Predictors of Hyporesponsiveness to Erythropoiesis-Stimulating Agents in Patients with Non-Dialysis-Dependent Chronic Kidney Disease (RADIANCE-CKD Study) Open Access

Hyporesponsiveness to erythropoiesis-stimulating agents (ESAs) has been associated with increased mortality and cardiovascular events in patients with non-dialysis-dependent chronic kidney disease (NDD-CKD) or maintenance hemodialysis. Inflammation, iron metabolism, and nutritional disturbance have been implicated in this hyporesponsiveness [1, 2]. Although it has been reported that ESA-treated patients who rapidly achieve and maintain a target hemoglobin (Hb) level of 11 g/dL in a short period of time and maintain this target level have a low risk of renal death [3], excessive Hb increases have been associated with an increased risk of death and heart failure [4], and the target Hb level and rate of Hb increase are controversial. The recent introduction of hypoxia-inducible factor prolyl hydroxylase (HIF-PH) inhibitors has increased the treatment options for chronic kidney disease patients with anemia. HIF-PH inhibitors have the potential to improve anemia in ESA-hyporesponsive patients, but there are still unresolved issues related to ESA hyporesponsiveness, including concerns about adverse effects on malignancy and diabetic retinopathy [5].

The erythropoietin resistance index (ERI) is used as a measure of ESA responsiveness [2], and can be obtained simply by dividing the total weekly erythropoietin dose first by the patient’s weight (in kilograms) and then by the patient’s Hb level (in g/dL); it is expressed in units/week per kg per g/dL [6]. However, the ERI can only be calculated after the actual administration of an ESA. The epoetin beta-pegol formulation, or continuous erythropoietin receptor activator (CERA), has a significantly longer half-life in the blood than conventional ESA formulations and was shown to be significantly more effective in maintaining target Hb levels in a randomized controlled trial [7]. A multicenter, open-label, randomized, parallel-group study was conducted to evaluate the renal prognosis of different target Hb levels during CERA administration in patients with conservative ESA-hyporesponsive renal anemia and to determine the target Hb level during ESA administration [8]. Among the patients included in this previous comparative study, differences in response to ESA were evident. We conducted the present study as a sub-analysis of this comparative trial because we believe that if we could predict whether a patient would be an ESA hypo-responder when receiving ESA, it would be very useful in deciding on a treatment plan.

This study was a post hoc sub-analysis of the article “Renal prognoses by different target hemoglobin levels achieved by epoetin beta pegol dosing to chronic kidney disease patients with hyporesponsive anemia to erythropoiesis-stimulating agent: a multicenter open-label randomized controlled study” [8]. After this trial assignment, a landmark analysis [9] at 3 months into the trial was used to identify predictors of ESA-hyporesponsive patients.

The trial [8] enrolled Japanese NDD-CKD patients with ESA-hyporesponsive anemia who met all three of the following inclusion criteria: (i) Hb level ≥8 g/dL but <11 g/dL at eligibility and confirmed ESA-hyporesponsiveness; (ii) dialysis initiation not planned for at least 6 months from the start of study treatment; and (iii) at least 20 years of age at the time of consent. Enrolled patients were assigned to an active treatment group (n = 190 patients) with the goal of achieving an Hb level of 11 g/dL or higher with CERA and a maintenance treatment group (n = 192 patients) with the goal of maintaining the Hb level at enrollment ±1 g/dL. After the exclusion of ineligible patients, a final group of 362 cases (183 and 179 from the active and maintenance treatment groups, respectively) were analyzed in the trial.

