Replies to post #35097 on Biotech Values

[walldiver]

The HR is not nearly as good in 9902A as it was in 9901, but if you flip the numerator and denominator and measure it the same way as most biopharmas do, you get an HR of 0.77, which represents a 23% lowering of the risk of death. For a 98-patient trial, I'm betting that the FDA sees this result as sufficiently supportive for approval, especially with Provenge's very mild toxicity profile.

As for 9901, off the top of my head DNDN measured the HR as 1.7, or 0.59 as most companies measure it. So the question is, how will CBER view a BLA which had the primary Phase III come back with an HR of 0.59 and the supportive Phase III with a score of 0.77, plus a very mild toxicity profile, plus the fact it's for a terminal oncology indication?

We already know how much the prespecified, secondary Cox regression analysis improved the HR and the p value in both trials.

[genisi]

Re: Median values not indicative of hazard ratio

Thanks a lot for the educated explanation.

genisi

[nerdseeksblonde]

medians throw out outliers,
otherwise they don't do a lot- hard to factor into
analyses, point quantity, etc. The hazard ratio,
while integrating most of the data is still just
a single number. If the curves being compared don't
fit the model or differ qualitatively it won't
help a lot. Hazard ratio could vary with time,
quoting from the cited work,
"The hazard rate is the probability that if the event in question has not already occurred, it will occur in the next time interval, divided by the length of that interval."
( the time interval goes to zero and you get a point
value for each point )

Most of the time this isn't a big deal
but the "point" is that there is no substitute
for looking at the graph ( similar to watching weather
guy on TV versus looking out the window).