If the study reveals that Stimuvax fails to improve lung cancer survival, as I suspect it will, Oncothyreon stock will drop 80% to cash value of $2 per share. I believe there is a high likelihood that the trial will be stopped for futility at the second interim analysis in a few months. If not, the study will fail at the final analysis which should be about 9 months from today. My hedge fund is short ONTY common stock and long various ONTY put options.
There are two smoking guns that support my case: first, Stimuvax is not a well-designed drug and is likely to be inert in humans. The second is that the phase II data are not suggestive of a benefit, and the phase III is highly unlikely to succeed. We will prove that recent analyses speculating that the length of the trial is somehow predictive of Stimuvax success are illegitimate and only serve to increase the risk/reward ratio for short sellers.
Mechanism of Action
Stimuvax is often referred to as a MUC1 “cancer vaccine”. The problem with this terminology is Stimuvax is not a vaccine in any way, shape or form. Vaccines typically stimulate an immune response against an introduced antigen. This allows the immune system to fight off this antigen in the future.
Without some leukopheresis/immune-training process, a-la Dendreon or a specialized adjuvant, it is difficult to generate a vaccine to an endogenously found peptide. A MUC1 antibody or small molecule antagonist would make a lot more sense. Indeed, several companies are attempting to improve the immune response for prospective MUC1 vaccines as well as attempting targeted payload strategies for MUC1. MUC1 is certainly a good marker for cancer, but it still isn’t useful. MUC1 tests are no longer used. MUC16, which you may know better as CEA125, is widely used to determine ovarian cancer staging and status. Antibodies, armed antibodies and vaccines to CEA125 went nowhere.
According to the NIH’s protein database, mucin 1 (MUC1) is a 1255 amino acid protein. Stimuvax is the common repeat found in mucin 1. I wonder if the inventors assumed this will serve as an antigen to the immune system and T cells will destroy any cell with mucin 1 expression. It doesn’t take a cell biologist to appreciate this is not a good idea given the widespread expression of MUC1.
Furthermore, without legitimate immune response data and with no side effects, my supposition that Stimuvax is inert is likely correct. The tandem repeat found in Stimuvax is no different from native MUC1. While the glycosylation patterns may be different, it’s hard to see MUC1 inducing a potent immune response. Indeed, the failure of the similar MUC16 to generate a viable anticancer candidate bodes poorly for MUC1 and Stimuvax.
Getting back to the question of whether Stimuvax is an effective vaccine, we noticed that only 21% of Stimuvax patients in the Phase IIB study had an immune response to Stimuvax. Because the only way Stimuvax could have any anti-cancer effect would be through immune recognition of the MUC1 tandem repeat, and a subsequent immune response strong enough to destroy MUC1 expressing cells (a long shot), a 21% immune response (forget evidence of a CR/PR) is not promising.
Most good vaccines have to have 100% immune responses or close to it. The immune response data is uncontrolled and hazy at best. For a drug billed as a vaccine, Stimuvax doesn’t do a great job of proving it is a vaccine. Even in a small n=22 trial that tested a new formulation of Stimuvax, immune response results were not disclosed.
Let’s move on to what everyone really cares about. ONTY has recently risen dramatically because there is speculation that the START study will succeed. “START” is the Phase III study Merck KGAA is conducting to determine if Stimuvax extends overall survival in lung cancer. It is important for anyone trading ONTY to understand, that it is much more likely any putative “extension” or “delay” in the combined overall survival of the START study, is much more likely to be a result of rising survival expectations for this tumor type and a miscalculated overall survival, than it is to be a miracle performance from Stimuvax.
I’ve seen this movie before, and it never ends well. The long trade has a very poor risk/reward characteristic and it is mathematically inept to assign a higher probability of success to START because the combined median overall survival is higher than originally expected. Unless you can perfectly estimate the control group overall survival of a new study by looking back at literature, and taking into account the several new drugs approved for this disease, you are making a mental assignment error of hoping you’ve stumbled onto a free lunch when you’ve only found yourself in a game of speculative musical chairs.
