Roxadustat

Roxadustat Treatment for Anemia in Patients Undergoing Long-Term Dialysis

ABSTRACT
BACKGROUND
Roxadustat is an oral hypoxia-inducible factor prolyl hydroxylase inhibitor that stimulates erythropoiesis and regulates iron metabolism. Additional data are needed regarding the effectiveness and safety of roxadustat as compared with standard therapy (epoetin alfa) for the treatment of anemia in patients undergoing dialysis.In a trial conducted in China, we randomly assigned (in a 2:1 ratio) patients who had been undergoing dialysis and erythropoiesis-stimulating agent therapy with epoetin alfa for at least 6 weeks to receive roxadustat or epoetin alfa three times per week for 26 weeks. Parenteral iron was withheld except as rescue therapy. The primary end point was the mean change in hemoglobin level from baseline to the average level during weeks 23 through 27. Noninferiority of roxadustat would be established if the lower boundary of the two-sided 95% confidence interval for the difference between the values in the roxadustat group and epoetin alfa group was greater than or equal to −1.0 g per deciliter. Patients in each group had doses adjusted to reach a hemoglobin level of 10.0 to 12.0 g per deciliter. Safety was assessed by analysis of adverse events and clinical laboratory values.

RESULTS
A total of 305 patients underwent randomization (204 in the roxadustat group and 101 in the epoetin alfa group), and 256 patients (162 and 94, respectively) completed the 26-week treatment period. The mean baseline hemoglobin level was 10.4 g per deciliter. Roxadu- stat led to a numerically greater mean (±SD) change in hemoglobin level from baseline to weeks 23 through 27 (0.7±1.1 g per deciliter) than epoetin alfa (0.5±1.0 g per deciliter) and was statistically noninferior (difference, 0.2±1.2 g per deciliter; 95% confidence inter- val [CI], −0.02 to 0.5). As compared with epoetin alfa, roxadustat increased the transferrin level (difference, 0.43 g per liter; 95% CI, 0.32 to 0.53), maintained the serum iron level (difference, 25 μg per deciliter; 95% CI, 17 to 33), and attenuated decreases in the transfer- rin saturation (difference, 4.2 percentage points; 95% CI, 1.5 to 6.9). At week 27, the de- crease in total cholesterol was greater with roxadustat than with epoetin alfa (difference,−22 mg per deciliter; 95% CI, −29 to −16), as was the decrease in low-density lipoprotein cholesterol (difference, −18 mg per deciliter; 95% CI, −23 to −13). Roxadustat was associ- ated with a mean reduction in hepcidin of 30.2 ng per milliliter (95% CI, −64.8 to −13.6), as compared with 2.3 ng per milliliter (95% CI, −51.6 to 6.2) in the epoetin alfa group. Hyperkalemia and upper respiratory infection occurred at a higher frequency in the roxa- dustat group, and hypertension occurred at a higher frequency in the epoetin alfa group.CONCLUSIONS:Oral roxadustat was noninferior to parenteral epoetin alfa as therapy for anemia in Chi- nese patients undergoing dialysis. (Funded by FibroGen and FibroGen [China] Medical Technology Development; ClinicalTrials.gov number, NCT02652806.)

N CHINA, 120 MILLION PERSONS HAVE chronic kidney disease, a prevalence that is projected to increase.1,2 Anemia is present in more than 90% of the 500,000 patients who undergo dialysis3 and is a complication that con- tributes to increased morbidity and mortality.4 Treatment for anemia is recommended by clinical practice guidelines.5-8 However, studies link the use of high-dose erythropoiesis-stimulating agents to increased risks of cardiovascular events and death.9-11 Only half the patients undergoing dialy- sis in China reach a hemoglobin level of 10.0 g per deciliter or greater using recombinant erythropoie- tin therapy. This apparent undertreatment may re- sult from the cost of the medication, hyporespon- siveness due to inflammation, or iron depletion.12 The kidneys of patients with kidney disease retain the ability to produce erythropoietin.13,14 Levels of hypoxia-inducible factor (HIF) change according to changes in oxygen tension through oxygen-sensing prolyl hydroxylase enzymes.15 When oxygen levels decrease, prolyl hydroxylase enzyme activity decreases, resulting in the ac- cumulation of HIF-α subunits and an increase in HIF transcriptional activity, which induces the expression of erythropoietin, erythropoietin re- ceptors, and proteins that promote intestinal absorption of iron and recycling of iron from the macrophage iron storage system.16

Roxadustat (FG-4592) is a potent, reversible, HIF prolyl hydroxylase inhibitor that mimics the natural response to hypoxia. The intermittent dosing strategy with roxadustat for the treat- ment of anemia17 in patients with chronic kidney disease was developed to permit durable mainte- nance of effect.18-20 With a half-life of approxi- mately 10 hours,21 roxadustat, administered three times per week, enables HIF transcriptional activity to return to baseline between doses, which results in the intermittent induction of hypoxia-inducible target genes involved in eryth- ropoiesis.16,22,23 Previous phase 2 trials tested the efficacy of roxadustat in patients in China who had chronic kidney disease–related anemia.24 We now report the results of a 6-month, phase 3 trial involving patients undergoing dialysis in China.ating the efficacy and safety of roxadustat for the treatment of anemia in patients undergoing dialysis in China. The protocol, available with the full text of this article at NEJM.org, was ap- proved by regulatory authorities and ethics com- mittees, and the trial was conducted in accordance with local regulatory and ethics requirements.The trial was designed by the first two authors and the sponsor (FibroGen). The sponsor pro- vided financial support and was responsible for data collection and analysis. All the authors had full access to the trial data and analyses and contributed to data analysis and interpretation and to the conduct of the trial. An author who is an employee of the sponsor wrote the first draft of the manuscript.