In the present study, we performed a sub-analysis of the active treatment group of this RCT. The landmark point was set 3 months after study entry, and 161 patients in the active treatment group who were still receiving treatment at that time were included in the landmark analysis [9]. Cases were stratified into tertiles according to the ERI [6] at 3 months after study entry. A renal prognostic event was defined as any of the following: (i) conversion to renal replacement therapy (dialysis or renal transplantation); (ii) a decrease in estimated glomerular filtration rate (eGFR) to <6 mL/min/1.73 m²; or (iii) a decrease in eGFR of >30% from baseline. Renal survival was defined as the time from the start of the trial to the earliest posttrial event, and the incidence of renal prognostic events and the incidence of conversion to renal replacement therapy were evaluated in the low, middle, and high tertile ERI groups, respectively.

Furthermore, the high ERI-tertile group was defined as the ESA hypo-responder group, and the low and middle ERI tertiles were defined as the non-ESA hypo-responder group. A multivariate logistic regression model was used to assess the contribution of each of the following items that may influence ESA hyporesponsiveness and to estimate ESA hyporesponsiveness based on the area under the receiver operating characteristic (ROC) curve (AUC): 10 background factors (age, weight, Hb level, eGFR, systolic blood pressure, serum albumin level, urine protein, serum ferritin level, transferrin saturation [TSAT], and C-reactive protein [CRP]), 9 items including gender and comorbidity (gender, primary disease, smoking history, heart disease, brain disease, diabetes, hyperlipidemia, diabetic retinopathy, and hypertensive retinopathy), and 5 serum test items that affect anemia (serum vitamin B12, serum vitamin B6, and serum folic acid as factors related to red blood cell production; serum parathyroid hormone [PTH] and serum 25-hydroxyvitamin D [25(OH)D] as factors related to parathyroid function). Among the existing risk factors, i.e., 10 background factors and 9 comorbidities, combination of the 6 factors with the largest χ² statistics were selected by multiple logistics regression to calculate AUC and others.

Items for risk factors at the start of the trial are presented as the mean ± standard deviation, median (interquartile range), or a percentage. Differences in each item between the ESA hypo-responder and non-ESA hypo-responder groups were tested by t test, Wilcoxon rank sum test, and Fisher’s exact test. To construct a predictive model for ESA low responder cases, the background factors and serum test items described in the previous section at the start of the study were included in the logistic regression model to estimate the AUC and its 95% confidence interval (CI). In addition, sensitivity and specificity were calculated using the point with the smallest distance from the upper left corner of the ROC curve as the threshold. Values of p < 0.05 were considered statistically significant. The statistical software used was SAS, version 9.4 (SAS Institute, Cary, NC, USA).

Subjects were stratified into tertiles by ERI at the landmark time point of 3 months into the trial. We classified 53 patients in the low tertile (mean 0.027 units/week per kg per g/dL), 54 in the middle tertile (0.048), and 54 in the high tertile (0.085) group (Table 1). Figure 1 shows the changes in Hb levels for each of the three groups from the start of the study to the landmark time point at 3 months.

Based on the ERI at 3 months, we evaluated risk factors at the start of the trial for the true ESA hypo-responder group, i.e., the group with high ERI tertile at 3 months (n = 54), as well as for the non-ESA hypo-responder group (n = 107). Examination of 10 background factors (Table 2) showed that the ESA hypo-responder group had significantly higher CRP and lower weight and Hb levels than the non-ESA hypo-responder group. The mean dose of CERA at the landmark was 45.6 μg/week in the ESA hypo-responder group and 22.8 μg/week in the non-ESA hypo-responder group, and the mean Hb level was 10.3 g/dL in the ESA hypo-responder group and 10.8 g/dL in the non-ESA hypo-responder group, both of which were significantly different between the two groups. In the examination of 9 items (Table 2) including gender and comorbidities, the ESA hypo-responder group had a significantly higher proportion of individuals with a history of smoking compared with the non-ESA hypo-responder group. Examining 5 serum test items (Table 2) showed no significant differences between the two groups.

