Read a lot of the papers on cognitive decline after CABG. My primary point of interest was why there are such wide variations in results. Most papers find something (most commonly attention, memory or Trail-making), but rarely the same thing and maybe 1/3 find nothing at all of interest to the patient population (audio response time would, to me (and the FDA?), fall into that category).
Well, the answer turns out to be simple. Many tests had:
a) no control group so no control over the training effect (the second time you take these kinds test everyone improves substantially but without a control to calibrate that effect out it utterly obscures the CABG deficit).
b) poor administration of psych tests, the effect of which is absolutely horrible measurement uncertainty (one paper that found no deficit had uncertainties for each individual psychoneuro test of between 30 and 50 percent of the mean values. In contrast the best run trial had uncertainties about 5% of the mean values of the tests. Almost certainly the poor uncertainties are due to a lack of control over the test run-in. (When a psych test is first administered it is very important to have some dry runs with each testee to ensure the testee is not confused about it)).
The horrible uncertainties paper:
ncbi.nlm.nih.gov
The paper for the best trial I've seen so far:
ncbi.nlm.nih.gov
Note that as measurement uncertainty goes up, the trial size needs to go up as a square. So 6 times the uncertainty means you would need 36 times the number of testees to see the same result. (Note that the bad paper doesn't actually define what their error bounds mean, but at best it is 2SD (+/-95%) and still 3-5 times worse than the good paper)
c) poor statistical analysis. There are lots of ways to do it poorly, but one of the most common is to count as a deficit only someone who upon retest is more than 1 population SD below where he was before for some particular test or tests. Instead, one should count as a deficit someone whose change from test to test is more than one SD of the amount of change that everyone sees. For instance on an intelligence test 1 SD is 15 points and they are repeatable to a lot better than 15 points. I suspect we'd all feel slow if whatever our IQ's measure dropped only 7 points. So the test upon which many negative findings are based is biased to not report as 'impaired' any but the most extreme cases.
So, what does all this mean for DEX in CABG? First, several more pieces of info: It is essentially incontrovertible that CABG produces many many times more pieces of micro and mini flotsam going up to the brain. Multiple papers confirm this. None that actually tried to measure it passing found otherwise. And it is a very good bet that many of the CABG have small strokes - many (albeit not all) MRI find much more evidence of mini strokes after CABG than after other surgeries (including angioplasty). Finally, whole brain perfusion after CABG seems to be impaired - presumably as a result of the literally hundreds of pieces of micro and mini flotsam getting jammed in the vessels. Are there other effects from CABG? Maybe, but certainly this effect should not be discounted. The last piece of info is that DEX seems to perform very well in animal models of significant stroke and they had some confirmation of carryover to humans in the TBI PII, so it is a good bet (but by no means a certainty - e.g. do ministrokes have the same extended damage effects as bigger strokes?) that it works well in all of the ministrokes and microstrokes associated with CABG.
The real question is whether Pharmos avoided all of the pitfalls above. You'd think it was easy, but it isn't. Getting "test run in" done well is probably tricky. Statistics are not something that most biochemists do well (e.g. a PhD microbiologist friend of mine once said "there is a Math For Biologists class, but there is not a Biology For Mathmaticians class"). I'd guess 70% chance that if Pharmos ran their trial as well as the best study done above they will easily see statistically significant results. But that is a big if! So many have done it so poorly in the past. So I still give it a 50/50 chance of seeing results of any kind - and in the event of 'trending' I'll be looking to see if they are getting the right lessons learned to do a proper PIII.
Nothing like writing up one's thought as a method of organizing. Comments?
Clark |
