Target Hemoglobin Level for EPO Therapy in CKD

THE ANEMIA OF chronic kidney disease (CKD) is caused by a deficiency of erythropoietin (EPO), and it can be corrected by administration of recombinant human EPO (rHuEPO). Initially, rHuEPO was approved for use in patients with CKD with a target hemoglobin level of approximately 11 g/dL (110 g/L) because this hemoglobin level reduced the need for blood transfusions and improved quality of life.1 Subsequently, controversy has existed concerning the optimal target hemoglobin level and whether higher targets, similar to physiological levels, should be used in clinical practice.

Observational studies have shown a strong inverse relationship between mortality and hemoglobin level. In this issue of the American Journal of Kidney Diseases, Volkova and Arab2 assessed 13 observational studies in their evidence-based systematic literature review of hemoglobin and/or hematocrit values and all-cause mortality in dialysis patients. They concluded that they were heterogeneous in design, used varying exposure categorizations, controlled for different covariates, and were supportive of increased mortality associated with a hemoglobin level less than the reference range. In the 2 largest studies of incident hemodialysis patients (66,761 patients in the study of Collins et al3 and 50,579 in that of Li and Collins4), the reference range was 11 to 12 g/dL (110 to 120 g/L). However—and this is an important observation—the evidence of risk of hemoglobin levels below normal but greater than the reference range (moderate anemia) was not strong.

Observational studies, by their design, are not able to discern cause from effect; they can assess anemia as a risk factor, but not the efficacy of treating anemia. It is possible that lower hemoglobin levels in dialysis patients reflect adverse comorbidity, with the latter predisposing to death, not the former. Consequently, observational studies must control for relevant comorbidities, another important conclusion made by Volkova and Arab.2 Only 7 of the 13 studies they assessed controlled for comorbidity. An additional point is that a study designed to determine risk should include incident patients at the same phase of their disease. This is particularly important in dialysis cohorts because of their high mortality. Investigation of prevalent patients includes only survivors, and this survivor bias prevents adequate assessment of risk. Only 4 of the 13 studies assessed were of incident patients.2

The biological basis of how anemia could predispose to earlier death has been studied in a Canadian echocardiography observational study.5, 6 An incident cohort of 261 hemodialysis and 171 peritoneal dialysis patients had serial echocardiograms performed. It was concluded that anemia may predispose to left ventricular dilation and hypertrophy, which could predispose to heart failure and, subsequently, death. The reference range used to determine risk limited these conclusions; average baseline hemoglobin level in this study was approximately 9 g/dL (90 g/L), and only a small number of patients had hemoglobin levels greater than 12 g/dL (>120 g/L). Therefore, it is unclear whether hemoglobin levels below normal but greater than 11 to 12 g/dL (>110 to 120 g/L) engender the same cardiac risk as those less than this reference value.

To determine whether correction of anemia can improve clinical outcomes, randomized controlled trials (RCTs) are necessary. No RCT focusing on hard clinical outcomes, such as mortality and cardiovascular clinical events, has been performed in which partial correction of anemia with rHuEPO (to a hemoglobin level of 11 to 12 g/dL [110 to 120 g/L]) was compared with no treatment. In the article by Volkova and Arab,2 5 RCTs were reviewed that compared normalization of hemoglobin level with partial correction with rHuEPO. Four of these RCTs were underpowered to study mortality and enrolled different patient populations.

The study of Besarab et al7 of 1,233 hemodialysis patients with preexisting symptomatic cardiac disease showed that the greater hemoglobin target caused harm in that significantly higher numbers of patients in the higher-hematocrit group experienced vascular access clotting and loss than in the lower-hematocrit group. This observation, together with a hazard ratio for the primary outcome (death or myocardial infarction) of 30% greater (95% confidence interval, 0.9 to 1.9) in the high-hematocrit group, induced early termination of the trial. Conclusions from this RCT were distorted by a post hoc analysis, which showed that the greater achieved hematocrit in the intervention group was associated with better outcomes. However, this analysis is not interpretable because of survivor bias and the inability to determine whether the better outcome in those who achieved greater hematocrits was the result of the intervention or of comorbidity. The discrepancy of the as-treated with the intention-to-treat analysis is analagous to a similar discrepancy in the HEMO RCT of increased dialysis dose.8

