- Big pharma partnership does not ensure success.
- Isis management tends to err on the side of omission in data release.
- There isn't enough data from newer short-term antisense studies to derive long-term safety conclusions.
- Even an optimal ISIS 2.0 sequence can still generate an immune response in mice.
After reading the comments on my initial article on Isis (NASDAQ:ISIS), I thought it might be interesting to try to address some of the intelligent issues raised by readers. I will try to address some of the recurring arguments below.
Argument #1 - ISIS technology is good because they have big pharma partners
This is the weakest bull argument. I don't find it persuasive at all by the fact that ISIS has partnerships with big pharma. These partnerships are not necessarily validation that ISIS' technology is "good". There are many examples where this validation ends in disaster - see Bristol-Myers (NYSE:BMY) and Inhibitex (NASDAQ:INHX) as just one famous example. A more recent example is the Merck (NYSE:MRK) and Endocyte (NASDAQ:ECYT) partnership. There are many, many examples where partnerships with a bigger partners did not result in success.
Argument #2 - Investors believe that newer ISIS 2.0 products have an improved safety profile because ISIS management says so.
Listening to ISIS management in regards to tolerability is, in my opinion, the least reliable way to get information. I am not aware, to date, that ISIS management has ever acknowledged that Kynamro has tolerability issues of any sort. They insist that Kynamro's long-term tolerability is similar to other injectable drugs. If they won't acknowledge that Kynamro has any issues today, do you think that they would do this for compounds with a few weeks of data? Perhaps ISIS management's definition of tolerable is different from others, but if Kynamro is considered tolerable, I would think that the company has set a very low bar for the definition of tolerability.
Argument #3 - Current data from more recent Phase II programs indicate newer Isis antisense compounds are better tolerated
I maintain that it is difficult to ascertain long-term tolerability from short-term studies, particularly when there is a precedent of adverse events. As shown by the data in my previous article, the rate of flu-like symptoms increase with time. Thus any small, short-term study would have a low rate. As a reminder, the dropout rate for flu-like symptoms in Kynamro was 2.7% for the first 6 months and increased to 25% in open label. In my opinion, you will need at least 1 year of data to determine if a newer compound has a different side effect profile than Kynamro. Treatment for a few weeks or months of ISIS antisense should be well tolerated; else the compound would be really intolerable. I'd also note that ISIS has consistently maintained that Kynamro is very well tolerated, so I would take current management comments on subjective tolerability with a grain of salt.
ISIS has not revealed if any of the patients receiving new antisense molecules become anti-drug antibody positive. Note that, while it takes time for anti-drug antibodies to arise, we should see a few patients starting to test faintly positive after 3 months. I would feel a lot better if at 6 months, there were no anti-drug antibodies. I agree if none of the patients that receive the new antisense generate antibodies to the drug at 6 months, this would be extremely important in validating the bull argument that changing the sequence completely changes the immunogenicity of the antisense. Seroconversion rates would be an excellent objective measure of immunogenicity of the newer antisense candidates.
ISIS has a historical tendency to disclose less than more
I think an important exercise is to examine how ISIS generally communicates clinical data. An example of how ISIS communicates safety data can be taken from their 2007 press release on Kynamro (here). In this press release, they discuss the "integrated safety analysis including data from more than 250 subjects treated with mipomersen in phase 1 and phase 2 studies." Nowhere in this press release is there any mention of any drug related dropouts. ISIS' press release is factually 100% correct, they simply did not confirm or deny any drug related dropouts. They must feel this type of information is not relevant. Keep in mind that Kynamro ultimately had a 60% two-year dropout rate in their open label study (see previous article).
If you look at the dropout disclosure for the most recent molecules, Factor XI, ApoCIII, and GCGR, none of the press releases indicate that there are any dropouts - drug related or otherwise. In fact, I could not find any indication of dropouts in their extensive R&D day presentation on May 22, 2014 (sign up to download the slides). However, GCGRrx is the only program that ISIS presented at a scientific conference, and here, physicians do care about dropouts. You find on slide 32 of the ISIS GCGR presentation that 11/49 GCGRrx treated patients dropped out the trial (combined 2 doses), which is a 22.5% rate. Not all were due to GCGR related events, but there were 4 patients, or 8.1% of the GCGRrx treated patients, that did drop out due to drug-related adverse events. Of these 4, one did drop out due to a "moderate" injection site reaction. This is probably relevant, given that injection site reactions are an issue with Kynamro. I argue that it is hard to point to the recent results of the GCGRrx data and conclude that the dropout rate is much better than Kynamro, since these overall rates still seem high for a 13-week study, in my opinion. To my knowledge, we only know the final patient disposition for ISIS-GCGRrx phase 2 and not for the other two programs. My point is that ISIS generally reserves the dropout data for scientific forums and not investor press releases. Thus, I conclude that it is dangerous to assume that if ISIS does not mention dropouts in the press release, that there were none.
