Innovation Pharmaceuticals Inc (IPIX)

[KMBJN]

Really appreciate your nice statistical tutorial on censoring in KM curves / survival / time to disease onset trials and explanation of Leo's/IPIX's/our situation. I knew I should have paid more attention in my statistics classes - and this is going to take a while to digest!

Luckily, one can still see nice lectures on the internet, such as this one from Stanford:

https://www.stanford.edu/~kcobb/hrp258/lecture9_na.ppt

With a nice definition of reasons for censoring, similar to the ones LR posted (thanks LR):

Subjects are said to be censored if they are lost to follow up or drop out of the study, or if the study ends before they die or have an outcome of interest. They are counted as alive or disease-free for the time they were enrolled in the study

and here a statistical explanation in the event that the censoring was due to inexact knowledge of the time of the event (onset of SOM) within some interval:

https://support.sas.com/resources/papers/proceedings10/257-2010.pdf

Anyways, I guess the number of events (onset of SOM) was 7 of 19 in the treated/brilacidin group, but there were 8 censors. These 8 censors were assumed to not have developed SOM, even though they dropped out or were lost to follow-up before the trial ended. If these were dropouts, it's still very confusing, because they started with 46 that met the cumulative radiation dosing criterion (mITT group), but 7 of these didn't meet the per-protocol criteria (which is why we have 39 in the PP group, 19 brilacidin and 20 placebo). Not to mention, what do you do with the 15/61 randomized subjects that didn't achieve the full radiation dose - are they excluded altogether from the analysis? Are they considered dropouts - and we would then have 3 groups of dropouts?

So, there were 7 not-per-protocol subjects (which could be dropouts/withdrawn/died? assume for some reason didn't complete the full course of treatment), and 13 censors (8 in brilacidin, 5 in placebo) (which could also be dropouts/withdrawn/died or lost-to-followup). Why would one dropout be not-per-protocol, and the other per protocol but censored and included?

Very confusing.

Is this why the FDA requires the use of mITT, so all dropouts/withdrawn/died/lost-to-follow-up are handled the same, and counted as a censor on KM/Cox proportional hazards analysis?

Thanks again for helping me understand and think through this.

Maybe I should just focus on the 4 year old nephew explanation on KM curve and pancakes instead :)