All the authors reviewed the manuscript. The authors vouch for the accuracy and completeness of the data and for the fidelity of the trial to the protocol.Eligible patients were 18 to 75 years of age, had end-stage kidney disease, had received dialy- sis for at least 16 weeks, had been receiving stable doses of epoetin alfa for at least 6 weeks, and had a mean hemoglobin value (from the last two screening assessments) of 9.0 to 12.0 g per deciliter. A list of inclusion and exclusion criteria is provided in Table S1 in the Supplementary Ap- pendix, available at NEJM.org.Eligible patients underwent randomization in a 2:1 ratio to receive either oral roxadustat or par- enteral epoetin alfa (ESPO, Kyowa Hakko Kirin) three times per week for 26 weeks. (Kyowa Hakko Kirin had no role in the trial.) Random- ization was performed centrally in sequence, stratified according to the dose of epoetin alfa at baseline (<8000 IU or ≥8000 IU per week) and dialysis method (hemodialysis or peritoneal dialysis).The starting dose of roxadustat was either 100 mg (in patients weighing 45 to <60 kg) or 120 mg (in patients weighing ≥60 kg). Patients who had been randomly assigned to receive epoe- tin alfa continued their prerandomization doses. Both epoetin alfa and roxadustat were supplied by the sponsor. Doses were adjusted so that the patient would have a hemoglobin level of 10.0 to12.0 g per deciliter (Table S2 in the Supplemen- tary Appendix). The use of oral iron therapy was allowed; intravenous iron therapy was prohibited except as rescue therapy. Rescue therapy included intravenous iron, blood transfusion, or erythro-poiesis-stimulating agents (or a combination of these treatments) in patients who had a hemo- globin level of less than 8.0 g per deciliter or in patients who had a hemoglobin level of less than9.0 g per deciliter as well as a confirmed de- crease from baseline of more than 1.0 g per deciliter. The primary efficacy end point was the mean change in the hemoglobin level from baseline to the average level during weeks 23 through 27. The prespecified noninferiority analyses were conducted in the full analysis set–intention to treat population and the per-protocol population, according to Chinese regulatory guidance.25 The full analysis set–intention to treat population (hereafter, the intention-to-treat population) in- cluded all the patients who had undergone ran- domization and had baseline and postbaseline hemoglobin values assessed during treatment. The intention-to-treat analyses were performed according to the randomly assigned treatment group. The per-protocol population included all the patients who had undergone randomization, received at least 2 weeks of treatment, had base- line and postbaseline hemoglobin values assessed without the use of rescue therapy in the preceding 6 weeks, and had no major protocol violations. Secondary efficacy end points were examined in both the intention-to-treat population and the per-protocol population; results from the intention- to-treat population are presented unless other- wise specified. The secondary efficacy end points were the following: the proportion of patients with a hemoglobin response (defined as a mean hemoglobin level, averaged over weeks 23 through 27, that was no lower than 1.0 g per deciliter below baseline); the proportion of patients with a mean hemoglobin level, averaged over weeks 23 through 27, of at least 10.0 g per deciliter; the mean change from baseline in the total choles- terol level, averaged over weeks 25 through 27; the mean change from baseline in iron biomarker levels at week 27; the first exacerbation of hyper- tension in a time-to-event analysis; and the mean change from baseline in the mean arterial blood pressure measured before the start of a dialysis session, averaged over weeks 23 through 27. Ex- ploratory analyses of the hemoglobin treatment effect on the basis of inflammatory status, as assessed by the C-reactive protein level, were conducted, as specified in the protocol. For the primary efficacy analysis, we calculated that the inclusion of 300 patients would provide the trial with 90% power to test the noninferior- ity26 of roxadustat to epoetin alfa (margin for hemoglobin level, 1.0 g per deciliter, as estab- lished in a previous phase 3 trial of erythropoi- esis-stimulating agents27). The mean change in the hemoglobin level from baseline to the aver- age level during weeks 23 through 27 was com- pared with the use of the mixed-model, repeated- measure model. The model included treatment, visit (class effect), treatment by visit, baseline dose of epoetin alfa (<8000 IU or ≥8000 IU per week), and dialysis method as fixed effects and the baseline hemoglobin level as a covariate, with an unstructured covariance matrix within each treatment group for a repeated-measures covariance structure. The baseline hemoglobin level was defined as the mean of the last three hemoglobin levels before the first dose of a trial drug. The 95% confidence interval for the treat- ment difference was constructed with the use of least-squares means. In order for the