Debate - what is the relative importance of sequence vs. modification
Some investors espouse the view that the key immunological feature of DNA is the sequence. While there is no doubt that sequence is an extremely important determinant in the short-term immunogenicity of DNA, these people neglect the fact that the anti-drug antibodies generated to Kynamro have specificity to the MOE modification. Oligonucleotides which do not have this modification do not bind antibody. This data argues that the sequence is not a major determinant in the generation of antibody binding in the case of Kynamro because these antibodies can bind independent of sequence (page 27-28, FDA Other Reviews). High affinity antibody formation is a multi-step process, and it takes time. Anti-Kynamro antibodies take months to develop. How a short term in vitro assay could predict this seems a stretch. A real preclinical test for the potential for antibody formation would be after repeated injection in rabbits, since we know rabbits can generate high affinity antibodies to at least Kynamro (page 28, FDA Other Reviews). I argue the real question the bulls should be asking is whether long term administration of all ISIS 2.0 MOE antisense in rabbits result in high affinity antibody formation. Ultimately, this argument will only be settled by long term human data from other ISIS programs. If antibodies to MOE are generated with the other ISIS 2.0 antisense, this would be a strong argument for the role of MOE in the generation of an immune response independent of sequence. If antibodies are not generated to the other ISIS 2.0 MOE antisense, then I would be incorrect. However, I'm betting that we will see this MOE specific antibodies again.
Published ISIS preclinical data indicates that ISIS 2.0 chemistry can still generate an immune response - even a optimally silent one
For those really interested in the science, ISIS scientists actually did a study which compared an immunologically active 1st generation antisense compared to an optimally immunologically silent ISIS 2.0 (Jenn et. al.). Note that while this study is old, it doesn't make it less true. In these series of experiments, Isis scientists demonstrated that even an ISIS 2.0 antisense completely devoid of the "CG danger motif" was still able to increase spleen weight, generate monocytic infiltrates in the liver, and induce cytokine production in the liver of mice. The scientists conclude that there is about a 10-fold improvement in immunogenicity between the 1st generation vs. 2nd generation antisense. Note that these experiments compared an immunogenic sequence using 1st generation technology compared to an immunologically silent 2nd generation technology, thus likely representing a best case scenario.
I will point out that due to the 1-2 month half-life of Kynamro and the highly variable metabolism of ISIS 2.0, some patients can clearly accumulate very high levels of drug (see highest trough vs. normal trough discussion in my previous article). Remember that the trough level in long-term treated patients could vary by over 150-fold, which would easily eat up a 10-fold difference very quickly.
After considering comments to my ISIS short investment thesis and conducting additional research, I still think the data leans toward the idea that ISIS 2.0 is immunologically active. I do not believe that by simply changing the sequence of the antisense will markedly change the long-term safety because I believe that the immune system is seeing the MOE modification and not the sequence. Keep in mind that the antibodies generated to Kynamro only bind to DNA which have this signature ISIS modification, which is found throughout ISIS 2.0. In addition, I think it is also clear that even in ISIS' hands, higher concentrations of an immunologically silent ISIS 2.0 sequence can still be seen by the immune system at higher concentrations, as evidenced by Jenn and colleagues. While there may be a safety margin, Kynamro has such wildly variable pharmacokinetics that a 10-fold safety margin may be swallowed up by the 150-fold difference in drug trough levels. Lastly, and most importantly, I don't think that you can view the omission of certain tolerability data from this company as affirmation to the positive in the more recent programs. For instance, there were 4 drug related adverse event dropouts in the GCGRrx phase 2 study, with one due to a "moderate" injection site reaction. Whether this indicates an "improvement" in the newer antisense sequence selection compared to Kynarmro, I can't say for sure. However, I can say, to those that believe that there were no drug related discontinuations in the recent phase 2 studies as a sign of improved tolerability, that they probably need to look at the data harder. This is definitely not true in the case of GCGRrx.