Before going into a point by point I'll say that the primary difference between he and I is that he believes we actually really understand the systems involved in atherosclerosis. Whereas I think we have, at best, a weak understanding - witness the recent Torc trial implosion or the fact that the majority of statin efficacy is not LDL lowering.
a) "CART-II showed 1067 raising LDL by 13% relative to placebo". Given the epidemiological data on CETP malfunctions and the recent Torc implosion you have to ask whether you would rather: 1) raise CETP, raise LDL and lower HDL (like 1067) or 2) lower/inhibit CETP, lower LDL and raise HDL. I'd choose number 1 - although it isn't a strong preference because I consider the biological systems so poorly understood in atherosclerosis. Note also that in animal models it was shown to decrease atherosclerosis even when it effected LDL/HDL the wrong way. Disagree with him
b) "Doctors won't prescribe something that raises LDL". Agree that this is concern. Hopefully Atherogenics is collecting data on a variety of markers and finds one with a strong correlation between 1067 effect on it and lower event rate. But I am not holding my breath. Agree.
c) "1067 decreased HDL by 14%". See my comments under a. Disagree with him.
d) "No effect on CRP". Another MOA argument. It does have efficacy, albeit limited, in other inflammation markers known to correlate to future risk of heart attack. Neutral.
e) "1067 is not behaving like Probucol inre LDL". This goes to whether 1067 is really a follow on drug with the high success rate that goes with being such. My only comment is that it is very clear that the primary MOA of Probucol was NOT LDL lowering. It provided only modest lowering and yet blew the socks off statins in preventing revascularization. But all in all I agree mildly with him.
f) "The lower event rate seen in CART-II could have been caused by the imbalance in previous MI patients". Or another imbalance. Not a chance that the small difference in pct patients with previous MI caused the observed event rate difference. Too small an imbalance. But could it be an imbalance in some unknown prognostic? Absolutely. But that is what p values are for and the p value was less than 0.2 or only a 10% chance that this occurred by chance. Conclusive? Certainly not. But all in all reasonably supportive.
g) "CART-II IVUS data was post hoc and ITT was trending the other way". Partially valid. But not really. All IVUS trials outside of stenting restenosis studies exclude patients with calcification. CART-II as a restenosis study initially did not exclude such patients. As long as the later exclusion was done in a blinded fashion, which it was (by CCF) the results should be comparible to other studies. Should Atherogenics have thought of this when they changed the trial? Absolutely - they are sloppy. But the reality is that I personally wouldn't care much about the unfiltered data even if it had been positive just as I don't care much about the Regression study they did on the filtered data. Effectively mostly disagree.
Clark (Out of the closet although I can't figure out how to change my handle)
Market Data 
Markets 
AI
