Treatment of Anemia With Darbepoetin Alfa in Heart Failure

Authors

  • William T. Abraham MD,

    1. From the Division of Cardiovascular Medicine, The Ohio State University, Columbus, OH ; 1Heart Failure Program, Veterans Administration Medical Center and University of Minnesota, Minneapolis, MN ; 2Department of Cardiology, New Jersey Medical School, Newark, NJ ; 3Department of Heart Disease, Medical University, Military Hospital, Wroclaw, Poland ; 4Biostatistics, Amgen Inc., Thousand Oaks, CA ; 5Global Development, Amgen Inc., Thousand Oaks, CA ; 6 and Cardiology, University Medical Center Groningen, Groningen, the Netherlands7
    Search for more papers by this author
  • 1 Inder S. Anand MD, DPhil,

    1. From the Division of Cardiovascular Medicine, The Ohio State University, Columbus, OH ; 1Heart Failure Program, Veterans Administration Medical Center and University of Minnesota, Minneapolis, MN ; 2Department of Cardiology, New Jersey Medical School, Newark, NJ ; 3Department of Heart Disease, Medical University, Military Hospital, Wroclaw, Poland ; 4Biostatistics, Amgen Inc., Thousand Oaks, CA ; 5Global Development, Amgen Inc., Thousand Oaks, CA ; 6 and Cardiology, University Medical Center Groningen, Groningen, the Netherlands7
    Search for more papers by this author
  • 2 Marc Klapholz MD,

    1. From the Division of Cardiovascular Medicine, The Ohio State University, Columbus, OH ; 1Heart Failure Program, Veterans Administration Medical Center and University of Minnesota, Minneapolis, MN ; 2Department of Cardiology, New Jersey Medical School, Newark, NJ ; 3Department of Heart Disease, Medical University, Military Hospital, Wroclaw, Poland ; 4Biostatistics, Amgen Inc., Thousand Oaks, CA ; 5Global Development, Amgen Inc., Thousand Oaks, CA ; 6 and Cardiology, University Medical Center Groningen, Groningen, the Netherlands7
    Search for more papers by this author
  • 3 Piotr Ponikowski MD,

    1. From the Division of Cardiovascular Medicine, The Ohio State University, Columbus, OH ; 1Heart Failure Program, Veterans Administration Medical Center and University of Minnesota, Minneapolis, MN ; 2Department of Cardiology, New Jersey Medical School, Newark, NJ ; 3Department of Heart Disease, Medical University, Military Hospital, Wroclaw, Poland ; 4Biostatistics, Amgen Inc., Thousand Oaks, CA ; 5Global Development, Amgen Inc., Thousand Oaks, CA ; 6 and Cardiology, University Medical Center Groningen, Groningen, the Netherlands7
    Search for more papers by this author
  • 4 Debra Scarlata MS,

    1. From the Division of Cardiovascular Medicine, The Ohio State University, Columbus, OH ; 1Heart Failure Program, Veterans Administration Medical Center and University of Minnesota, Minneapolis, MN ; 2Department of Cardiology, New Jersey Medical School, Newark, NJ ; 3Department of Heart Disease, Medical University, Military Hospital, Wroclaw, Poland ; 4Biostatistics, Amgen Inc., Thousand Oaks, CA ; 5Global Development, Amgen Inc., Thousand Oaks, CA ; 6 and Cardiology, University Medical Center Groningen, Groningen, the Netherlands7
    Search for more papers by this author
  • 5 Scott M. Wasserman MD,

    1. From the Division of Cardiovascular Medicine, The Ohio State University, Columbus, OH ; 1Heart Failure Program, Veterans Administration Medical Center and University of Minnesota, Minneapolis, MN ; 2Department of Cardiology, New Jersey Medical School, Newark, NJ ; 3Department of Heart Disease, Medical University, Military Hospital, Wroclaw, Poland ; 4Biostatistics, Amgen Inc., Thousand Oaks, CA ; 5Global Development, Amgen Inc., Thousand Oaks, CA ; 6 and Cardiology, University Medical Center Groningen, Groningen, the Netherlands7
    Search for more papers by this author
  • and 6 Dirk J. Van Veldhuisen MD 7

