||58 (29%)||59 (30%)||0.002|
| Use of diuretics||n (%)||76 (39%)||78 (40%)||101 (53%)||0.007|
|hsCRP||(mg /L)||1.6 [1.5–1.8]||1.9 [1.7–2.1]||2.7 [2.4–3.0]||0.001|
| Glycolized Hb||(%)||6.5 [6.4–6.6]||6.3 [6.3–6.4]||6.5 [6.4–6.6]||0.22|
| Diabetes||n (%)||36 (18%)||29 (15%)||39 (20%)||0.33|
| Duration of diabetes||(years)||5 [3–7]||5 [3–6]||4 [3–5]||0.64|
| Glucose||(mmol/L)||4.9 ± 0.2||4.6 ± 0.2||4.9 ± 0.1||<0.001|
| Triglycerides||(mmol/L)||2.05 [1.99–2.09]||2.14 [2.11–2.18]||2.37 [2.33–2.41]||<0.001|
| Total cholesterol||(mmol /L)||5.67 ± 0.11||5.66 ± 0.10||5.54 ± 0.09||<0.001|
| HDL – cholesterol||(mmol /L)||1.14 ± 0.05||1.13 ± 0.04||1.07 ± 0.04||<0.001|
| Creatinine clearance||(mL /min /1.73 m2)||59 ± 20||58 ± 20||53 ± 20||0.02|
| Urine protein excretion||(g /day)||0.3 [0.3–0.3]||0.4 [0.3–0.4]||0.4 [0.3–0.4]||0.16|
In a backward linear regression analysis, use of azathioprine, age, hsCRP, MCV, triglycerides and diuretics remained as significant positive independent determinants of continuous EPO levels, while use of RAAS inhibiting medication, Hb, total cholesterol, ferritin and cyclosporine remained as significant negative independent determinants of continuous EPO levels (Table 2). R2 of the final model was 0.25.
Table 2. Backward linear regression model for determinants of log transformed EPO levels
|Multivariate determinants of log transformed EPO levels|
| ||β||95% CI||Standardized beta||p-Value|
|Use of azathioprine||0.11||0.06; 0.15||0.21||<0.001|
|Use of RAAS inhibition||−0.08||−0.12; −0.04||−0.16||<0.001|
|hsCRP (log)||0.06||0.03; 0.09||0.14||0.001|
|Total cholesterol (log)||−0.41||−0.64; −0.19||−0.14||<0.001|
|Ferritin (log)||−0.08||−0.13; −0.04||−0.14||<0.001|
|Triglycerides (log)||0.10||0.01; 0.19||0.09||0.027|
|Use of cyclosporine||−0.05||−0.08; −0.01||−0.09||0.024|
|Use of diuretics||0.04||0.00; 0.08||0.08||0.050|
Of the 568 subjects in the study, 121 (21%) died during the follow-up of 3647 person-years. A total of 64 (11%) of these deaths were of cardiovascular origin. Subsequently, we performed univariate Cox-regression analyses corresponding to the Kaplan–Meier curves presented in Figure 1. For mortality, HRs for the second and third age- and gender adjusted tertiles versus the first tertile were 1.18 (95% CI 0.72–1.94, p = 0.52) and 1.66 (1.03–2.67, p = 0.04) respectively. For cardiovascular mortality, HR for second tertile was 1.31 (0.63–2.76, p = 0.47) and 2.23 (1.12–4.47, p = 0.02 for the third tertile).
Figure 1. Total (121 events) and cardiovascular (64 events) mortality according to Kaplan–Meier analyses for age-adjusted, gender-stratified tertiles of EPO.*Significance tested with log rank test. Cut-off levels for tertiles of EPO were: tertile 1: EPO < 13.9 IU/L; tertile 2: EPO 13.4–22.4 IU/L; tertile 3: EPO 21.0–182.0 IU/L.
