Biotech Values

[Robert C Jonson]

Thanks, Dew. You are correct that for most of the regular posters on the PPHM board, HR discussions are new and unfamiliar. I appreciate you, iwfal, and jq1234 trying to educate me (and others) on this matter.

In fact, HR always matters; when the FDA and other regulatory agencies evaluate a clinical variable, they look at the HR and pay little or no attention to the median value for the reason discussed in #msg-13857709.

In trying to understand the HR concept more fully, the best reference I've come up with so far is this:

http://journals.lww.com/jto/Fulltext/2011/06000/Biostatistics_Primer__What_a_Clinician_Ought_to.2.aspx


In it I found the part below that seems to flesh out a little more of the relationship between OS data and HR data, and in fact belies (at least to me) that your assertion that HR always matters is correct:

Magnitude Matters

The HR is a relative measure. Hence, a statistically significant p value (p < 0.05) associated with the HR (E versus C) = 0.75 may be obtained, meaning that (i) the experimental treatment is superior or inferior to the control arm and (ii) there is at most a 5% chance of observing an effect of this magnitude or more extreme (e.g., <0.75) by chance if there is no difference between the treatments. This would seem a positive result for patients. However, whether this is clinically significant also needs to be evaluated. To do this, clinicians will likely need to assess the absolute improvement in survival time by examining absolute measures such as the survival rates at fixed time points (e.g., 1-, 2-, and 3-year survival probabilities) and the median survival. These absolute measures only focus on one point of the survival curve and so can be misleading if they are considered individually. However, collectively they describe the KM curve. Hence, clinicians should look for consistent clinically meaningful improvements across these absolute measures. For example, if a HR = 0.75 corresponds to an increase in the 1- and 2-year OS rate of 10% and 20%, respectively, between the treatment arms in an advanced non-small cell lung cancer trial, these might be deemed clinically meaningful improvements. If the median survival difference between arms is next considered, then an improvement of 50 days may also be regarded as clinically significant, whereas an improvement of approximately 10 days may not. Only if the descriptive absolute measures reveal a fairly consistent pattern of clinically meaningful improvements should a statistically significant HR be hailed as a clinical advancement.

Finally, I asked you if HR would be affected by a small n trial, since stat. sig. would be difficult, and you replied that it wouldn't. Yet jq1234, in a post to me said this:

HR determines statistical significance, not median.

(http://investorshub.advfn.com/boards/read_msg.aspx?message_id=78625086)

So it seems to be HR would not be meaningful in a low n trial. Admittedly I have much to still learn about HR, but I still think you're underplaying the importance of the survival data in PPHM's current 2nd-line NSCLC trial.