    1. From the Division of Cardiovascular Medicine, The Ohio State University, Columbus, OH ; 1Heart Failure Program, Veterans Administration Medical Center and University of Minnesota, Minneapolis, MN ; 2Department of Cardiology, New Jersey Medical School, Newark, NJ ; 3Department of Heart Disease, Medical University, Military Hospital, Wroclaw, Poland ; 4Biostatistics, Amgen Inc., Thousand Oaks, CA ; 5Global Development, Amgen Inc., Thousand Oaks, CA ; 6 and Cardiology, University Medical Center Groningen, Groningen, the Netherlands7
    Search for more papers by this author

William T. Abraham, MD, Division of Cardiovascular Medicine, The Ohio State University Medical Center, Room 110P Davis Heart and Lung Research Institute, 473 West 12th Avenue, Columbus, OH 43210
E-mail: william.abraham@osumc.edu

Abstract

Anemia is common in heart failure (HF) patients. A prespecified pooled analysis of 2 randomized, double-blind, placebo-controlled studies evaluated darbepoetin alfa (DA) in 475 anemic patients with HF (hemoglobin [Hb], 9.0–12.5 g/dL). DA was administered subcutaneously every 2 weeks and titrated to achieve and maintain a target Hb level of 14.0±1.0 g/dL. By week 27, mean (SD) Hb concentrations did not increase with placebo but increased with DA from 11.5 (0.7) to 13.3 (1.3) g/dL. Hazard ratios (HRs) for DA compared with placebo for all-cause death or first HF hospitalization (composite end point), all-cause death, and HF hospitalization by month 12 were 0.67 (95% confidence interval [CI], 0.44–1.03; P=.067), 0.76 (95% CI, 0.39–1.48; P=.419), and 0.66 (95% CI, 0.40–1.07; P=.093), respectively. Incidence of adverse events was similar in both groups. In post hoc analyses, improvement in the composite end point was significantly associated with the mean Hb change from baseline (adjusted HR, 0.40; P=.017) with DA treatment. There was no increased risk of all-cause mortality or first HF hospitalization with DA in patients with reduced renal function or elevated baseline B-type natriuretic peptide, a biomarker of worse HF. These results suggest that DA is well tolerated, corrects HF-associated anemia, and may have favorable effects on clinical outcomes.  Congest Heart Fail. 2010;16:87–95. © 2010 Wiley Periodicals, Inc.

Anemia is common in patients with chronic heart failure (HF); its prevalence appears to be greater in older patients, those with more severe HF, and patients with other comorbidities such as chronic kidney disease (CKD) or diabetes.1,2 Anemia has been shown to be an independent risk factor for increased morbidity and mortality in patients with HF, and even slightly lower hemoglobin (Hb) concentrations are associated with worse outcomes.2–6 Change in Hb over time has been shown to be inversely associated with the risk of mortality and morbidity, suggesting that treating anemia may improve outcomes.7

Preliminary interventional studies have shown that increasing Hb concentrations using erythropoiesis-stimulating agents (ESAs) may improve New York Heart Association (NYHA) functional class, cardiac function, and exercise capacity and reduce the need for diuretics and hospitalization in anemic patients with HF.8–11 However, larger randomized studies are needed to confirm whether treatment of anemia with ESAs is well tolerated and can improve clinical outcomes in the anemic HF population. This is particularly relevant given safety concerns arising from recent studies of ESAs targeting higher Hb levels in patients with CKD or cancer.12–15 Of note, these trials in other populations found that higher Hb targets were associated with an increased risk of adverse outcomes,13,14 including death, arteriovenous access thrombosis, and thrombotic events.15–17

In a small randomized study of anemic patients with symptomatic HF (Clinical Trials.gov number NCT00117234), administration of the ESA darbepoetin alfa (DA) increased and maintained Hb concentrations and was associated with improvements in health-related quality of life as assessed by the patient’s global assessment but did not increase peak VO2 and/or exercise duration significantly.18 Notably, 3 clinical trials—all of which were randomized, double-blind, and placebo-controlled—have been carried out to investigate the effects of DA in anemic patients with symptomatic HF.18–20 Although these studies showed that DA was well-tolerated with an adverse event profile similar to that of the placebo, they were nevertheless of relatively short duration and involved a comparatively small number of patients.