The enrollment curve that was released at ASCO has helped us determine that the START trial is exactly on schedule. The control group should register a 33 month median overall survival. We also think the Stimuvax arm will register a 33 month median overall survival. This data is backed by recent clinical results (Kelly et al, Franca et al, Gadgeel et al and Chang et al) as well as the recent innovations in lung cancer.
Again, nothing can be determined about the performance of the individual arms from the data released thus far. We only have two data points from the trial: the enrollment curve, and the timing of the first interim analysis. One can relatively easily impute that the combined overall survival of the study is 33 months. It is tempting to look into the scientific literature and see that there are some lung cancer studies with 25 months of overall survival.
It is even more tempting to assume Stimuvax must be on track for 37 months of overall survival given the 2:1 randomization of the study, or any other sizeable benefit. There’s a lot of hand-waving that can be done here. Instead of retrofitting past data to speculate on the potential clinical benefit of a drug in a new trial, investors are better served assuming the null hypothesis that the combined median overall survival of the trial fairly represents today’s standard of care and the investigational and control groups do not differ.
It is downright silly to assume the Stimuvax arm is outperforming the control arm without any a priori evidence of this. As we’ll see later, the only basis for assuming Stimuvax success is a single specious Phase II study. Stage III NSCLC is highly heterogeneous and survival can be well in excess of the combined observed 33 months in START.
The recent advances include Avastin, Tarceva, Alimta and other drugs, as well as improvements in radiation techniques and genetic sequencing of tumors altering treatment paradigms render looking at past data to determine the likely performance of the control group of START a waste of time. I posit that when all is said and done with START, the Kaplan-Meier curves of the two arms will be found to overlap. History tells us a cancer trial taking longer than expected is not suggestive of a benefit. Recent examples of this include Novelos, Genvec and Telik.
Buying ONTY is a bad idea, especially if you think the recent statistical analyses by “biostatisticians” suggest a successful study. I recall fondly Northfield Laboratories where we hired 10 separate biostatisticians (they’re cheap) to analyze that pivotal trial. These folks are prone to biases and like stock-pickers, it’s hard to find good ones that are impartial and thorough. We have learned enough about statistics in our collective experience to dismiss the recent analyses as uninformed opinion to take advantage of.
Prior Stimuvax Data
The prior Stimuvax data is not encouraging. This data has serious flaws with a small n, open-label nature and a large background imbalance that would have favored Stimuvax. There is a rich history of positive phase 2 lung cancer trials failing to replicate in phase 3 (Affinitak, CPG7909, ASA404, figitumumab, Novelos and others). Subgroup analyses with no rational basis are done to support Stimuvax efficacy. Even these subgroup analyses don’t rise to the level of p<0.05.
Another subtle but important finding from the Phase II data is the separation of the Kaplan-Meier curves after overlapping for 12 months. Most legitimate cancer drugs have quick separation of K-M curve. This “late” effect can be ascribed to baseline characteristics. Metastasis kills cancer patients. Stimuvax doesn’t seem to have an anti-metastasis mechanism of action, but the data bears out its late effect. It’s much more likely there was a baseline imbalance, and indeed there is substantial evidence for that.
Even if Stimuvax works, which I am positively sure it does not, the valuation of ONTY isn’t tremendously compelling. With 60,000 applicable lung cancer patients (Stage III is rare), 30,000 on treatment at $50,000, total end-user sales of $1.5bn, a royalty to ONTY of $150m, net profits of $100m, 10x earnings results in a $1 billion market cap for ONTY.
This is roughly a double for the stock in the best case scenario. The odds are not 50-50. They’re more like 5%. You lose almost all of your investment if the trial fails. Most lung cancer phase IIIs fail. Without a rationale design, reasonable phase II data or a better reason to invest, I would walk away from ONTY or short it. Our fund is short and owns puts.
Disclosure: We are short ONTY and may reverse, hedge, liquidate or otherwise modify our position at any time without notifying SeekingAlpha.