The ROC curves for models 4 and 5 are shown in Figure 2. Renal events occurred in 29 of 54 patients (54%) in the ESA hypo-responder group and 55 of 107 patients (51%) in the non-ESA hypo-responder group. In addition, 17 patients (32%) in the ESA hypo-responder group and 24 patients (22%) in the non-ESA hypo-responder group crossed over to renal replacement therapy. Neither renal events nor the incidence of switching to renal replacement therapy were significantly different between the two groups.

Previous studies have shown that ESA-hyporesponsive anemia is associated with increased mortality and cardiovascular events, and inflammation, iron metabolism, and nutritional disorders have been implicated in ESA hyporesponsiveness [1, 2, 10, 11]. This study included a group of NDD-CKD patients with preexisting ESA resistance who were highly ESA resistant after 3 months of active ESA administration and a group of patients who were not. In these patients, we investigated whether risk factors prior to the initiation of ESA therapy could be used to predict whether patients would have ESA hyporesponsiveness when ESA administration. This study also examined the incidence of renal events in true ESA-refractory patients enrolled in the trial who were assigned to the active treatment arm but did not have an adequate Hb increase at 3 months.

When a prediction equation was created using 19 existing risk factors, including 10 background factors and 9 complications, the AUC was found to be 0.783, effectively predicting the ESA hypo-responder group. Appropriate combinations of variables were explored to reduce the number of items in the prediction equation. The prediction equation created using 6 existing risk factors (Hb level, systolic blood pressure, weight, gender, smoking, and presence of hypertensive retinopathy) and 5 serum test items that affect anemia (vitamin B12, vitamin B6, folic acid, PTH, and 25(OH)D), for a total of 11 variables, had sufficient predictive power with an AUC of 0.736. If these prediction equations can be used to predict whether a patient is an ESA hypo-responder when ESA administration, then the use of non-ESA agents such as HIF-PH inhibitors as therapeutic agents can be seriously considered. In addition, when administering ESA, it is natural to be cautious about elevated blood pressure and the induction of thromboembolism associated with rapid and excessive improvement of anemia, but for decisions such as whether to set a higher initial dose in advance or to increase the dose earlier, it may be helpful to know that a patient is predicted to be an ESA hypo-responder. Since, to our knowledge, no previous report has shown that hyporesponsiveness to ESA can be predicted from a combination of existing risk factors and serum test items, we believe the present report makes a unique contribution.

There are scattered reports of poor renal [1, 3, 12] and life prognosis [10] in patients with NDD-CKD whose anemia does not improve sufficiently with the use of ESA preparations, although it should be noted that these studies used different definitions of ESA hypo-responders. For the CERA administered in this study, a pooled analysis of clinical trials with this drug reported that it was associated with decreased renal survival in patients with ESA hyporesponse, consistent with results with conventional ESA preparations [13]. In the trial whose dataset was analyzed here [8], there was no predominant difference in the incidence of renal prognostic events, including renal replacement therapy, between the intensive ESA treatment group and the conservative treatment groups. In addition, in this previous trial divided the study population of ESA hypo-responders into a group of true ESA hypo-responders who showed ESA resistance despite 3 months of active treatment, and a group of the remaining (“non-true”) ESA hypo-responders, and found no significant difference in the incidence of renal prognostic events between the two groups, similar to the results of the present study. The patients included in our present analysis had a mean age of 75 years, of whom 43% had diabetes mellitus, and a mean eGFR of 15.3 mL/min/1.73 m² at trial initiation. Compared to previous reports [1, 3, 12], the mean age was higher and renal function was lower at baseline. The higher mean age, lower renal function at baseline, and more importantly, the enrollment of a group of patients with preexisting ESA resistance may have contributed to the lack of significant differences in renal outcomes in this study. As this study did not investigate the relationship between ESA dose and renal prognosis, the effect of high ESA dose on renal prognosis is unknown. In a previous report, the ESA hypo-responder group was found to have a poor nutritional status as indicated by low BMI and hypoalbuminemia, high CRP, and significantly low serum iron and TSAT, i.e., iron deficiency [14]. In our present analysis, which examined ESA resistance at 3 months, 3 of the 10 background factors showed significant differences between ESA hypo-responders and non-ESA hypo-responders – namely, the former group had significantly lower body weight and Hb levels and significantly higher CRP, similar to previous reports, but there were no significant differences in albumin. As markers of iron metabolism, ferritin and TSAT were not significantly different.

