Roxadustat for CKD Anemia - Starting the Jigsaw Puzzle, What Will the Finished Picture Show?

See Clinical Research on Pages 613, 624

Anemia is a “classic” complication of advanced chronic kidney disease (CKD) and used to constitute a major unmet need in that no effective chronic treatments were available. In 1988, roughly 75% of dialysis patients had a hematocrit of < 30%. The ensuing symptom burden led to the frequent administration of blood transfusions, often resulting in systemic iron overload, especially in persons undergoing maintenance hemodialysis. In 1989, this changed dramatically with the introduction of the biologic, epoetin alfa, a first-in-class erythropoiesis-stimulating agent (ESA) that was approved via an orphan drug designation for the indication, “to elevate the red blood cell level […] and to decrease the need for transfusions […].” Subsequent uptake of ESAs was rapid, facilitated by a coverage determination from Medicare, and ESAs have since been a cornerstone of anemia treatment in persons with advanced CKD including those on dialysis.

While transfusion avoidance was the labeled indication another hypothesized, but at the time unproven, benefit of treating anemia was to ameliorate its downstream harms to cardiovascular health. In fact, as persons with advanced kidney disease or kidney failure were known to be at excess risk of fatal and non-fatal cardiovascular events, anemia was considered a contributing cause and, hence, treatment of anemia with ESAs was expected to reduce this excess cardiovascular risk. However, when ESA trials were finally conducted in persons with kidney failure on hemodialysis, as well as in patients with moderate to advanced CKD, no improvements and in some trials even increases in cardiovascular event rates were found.3, 4, 5, 6 As a result, dramatic changes in ESA labeling were implemented with boxed warnings about these cardiovascular risks, and there was a marked shift towards much more conservative recommendations for initiation and maintenance of anemia treatment using ESAs. This, combined with a landmark expansion of the bundled reimbursement of dialysis care in the USA in 2011, has subsequently led to a dramatic reduction in ESA use in both dialysis and advanced CKD populations. Thus, patients were once again left to struggle with the symptoms of uncorrected anemia and face the risks of blood transfusion while still depending on treatments with potentially unsafe drugs. Treating CKD anemia had once again become an unmet need.

The current issue of Kidney International Reports contains two original research reports on the efficacy and safety of roxadustat, a novel anemia treatment that inhibits prolyl hydroxylase, a key enzyme regulating the activity of hypoxia-inducible factor. Roxadustat thus works through manipulation of an ancient biological system evolutionally selected towards providing protection of organisms from acute and chronic hypoxia. Roxadustat is a first-in-class oral medication for the treatment of CKD anemia, with other agents in this class also in advanced stage trials, including vadadustat, daprodustat, enarodustat, and molidustat. The two investigations in the current issue concern the longer-term efficacy of roxadustat to treat anemia in patients with CKD, as well as a secondary, pooled cardiovascular safety analysis of three trials in persons relatively new to dialysis.

In the former, 922 persons with CKD Stage G3-G5 (not on dialysis) who had anemia (hemoglobin ≤ 10 g/dl) were 2:1 randomized to receive thrice weekly roxadustat or matching placebo. Key efficacy endpoints included change from baseline in average hemoglobin concentration during weeks 26-52 and proportion of patients with hemoglobin response in two consecutive visits within the first 24 weeks. In brief, both efficacy endpoints were decidedly met, with the roxadustat group achieving a 1.85 g/dl greater least-squares mean difference from baseline in hemoglobin concentration versus placebo (P < 0.0001). In the roxadustat group, 86% achieved a prespecified measure of hemoglobin response during two consecutive visits within the first 24 weeks, without needing rescue therapy, whereas only 6.6% did so in the placebo group (P < 0.0001). Substantially fewer subjects in the roxadustat group required blood transfusion or intravenous iron as rescue therapies.

While these results indicate impressive efficacy for the treatment of anemia, and are very much in line with a previous, similar trial of smaller size and shorter duration in China, there are certain aspects of the present trial that make interpretation more complicated. First and foremost, the rates of treatment discontinuation were quite high, and differential between groups, at 43% among roxadustat and 68% among placebo-assigned subjects. These rates seem to refer to the full trial follow up of up to 4.5 years but appear to have been approximately 20% and > 50% already by week 52, respectively, the time relevant for this efficacy evaluation. Thus, the results rely on multiple imputation techniques, and the associated “missing-at-random” assumptions, to address this type of informative censoring. Second, the inclusion criteria with regard to baseline iron status were extreme, with patients allowed into the trial if they had a ferritin concentration ≥ 30 ng/ml and a transferrin saturation of ≥ 5%. In fact, 40.7% of patients were iron deficient (language used: “not iron replete”) at baseline. While oral iron was “encouraged,” administering intravenous iron was considered rescue therapy. International guidelines for the treatment of CKD-anemia and ESA labels were unambiguous in suggesting an iron first approach, only initiating ESAs in iron replete persons., While the authors argue that placebo as a comparator reflected the standard of care, correctly stating that the majority of patients with advanced CKD do not receive anemia treatment, we are not sure whether enrolling persons with significant iron deficiency and, by design, leaving them in that state can be considered “standard of care” (although it may well reflect “usual care” in some places). Iron deficiency itself, regardless of anemia, is independently associated with poor outcomes in persons with CKD. While improving iron deficiency and utilization through increased absorption and reduced sequestration are probable mechanisms of action of roxadustat, a higher minimum iron threshold (perhaps a ferritin of > 100 ng/ml and/or a TSAT of > 20%, consistent with earlier ESA labels and guideline recommendations,) would have made the trial more clinically relevant.