In a recent meta-analysis that included a literature search and results from clinical trials through July 2008, van der Meer and coworkers21 compared ESA treatment with placebo or usual care. In contrast to other recent studies in CKD patients13,14 van der Meer and coworkers21 found that in HF patients, ESA treatment was not associated with a higher mortality rate or more adverse events, and a beneficial effect on HF hospitalization was observed.

The present study reports the results of a prespecified pooled analysis of 2 trials (ClinicalTrials.gov numbers NCT000 49985 and NCT00086086) to evaluate the effect of DA on the combined end point of all-cause mortality or HF hospitalization.

Methods

Study Design.  Full details of the designs of the individual studies on which these pooled analyses are based have been reported elsewhere.19,20 Briefly, following a 2-week screening period, anemic patients with symptomatic systolic HF were randomized in a 1:1 ratio to receive subcutaneous DA (Aranesp®; Amgen Inc., Thousand Oaks, CA) at a starting dose of 0.75 μg/kg or placebo, every 2 weeks (Q2W) for 52 weeks19 or in a 1:1:1 ratio to receive subcutaneous DA at a starting dose of 0.75 μg/kg (weight-adjusted) or 50 μg (fixed), or placebo, Q2W for 26 weeks (Figure 1).20 Other than these differences in starting dosing regimen and duration of follow-up, the 2 studies were designed to be nearly identical in terms of inclusion/exclusion criteria and study procedures in order to permit this pooled analysis. In both studies, DA doses were titrated to gradually achieve and maintain a target Hb concentration of 14.0±1.0 g/dL according to a predefined dosing schedule. Randomization was performed using a central, interactive voice response system. Both patients and study site personnel were blinded to treatment group assignment.

Figure 1.

 Study design. Nine participants were enrolled in both studies but were considered to have participated in only study NCT00049985 for the analysis of the pooled data. Q2W indicates once every 2 weeks.

Study Population.  Eligible patients were 21 years or older, with a ≥3-month history of symptomatic HF (NYHA class II to IV) at the time of enrollment, left ventricular ejection fraction ≤40%, and Hb concentration of 9.0 to 12.5 g/dL. Patients were also required to have a transferrin saturation ≥15.0%, to have serum vitamin B12 and folate levels at or above the lower limit of the normal range, and to have received treatment for HF with an angiotensin-converting enzyme inhibitor and/or angiotensin receptor blocker for at least 8 weeks (unless not tolerated). β-Blocker use was strongly encouraged for this study and, if used, had to have been taken for at least 8 weeks. Exclusion criteria for these studies are detailed elsewhere.19,20 Of note, the use of ESAs within 12 weeks of randomization excluded participants from these studies, as well as a serum creatinine >3.0 mg/dL or current malignancy.

The studies were approved by the ethics committee of each study center and all participants gave written informed consent.

Study Assessments.  Hb concentration, hospitalizations, and mortality were recorded throughout the study period. Adverse events occurring during the treatment period or within 30 days of the last administration of study drug were also recorded. Hospitalizations and deaths were adjudicated by an independent end point adjudication committee that was blinded to treatment group assignment.

Statistical Analyses.

Pooled Analyses. In the pooled analyses of data, baseline characteristics, Hb concentration over time, and the proportion of participants in whom target Hb was achieved (defined as participants with at least 2 consecutive Hb concentrations between 13.0 and 15.0 g/dL and a ≥1.0 g/dL increase from baseline by week 27) were assessed in both DA and placebo-treated participants. All-cause mortality, HF-related hospitalizations, and the incidence of adverse events were also analyzed by treatment group. The pooled morbidity and mortality analysis was prespecified. Statistical analyses were conducted at Amgen Inc. Data were interpreted by the study investigators, who had full access to the primary data, participated in the statistical analyses, and were not limited with regard to any statements made in this report.