Download figure to PowerPoint
Subsequently, Cox regression models were built for EPO levels as a continuous variable, with the determinants of EPO levels on univariate linear regression analysis and factors known to influence EPO levels or cardiovascular risk included to test whether inclusion of the factors in the model might alter the association between EPO and cardiovascular and total mortality (Table 3). The first was crude and in a second model, age and gender were added. The third model additionally contained hemoglobin, MCV and ferritin levels, while the fourth added instead significant determinants of inflammation; hsCRP and use of azathioprine and cyclosporine, in the fifth renal parameters were included, as known determinants of both mortality and EPO levels. The sixth model contained age, gender, waist circumference, triglyceride levels and classical Framingham risk factors (Table 3). Neither of the adjustments did materially change the significant association of EPO with outcome.
Table 3. Cox regression analyses in cardiovascular and total mortality for EPO levels (per 10 IU/L increase)
| ||All-cause mortality||Cardiovascular mortality|
|(121 events)||(64 events)|
|Model 1||1.13 [1.05–1.22]||<0.001||1.16 [1.05–1.27]||<0.001|
|Model 2||1.15 [1.04–1.28]||0.005||1.20 [1.06–1.36]||0.003|
|Model 3||1.16 [1.05–1.28]||0.003||1.21 [1.06–1.38]||0.005|
|Model 4||1.13 [1.02–1.27]||0.03||1.20 [1.05–1.37]||0.006|
|Model 5||1.11 [1.01–1.23]||0.05||1.15 [1.01–1.31]||0.03|
|Model 6||1.16 [1.04–1.29]||0.01||1.22 [1.06–1.40]||0.005|
Finally, EPO was a consistent determinant of both total and cardiovascular mortality in the forward and backward Cox regression analyses, independent of CRP, age, creatinine clearance, urinary protein excretion, presence of diabetes, use of MMF and smoking status (Table 4). The associated hazard ratio was comparable to that of CRP.
Table 4. Forward and backward regression analysis, including all variables
| ||All-cause mortality||Cardiovascular mortality|
|(121 events)||(64 events)|
|HR (95% CI)||p-Value||HR (95% CI)||p-Value|
|Age (per year)||1.07 (1.05–1.09)||<0.001||1.08 (1.05–1.11)||<0.0001|
|Creatinine clearance (per mL/min/1.73 m2)||0.97 (0.96–0.98)||<0.001||0.96 (0.94–0.98)||<0.001|
|Protein excretion (per g/24 h)||1.25 (1.08–1.43)||0.003||….||…..|
|Diabetes (yes/no)||1.74 (1.15–2.62)||0.008||2.17 (1.27–3.73)||0.005|
|Use of MMF (yes/no)||1.65 (1.09–2.49)||0.02||2.84 (1.53–5.28)||0.001|
|Smoking at any time (yes/no)||1.70 (1.11–2.61)||0.02||2.12 (1.17–3.87)||0.01|
|EPO (per 10 IU/L)||1.13 (1.02–1.25)||0.03||1.17 (1.02–1.35)||0.02|
|CRP (per doubling)||1.11 (1.01–1.24)||0.03||1.18 (1.02–1.37)||0.02|
- Top of page
In this study in 568 RTR we showed endogenous EPO levels to be associated with total and cardiovascular mortality. To our knowledge, this is the first study in renal patients to show that endogenous EPO levels are linked to worse outcome. Even after correction for hematopoietic, renal and inflammatory markers, EPO was significantly associated with mortality in this renal transplant population.
The role of EPO has been well established in both chronic kidney disease (CKD) and end-stage renal disease (ESRD) patients, where substantial evidence from large cohort studies (4,5) has shown that correcting anemia, unless severe, with recombinant EPO (rHuEPO) is not beneficial and may in fact have strong adverse effects, which include increased risk for cardiovascular morbidity and mortality. The highest risk was seen in patients requiring the highest rHuEPo dosage, most likely reflecting resistance to EPO through inflammation (5). In line with these findings, we found increased risk for mortality in subjects with higher EPO levels. This may suggest EPO resistance plays a role in the increased risk for cardiovascular mortality in these subjects; however EPO levels predicted mortality independent of C-reactive protein, renal function, hematological parameters and classical cardiovascular risk factors, suggesting involvement of either EPO as such, or other pathways in the association between EPO and cardiovascular risk.
