Talking with other members of the investment community, it seems to me that the main bull case for Atherogenics (AGIX) is "if it works, it will be huge!" If I had a nickel for every time I heard that about some questionable biotech stock or drug, I'd be partying with Maria Bartiromo at the Yellowstone Club.
Usually when a drug reaches Phase III there is some sort of basis for efficacy, but in the case of AGI-1067 there really seems to be no basis at all. I understand that is a pretty strong statement and I myself was a bit cautious about actively shorting AGIX until I saw the CART-2 study results that were finally published in a little known journal called Atherosclerosis (after a two year wait, an extremely long time for a study to get published) last month.
When the CART-2 results initially came out, they actually seemed quite positive. While there was not a benefit versus the standard of care, compared to baseline AGI-1067 (in addition to the standard of care) did seem to have a significant effect upon plaque volume, a potential revolutionary occurrence. Now that the full results have been published however, that initial data release seems like no more than hocus pocus:
• AGI-1067 caused a highly statistically significant increase (P<0.0001) in LDL (4% compared to a 9% decrease in the placebo arm). Considering 97.5% of AGI-1067 patients were on lipid lowering agents, it appears that AGI-1067 may actually counteract their effect. Besides lowering the probability of success of ARISE, this also may lower the market opportunity as doctors may be uncomfortable in using a drug that has such an effect.
• AGI-1067 caused a highly statistically significant decrease (P<0.0001) in HDL (14% compared to 1% in the placebo arm). While an attempt is made to argue that this is a positive based on animal models, this is highly counterintuitive and I put little value to this attempt at an explanation.
• The fact that hs-CRP was not significantly different between groups despite AGI-1067 being touted as an anti-inflammatory agent puts into question the mechanism and if AGI-1067 actually does what the company thinks it does.
• AGI-1067 is supposedly a derivative of Probucol, which had been pulled off the market for safety reasons. Given that Probucol had shown efficacy in the past, this would strengthen the case for AGI-1067. However, based on the numbers I have seen, Probucol is supposed to lower LDL by 5-15%, not raise it like AGI-1067 and completely negate the effects of the LDL lowering properties of statins. This completely puts into question any argument that AGI-1067 would work just because Probucol worked. Maybe they are related but they seem to be completely different in terms of effect.
• While there was a higher number of acute coronary events in the placebo arm (9.9% compared to 7%, 4% and 7.8% in the three AGI-1067 groups), this could potentially be explained by certain imbalances in the arms. Most importantly, 35.1% of those in the placebo group had a prior MI, compared to only 28% in the AGI-1067 arm. As those with prior heart attacks are more likely to have another one, this imbalance and not AGI-1067 could be the reason for the different coronary event rates. Other imbalances that favor AGI-1067 include a higher number of patients taking Lipid lowering agents (97.5% vs. 93.7%), a higher percentage of patients with diabetes in the control arm (18.9% vs. 15.3%) and a higher percentage of patients with hypertension in the control arm (63.1% vs. 59.6%).
• In the table with the plaque volume quartile breakdown for IVUS patients with both baseline and follow up data, AGI-1067 actually showed trends towards inferiority in two of the three groups listed. As the reported benefit versus baseline was initially only reported in a smaller group of those with “technically adequate recordings”, it seems as if the top line CART-2 results were no more than post-hoc subset analysis.
• It took two years for the CART-2 results to be published in a relatively obscure journal and without Steve Nissen listed as a co-author. Clearly other scientific journals and Steve Nissen had an issue with the validity of the conclusions or presentation of the results.
• Besides AGI-1067, the company only has AGI-1096 partnered with Astellas for transplant rejection, which has been in PhI since 2002. They also have a net debt position of $116m ($172m in cash minus $286m in debt) with $86m of the debt coming due on September 1, 2008. With only about $6m in property, plant and equipment, they do not have many assets besides the IP of their drugs and the cash offsetting this debt. Therefore, if ARISE fails, AGIX could very easily trade below $1 (which would still translate into an enterprise value of over $150m) and possibly well below that as the debt starts to come due.
Let us also not forget the comments from AstraZeneca's CEO made in December that AGI-1067 has a "low probability for success". Coming from a partner, that is just stunning, though hardly surprising. After all, AstraZeneca had spent only $50m on this program and had the rights to this potential blockbuster if it worked. Given the average cost of developing a drug is $800m, this is chump change. Essentially, a rounding error in the world of large pharma.
Also, while we are on the conviction that AstraZeneca has, why doesn't AGI-1067 have a numerical designation from AstraZeneca? Historically by this point, they would have one. Even Cerovive from Renovis (which failed recently) got one.
I do understand investor logic to have some lottery tickets in the portfolio. But at an enterprise value of over $500m, AGIX is one hell of an expensive lottery ticket with little chance of paying off.
Disclosure: Author is short AGIX
AGIX 1-yr chart: