Replies to post #114784 on Biotech Values

[poorgradstudent]

If I was a betting man, I'd say it is mostly due to the inherent bias when the people on site are reading the scans. But also see the possible cause for discrepancy cited by biomaven in post 112748.

Note that these ambiguities are based on different people reading the same scan. So it's not a question of the people on site having one CT scan to look at and those at the central review looking at another one of different quality. It really comes down to how the viewer assesses the situation.

Since both analyses in the rida trial yielded strong p values, I don't think we're in trouble here. Also note that the Hazard Ratio changes little from the central to the investigative site analysis despite the relatively large shift in the rida arm's median PFS. This HR is a more thorough analysis that suggests the difference between the curves over the duration of the trial is relatively constant despite where the median events occur in these two analyses. This also suggests to me that the majority of the separation between the curves is happening later on during the trial* so there may be merit to the idea that a subset is really pulling the curves apart at later points.


* This isn't overly insightful given the short duration to median PFS, but it's still important when trying to visualize what these curves look like. I suspect that when we eventually see them we won't be overwhelmed by how far apart the curves are right at the median.

[biomaven0]

>>ARIA site vs IRC

Someone who I respect hypothesized that the difference between the central and the local review was because some small peripheral sarcomas (e.g. nerve sheath) are tricky to measure radiographically, and consequently the expert centers, who have seen a much greater volume of scans, detect progression earlier. She said she was not at all surprised by the discrepancy, and did not believe it indicated bias.

If this is correct, it suggests that although there was subtle progression in the discrepant scans, it was likely not clinically meaningful progression.

Peter