trial to show the noninferiority of roxadustat to epoetin alfa, the lower boundary of the 95% confidence interval for the treatment difference in the change in hemoglobin level had to be greater than or equal to −1.0 g per deciliter. Sensitivity analyses with analysis of covari- ance (ANCOVA) were performed in which miss- ing hemoglobin values were imputed with the use of the Markov chain Monte Carlo method,28,29 which assumes multivariate normal distribution of hemoglobin values, to create a data set of ob- served plus imputed data. We used the ANCOVA model to analyze the change from baseline aver- aged over weeks 23 through 27 from each impu- tation with the same covariates as the mixed- model repeated-measures analysis. This process was repeated 1000 times to generate many data sets with imputed data, and the results of the analysis were summarized with adjustment for the variances from the imputations with the use of a multiple-imputation technique.The binary response end points for hemoglo- bin values were assessed for noninferiority with the approach of Miettinen and Nurminen,30 with adjustment for randomization stratification fac- tors on multiple-imputation data. The number of patients, proportion of patients with a response, and treatment differences were averaged from the multiple imputations. The 1000 multiple- NEJM.ORG imputed data analysis results were summarized for the treatment comparison with the use of the multiple-imputation technique. Noninferiority testing (with a noninferiority margin of 15 per- centage points) was prespecified for the two hemoglobin secondary end-point measures (the proportion of patients with hemoglobin response and proportion with a mean hemoglobin level of ≥10 g per deciliter). Superiority testing was planned for the other secondary end-point measures. We used the same mixed-model repeated- measures method to analyze the mean changes from baseline in the iron biomarker levels at week 27 and in the total cholesterol level aver- aged over weeks 25 through 27. We calculated the 95% confidence interval for the treatment difference that was based on the least-squares means from the mixed-model repeated-measures analysis. Since the analyses of secondary end points were not adjusted for multiple compari- sons, we report point estimates and 95% confi- dence intervals without P values. The 95% con- fidence intervals have not been adjusted for multiple comparisons, and inference drawn from them may not be reproducible. Safety was monitored by assessment of ad- verse events and serious adverse events during treatment for 28 days after the discontinuation of trial drug, by review of clinical laboratory values, and by physical examinations. The safety information is reported up to and including 2 days after the discontinuation of the trial drug (approximately four half-lives of roxadustat). The number and proportion of patients who received rescue therapy during trial treatment and the time to rescue therapy from the first dose during treatment were also reported. From December 2015 through June 2016, a total of 305 patients underwent randomization (204 patients to the roxadustat group and 101 to the epoetin alfa group) (Fig. S1 in the Supplemen- tary Appendix). One patient in the epoetin alfa group did not receive treatment, so 304 patients were included in the full analysis set (intention- to-treat population). The per-protocol population comprised 196 patients in the roxadustat group and 98 patients in the epoetin alfa group. A total of 48 patients (42 in the roxadustat group and 6 in the epoetin alfa group) discontinued the assigned medication. A total of 256 patients (162 in the roxadustat group and 94 in the epoetin alfa group) completed treatment, for a total of 88.3 patient-years in the roxadustat group and 48.1 patient-years in the epoetin alfa group. The baseline characteristics of the patients were similar in the two groups (Table 1). Over- all, the mean hemoglobin level of the patients was 10.4 g per deciliter, and the mean dose of epoetin alfa was approximately 7500 units per week. Approximately 80% of the patients (247 patients) had a transferrin saturation (the per- centage of transferrin, an iron-carrier protein, occupied by iron) of at least 20%, and 65% of the patients (198) had a ferritin level of at least 200 μg per liter. Approximately 20% of the patients (66) had a C-reactive protein level above the upper limit of the normal range. Roxadustat treatment resulted in a numerically greater mean (±SD) increase in the hemoglobin level of 0.7±1.1 g per deciliter than did epoetin alfa treatment (0.5±1.0 g per deciliter) and was noninferior to epoetin alfa in both the intention- to-treat population and the per-protocol popula- tion (treatment difference in the intention-to-treat population, 0.2±1.2 g per deciliter; 95% confi- dence interval [CI], −0.02 to 0.5) (Fig. 1A). The percentage of patients with a hemoglobin re- sponse (hemoglobin level not <1.0 g per deciliter below the baseline value) was 92.5% in the roxa- dustat group (189 patients) and 92.5% in the epoetin alfa group (92 