Safety of erythropoiesis-stimulating agents in patients with end-stage kidney disease: data are safer than extrapolations

Keywords::

• chronic kidney disease
• end-stage renal disease
• erythropoiesis-stimulating agents
• erythropoietin
• safety

Anemia is a common complication of chronic kidney disease (CKD) as production of endogenous erythropoietin declines in parallel with progressing disease. Erythropoiesis-stimulating agents (ESA) have been shown to correct and maintain hemoglobin (Hb) levels, thus improving oxygen supply to tissues and organs and reducing fatigue, weakness, and shortness of breath.

The hypothesis that ESA treatment targeting Hb levels at the lower end of the normal range (13.0 – 13.5 g/dL) may reduce cardiovascular events and death, and delay progression to kidney failure and dialysis, was investigated in a rather heterogeneous set of three large randomized controlled trials (RCTs) involving CKD-3/4 patients: CREATE [1], CHOIR [2], and TREAT [3]. The normalization hypothesis did not bear out in this population of renal patients not yet requiring dialysis. Although signals were not consistent in presence and strength across the three studies, targeting higher Hb levels may be associated with an increased risk of cardiac events or death [2], stroke [3,4], progression to renal failure [1], and cancer-related mortality in patients with a history of malignant disease at baseline [3].

The finding that targeting higher Hb levels does not have a therapeutic benefit is important. It puts the question of the optimal level of Hb correction in predialysis patients to rest—albeit without giving the answer for which the nephrology community has been looking eagerly. The likelihood of a target-finding trial is virtually nil as the cost would be (close to) prohibitive. In addition, one might argue that such a trial is redundant given that a safe Hb target can be inferred from the three RCTs: at the sub-normal Hb range. However, as Del Vecchio and Locatelli advise in this issue of the journal, “there is a grey area of no evidence either way for intermediate levels (11.5 – 13 g/dL)” relative to higher or lower Hb levels (Abstract) [5]. In their open and balanced review of key safety issues concerning ESA therapy in renal patients, the authors offer a welcome reprieve and needed alternative to the many post-TREAT comments and editorials that are high on (alarming) opinions but markedly low on (careful) analysis. The authors thoughtfully distance themselves from the overemphasis on the safety signals that have come from, in particular, TREAT.

The U.S. Food and Drug Administration took note of the safety signals from CHOIR and TREAT, combined these with the findings from the older Normal Hematocrit Study (NHS) [6], and added a warning to the ESA label. It now states that “in controlled trials with CKD patients, patients experienced greater risks for death, serious adverse cardiovascular reactions, and stroke when administered ESAs to target a hemoglobin level of greater than 11 g/dL” and that “no trial has identified a hemoglobin target level, ESA dose, or dosing strategy that does not increase these risks” [7].

All this makes for rather broad if not unwarranted extrapolations to the dialysis (CKD-5) population that go well beyond what could (and should) be inferred from these rather different studies on CKD-3/4 patients not yet requiring dialysis. CHOIR enrolled CKD-3/4 patients. About half of them had diabetes. TREAT involved only CKD-3/4 patients with diabetes. The NHS study included CKD-5 patients with congestive heart failure or ischemic heart disease, about half of whom had diabetes. Add to this that only about a quarter of the patients in CREATE, which did not find a higher risk of cardiovascular events or mortality, had comorbid diabetes.

What extrapolations from the predialysis to the dialysis setting, if any, do these study samples allow? Diabetes is a risk factor, certainly in the predialysis setting but most likely also in the hemodialysis setting. In the latter, although, we only have evidence for patients with congestive heart failure or ischemic heart disease, or an estimated 15.1% and 25.4%, respectively, of the hemodialysis population [8]. CHOIR and NHS found that normalizing ESA therapy is associated with higher cardiovascular and mortality risks in CKD-3/4 and CKD-5 population, respectively, but CREATE and TREAT did not confirm this in the CKD-3/4 population. There is a convincing signal in TREAT for an increased risk of stroke, but the question remains why this was not observed in the other studies. There is a weak indication from CREATE that normalizing treatment may accelerate progression to end-stage renal disease, but no such signals from the two other CKD-3/4 studies. Lastly, there is a faint signal from TREAT of an increased cancer-related mortality risk among patients with a history of cancer at baseline. With the exception of the NHS, all this information concerns the predialysis population.