Whether normalization of hemoglobin level has a beneficial impact on cardiac structure was studied in RCTs by Foley et al9 and a recent publication by Parfrey et al.10 Higher hemoglobin levels failed to induce regression of established left ventricular dilation in hemodialysis patients compared with lower levels.9 In the recent report of 596 incident hemodialysis patients without symptomatic cardiac disease or left ventricular dilation enrolled in an RCT with 2 years of follow-up, percentages of change in left ventricular volume indices between baseline and last value were similar (7.6% in the higher- and 8.3% in the lower-target group), as were changes in left ventricular mass index (16.8% versus 14.2%).10 Thus, 1 biological basis of how anemia could predispose to earlier death is not tenable for moderate anemia, when patients maintained with an average hemoglobin level of 10.9 g/dL (109 g/L) were compared with those maintained with an average hemoglobin level of 13.3 g/dL (133 g/L).

Normalization of hemoglobin level probably improves quality of life, but in the recent Canada-Europe study, improvements in vitality scores seen in the higher-hemoglobin group were of modest clinical significance and waned in the second year.10 Furthermore, costs involved in achieving normal hemoglobin levels are extremely high; for the target of 12 to 12.5 g/dL (120 to 125 g/L) compared with 11.0 to 12.0 g/dL (110 to 120 g/L), costs per quality-adjusted life-year gained were US $613,015, and for a target of 14.0 g/dL (140 g/L) compared with 12.0 to 12.5 g/dL (120 to 125 g/L), US $828,215.11

Clinical decisions to prescribe rHuEPO at target hemoglobin levels higher than 11 to 12 g/dL (110 to 120 g/L) also must take account of potential adverse outcomes: higher vascular access clotting and loss reported in the study by Besarab et al7 and significantly higher cerebrovascular events in the Canada-Europe study10 echo concerns about thrombotic events identified in oncology patients treated with rHuEPO.12, 13 These side effects are not negligible and require further study in large groups of patients.

In patients with CKD not at end stage, observational studies suggested that anemia is a risk factor for cardiovascular events and stroke.14, 15 Whether correction or prevention of anemia with rHuEPO can prevent death or cardiovascular events is unknown. The Cardiovascular Risk Reduction by Early Anaemia Treatment With Epoetin (CREATE) RCT in Europe was underpowered to assess major clinical end points (N = 600),15 and the Correction of Hemoglobin and Outcomes in Renal Insufficiency (CHOIR) study in the United States was terminated early for reasons that have not yet been made public.

At present, there is sufficient uncertainty regarding the risks and benefits of rHuEPO therapy in patients with CKD and anemia to warrant major RCTs. The Trial to Reduce Cardiovascular Events with Aranesp Therapy (TREAT) RCT has started in anemic patients with diabetes and CKD (estimated glomerular filtration rate, 20 to 60 mL/min [0.33 to 1.00 mL/s]). Sample size is 4,000. In the intervention group, darbepoetin alfa will be prescribed to patients with a hemoglobin level less than 11 g/dL (<110 g/L), using a target hemoglobin level of greater than 13 g/dL (>130 g/L), and the control group will be administered a placebo, with hemoglobin level allowed to decrease to 9 g/dL (90 g/L), at which level darbepoetin alfa therapy will be started.17 The primary end point is death or cardiovascular event. TREAT is unique in that it is an event-driven study with a projected event rate of 1,200.16

The conclusion made by Volkova and Arab2 that there is a need for better designed RCTs of rHuEPO therapy is incontrovertible. Current guidelines for target hemoglobin level rely on modest data. The research priority should be on large RCTs using major clinical events as a primary outcome and enrolling patients with end-stage kidney disease or those in the earlier stages of CKD. Until these trials are completed, target hemoglobin levels greater than 11 to 12 g/dL (>110 to 120 g/L) cannot be recommended for rHuEPO therapy because of lack of evidence for benefit, potential for increased risk for thrombogenic events, and high cost of small improvements in quality of life. In practice, our dialysis unit targets hemoglobin levels of 11 to 12 g/dL (110 to 120 g/L); when levels are greater than target, we respond quickly to decrease rHuEPO dose because of cost and side-effect concerns. When levels are less than target, responsiveness is driven by the impact on quality of life, which is modest for small decreases in hemoglobin level.