While mOS cannot be determined for each cohort with absolute precision, the statistical data that has already been released to the market contains some clues about the likely survivability for the 1200mg cohort. Of course, only within the limits of the assumptions being made.
In saying that, according to my analysis it is more likely that the increase in mOS from 17.1 to 19.7 months is entirely attributable to participants in the 400 and 800 cohorts, not the 1200mg cohort. I also anticipate survival in the 1200mg cohort to be much higher than the 400 and 800mg cohorts, as I explain below.
First, we need to pay careful attention to the trial recruitment period for each cohort so we can overlay these with the dates the results were reported to the market:
Enrolment into the trial for the 400mg and 800mg groups (hereafter referred to as cohorts 1 & 2) was from November 2017 – June 2019.
Enrolment into the trial for the 1200mg group (cohort 3) was from August 2019 – February 2020.
Second, we need to consider mOS and number of deaths at each reporting period, as well as the dates this information was released:
An announcement to the market on 14th February 2020 reporting interim trial data for cohorts 1 & 2 showed a mOS of 17.1 months. This data was also published in the Journal of European Urology Oncology and showed mOS of 17.1 months (95% CI 6.5–27.1) and 17/32 men had died.
The most recent announcement to the market was the presentation at the ASCO meeting in February 2021 which showed 31/56 had died and mOS of 19.7 months (95% CI 10.7-28.7).
A further announcement to the market on 15th of February provided updated figures showing 37/56 had died and mOS of 19.7 months (95% CI 9.5-30.0).
Now since the abstract deadline for the February 2021 ASCO meeting was October 13, 2020, it is fair to assume that the information presented was known to the investigators around August 2020. So if you consider the enrolment dates for cohort 3 this will mean the men from the 1200mg group had only been involved in the trial for between 6-12 months, so they couldn’t have possibly contributed to the upper end of the distribution curve.
Another way of looking at it is since mOS is the point where 50% of men have died and 50% are still alive, the shift from 32 trial participants to 56 meant an additional 12 deaths first had to be registered before the new events-driven mOS measurement could be calculated. Based on the confidence intervals it is highly probable that every single one of these 12 additional deaths arose solely from the 400 and 800mg cohorts. This would mean that all men from the 1200mg cohort were still alive 6-12 months after recruitment and potentially all 19 men still alive (at least 12-18 months after recruitment) are also part of the 1200mg cohort.
So if we assume the remaining 19 men had been alive for a median of 16 months as of February 2021, there is a very high probability that the 1200mg cohort will achieve a mOS of greater than 20 months, and even as high as 22.8 months mOS as suggested by @Stayso 54.
(20min delay)