Biotech Values

[iwfal]

If the progression assessments for the two metrics are made on the same schedule for every patient, it shouldn’t be possible for PFS to be greater than TTP. A death in between progression assessments ought not to cause a patient to be deleted from the TTP dataset, but rather for TTP to be censored at the time of death.

Depends entirely on what the censoring rules are. E.g. if the censoring rule is "Censor at the last scheduled scan unless a later scan shows progression free or an event occurs within 2 week of the scheduled date" then you can get the following results (the below example is about 75 percentian, versus median, just because it takes fewer patients to explain):

2 patients, patient a has progression at 6.5 week scan, patient b misses their scan but dies at 7 weeks.

TTP results: at 5 weeks 100% KM curve and 2 patients on curve. At 6 weeks patient b is censored so there is only one patient on curve. At 6.5 weeks the patient a progresses and curve drops to zero. 75 percentian TTP is 6.5 weeks.


PFS results: as 5 weeks 100% KM curve and 2 patients on curve. At 6.5 weeks patient a progresses and curve drops to 50% with one patient left on curve. At 7 weeks curve drops to zero when patient b dies. 75 percentain PFS is 7.0 weeks.


There may be a standard for censoring - but I haven't discovered it yet and the rules I have seen have been pretty arcane and thus have lots of potential for small oddities.