Wednesday, August 7, 2019

FDA: Novartis “mismanaged” or “manipulated" approval data for gene therapy

The drug maker Novartis concealed manipulated data from the Food and Drug Administration while applying for approval of an extremely expensive gene therapy treatment and then delayed reporting the issue, the agency said on Tuesday. Officials said the inaccurate data, which involved testing in mice of two different strengths of the treatment, did not affect the safety or efficacy of the therapy, Zolgensma, used to treat a rare, often fatal genetic disease called spinal muscular atrophy. 
Approved in May, the treatment’s price — set at $2.1 million — stoked concerns about the astronomical costs of potential cures for rare diseases and upset parents who initially could not get insurance coverage for the breakthrough treatment. The F.D.A. said patients were not at risk, and that the treatment could still be sold.... 
...The F.D.A. said it was notified of the data manipulation issue on June 28, more than one month after Zolgensma was approved, even though officials at AveXis, the Novartis unit developing the product, learned of the problem in March. 
The problems involved experiments on mice used in early phases of the research. An F.D.A. inspection report dated July 24 to Aug. 2, 2019 noted lapses and discrepancies in record-keeping by the company, and improper procedures in quality control in gathering data on the mice. In some instances, the report said, records stating how long the mice lived “were different from the actual value,” and in four cases “discrepancies of greater than one day were noted (ranging from 2 to 19 days).” 
The F.D.A. said the data were “mismanaged” or “manipulated,” and declined to say whether the information was deliberately falsified. “It’s unclear to us, at this point, exactly why this occurred,” Dr. Marks said. In data manipulation cases, he added, the motive is not always clear. “Many times people do things for stupid reasons, because if they would have left well enough alone, everything would have been O.K.”
I wonder how difficult it was to discuss the data manipulation episode within the company? I can't imagine the meetings about that were easy, and I could imagine heads rolling as a result. Here's hoping that there is an innocent answer, for the sake of the relevant people...


  1. When I used to see this kind of stuff working in pharma, it was much more innocent. Genuine out-of-spec batches due to manufacturing errors were extremely rare, so the QC techs didn't trust themselves when something came out OOS and would fudge it to make it pass - not to deliberately pass bad product, but because OOS lots almost always turned out to be chemist errors on closer inspection. People would also sometimes get caught fudging dates in their notebooks if they ran out of time and documented something a day after carrying out the work.

  2. Could be as simple as a biologist who didn't want to get out of bed on Sunday to go into the lab and check on their mice, and on Monday realized they probably should cover their tracks.

  3. when you submit for approval, you need to provide FDA every little piece of data from studies done with the drug on animals - even exploratory research non-relevant to the safety/efficacy, in handful of mice - as long as the research was done with the drug that is to be approved. This is a major nuisance: when I was at SUGEN, we were told not to do unnecessary exploratory mechanism-of-action research on our clinical candidate in animals, and pick instead some closely related analogue, so as to avoid the paperwork and possible delays from having to explain this stuff to FDA.

    So it is possible that there was some someone "tidied up" messy old data for the benefit of FDA, from a peripheral little rodent experiment and she was caught because someone knew about the data doctoring and reported it as a concern to the CDER/FDA - for example an unhappy ex-employee


looks like Blogger doesn't work with anonymous comments from Chrome browsers at the moment - works in Microsoft Edge, or from Chrome with a Blogger account - sorry! CJ 3/21/20