Abstract
Background Exposure to high doses or a high cumulative dose of erythropoiesis-stimulating agents (ESAs) may contribute to cardiovascular events in patients with CKD and anemia. Whether using a low fixed ESA dose versus dosing based on a hemoglobin-based, titration-dose algorithm in such patients might reduce risks associated with high ESA doses and decrease the cumulative exposure—while reducing the need for red blood cell transfusions—is unknown.
Methods In this phase-3, randomized trial involving 756 adults with stage-3 to -5 CKD and anemia, we evaluated incidence of red blood cell transfusions for participants randomized to receive darbepoetin given as a fixed dose (0.45 µg/kg every 4 weeks) versus administered according to a hemoglobin-based, titration-dose algorithm, for up to 2 years. Participants received transfusions as deemed necessary by the treating physician.
Results There were 379 patients randomized to the fixed-dose group, and 377 to the titration-dose group. The percentage of participants transfused did not differ (24.1% and 24.4% for the fixed-dose and titration-dose group, respectively), with similar time to first transfusion. The titration-dose group achieved significantly higher median hemoglobin (9.9 g/dl) compared with the fixed-dose group (9.4 g/dl). The fixed-dose group had a significantly lower median cumulative dose of darbepoetin (median monthly dose of 30.9 µg) compared with the titration-dose group (53.6 µg median monthly dose). The FD and TD group received a median (Q1, Q3) cumulative dose per 4 weeks of darbepoetin of 30.9 (21.8, 40.0) µg and 53.6 (31.1, 89.9) µg, respectively; the median of the difference between treatment groups was −22.1 (95% CI, −26.1 to −18.1) µg.
Conclusions These findings indicate no evidence of difference in incidence of red blood cell transfusion for a titration-dose strategy versus a fixed-dose strategy for darbepoetin. This suggests that a low fixed dose of darbepoetin may be used as an alternative to a dose-titration approach to minimize transfusions, with less cumulative dosing.
Since their approval, erythropoiesis-stimulating agents (ESAs) have been widely used in both patients with CKD on dialysis and those not on dialysis (ND) to raise hemoglobin concentrations and reduce the need for red blood cell (RBC) transfusions. Observational studies showed an association between anemia and increased cardiovascular (CV) morbidity and mortality in patients with CKD who were ND (ND-CKD),1⇓⇓⇓⇓–6 and clinical trials were conducted in this population to evaluate potential CV benefits of ESAs. However, two large clinical trials, CHOIR and TREAT, in patients with anemia and ND-CKD, did not demonstrate a CV benefit when anemia was treated with any ESA.7,8 In the TREAT study, an increased risk of stroke was observed when compared with the placebo group (hazard ratio, 1.92; 95% CI, 1.38 to 2.68).
The mechanism for the increased risk for CV events observed when targeting high hemoglobin concentrations has not been determined. It has been suggested that exposure to high ESA doses, large hemoglobin excursions, and rapid hemoglobin rises might contribute to these risks. Thus, there is a desire to define a dosing regimen that can retain the transfusion-reduction benefits of ESA treatment, while minimizing the potential contributing factors listed above.
This START-CKD study, a Food and Drug Administration postmarketing requirement, was designed to describe the effects of a new novel treatment strategy using a fixed dose (FD) of darbepoetin. In this study, subjects with anemia and ND-CKD, without recent ESA use, were randomly allocated to treatment with an FD of darbepoetin or to treatment with darbepoetin using a hemoglobin-based titration strategy, as recommended in the United States prescribing information. This study was designed to estimate the subject incidence of RBC transfusions, and the difference in incidence of RBC transfusions between the two groups.
Methods
Study Design and Patients
A total of 126 sites across the United States conducted this double-blind, parallel-group, phase-3 trial in accordance with the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use Guidelines for Good Clinical Practice.
Subjects with anemia (hemoglobin <10 g/dl) and ND-CKD, without use of ESAs within 4 weeks of screening, were randomized to one of two dosing strategies. The trial protocol was approved by central and local institutional review boards. All trial participants provided written, informed consent before undergoing any study procedure. Amgen designed the trial, was responsible for trial oversight, and conducted the data analyses according to a prespecified statistical analysis plan. Brigham and Women’s Hospital Clinical Endpoint Center served as the external, independent, monitoring committee to adjudicate blinded safety data and the prespecified primary and secondary clinical end points. The decision to transfuse RBC was left to the discretion of the site investigator or clinician. We also captured the reasons and context for RBC transfusions.
Subjects aged ≥18 years who had two hemoglobin concentrations <10.0 g/dl at least 2 weeks apart, were not receiving dialysis, had an eGFR of <45.0 ml/min per 1.73 m2 during screening, were iron replete, and deemed clinically stable by the investigator were eligible.
In the hemoglobin-based, titration-dose (TD) group, darbepoetin doses were titrated to maintain hemoglobin ≥10.0 g/dl, with dose reduction if hemoglobin exceeded 10.5 g/dl or if the hemoglobin rate of rise (ROR) exceeded 1.0 g/dl over 4 weeks. In the FD group, subjects received an FD of darbepoetin (0.45 µg/kg). All subjects received darbepoetin as a subcutaneous injection once every 4 weeks. Randomization was stratified by RBC transfusion received within 12 months before randomization and site practice setting.
To maintain blinding, subjects’ hemoglobin was measured once every 4 weeks using a modified hemoglobin point-of-care device, which provided coded values that were entered into an interactive voice response/interactive website response system at each scheduled dosing visit. The interactive voice response/interactive website response system determined and assigned the appropriate dose and individual box(es) of investigational product (darbepoetin or placebo) for that specific visit on the basis of the hemoglobin concentration and the dosing algorithms. The starting dose of darbepoetin was 0.45 µg/kg for both groups, using the subject’s body weight at screening visit 1. Subjects randomized to the hemoglobin-based TD group had their investigational product titrated on the basis of the hemoglobin concentration on the date of the visit, the corresponding hemoglobin ROR, and the previously assigned dose (Table 1). The hemoglobin ROR was determined by the protocol-specified hemoglobin concentrations obtained from the previous dose to the current dose. When darbepoetin therapy was withheld per the dosing algorithm, placebo was administered. Once the hemoglobin fell to <10.0 g/dl, darbepoetin therapy resumed at the next lower protocol-specified dose.
Dosing algorithm for the hemoglobin-based TD group
Subjects randomized to the FD group received the same dose for the duration of the treatment period, with one exception: if the hemoglobin was >12.0 g/dl, darbepoetin therapy was withheld and placebo was administered. If the hemoglobin fell to <10.0 g/dl, darbepoetin therapy resumed at the same dose. If the subject’s mean systolic BP was ≥160 mm Hg for two consecutive visits, the dose of investigational product was withheld.
Supplemental iron therapy was administered as needed and as tolerated by the investigator, following the National Kidney Foundation Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines (transferrin saturation should be maintained at ≥20% and ferritin at ≥100 ng/ml).9
Statistical Analyses
The primary end point was receipt of one or more RBC transfusions, which was summarized by the percentage of subjects receiving at least one RBC transfusion. Secondary end points included the following: total number of units of RBC transfused, time to first RBC transfusion, average achieved hemoglobin concentration, average achieved hemoglobin concentration while receiving investigational product, and cumulative dose of darbepoetin. Safety assessments included monitoring and collection of adverse events and time to major clinical events. To ensure the 95% CI had a width of ≤10%, 300 subjects per group were needed to enroll into the study. To ensure this precision could be maintained for sensitivity analyses, we accounted for a 20% dropout, and planned for an enrollment of 375 subjects per treatment group. This total sample size of 750 subjects also ensured the width of the 95% CI of the difference in transfusion incidence rates was ≤14% (i.e., ±7% around the point estimate), assuming the proportion of subjects receiving at least one RBC transfusion during the study in both treatment groups was 25%. The assumption of 25% of subjects receiving at least one RBC transfusion was made on the basis of post hoc analyses in both treatment groups of TREAT—of subjects who met the eligibility criteria for this study (Amgen data on file)—and transfusion rates observed in other studies.10,11 This study was not designed as an inferiority trial and no formal hypothesis was tested.
The analysis of the primary end point was conducted using the full analysis set during the on-treatment period, which began from the date of randomization. Subjects were censored at the last dose period of investigational product plus 3 months, or end of study, whichever was earlier (i.e., from date of randomization to 3 months after the last dose or end of study). The proportion of subjects in each treatment group receiving at least one RBC transfusion and the difference between two groups are presented with two-sided 95% CIs from normal approximation. The relative risk ratio and its 95% CI are described using the Cochran–Mantel–Haenszel statistic, controlling for the two stratification factors (transfusion within 12 months of randomization and site practice setting). A risk ratio less than one indicates a lower event rate for the FD arm relative to the TD arm.
Sensitivity analyses of the primary end point were repeated using: (1) all randomized subjects during the on-study period (from date of randomization to end of study); and (2) completer analysis set (subjects who completed investigational product or ended due to death) during the on-study period. Sensitivity analyses adjusting for potential differential follow-up time between the two treatment groups were done using exposure-adjusted subject incidence and 95% CIs from chi-squared approximation to the Poisson distribution.
The analyses for secondary end points include: (1) The total number of units of RBC transfused per subject were analyzed using a negative binomial regression model, accounting for stratification factors and subject exposure time. (2) Time to first RBC transfusion was analyzed using Kaplan–Meier and Cox proportional hazards models. (3) Average achieved hemoglobin concentration while receiving investigational product were described using the area under the curve method. The Hodges–Lehmann median difference and nonparametric Wilcoxon rank-sum 95% CI were provided. As a sensitivity analysis, the above analyses were repeated, excluding hemoglobin concentrations measured within 3 months (84 days) from an RBC transfusion.4 Cumulative doses of darbepoetin during the study and average per 4 weeks (to adjust for investigation product exposure time) were summarized.
Subgroup analyses were repeated by subgroups of stratification factors and baseline subgroups of age, sex, race, baseline hemoglobin concentration, baseline eGFR, and diabetes. An exploratory subgroup analysis on the basis of initial hemoglobin response, defined as the change of hemoglobin concentration from baseline at week 5, is also provided. This subgroup, although assigned postbaseline at week 5, was evaluated because the starting dose in the first 4 weeks is the same for both dosing groups.
The safety analysis included all of the patients who underwent randomization and received at least one dose of darbepoetin. Clinical events collected for adjudication included all-cause mortality, stroke, transient ischemic attack, myocardial infarction, decompensated heart failure, thromboembolism, vascular access thrombosis, and the clinical context surrounding RBC transfusions. Time to adjudicated major clinical events was summarized by each treatment group using the Kaplan–Meier method.
Analyses were conducted using SAS software, version 9.4 (SAS Institute).
Results
Patient Disposition and Baseline Characteristics
The first subject was enrolled into the study on July 30, 2012; the last subject was enrolled on November 18, 2015; and the end of study date for the last subject was October 19, 2017. A total of 2443 subjects were screened, 756 subjects were enrolled, and 458 subjects completed the study. A total of 379 subjects were randomized to the FD group, and 377 were randomized into the TD group. Two subjects from the FD group did not receive investigational product. The full analysis set therefore comprises 754 subjects.
A total of 406 subjects (203 from each treatment group) discontinued from the investigational product. The most frequently cited reasons were initiation of dialysis (90 [23.8%] subjects in each treatment group) and subject request (42 [11.1%] subjects in the FD group and 40 [10.8%] in the TD group).
A total of 458 subjects (226 [59.6%] from the FD group and 232 [61.5%] from the TD group) completed the study. For the overall 298 subjects who discontinued from the study, the most frequently cited reasons were withdrawal of full consent (84 [22.2%] in FD group and 76 [20.2%] in TD group) and death (39 [10.3%] in FD group and 36 [9.5%] in TD group). Among the patients who had not died, were transplanted, or progressed to dialysis, 76% of patients in both arms were receiving treatment at 2 years, which was comparable to the TREAT study (Supplemental Figure 1, Supplemental Table 1)
Baseline demographics and clinical characteristics were balanced between the groups (Table 2).
Baseline demographics and characteristics
In the TD arm, 68% of subjects received iron supplementation, compared with 72% in the FD arm. Ferrous sulfate was prescribed in 41% versus 47%, ferumoxytol in 5% versus 6%, and iron saccharide in 15% versus 13% of patients in the TD versus the FD arm, respectively.
RBC Transfusion
In the FD group, 91 of 379 subjects (24.1%) compared with 92 of 377 subjects (24.4%) in the TD group received at least one RBC transfusion. The difference (FD−TD) in the proportions was −0.27% (95% CI, − 6.39% to 5.85%), and the stratified risk ratio (FD/TD) was 1.00 (95% CI, 0.78 to 1.29). Sensitivity analyses for exposure-adjusted subject incidence rate using an on-study approach with either all randomized subjects or with the completer analysis set showed similar results as the primary analysis (Supplemental Table 3).
Subgroup analyses (see Methods) did not show a difference in subject incidence rates of RBC transfusions between the two treatment groups (Supplemental Figure 2), except in baseline eGFR >30 ml/min per 1.73 m2, where the difference is 17.96% (95% CI, 5.83 to 30.09). The burden time to first RBC transfusion did not differ between the two groups (Figure 1, Table 3). The average hemoglobin at the time of first transfusion was similar between the two groups, as were the reasons and clinical situations for the transfusion events (Table 4).
Kaplan–Meier plot of time to first RBC transfusion shows similarity between FD and TD group. The hazard ratio and 95% CI estimates are obtained from the Cox proportional hazards model. Stratification factors are RBC transfusion received within 12 months before randomization, yes versus no; and site practice setting, nephrology versus non-nephrology.
Results for secondary end points
Clinical context around RBC transfusions
Cumulative Dose of Darbepoetin
The FD group and TD group received a median (25th percentile [Q1], 75th percentile [Q3]) cumulative dose per 4 week of darbepoetin of 30.9 (21.8, 40.0) µg and 53.6 (31.1, 89.9) µg, respectively; the median of the difference between treatment groups was −22.1 (95% CI, −26.1 to −18.1) µg (Table 3). For the duration of the study, the median (Q1, Q3) cumulative dose was 480.0 (220.0, 720.0) µg for the FD group and 740.0 (280.0, 1640.0) µg for the TD group, with a median of the difference of −285.0 (95% CI, −380.0 to −190.0) µg between the two groups. Exploratory analysis of cumulative dose by initial hemoglobin response at week 5 shows that the largest dose difference between the two arms is observed in subjects with low initial hemoglobin response, whereas the differences are smaller in groups with higher initial hemoglobin responses (Supplemental Table 2). With an additional response of >0.8 g/dl, the median dose in the TD arm was 560 versus 500 μg in the FD arm. However, quartiles <0.8 g/dl had greater median doses in the TD arm.
The histogram of cumulative dose for the TD arm as compared with the FD arm is shown in Figure 2. Compared with the FD group, a substantial proportion of patients in the TD group received cumulative doses more than the maximum seen in the FD group (1250 µg) during the study.
Histogram of cumulative dose of darbepoetin during the study shows the titration-dose (TD) group receiving higher cumulative dose than the maximum seen in fixed-dose (FD) group. The cumulative dose is the sum of all darbepoetin doses received from all study visits. The median cumulative doses in the TD arm and FD arm are 740.0 and 480.0 μg, respectively. Hb, hemoglobin.
Achieved Hemoglobin Concentration
Subjects in the TD group achieved a higher median hemoglobin concentration from the standardized area under the curve method (median [Q1, Q3], 9.9 [9.4, 10.2] g/dl) during the treatment period than did those in the FD group (median [Q1, Q3], 9.4 [8.9, 10.0] g/dl), with a median of the difference of −0.34 (95% CI, −0.46 to −0.22) g/dl (Table 3). Hemoglobin concentration at each study visit is summarized in Figure 3. Consistent results were obtained in a sensitivity analysis that excluded hemoglobin concentrations within 3 months after an RBC. A total of 252 subjects (66.8%) in the TD group experienced at least one hemoglobin concentration of <9 g/dl as compared with 277 (73.5%) in the FD arm.
Hemoglobin concentration at each study visit (median and interquartile range) while receiving investigational product is numerically higher in TD group compared to FD group.
Safety Evaluation
The incidence of adverse events was similar between both treatment arms and no new safety signal was identified (Table 5). The time to major adjudicated clinical events were similar in both treatment groups (Table 6).
Treatment-related adverse events system organ class with >1% incidence in any treatment group
Time to major clinical events
Discussion
The optimal strategy for dosing ESA in patients with CKD is not well defined. Our clinical trial provides a unique observation indicating that a dosing strategy using a low FD of ESA is as effective as using a titrated approach in minimizing the risk for RBC transfusions, over the course of approximately 2 years, for patients with CKD not on hemodialysis. Moreover, our results indicate a lower cumulative dose of ESA and fewer excursions >10 g/dl, with a similar degree of safety and tolerability with the FD strategy. In this study, we collected data on the context of the RBC transfusion. The main reason for transfusion was symptoms that were principally constitutional. Bleeding accounted for approximately 20% of transfusions. The hemoglobin at the time of transfusion was consistent with transfusion practice guidelines. Although the average hemoglobin was modestly higher in the titrated group, this is not surprising given the higher cumulative dose of darbepoetin administered to the titrated group.
Before the approval of ESA in 1989, RBC transfusions were the primary treatment to alleviate the symptoms of fatigue and lassitude in patients with CKD. After the approval of ESAs, transfusion rates declined substantially so that, instead of nearly 60% of patients on dialysis receiving transfusions, it is now typically <10%.12⇓–14 In the early 1980s, RBC transfusions were considered by many experts to be helpful in improving kidney transplant graft survival.15 Over time, however, it became apparent there were numerous risks associated with RBC transfusions, including allosensitization, transmission of blood-borne diseases, and the potential for iron, potassium, and volume overload.16⇓⇓⇓⇓⇓⇓–23 Consequently, strategies to prevent the need for RBC transfusion is of great importance in patients with advanced CKD.
Because clinical trials in patients with CKD and anemia using ESAs to improve hemoglobin have an increased risk for CV morbidity and mortality,7,8 most guidelines and regulatory agencies encouraged hemoglobin targets in the range of 10–11 g/dl. This lower target hemoglobin goal has been associated with increased RBC transfusion rates among patients on hemodialysis.24⇓⇓–27 Thus, there is a need to explore more optimal strategies for using ESAs to prevent RBC transfusions, preferably with lower doses of ESA, and with less fluctuation of hemoglobin levels.
Our study also provides new information on the setting of CKD surrounding the clinical indication for an RBC transfusion (Table 4). Thus, it provides the opportunity to examine how to optimally manage hemoglobin and avoid transfusions in patients with advanced ND-CKD. However, we cannot assume that we would see similar results in patients with CKD on dialysis.
Subgroup analyses demonstrated a somewhat higher incidence of RBC transfusion in those assigned to the FD group with a baseline eGFR >30 ml/min per 1.73 m2. Interestingly, the largest ESA cumulative dose difference between the two different treatment strategies was observed in patients with a lower initial hemoglobin response, whereas the differences were smaller in groups with higher initial hemoglobin responses to ESA. Thus, patients who are more ESA resistant may benefit the most when using a low FD approach, as opposed to titrating the dose of ESA to a much higher level.
Our study has some limitations. First, a relatively high fraction of participants discontinued the study drug before completion of the study, primarily due to development of ESKD requiring RRT. We could not control for the event, which was anticipated during the design of the study. It was not considered practical to continue the study protocol after this end point occurred. Another cause of discontinuation was withdrawal of consent, which was much higher than anticipated. We are uncertain as to why the rate was so high given that many of these patients would have been treated with a subcutaneous injection of an ESA as part of the routine clinical care including repeated measures of hemoglobin and ESA dosing. We did not prescribe nor control for the administration of RBC transfusion. Finally, because the study sample size was planned to allow for estimations with acceptable precision, it was not powered for hypothesis testing to detect a difference between groups. Still, the observed proportion of subjects receiving RBC transfusion in the study turned out to be in agreement with the assumption behind this sample size (25%), lending credibility to our findings.
In conclusion, we found that an FD versus titration-based dosing regimen of darbepoetin in patients with ND-CKD and anemia was associated with a similar rate of RBC transfusion and resulted in a lower cumulative dose of darbepoetin. Our study results are timely and important in the current healthcare environment, emphasizing improved patient outcomes and better utilization of healthcare resources, and represent an important step in better understanding optimal utilization of ESA in advanced CKD to avoid risks associated with RBC transfusions.
Disclosures
J. Berns reports being the Nephrology Subspecialty Board Chair of ABIM and the Chair of the ABIM Council, receiving honoraria from ABIM and UpToDate, and being the deputy editor of the American Journal of Kidney Diseases. E.F. Lewis reports receiving grants and personal fees from Amgen, and grants from Akebia, outside the submitted work. J. Petersen and Q. Tran are Amgen employees and stockholders. R. Toto reports serving as a consultant to Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, IQVIA/Akebia, Medscape, Quest Diagnostics, Quintiles, Reata, and Relypsa. M. Weir reports being on the scientific advisory board for AbbVie, Amgen, AstraZeneca, Boehringer Ingelheim, Janssen, MSD, Relypsa, and Vifor, and reports receiving personal fees from Akebia/Otsuka, Boston Scientific, and Merck, outside the submitted work.
Funding
The trial was funded by Amgen. M. Weir was supported by National Institutes of Health grants R01DK066013, R01HL147422, R01132732, U01DK106102, and U01116095.
Acknowledgments
We thank the patients and their caregivers; the investigators and their teams who contributed to this trial; and Dr. Jerome Rossert, Dr. Maple Fung, Dr. Rajani Dinavahi, Dr. Wendy Carroll, and Dr. Kurt Olsen for contributions to the trial design, execution, and analysis planning of the trial.
Dr. Qui Tran analyzed the data; Dr. Jeffrey Petersen, Dr. Robert Toto, and Dr. Matthew R. Weir drafted the article; Dr. Jeffrey S. Berns and Dr. Eldrin Foster Lewis revised the article; and all authors approved the final version of the manuscript.
Supplemental Material
This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2020050556/-/DCSupplemental.
Supplemental Figure 1. Subject disposition (CONSORT diagram).
Supplemental Figure 2 Forest plot of subgroup analysis of subject incidence of transfusion.
Supplemental Table 1. Subject disposition with discontinuation reason.
Supplemental Table 2. Subgroup of cumulative dose by initial hemoglobin response.
Supplemental Table 3. Primary and sensitivity analysis of the primary endpoint of subject incidence of red blood cell transfusion.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
- Copyright © 2021 by the American Society of Nephrology
References
In this issue
Jump to section
More in this TOC Section
Cited By...
- No citing articles found.