The pooled intent-to-treat analysis population was defined as all participants who were randomized in the individual studies. Nine participants participated in both trials but were considered to have participated in only 1 trial19 in this pooled analysis. For these participants, data from the first study were included in the primary analysis, and there were no mortality or additional HF-related hospitalization data for them during the second study. Sensitivity analyses were performed including data from the second study for these 9 participants and showed similar results. Continuous variables were reported as mean, standard deviation (SD), and standard error (SE). Categorical variables were reported as frequencies and percentages. Mortality and morbidity were evaluated using Kaplan–Meier analysis and proportional hazards model. These survival analysis methods accounted for event status and the different follow-up time for participants in the 2 studies. Log-rank test and hazard ratios (HRs) were summarized to compare treatment groups. In these analyses, patients who did not experience an event during each of the individual studies were censored at the time they ended the study. Mortality and morbidity analyses were based on the intent-to-treat analysis population. Results were considered statistically significant if P<.05. The pooled safety analysis set was defined as all participants who received at least 1 dose of study medication. The pooled safety summary includes adverse event data reported during each study.

Post Hoc Analysis.  Several prior randomized studies with ESAs in CKD found that targeting higher Hb levels was associated with an increased risk of adverse outcomes.13,14,16 However, observational data and retrospective analyses suggest that higher achieved (vs targeted) Hb levels may be associated with lower mortality rates.16,22–26 To further explore this distinction between achieved and targeted Hb levels in HF, a subsequent post hoc analysis was performed to evaluate whether morbidity and mortality outcomes were associated with change in Hb from baseline. The subgroup for this post hoc analysis was DA-treated participants who reached study week 17. As treatment of anemia with DA typically requires 12 to 14 weeks to reach the Hb target, week 17 was chosen to allow for sufficient time to achieve target Hb. Hb response was categorized by mean Hb (between week 17 and end of study) change from baseline of either ≥1.0 g/dL or <1.0 g/dL. Cox proportional hazards models were used to compare differences between categorical changes in Hb, adjusting for significant baseline covariates. The stepwise selection method was applied to select significant (P<.05) baseline covariates in the final models. Baseline covariates that were considered included age, sex, ethnicity, smoking history, duration of HF, NYHA class, β-blocker use, left ventricular ejection fraction, geographic region, Hb level, B-type natriuretic peptide (BNP) value, etiology of HF, estimated glomerular filtration rate (eGFR), use of implantable cardioverter-defibrillator and/or cardiac resynchronization therapy, presence/absence of diabetes, and presence/absence of valvular disease.

Data from all participants were used to evaluate morbidity and mortality outcomes in subgroups of eGFR (eGFR <60 and eGFR ≥60 mL/min/1.73 m2; an eGFR <60 mL/min/1.73 m2 is defined by the National Kidney Foundation as the cutoff for moderate to severe CKD27) and BNP (BNP <167.4 and BNP ≥167.4 pg/mL [median value of BNP in the present study]; BNP concentrations >100 pg/mL are considered to be an indicator of HF28,29). Stratified Kaplan–Meier analyses were used to determine outcome differences between treatment groups for each subgroup. Cox proportional hazards models were used to compare differences between treatment groups, adjusting for significant baseline covariates. The stepwise selection method was applied to select significant (P<.05) baseline covariates in the final models. Baseline covariates considered in these analyses were identical to those listed above.

Results

Pooled Analyses. Study Population. Four hundred seventy-five participants were randomized to receive DA (n=266) or placebo (n=209). Overall, demographic and baseline characteristics were similar in the pooled study population between the DA and placebo groups (Table I). The mean (SD) age of participants was 69.5 (10.7) years and 61% were men; HF had been diagnosed a mean (SD) of 6.0 (5.8) years earlier, and they had a mean (SD) left ventricular ejection fraction of 33.4% (10.1%) (Table I). The majority of participants in both groups were receiving treatment with angiotensin-converting enzyme inhibitors and/or angiotensin receptor blockers, β-blockers, and diuretics (Table I). Of the 475 randomized participants, all but 2 (in the DA group) received study medication and 376 (79%) completed the study (217 [82%] in the DA group and 159 [76%] in the placebo group). The most common reasons for study discontinuation were death (n=34; 7%), withdrawn consent (n=22; 5%), and adverse events (n=17; 4%). The average time in the study was 40.1 weeks (range, 0.9–59.7 weeks).