The fact that EPO was associated with mortality independent of CRP does not exclude a role for inflammation in the association between EPO and cardiovascular risk. There are several inflammatory pathways that do not necessarily include CRP (12). A link between EPO and cardiovascular risk can, however, also lie outside pathways involving chronic low grade inflammation. One of the most obvious of these potential pathways may be stimulation of aberrant angiogenesis with associated plaque rupture (13).
It is known that rHuEPO promotes angiogenesis (14). Angiogenesis may lead to unstable plaques via intraplaque hemorrhages, which in turn causes plaque instability (13) and may as such lead to thrombotic and hemorrhagic events. Remarkably, we also found a negative association between EPO levels and cholesterol levels in this study. Exogenous EPO is known to improve lipid abnormalities (15); how this may affect the process of plaque formation should be investigated.
Another potential pathway involved in the cascade between EPO, anemia, inflammation and mortality may be renal hypoxia. Both renal vascular dysfunction and decreased ejection fraction may cause renal hypoperfusion, leading to compensatory increased EPO production along with RAAS and sympathic nervous system activation with subsequent sodium and fluid retention, known as the cardiorenal axis, but angiotensin II activity may also upregulate EPO levels (16), which explains why EPO levels were negatively associated with use of RAAS inhibiting medication (17). As increased RAAS activity has been shown to be associated with cardiovascular risk (18), it remains unclear to what extent EPO plays a role in this cascade. In combination with RAAS inhibiting medication in native CKD, diuretics may lower EPO levels (17), possibly due to reduced oxygen demand via reduced sodium reabsorption. In variance, in the current RTR population diuretics were more often used by subjects with the highest EPO levels. This may partly be due to less use of RAAS inhibiting medication, whereby increased aldosterone levels stimulate oxygen consumption. Furthermore, it has been shown that in hypoxic kidney with reduced blood flow, diuretics do not decrease oxygen demand (19). Although information on renal blood flow was not available, GFR was reduced in subjects with highest EPO levels.
EPO levels were also associated with use of azathioprine and negatively with use of cyclosporine and hemoglobin levels. This may suggest the mild bone marrow suppression, caused by azathioprine, due to its antiproliferative effect, may give rise to a compensatory increase in EPO levels. A change in immunosuppressants to cyclosporine may reduce this bone marrow suppression and allow for restoration of EPO levels to normal values. Whether this indicates that EPO levels may be indicative of the extent of bone marrow suppression should be further investigated.
Furthermore, EPO levels were negatively associated with ferritin levels. Although ferritin depleted subjects were excluded from the analysis, ferritin was thus still a determinant of EPO levels, suggesting that the inhibiting effect of ferritin on erythropoiesis is still there, even when an absolute deficiency is not present. The association between MCV and EPO levels, however, suggests there may well be folate and vitamin B12 deficiencies, unfortunately data on folate and B12 were not available. The compensatory increased endogenous EPO levels may lead to thrombocytosis and thus to increased risk of thrombotic events.
Higher EPO levels may also affect cardiovascular risk through nonhematopoietic effects. rHuEPO may give rise to hypertension (20); in our study EPO levels were however not related to mean arterial pressure or the number of antihypertensives prescribed.
Thus, there is much unclear still about the mechanisms behind the association between EPO levels and cardiovascular disease. Nevertheless, EPO was independently associated with cardiovascular disease in RTR. Endogenous EPO levels may be helpful in identifying resistance to exogenous EPO, precluding inadvert use of exogenous EPO. Prospective interventional studies could help to address the possible underlying mechanisms, such as hypoxia, inflammation and angiogenesis.
The lack of data on baseline cardiac status, other than clinical assessment, is one of the main limitations of our study. Furthermore, data are observational and EPO levels were only measured at baseline. Repeated measurements could have improved power of our findings and could have elucidated the mechanisms behind the association between cardiovascular risk and EPO levels. Thus, this study looks at only a small part of the relationship between CVD, EPO and survival in RTR and would require more thorough prospective studies to support its major findings.
In conclusion, EPO levels were associated with overall and cardiovascular mortality in renal transplant recipients. Further research to elucidate underlying mechanisms in the association between EPO and mortality may help improve our understanding of the pathophysiology that explains the excess cardiovascular risk in this population.