There are only two important endpoints in randomized trials, experts say:
- Overall survival (OS)
- Quality of life (QoL)1
This list is too long. Yes, both OS and QoL are two important concepts we should care about — but only OS is a robust endpoint. On the other hand, any measure of quality will by definition be its surrogate and not the real thing.
Here we come to an important muddling of terms. In oncology, an endpoint is deemed a surrogate if it only measures something that is related to a thing patients care about, but not the thing itself. Classically, progression-free survival (PFS) is a surrogate endpoint of OS because (in some cancers and with some treatments) longer PFS implies longer OS. The reason you use PFS is that it takes less time to measure, not that it’s impossible to measure OS. Furthermore, OS is much more precise: most people’s date of death is easy to determine, and what exactly constitutes death is not ambiguous.2
Not so with QoL, which is actually honest-to-goodness impossible the measure, having “quality” and not “quantity” right there in the title. What is quality of life? And while we’re at it, what is quality (supposing that we can agree on what constitutes life2)? “Establishing differences in quality of life between the experimental and control groups” is a fine study objective, but QoL isn’t a proper endpoint. QoL scores are, but they are but surrogates of what we care about, as maleable, pliable, and subject to the investigators’ (and participants’) whims as poor3 old PFS.
Which is to say: when interpreting QoL scores, be as alert as you are with any other surrogate outcome.
Yes, the actual term used in the slide was “quality of survival” which I don’t like. People with cancer are living their life not surviving a battle. ↩
Although considering how much weight the FDA has given to PFS for definitive cancer drug approvals, “poor” is a poor choice of words. ↩