Table I.   Demographic and Baseline Characteristics (All Randomized Participants)
 Placebo (n=209)Darbepoetin Alfa (n=266)All Participants (N=475)
  1. Abbreviations: ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; eGFR, estimated glomerular filtration rate; Hb, hemoglobin; HF, heart failure; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; TSAT, transferrin saturation. an=265; bn=474; cn=264; dn=473; en=208. Values are mean (SD) unless otherwise indicated.

Sex, No. (%)
 Men139 (67)153 (58)292 (61)
Ethnicity, No. (%)
 White181 (87)221 (83)402 (85)
 Black20 (10)27 (10)47 (10)
 Other8 (4)18 (7)26 (5)
Age, y69.6 (10.0)69.5 (11.2)69.5 (10.7)
Medical history, No. (%)
 Ischemic heart disease162 (78)193 (73)355 (75)
 Hypertension134 (64)167 (63)301 (63)
 Myocardial infarction126 (60)162 (61)288 (61)
 Diabetes93 (44)132 (50)225 (47)
HF duration, y6.1 (6.3)5.9 (5.5)a6.0 (5.8)b
Primary cause of HF, No. (%)
 Ischemic heart disease146 (70)173 (65)319 (67)
 Acquired cardiomyopathy36 (17)50 (19)86 (18)
 Hypertension12 (6)22 (8)34 (7)
 Valvular heart disease12 (6)10 (4)22 (5)
 Missing0 (0)1 (<1)1 (<1)
NYHA classification, No. (%)
 Missing1 (<1)1 (<1)2 (<1)
 Class I4 (2)4 (2)8 (2)
 Class II75 (36)100 (38)175 (37)
 Class III123 (59)156 (59)279 (59)
 Class IV6 (3)5 (2)11 (2)
LVEF, %34.3 (10.3)32.6 (9.9)33.4 (10.1)
LVEF, No. (%)
 Missing16 (8)17 (6)33 (7)
 ≤25%40 (19)64 (24)104 (22)
 >25%153 (73)185 (70)338 (71)
HF medication, No. (%)
 ACE inhibitors and/or ARBs189 (90)243 (92)c432 (91)d
 β-Blockers165 (79)214 (81)c379 (80)d
 Diuretics184 (88)245 (93)c429 (91)d
Laboratory values
 Hb, g/dL11.3 (0.8)11.5 (0.7)a11.4 (0.8)b
 Serum ferritin, μg/L191.2 (208.2)182.1 (197.5)a186.1 (202.1)b
 TSAT, %26.1 (11.6)26.8 (10.8)a26.5 (11.2)b
 eGFR, mL/min/1.73 m252.5 (21.1)e53.9 (22.9)a53.3 (22.1)d
 Creatinine, mg/dL1.5 (0.5)1.5 (0.5)a1.5 (0.5)b

Hb Concentration.  Baseline Hb concentrations were similar between the DA and placebo groups (Table I); the mean (SD) baseline Hb concentration overall was 11.4 (0.8) g/dL. The Hb target range was achieved in 75% of participants receiving DA (n=197) compared with 15% of those (n=31) receiving placebo. Mean Hb concentrations in the placebo group showed no clinically meaningful increase up to week 27, but Hb concentrations in the DA group increased to a mean (SE) of 13.3 (0.1) g/dL over this period, compared with a mean (SE) of 11.8 (0.1) g/dL in the placebo group at week 27. Mean Hb concentrations for participants receiving DA who remained in the first study for longer than 27 weeks continued to remain stable, reaching a mean (SE) of 13.5 (0.1) g/dL at week 53 (n=121).

