In a stunning revelation, FibroGen admitted to presenting roxadustat data manipulated to make the anemia drug look safer than it is.
The late-Tuesday announcement sent FibroGen shares tumbling and prompted analysts to reassess the potential first-in-class therapy ahead of an FDA decision.
The company changed parameters used to analyze heart safety data for roxadustat in patients with anemia from chronic kidney disease, FibroGen acknowledged Tuesday. The false criteria yielded more flattering data, which FibroGen and partner AstraZeneca disclosed to the public in late 2019.
FibroGen’s stock price plunged by more than a quarter in after-hours trading Tuesday. Investors now question the drug’s prospects for FDA approval. Even if approved, the drug’s use could be restricted. Either way, previously sales estimates are overly optimistic, analysts said—and lawsuits might follow.
Jefferies analyst Michael Yee significantly dialed down his estimates for FibroGen’s share of peak sales to just over $1 billion from the previous $3 billion.
FibroGen CEO Enrique Conterno immediately distanced senior management from the data doctoring. He said the execs only found out about the changes, which affected the factors used to group patients for analysis, while preparing for an FDA advisory committee meeting the agency requested last month, much to pharma watchers’ surprise.
If that’s indeed the case, it may clear FibroGen management of knowingly presenting false data, but it casts a different shadow over their performance: They overlooked a damning problem in a do-or-die FDA filing.
Roxadustat belongs to a new class of drugs called HIF-PHI. Other than its convenient oral dosing, a key selling point for roxadustat compared with standard-of-care erythropoietin injectables such as Amgen and Johnson & Johnson’s Epogen/Procrit has been its heart safety.
And FibroGen seemed to have the data to support that marketing edge. A pooled safety analysis of six phase 3 trials showed roxadustat was comparable to placebo on a composite of major cardiac events among nondialysis patients, comparable to Epogen/Procrit in dialysis-dependent patients and superior to the erythropoietin drug in the incident dialysis subpopulation, which includes people new to dialysis.
The hazard ratios for those three populations between roxadustat and Epogen/Procrit were 1.08, 0.96 and 0.7, respectively, according to data unveiled at an American Society of Nephrology event in November 2019. Those ratios measure the relative risks of the two drugs; higher numbers indicate greater risk for roxadustat.
Now we know those ratios were based on stratification factors altered after the data were unblinded to FibroGen. The actual numbers by pre-specified standards were 1.1, 1.02 and 0.83, respectively, suggesting higher relative risks for roxadustat.
Conterno said the new numbers don’t change the previous noninferiority conclusions for roxadustat in dialysis and nondialysis patients, but the company can no longer say its drug was safer than epo in incident dialysis patients. The company doesn’t have an agreed statistical margin with the FDA to define noninferiority, he added.
The fact that roxadustat’s no longer superior in that patient group marks “a material change to the profile and one of the key prior advantages for patients,” Jefferies’ Yee said in a Wednesday note. He lowered the possibility of an FDA approval with a favorable heart safety label to 25% to 30%, down from the previous 50%, “given the benefit/risk profile of roxa has become less compelling.”
Modifications to the measures used to divide patients for analysis run the gamut from the cutoff point for baseline kidney function to the definition of geographic regions, as well as variables such as sex, race and body mass index, FibroGen Chief Medical Officer Mark Eisner, M.D., told investors on a conference call Tuesday.
The analyses also include six more for all-cause death and a broader medley of cardiac events that includes hospitalization. The fact that all nine analyses across the patient groups looked less favorable for roxadustat after the change raises the suspicion that someone within FibroGen carefully selected the new criteria to make roxa’s profile look better.
Conterno said the firm is investigating the data manipulation.
The incident doesn’t affect roxa’s profile with the FDA at least; according to Conterno, data from both stratification measures were included in its new drug application to the agency, with details on the differences. The company spoke with the FDA late last week to clarify the data in what Eisner described as a “cordial and collaborative” discussion.
But FibroGen’s public credibility is without doubt seriously damaged, which put another layer of uncertainty on the upcoming FDA advisory committee meeting. All the outside experts invited to review the approval application had previously received the same manipulated data. The conference is now scheduled for July 15, FibroGen said Tuesday.
What’s more, given that the mess touches upon company disclosures of false data, it might invite investor lawsuits alleging breaches of securities laws.
Roxadustat is already approved in China and Japan under the brand Evrenzo. AZ is FibroGen’s partner in the U.S. and China, while Astellas is responsible for Japan and the EU. The current data problem doesn’t affect the two existing approvals because they were based on different sets of data, and the ongoing review of its application with the European Medicines Agency is also unaffected, Eisner said.