patients) in weeks 23 through 27, resulting in a treatment difference of 0.2 percentage points (95% CI, −7.1 to 7.6). The percentage of patients with a mean hemo- globin level of at least 10.0 g per deciliter was 87.0% in the roxadustat group (178 patients) and 88.5% in the epoetin alfa group (88 patients) in weeks 23 through 27 (treatment difference, −0.1 percentage point; 95% CI, −8.6 to 8.5). The treat- ment difference in the change in hemoglobin level was similar according to the ANCOVA mul- tiple imputations (increase in hemoglobin level, 0.7±1.1 g per deciliter in the roxadustat group and 0.5±1.0 g per deciliter in the epoetin alfa group).patients undergoing hemodialysis or peritoneal dialysis in China. The percentage of patients who received rescue therapy or who did not reach the lower end of the hemoglobin target range (10 g per deciliter) did not differ substantially between groups.Patients in the epoetin alfa group who had elevated C-reactive protein levels had lower hemo- globin responses than those with normal C-reac- tive protein levels, despite receiving higher doses of epoetin alfa — a finding that is consistent with results in published studies showing that inflammation suppresses response to erythro- poiesis-stimulating agents.31,32 In contrast, and in a finding consistent with results in phase 2 stud- ies of roxadustat, it was suggested that apparent inflammation, as assessed on the basis of C-reac- tive protein levels, did not appear to affect the hemoglobin response with roxadustat.33 In the present trial, among patients with elevated C- reactive protein levels, patients in the roxadustat group had a greater increase in the hemoglobin level than those in the epoetin alfa group. In- flammation is known to increase the hepcidin level, resulting in functional iron deficiency. We speculate that the hepcidin level–lowering effect that has been associated with roxadustat and the mobilization of internal iron stores may have contributed to these findings.The use of intravenous iron therapy was re- stricted in both groups — a design that was based on the previous observation that oral iron therapy provided results equivalent to those of intravenous iron therapy with roxadustat.34 The mechanism of action of epoetin alfa is limited to stimulation of the erythropoietin receptor; oral iron is expected to be ineffective relative to intra- venous iron with epoetin alfa in the treatment of anemia in patients undergoing dialysis.35 Overall, changes in iron biomarker levels showed improve- ment with roxadustat as compared with epoetin alfa. The serum iron level is strongly affected by the serum transferrin level, which is increased with roxadustat. The attenuation of the decrease in transferrin saturation with roxadustat as com- pared with epoetin alfa, despite the increase in transferrin level, supports an effect on enteric iron absorption with roxadustat. Improvements in iron delivery to the bone marrow could result in a reduced use of intravenous iron therapy and an increased efficacy of oral iron therapy.The adverse events during treatment that we observed are consistent with those expected in patients undergoing dialysis. Hyperkalemia was reported more often in patients who received roxadustat than in those who received epoetin alfa. Analyses of central laboratory data did not show any clinically significant changes in the mean potassium levels over time or between groups. Although it is possible that the report- ing of hyperkalemia might reflect a potential bias inherent in open-label trial design,36 in a double-blind trial comparing roxadustat with placebo in patients with chronic kidney disease not undergoing dialysis, hyperkalemia and meta- bolic acidosis were reported more frequently in the roxadustat group.37 The intermittent central laboratory monitoring may not have detected po- tassium elevations, and therefore continued evalu- ation will be important in presently ongoing trials (ClinicalTrials.gov numbers, NCT02052310 and NCT02273726) and as wider experience with roxadustat occurs. More patients receiving roxa- dustat discontinued treatment owing to adverse events than did patients receiving epoetin alfa. Furthermore, we speculate that the between- group difference in the percentage of patients who discontinued may have been due to the open- label trial design, given that the comparator, epoetin alfa, was the only approved treatment option for anemia in patients with chronic kid- ney disease in China, so there may have been concerns regarding the use of an unfamiliar therapy. No clustering of severe adverse events during treatment was observed in either group. However, the small sample size of this trial and the placebo-controlled trial37 relative to the larger, international phase 3 trials (NCT02052310 and NCT02273726) should be considered. Long- term safety will also need to be assessed in the international trials. In conclusion, this phase 3 trial comparing 26 weeks of roxadustat therapy with epoetin alfa therapy in patients undergoing dialysis showed the noninferiority of roxadustat in the treatment of anemia.