Hospitalization and Mortality.  Forty of 266 participants (15%) in the DA group and 47 of 209 participants (22%) in the placebo group died or had an HF-related hospitalization during the study period. Kaplan–Meier analysis demonstrated that DA-treated participants had a reduced risk of HF hospitalization or all-cause mortality compared with the placebo group (log-rank test, P=.064; Figure 2A). Proportional hazards model analysis revealed that treatment with DA was associated with a trend for lower composite risk of all-cause mortality or HF hospitalization compared with placebo by month 12, although the P value was not significant (HR, 0.67; 95% confidence interval [CI], 0.44–1.03; P=.067; Figure 2B). The individual components of the composite outcome were directionally similar, suggesting reduced risk with DA treatment compared with placebo by month 12: all-cause mortality, with an HR of 0.76 (95% CI, 0.39–1.48; P=.419; Figure 2B); and HF hospitalization, with an HR of 0.66 (95% CI, 0.40–1.07; P=.093; Figure 2B).

Figure 2.

 All-cause mortality or first heart failure (HF)–related hospitalization (intent-to-treat analysis population). (A) Kaplan–Meier plot for the composite end point (all-cause mortality or first HF-related hospitalization). (B) Hazard ratios (95% confidence interval [CI] of all-cause mortality, HF-related hospitalization, and the composite end point for darbepoetin alfa relative to placebo. P values were from the proportional hazards models.

Safety Assessments.  The overall incidence of adverse events and the occurrence of individual adverse events were similar in both treatment groups (Table II). The incidence of adverse events of specific interest, particularly those related to thrombotic events, was similar between the 2 treatment groups. There was a slightly lower incidence of worsening HF reported as an adverse event in the DA group (19%) than in the placebo group (25%; Table II). Exposure-adjusted incidence rates for adverse events were also similar in both treatment groups.

Table II.   Summary of Adverse Events (Safety Analysis Population)
CategoryPlacebo (n=209)Darbepoetin Alfa (n=264)
  1. Values are No. (%).

Any adverse event184 (88)231 (88)
Serious adverse events94 (45)101 (38)
Treatment-related adverse events20 (10)32 (12)
Study discontinuation due to adverse events9 (4)7 (3)
Adverse events of specific interest67 (32)69 (26)
 Worsening heart failure53 (25)50 (19)
 Hypertension12 (6)15 (6)
 Myocardial infarction5 (2)6 (2)
 Deep vein thrombosis2 (1)0 (0)
 Pulmonary embolus1 (<1)0 (0)
 Cerebrovascular disorder4 (2)9 (3)
 Seizure2 (1)1 (<1)
Deaths on study18 (9)17 (6)

Post Hoc Analysis.  Demographic and baseline characteristics by categorical Hb change from baseline (≥1.0  vs <1.0 g/dL) for participants receiving DA who reached study week 17 were similar between Hb change groups; the only statistically significant differences were in eGFR (mean [SD] 55.7 [23.6] mL/min/1.73 m2 vs 45.6 [18.7] mL/min/1.73 m2, respectively; P=.0027) and in valvular heart disease as the primary cause of HF (2% vs 11% of participants, respectively; P=.0197). Adjusted HRs and 95% CIs for the composite end point (all-cause mortality or first HF hospitalization), all-cause mortality, and HF hospitalization in participants who received DA showed that an Hb concentration increase of ≥1.0 g/dL was associated with statistically significant (except for mortality) lower risk compared with an Hb concentration increase of <1.0 g/dL (HR, 0.40; 95% CI, 0.19–0.85; P=.017; HR, 0.34; 95% CI, 0.10–1.21; P=.097; and HR, 0.38; 95% CI, 0.16–0.90; P=.028, respectively; results adjusted for baseline left ventricular ejection fraction, BNP, diabetes, and valvular heart disease; Figure 3). Sensitivity analyses adjusted for baseline eGFR and presence of valvular heart disease showed similar results, as did sensitivity analyses with the DA subpopulation who reached study week 12 (data not shown).

Figure 3.

