Hi all - Just figured I'd sum up all of my DD in case anyone is interested:
Science:
1) I do not like the fact that the top two tiers of GOS combined actually showed statistical dead heat performance between placebo and DEX for the parts of PII that they published (and they edited the third arm so the data never really saw the light of day). Their extensive retrospective focus on the improvement in the top category alone is worthy of some penalty, BUT mitigating my judgement on this is: It appears that DEX ran into a ceiling effect. Even in placebo almost everyone is in the top two categories (about 77%), so if indeed the Outcome for DEX is to be better, it cannot get much better in the number 2 category because there are not many patients below to improve into that category. And if the Outcome for DEX is significantly better in the best Outcome category the patients must come from the number 2 category since there are few patients in the lower categories. So the fact that the total number in the top two categories hasn’t significantly changed is more an artifact of the GOS score tops out too soon and is too coarse. Meaning we just don’t have a lot of data about the Phase II results on final Outcome. Moving to GOS-E (which was not in widespread medical use during PII timeframe) for the phase III (which was confirmed in their latest CC) should alleviate the coarseness problem but not the ceiling problem, so everything else being equal I’d expect better results, but still results not as good as they could be (or as good as they were in the GOAT – which I tend to take with a large grain of salt since I don’t have enough data). Also, they appear to be using substantially sicker patients for Phase III (thus minimizing the ceiling effect) and their tiering system for Phase III seems to be using a more accurate assessment of patient entry state to ensure balance between control and treated groups. See Jonmarin's notes on the yahoo boards at:
finance.messages.yahoo.com
2) I saw something on the boards that indicated that only 1% of the patients had died in the phase III. That would be bad because it would indicate either the standard of care has improved making DEX’s job a lot harder, or the patients are healthier in the Phase III and that exacerbates the GOS ceiling effect. I can find no confirmation of this 1% mortality, and Pharmos does indicate that 1% of the patients were lost to follow up. I suspect the poster, who I could not find with any more credible posts, has confused “1% lost” with “1% dead”.
3) Not happy that the company chose not to publish the third arm of their PII even combined with the first two. Retrospective culling of data is always bad (e.g. was one of the two arms not culled unfair the other direction but we can't tell because it wasn't separated out?). BUT if there is an excuse for culling that is most justifiable (still deserving of a statistical penalty, but not as large a one), it is that the randomization for one of their prospectively defined control variables was not adequate. They should have combined all the data; the bias in the third arm would have been largely averaged out. Yeah, overall GOS wouldn’t have shown the same effect, but they could have talked about the most severe patients and GOS, they could have talked about GOAT, they could have talked about ICP and it wouldn’t have the statistical penalty (or in layman’s terms, ‘what aren’t you telling me’ effect.)
4) Stratified results in Phase III. I still don’t understand the FDA well enough, but I think that if the overall numbers are positive that they will give approval regardless of stratification – although they might modify the label in some way. Not worried by that scenario. As for overall numbers shy of stat sig, but a stratified subgroup is stat sig, I suspect Pharmos will have to run another test. This is a moderate worry to me, but that is why it is priced for phase ii and not phase iii.
5) On the strictly positive side (vs mitigated negative), I think that the very significant decrease in ICP without a decrease in Blood Pressure is a very good sign. It is, of course, an assumption that ICP and GOS are causatively linked, not just both correlated to a third variable which DEX does not effect. But I think it is a good assumption (although like all assumptions there is a chance that it is wrong). And I think that unlike many drug trials, the animal versions of these trials are relatively faithful to the human version and the animal results were very good.
Financial
1) A little worried by the recent financing and the whole history of dribbling out the financing. But I figure between now and Phase III results there isn’t enough time for more dilution.
2) A little worried about Pharmos’ history of partnering (i.e. Bausch and Laumb). Not adequately incentivizing the partner and/or picking partners not willing to push through. But that is a small concern compared to others, and well, we should be so lucky as to have problems partnering a blockbuster drug.
Further notes and work:
1) I currently consider the CABG study results to be luck of the draw. I put it somewhat arbitrarily at 20% chance of statistically significant results in the Phase II, 30% chance of success less than statistically significant but trending towards success or significant only in a subgroup (e.g. more hours on the machine) and 50% chance of total lack of measurable effect. The effect of CABG on cognitition has still not been completely proven (e.g. the measured results to date could be an artifact of CABG patients having other problems leading to loss of faculties at a faster rate than the general population).
2) I will be looking into the possible off-label use of DEX for spinal cord injury. Size of the market? Probable utility? Likelihood that doctors will use it that off-label? I know the etiology of long lasting spinal cord damage is thought to be similar to the etiology of long lasting effects from TBI.
In summary I believe the likelihood of success in the P III TBI is fairly high because the P II showed such good success in ICP and it showed how to adjust the P III to be more likely to avoid ceiling effects and data coarseness in the metric upon which the FDA insisted (GOS). Of course the P II also provided hints of success in GOS as well and I consider those results informative, but I take them with a grain of salt as well for the reasons given above.
All JMO.
Clark |
