cvac patients posting trial progress, page-18

Hi Macenroc,

Well I’m hoping to be wrong as well. Its a shame you can’t access the presentation to judge for yourself.

Why I came to the conclusion I did was because in the presentation across various slides it is reported:

OS data expected Q4 2014

The figure showing the release dates for data has remained unchanged with Q4 for a update on OS

On another slide it states (in red bold font)the OS data is immature

And finally when referencing the ASCO paper ML says “Final PFS data has been accepted for ASCO.

I appreciate this is contradictory with the announcement. But I came to the view that if OS was in fact being updated ML would have mentioned this as something happening shortly ... and updated his presentation accordingly.

I suppose we’ll all just have to wait and see.


I always appreciate a well thought out rebuttal to any of my posts Digster. But I think we might have to agree to disagree about the strength of the signal coming out of Can-003.

Theres no need to relive all the woes of Can-003 blow by blow. Everyone is familiar enough with the results and the market reaction.

The purported compelling signal is based a non-statistically significant (p=0.09) PFS advantage for Cvac over SOC in second remission subjects. The 95th confidence interval around the hazard rate estimate of 0.41 is 0.14-1.21. Which means there is a 95% probability that the estimate is somewhere in this range. This is described in MLs presentation as “a strong efficacy signal”, “a strong clinical signal” and “clinically significant”. Everything except “statistically significant”.

Now what these numbers mean in practice is that no-one in their right mind would take this as a signal to move to P3. If I was at work I could calculate for you the probability estimates of P3 success – but they would be way under current levels of success rates in oncology P3 trials. Hence why Can-04 was stopped and PRR has gone back to “proof of concept”.

So it comes down ML’s argument about a homogenous patient group and finding the sweet spot.

Now funnily enough ... this argument has just cost PBT investors about $100m following the failure of its AD trial. PBT argued that by using Pet scans at baseline it could select a more homogenous patient group that were most likely to respond to PBT2 because in prodomal stage a high proportion of false positives had been incorrectly recruited to previous trials that couldn’t benefit from the treatment.

The results of this line of thought are there for all to see. And as it turns out this homogenous patient group argument is really a kiss of death. The reason for this is that it usually arises out of failure ... and the best predictor of future failure is past failure.

Its not that Can-3 has confirmed the existence of this “sweet spot” (even if the PFS results for Can-003 in second remission were in fact statistically significant which they were not). Rather Can-003 has generated a exploratory hypothesis that it might have found a patient group that Cvac works in which is now being confirmed (or not) in Can-004(2). There is a big difference between these two things.

The other reason the homogenous patient argument is often a disaster is more conceptual. The idea is that as you move from P1 to P2 to P3 trials become larger and the patient sample more heterogeneous. So in P1 subjects are very carefully hand picked on a whole bunch of “intuitive”, not easily able to be articulated factors. By P3 this is impossible for 800 patients being recruited from say 25 sites across 5 countries. So what happens is the effect size from your new treatment plummets in response to this heterogeneity.

So when companies say that they now want to concentrate on targeting a more homogeneous patient group ... the sweet spot .. really they are just yearning back to the good old days when small P1/2 studies were producing positive results as easily as falling off a log.

So how are positive trial results in P3 ever produced. Well the signal coming out of P1 and P2 has to be so strong that it survives everyones best attempt to bugger it up. And when you see this ... probably once or twice in your life .. its just qualitatively different. No matter which way things are sliced or diced, who did what where .. the result comes out strong. This is in contrast to the norm where an effect pops in and out highly dependent on context which no one can posit a good reason for.

Now if you follow the heavy weights on SA SmithonStocks argues that the cancer vaccine development process is characterised by stumbling failures towards success. This is the sweet spot idea ... that it takes a few attempts before you find the right patient group at the right time in a very complex chain of treatments. This is quite plausible .. except for the fact that meta-analyses are showing that from 100 vaccine cancer trials only 5 have produced positive results at p<0.05. Which is exactly bang on what you would expect by chance alone. Now it might be that it takes 20 stumbles before someone latches to a success ... but it should be improving over time ...and it doesn’t seem to be.

Anyway this is all a long winded blurb to the “absurd” comment ... which probably was a little bit over the top.

To end on a positive reframe .. if some good OS data can be produced ... either shortly (best case) or at the end of the year (most likely I think) everyone (ML, you, me) will be in complete agreement that the signal coming out of Can-003 is even more compelling than it was before.

Regards Southoz