 Hazard ratios (95% confidence interval [CI]) of all-cause mortality, heart failure (HF)–related hospitalization, and composite end point (all-cause mortality or first HF-related hospitalization) among patients who reached week 17 receiving darbepoetin alfa stratified by mean hemoglobin (Hb) increase (≥1.0 g/dL and <1.0 g/dL, n=203 and n=45, respectively). Mean Hb increase is the mean Hb value between week 17 and end of study minus baseline for each participant. The stepwise selection method was applied. Final results were adjusted for baseline left ventricular ejection fraction, B-type natriuretic peptide, diabetes, and valvular heart disease. P values were from the covariate adjusted Cox regression models.

In recent reports with ESAs, an increased risk of death and hospitalization for congestive HF has been reported at higher Hb levels.14 In the present study, utilizing patient-level data stratified by baseline eGFR subgroups (eGFR <60 [n=322] and eGFR ≥60 mL/min/1.73 m2 [n=151]), there was no increased risk with DA treatment, either in the composite end point of all-cause mortality or first HF hospitalization, in patients with renal insufficiency (Figure 4A and Figure 4B). Furthermore, no increased risk with DA treatment, compared to placebo, was detected for all-cause mortality or first HF hospitalization in patients with baseline plasma BNP levels (an indicator of cardiac dysfunction) either below (n=224) or above (n=225) the population median BNP level of 167.4 pg/mL (Figure 5A and Figure 5B, respectively). Patients with HF and baseline BNP levels >167.4 pg/mL trended toward a beneficial effect of DA therapy in the composite end point of all-cause mortality or first HF hospitalizations compared to placebo (Figure 5B; P=.063).

Figure 4.

 Kaplan–Meier plot for the composite end point (all-cause mortality or first heart failure (HF)–related hospitalization) stratified by estimated glomerular filtration rate (eGFR) subgroup. (A) Participants with baseline eGFR <60 mL/min/1.73 m2. (B) Participants with baseline eGFR ≥60 mL/min/1.73 m2. CI indicates confidence interval.

Figure 5.

 Kaplan–Meier plot for the composite end point (all-cause mortality or first heart failure (HF)–related hospitalization) stratified by baseline B-type natriuretic peptide (BNP) subgroup. (A) Participants with baseline BNP < median level. (B) Participants with baseline BNP ≥ median level; median BNP=167.4 pg/mL. CI indicates confidence interval.

In summary, in the present study, patients with HF and an eGFR <60 mL/min/1.73 m2 did not fare worse with DA treatment compared to placebo in terms of all-cause mortality or first HF hospitalization (Figure 4). In addition, participants with worse HF, as suggested by higher levels of baseline plasma BNP, did not appear to be at increased risk with DA treatment compared to placebo (Figure 5).

Discussion

Recent studies evaluating the effects of ESA therapy on clinical outcomes in patients with CKD and cancer have raised safety concerns about the targeting of higher Hb levels.12–15 In contrast, the results from the present analysis, in a different population—patients with chronic HF—suggest that targeting Hb concentrations of 14.0±1.0 g/dL with DA therapy is well-tolerated and may result in improved outcomes. This analysis in the HF population is particularly relevant, as cardiovascular risk with ESA therapy in CKD patients was in part attributed to HF hospitalizations.14

The present report, a prespecified pooled analysis of data from 2 randomized, double-blind, placebo-controlled studies, is the largest patient population to date to evaluate the safety and efficacy of DA for the treatment of anemia in patients with chronic HF. Treatment with DA gradually increased the mean Hb concentration from baseline to week 27, whereas mean Hb concentrations in patients receiving placebo did not meaningfully increase. The percentage of patients in whom target Hb levels were achieved in response to study drug treatment was significantly higher in the DA group (75%) than in the placebo-treated group (15%). Importantly, DA treatment appeared to be well tolerated in the present study, with similar rates of adverse events or study drug withdrawal between the DA treatment and placebo groups. Moreover, in these anemic HF patients, treatment with DA compared with placebo was associated with statistically insignificant but favorable trends toward improved outcomes.

