From the Kid, on ymb:
Bottom line answer is management can break the blind if they want to but it is unlikely they will. Second part is I agree CLSN has an idea of results but not for the reason you state.
More detail: One of the requirements for submitting an NDA is that there be at least one “well-controlled” study. For many studies being well controlled includes being blinded, partially, singly, doubly, triple etc. Most people know this. What they might not know is that in the clinical portion of the NDA submission there is a section where the sponsor needs to defend that the study is in fact well controlled.
Blinding is part of that control. Blinding is not easy to accomplish in the best of circumstances. For example comparing two drugs with different side effect profiles allows for personnel to be fairly well able to know which treatment is being given. In that sense TDOX is not unique.
So to be able to support the well-controlled claim, sponsors routinely set up procedures to ensure the blinding as well as they can, or do things in such a way as to minimize the effect of inadequate blinding. Examples in TDOX would be that shielding vials to hide color of drug, having independent scan review since treating physician will probably be able to guess based on labs and side effects, etc.
Such procedures could also include having separate personnel responsible for safety monitoring and efficacy monitoring. Putting such procedures in place and documenting them is an integral part of the NDA process. Having an experienced DMC responsible for helping ensure the trial is scientifically valid allows a company to anticipate potential problems in blinding. The alopecia and white counts are clear potential problems for blinding. A prudent company would set up SOPs to avoid these issues.
All that said, if a sponsor wants to break the code they can circumvent their SOPs and break the code. I suggest it would be easier to just look at the randomization list and get all the pt assignments rather than guess based on labs and AEs.
So I believe but don’t know that CLSN has procedures in place to address this. And I believe they would be foolish to jeopardize the trial to get knowledge of who is getting what.
But I do believe management has a good sense of the results for the same 2 reasons that I have that sense. First simple arithmetic using publically available enrollment data and times when event milestones were reached gives an overall PFS median of at least 20 months. This is far above published control rates and while not conclusive is highly encouraging.
Second the DMC recommendation at the interim to add additional look based on results at or better than the boundary. Note the potential impact on the conduct of the trial should be covered in the above mentioned well-controlled justification.
AI