The second report, by Provenzano et al., contains an evaluation drawn from a subgroup of the pooled cohort of three pivotal trials comparing roxadustat with epoetin alfa in patients with kidney failure on dialysis. The authors report on the comparative erythropoietic efficacy and cardiovascular safety from a prespecified subgroup of patients who were new to dialysis, as defined by no more than four months since dialysis initiation. Quite oddly, the primary studies from which this subgroup analysis is drawn, studies evaluating roxadustat’s cardiovascular safety and mandated by regulatory agencies, have yet to be published, even 15 months after their presentation as a late breaking trial at 2019 ASN Kidney Week. The main contributor to this pooled analysis is the HIMALAYAS trial, which specifically enrolled 1043 patients new to dialysis, which itself a major accomplishment given the historical difficulties in enrolling patients with incident kidney failure and during their early months on dialysis. The other trials contributing to the present pooled analyses did not specifically focus on patients new to dialysis, but contributed 71 and 416 patients who satisfied the inclusion criterion of no more than four months since dialysis initiation. For all three trials, the inclusion criteria specified ferritin concentrations of ≤ 100 ng/ml and a TSAT ≤ 20%; still, it appears that 21% of patients were not iron replete at baseline with at least one of the two iron parameters below the stated threshold.

Following randomization, both subjects assigned to receive epoetin alfa and those allocated to the roxadustat group had brisk erythropoietic responses, which yielded maximum mean hemoglobin concentrations by week 12 that were subsequently maintained during follow up. For the efficacy evaluation during weeks 28-52, persons allocated to roxadustat experienced a mean increase from a baseline average hemoglobin concentration of 8.8 g/dl by 2.1 g/dl and those allocated epoetin alfa from 8.9 g/dl by 1.9 g/dl. The difference from baseline in the roxadustat was greater by 0.22 (95% CI, 0.05-0.40) g/dl compared with the epoetin alfa group. Thus, it is appropriate to conclude that in terms of erythropoietic efficacy roxadustat proved to be non-inferior to epoetin alfa, and nominally superior, at least with their respective protocolized dosing algorithms. Rates of blood transfusion were essentially identical between the groups.

With regard to cardiovascular safety, the primary composite cardiovascular endpoint consisting of myocardial infarction, stroke, or all-cause mortality occurred in 74 (roxadustat) and 97 (epoetin alfa) patients for corresponding incidence rates of 6.7 and 8.2 events per 100 person years, respectively. The ensuing hazard ratio was 0.70 (95% CI, 0.51-0.96) favoring the roxadustat group. Adding hospitalized unstable angina or heart failure to the composite yielded a hazard ratio of 0.66 (95% CI, 0.50-0.89), whereas all-cause (hazard ratio, 0.76; 95% CI, 0.52-1.11) or cardiovascular mortality (hazard ratio, 0.63; 95% CI, 0.37-1.05) did not differ between the groups. These findings arose from analyses that included only events up to within seven days from the final treatment. Such an approach is not unusual for safety endpoints, especially ones acutely linked to exposure. However, such censoring is concerning as it may induce bias, especially in an open label trial, and with a therapy where the potential impact on outcomes is more likely to be chronic rather than acute. For example, health deterioration may lead to treatment discontinuation prior to overt adverse outcomes, especially in the open label experimental drug arm where the novelty and uncertainty may make patients and practitioners more concerned and watchful. Additional analyses that expand the window beyond seven days, perhaps to 28 days as done in the efficacy analyses, as well as a full intention-to-treat analysis, would be simple to perform and provide important safety information. Also, importantly, this subgroup analysis, even though of a prespecified subpopulation, must be considered exploratory and hypothesis generating and by no means definitive. These subgroup findings must be understood in the context of analyses of the overall pooled phase 3 population, which will hopefully appear soon in a peer reviewed publication. One can hope that these encouraging findings will swiftly lead to the launch of a definitive trial in early dialysis, or even in persons with kidney failure whose initiation of dialysis is imminent.

In conclusion, these first two manuscripts from the roxadustat global phase 3 program are now available. Myriad additional aspects from these trials await more detailed analysis and reporting, particularly about iron parameters and treatments, effectiveness in the setting of inflammation and prior ESA hyporesponsiveness, the size and importance of off-target effects, including on lipid metabolism and blood pressure, and ultimately – assuming market access will be granted – longer follow up of real-world patients is necessary to learn about any uncommon or late-term adverse or beneficial effects. Finally, additional agents in this new class of hypoxia-inducible factor prolyl hydroxylase inhibitors are currently undergoing phase 3 testing. Will the assumption of class homogeneity be met? What will the picture show when enough of the jigsaw pieces have been placed on the table?

Disclosure

WCW reports having served as a consultant and receiving honoraria from Akebia/Otsuka, AstraZeneca, Bayer, Daichii-Sankyo, Janssen, Merck, Reata, Relypsa, and Vifor FMC Renal Pharma during the past 36 months. CPW declared no competing interests.