To further explore the distinction between higher achieved (vs targeted) Hb levels in ESA-treated patients, we performed a post hoc analysis to evaluate outcomes by achieved change in Hb. This post hoc analysis evaluating achieved categorical Hb change and outcomes in DA-treated HF patients found a statistically significant improvement in composite mortality and HF hospitalization with a greater Hb change from baseline (adjusted HR, 0.40; P=.017).

Additional post hoc analyses were performed to further explore whether there was increased risk with DA treatment in subpopulations of patients with renal insufficiency and worse HF at baseline. Higher plasma levels (≥median value) of the biomarker BNP was used as an indicator of greater HF severity. The median BNP value of 167.4 pg/mL is consistent with prior diagnostic thresholds for symptomatic HF (ie, >100 pg/mL).28 In addition, renal insufficiency, as assessed by a decreased eGFR, as defined by the National Kidney Foundation,27 was used to categorize patients into subpopulations of stage 3 CKD, as well as more severe renal insufficiency at baseline. In brief, renal dysfunction was chosen because this comorbidity in patients with HF is associated with an increased risk of adverse outcomes. The association of renal dysfunction with HF and the co-occurrence of these conditions with anemia has recently been reviewed, with the suggestion that HF, CKD and anemia is additive in increasing mortality.30 In the post hoc analyses presented in this report, the results suggest that there is no increase in risk with DA treatment in patients with decreased renal function or in patients with increased BNP levels, a biomarker of HF; this is contrary to other recent reports.13,14

A recent small study in patients with advanced HF showed that treating anemia with epoetin beta and oral iron over a period of 1 year resulted in an increase in left ventricular systolic function, improvement in left ventricular remodeling, lower BNP levels, and decreased pulmonary artery pressure, compared with treatment with oral iron alone.31 Studies in other patient populations have also found an association between ESA treatment and improved clinical outcomes. A recent randomized placebo-controlled trial in critically ill patients found that epoetin alfa treatment was associated with increased Hb concentrations and lower mortality compared with placebo, although ESA treatment was also associated with a significant increase in the risk of thrombotic events.32 In a recent meta-analysis of anemia in chronic HF, it was reported that ESA treatment is not associated with a higher mortality rate or more adverse events and that a beneficial effect on HF hospitalization may be apparent.21 The prespecified pooled analysis presented in this report provides patient-level evidence consistent with the recent meta-analysis.21 Given the recent lack of clarity in reports on the safety of ESAs, further primary patient-level data from uniform experimental populations in a clinical trial setting, such as the one in the present study, provide significant value.

The present study has several limitations. First, neither the individual studies nor the prespecified pooled analyses were prospectively powered for the morbidity and mortality end points; there are a limited number of deaths and HF hospitalizations. Hence, these analyses are not definitive, but rather hypothesis-generating. In addition, the relationship between categorical Hb change and mortality and morbidity outcomes was explored by post hoc analyses where participants were categorized based on a post-randomization outcome (ie, Hb value) and thus must be interpreted cautiously.

Although the potential mechanism(s) of DA action in chronic HF remain(s) to be elucidated, potential roles ranging from correction of anemia to improved perfusion and reduced apoptosis are plausible and have been recently reviewed.33 In brief, these results suggest that anemia in HF is responsive to DA therapy in most patients. Correction of anemia in these patients with HF appeared to be well-tolerated and associated with trends toward improved patient outcomes. The findings of this pooled analysis are being tested in the ongoing Reduction of Events With Darbepoetin Alfa in Heart Failure (RED-HF) trial, a large phase 3 randomized controlled trial evaluating the effect of DA treatment on morbidity and mortality in this patient population. RED-HF aims to definitively determine the impact of treatment of anemia with DA therapy on clinical outcomes in patients with HF.34,35

Acknowledgments and disclosures:  The authors wish to thank Dikran Toroser, PhD, Amgen Inc, and Mandy Suggitt, on behalf of Amgen Inc, for editorial assistance. DS and SW are employees of Amgen, Inc.; IA and DJVV have received support from Amgen and are members of the executive committee of the RED-HF trial; and PP has received honoraria from Amgen, Inc. The studies included in this manuscript were funded by Amgen, Inc.

Ancillary