Going back to the FDA's warning [7], the Agency may be among those extrapolating beyond the data. The warning implies that the safety signals from CHOIR, TREAT, and NHS can be generalized to the entire CKD-3 to CKD-5 population. Further, in its double-negative that “no trial has identified [a treatment approach] that does not increase” the various risks observed across studies, does the FDA perhaps imply that all ESA therapy, normalizing or not, increases the risk of death, serious adverse cardiovascular reactions, and stroke? That would be counter to a recent meta-analysis of studies comparing higher and lower Hb targets across the predialysis and hemodialysis settings [9]. Increased risks of stroke, hypertension, and vascular access thrombosis were observed in the higher Hb target group, but not increased risks of mortality, serious cardiovascular events, or end-stage renal disease.

There are indeed safety signals of evident or potential concern, mainly for CKD-3/4 patients with diabetes or significantly cardiac-compromised CKD-5 patients. Some are strong and warrant clinical attention, while others are indicative but in need of further evidence—if not plausible explanations. This leaves two additional grey areas of evidence, one clinical and one statistical: first, whether the safety issues observed in CKD-3/4 patients and those in cardiac-compromised CKD-5 patients apply to the entire CKD-5 population, an issue also referred to by Del Vecchio and Locatelli [5]; and second, at what thresholds does a safety signal transition from faint to weak to strong to established? Both require primary data because extrapolations should only be done within not across populations. These primary data should come from observational studies, electronic health records, and other real-world data sources, in addition to (and if warranted by the cost to yield ratio) RCTs.

In conclusion, Hb normalization in predialysis and dialysis patients may increase the risk of specific serious adverse events to varying degrees depending upon disease stage and comorbid conditions. However, what is seen in CKD-3/4 patients should not be extrapolated to CKD-5 patients (and vice versa). In CKD-5 patients, Hb normalization has been associated with higher risk of cardiovascular events and mortality in patients with congestive heart failure or ischemic heart disease. Despite broad safety warnings against higher Hb targets in CKD patients (all stages), there is little evidence of associated risks in CKD-5 patients with no underlying cardiac disease. While ESA therapy targeting sub-normal Hb levels would constitute safe practice, Del Vecchio and Locatelli [5] fairly argue the lack of evidence in the “grey area” range of 11.5 – 13 g/dL. Only primary data will inform whether this range is safe. Until such primary data from the hemodialysis setting are available to fill this evidence gap, clinicians should assess for cardiovascular comorbidity in all of their CDK-5 patients, consider comorbid diabetes, and individualize ESA therapy by targeting Hb levels to the lowest sub-normal level sufficient for alleviating anemia-related symptoms [10].

Declaration of interest

I Abraham and K MacDonald are employees of Matrix45, which has received research grants and contracts related to erythropoietic proteins from Johnson & Johnson (Eprex®), Roche (NeoRecormon®, Mircera®), Amgen (Epogen®, Aranesp®), and Sandoz/Novartis (Binocrit®). By company policy, employees cannot hold equity in sponsor organizations except through mutual funds, and cannot receive direct personal benefits, financial or other, from sponsor organizations. Matrix45 is currently contracted for scientific consulting services with Sandoz/Novartis on an observational study on Binocrit® in the hemodialysis setting, and both authors are involved in this work. Matrix45 provides similar services to other biopharmaceutical companies without exclusivity constraints. I Abraham was supported as director of the Academic Fellowship Program in Clinical Outcomes and Comparative Effectiveness Research funded by the Bureau of Health Professions, U.S. Department of Health and Human Services, through the Arizona Health Education Centers Program.