FDA staff raise safety issues with FibroGen's CKD anaemia drug roxadustat ahead of AdCom

Shares in FibroGen slipped about 7% Tuesday after the FDA posted briefing documents ahead of an advisory committee meeting on July 15, flagging safety issues regarding the company's oral HIF-PH inhibitor roxadustat to treat anaemia in chronic kidney disease (CKD) for both non-dialysis-dependent (NDD) and dialysis-dependent (DD) patients. The drug, formerly known as FG-4592, is being co-developed with AstraZeneca.

According to the FDA, roxadustat's efficacy is "not in question." It said the application provides "substantial evidence" of efficacy – comparable to that of erythropoiesis stimulating agents (ESAs) – demonstrating that roxadustat boosts haemoglobin levels in patients who are on dialysis and not on dialysis. "The principal issue before the committee is the drug's safety, and safety with respect to the specific CKD patient populations," the FDA said.

Risk of thromboembolic events, seizures

Agency reviewers pointed to "important risks of serious thromboembolic events, as well as other risks" with roxadustat. Compared to ESAs, the regulator noted "analyses of adverse events in the dialysis population demonstrated higher risks for roxadustat with respect to thrombotic events, including thrombosis of vascular access, events for which ESAs carry a boxed warning… There was also a higher risk of seizures with roxadustat compared to ESAs, another adverse drug reaction for which there is a warning in ESA labelling."

The filing, submitted in late 2019, was based on a Phase III programme that spanned six main trials and included more than 8000 study participants. The three principal studies of roxadustat in the NDD population were ALPS, ANDES and OLYMPUS, which compared the HIF-PH inhibitor against placebo. For the DD population, the three principal studies were HIMALAYAS, SIERRAS and ROCKIES, and these compared roxadustat against epoetin alfa.

FibroGen and AstraZeneca have touted pooled analyses suggesting roxadustat had a similar cardiovascular safety profile to placebo among NDD patients, and a comparable, or in some cases better, profile than epoetin alpha in DD patients. However, earlier this year, FibroGen disclosed that while it was preparing for the advisory panel meeting, the company realised that it had submitted altered safety data to the FDA. It said updated analyses indicated that roxadustat's safety profile may not be as encouraging as it had previously reported (see ViewPoints: FibroGen's fumble amplifies roxadustat doubts).  

High drop-outs cloud safety interpretations for NDD

FDA reviewers said that while it was "clear-cut" to interpret the results of safety analyses for the dialysis population, the same could not be said for the NDD studies because of "considerable disparity" in drop-out rates between the roxadustat and placebo arms. According to the agency's pooled analysis of adverse events across the three non-dialysis studies, there is an "obvious signal" for serious thrombotic events with roxadustat, with a relative risk of 1.45 versus placebo, with myocardial infarction and stroke being the two main contributors.

The rate of seizures in the non-dialysis studies also caught the FDA's attention, with reviewers noting that while the number of cases was relatively small, at nine events for roxadustat and one for placebo, the relative risk of 5.4 "merits concern." Meanwhile, FDA staffers said the signal for serious infection was unexpected, but sepsis/septic shock is an "obvious concern," with an estimated relative risk of 2.37 versus placebo, "and it is reinforced by signals for serious urinary tract infections, bacterial infections, cellulitis and peritonitis."

Agency reviewers said thrombotic events and seizures were "again prominent" in their analysis of the three on-dialysis studies. Except unlike the NDD trials, where these events were detected against a placebo background, staffers said here they were evident against epoetin alfa, "which is itself known to pose these risks."

MACE analyses tip in favour of control arms

In assessments of major adverse cardiovascular events (MACE), a composite of myocardial infarction, stroke and all-cause mortality, the FDA said the objective was to demonstrate non-inferiority of roxadustat against placebo in the NDD population, or against the ESA in the DD population.

For the NDD population, results for time to first MACE revealed a "considerable difference" between the estimated hazard ratios (HRs) for both the primary analysis and the sensitivity analyses, with HRs of 1.10 and 1.38, respectively. However, the FDA suggested that because of the higher drop-out rate among placebo patients, and subjects in the roxadustat arm consequently being exposed to treatment longer, results might have been skewed against the drug.

The agency also found an imbalance favouring controls in the DD population. Comparing roxadustat to epoetin alfa, the estimated HR for time to first MACE was neutral in the primary analysis at 1.02, and was 1.14 in the sensitivity analysis, a "difference [that] was nearly statistically significant," the FDA said, driven by all-cause mortality and heart attacks.

The oral HIF-PH inhibitor, which is also partnered with Astellas, is currently approved in China and in Japan, under the name Evrenzo, for the treatment of anaemia in CKD in non-dialysis dependent and dialysis-dependent adults. Late last month, the drug also received a positive opinion by the European Medicines Agency's Committee for Medicinal Products for Human Use (CHMP).

To read more Top Story articles, click here.