A limitation of this study is that the subjects of the trial [8] on which it was based were Japanese NDD-CKD patients with preexisting ESA hypo-responsive renal anemia, i.e., this study was a stratified comparison among the ESA hypo-responsive population, which may explain why no difference was found between the ESA hypo-responsive group as we defined it and the other group. In addition, this study was designed to be completed 21 months after the start of the trial, resulting in a shorter observation period to assess differences in renal outcomes compared to some previously published studies, and the fact that the patients were older and had more impaired renal function than in previously published studies may have affected the results. In conclusion, our results suggest that NDD-CKD patients with ESA resistance can be predicted when ESA administration using existing risk factors and serum test items that affect anemia.

We would like to express our sincere gratitude to Dr. Yasuo Ohashi of the Department of Integrated Science and Engineering for Sustainable Society, Chuo University, for his valuable guidance in compiling this paper.

This study protocol was reviewed and approved by the Non-Profit Organization MINS Institutional Review Board and the Non-Profit Organization MINS Research Ethics Committee, approval number 72. The study was registered in the University Hospital Medical Information Network (UMIN) database (UMIN000008617). Written informed consent was obtained from study participants. This study included patients who were enrolled in a previously published randomized controlled trial [8] designed to evaluate renal prognosis in anemic patients with NDD-CKD with ESA hyporesponse by setting different target Hb levels. This study was a post hoc sub-analysis of randomized controlled trial (RCT). Human rights were well protected.

K.Y. reports receiving grants or contracts from Kyowa Kirin and Mitsubishi Tanabe Pharma, and payment or honoraria from AstraZeneca, Kyowa Kirin, and Mitsubishi Tanabe Pharma; H.Y. reports receiving payment or honoraria from Bayer Yakuhin, Chugai Pharmaceutical, Astellas Pharma, and Kyowa Kirin; K.T. reports receiving grants or contacts from Kyowa Kirin, Chugai Pharmaceutical, Torii Pharmaceutical, Bayer Yakuhin, and Mitsubishi Tanabe Pharma, consulting fees from Mitsubishi Tanabe Pharma, and Astellas Pharma, and payment or honoraria from Mitsubishi Tanabe Pharma, Kyowa Kirin, Astellas Pharma, Bayer Yakuhin, Kissei Pharmaceutical, Torii Pharmaceutical, and Chugai Pharmaceutical; M.N. reports receiving grants or contracts from Chugai Pharmaceutical, and payment or honoraria from Chugai Pharmaceutical; T.H. reports receiving payment or honoraria from Kyowa Kirin and Torii Pharmaceutical; and Y.U. reports receiving payment or honoraria from Daiichi Sankyo and Eli Lilly Japan, and support for attending meetings and/or travel from Chugai Pharmaceutical. All remaining authors have nothing to disclose.

The funding for this study was provided by Chugai Pharmaceutical Co., Ltd., on the basis of a commissioning contract, and was provided to EPS Corporation (Japanese contract research organization) that handled administration tasks, etc., for the study secretariat and data center. The funding party, Chugai Pharmaceutical Co., was involved with conception of the study and provision of information, but not with planning or conducting the study, or analyzing or interpreting the results.

Conceived and designed the study and analyzed the data: K.M., K.Y., and Y.U. Collected the data and contributed to the writing and editing of the manuscript: K.Y., H.Y., K.T., Hiroki Hase, S.N., M.N., T.W., T.H., Y.U., and Hideki Hirakata. K.M. wrote the first draft of the manuscript. All authors agreed with manuscript results, conclusions, and publication.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.