Replies to post #81353 on NorthWest Biotherapeutics Inc (NWBO)

[exwannabe]

Would you mind explaining what MST and "average survival" are, and why the latter is meaningless?

Trials normally report the median time (the time that half the patients live longer than) for OS (or PFS) for several reasons. This is MST (or MOS).

On the other hand, AVII77 is saying is this paper they printed mean (common arithmetic average) survival time. A very different number, even though both may be referred to loosly as averages.

It is not that much that mean is "useless", but it is clearly a different number. In this case the mean was about 3-1/2 months longer than the median. And they compared the longer mean OS to another trial's median OS.

Also, are the OS data recorded at the time of death regardless of the cause of death?

Of course. But the question is when did they start the clock ticking. They started the clock running post surgery in the NWBO trial, but the comparator clock was started later when treatment began.

That gave the NWBO data a couple months advantage.

[AVII77]

Would you mind explaining what MST and "average survival" are, and why the latter is meaningless? Also, are the OS data recorded at the time of death regardless of the cause of death?

Yes, OS is recorded at the time of death regardless of cause.
(This is in contrast to cause specific deaths like we see in cardiac trials, but there they are very specific in their definition of cardiovascular death).

Regarding MST and average, below are the data for the 8 rGBM patients (sorted from lowest to highest),

214
219
324
331
347
393
454
1319

The average of those numbers is 450 days

The MST (median survival time) is the time that half the patients were dead and the other half alive. That would be half way between the 4th and 5th patient above.

Regarding why average is meaningless: The average survival time is never reported. What researchers are interested in is how the average patient is expected to perform. Here, the average patient would be expected to live about 340 days (with half living longer and half living less). The average survival time is influenced by outliers; here by the 33 year old patient who lived 1319 days.

(There is also a statistical aspect to reporting medians. If the hazards are constant (rarely true but almost always assumed), if you know the median you can calculate the survival at any other time point.

For example, in the above case the MST is 11 months. If you wanted to know the expected percent surviving at 14 months you would use the following excel formula "=EXP(14*LN(0.5)/11)". Again, this assumes constant hazards. So, that one number, the MST, defines the entire survival distribution. The average does not.)