Mipomersen Efficacy Surprises to the Upside, on Track to Become First Commercially Meaningful Antisense Therapeutic

ISIS Pharmaceuticals and co-development partner Genzyme reported today top-line data from the final 2 out of 4 phase III studies with the ApoB-targeting antisense compound mipomersen for the treatment of hypercholesterolemia. The efficacy results came in above my lowered expectations and support mipomersen to be an attractive treatment option for the many patients with severe hypercholesterolemia for which current care is still inadequate and who are therefore at a high risk of suffering cardiovascular events. Further studies, however, will be needed to better understand the clinical relevance of the observed liver enzyme and fat elevations.

As I wrote earlier this year, the first 2 phase III studies in patients with homozygous (25% LDL-cholesterol reductions) and heterozygous (28% LDLc reductions) familial hypercholesterolemia raised concerns that the efficacy of mipomersen could be less than one might have expected based on the initial phase I/II trial results. Couple this with a number of safety and tolerability issues such as elevated liver enzymes, fat accumulation, plus wide-spread injection site and flu-like reactions, the overall profile may not have looked that attractive. The 36-37% reductions in LDLc reported today for the severe hypercholesterolemic and high-risk/high-cholesterol patient populations, however, put efficacy concerns to rest, and it is possible that intact LDL-receptor function contributes to LDLc lowering, somewhat similar to statins.

Actually, due to the high drop-out rates in the latest studies (31-41% for those on drug), the LDLc reductions are likely to be quite a bit higher, at least on par or superior to LDL-apheresis which many of these patients are eligible for, but not widely used due to cost and safety/tolerability/patient convenience reasons. Moreover, the on-target efficacy, and not 2^nd gen antisense as a class, may indeed be responsible for some of the safety issues as the companies strongly suggest that the liver enzyme and fat elevations were in fact correlated with the speed and degree of the LDLc reductions. In some cases apparently more than 80% reductions in LDLc were achieved!

If this were the case, one can easily imagine that in real life, patients could be simply monitored for liver enzymes and fat content as they go on drug, similar to what is already routine for other widely used drugs such as warfarin. Although clearly something that regulatory agencies will want to see studied further (e.g. the ~200 persons that have received full 26-week course of drug may alone not be sufficient to argue against liver tox based on Hy’s Law), at this point it is actually not clear to what degree ApoB knockdown-related enzyme and fat elevations would be clinically relevant. The strategy of the companies to pursue a step-wise filing strategy starting with the highest risk patient populations (planned for early 2011) before seeking to expand the label is therefore very reasonable. For commercial success, the severe hypercholesterolemia population should be key and could already be part of the first filing cohort.

SNALP-ApoB in light of mipomersen data

Now a few words on how the mipomersen results may impact Tekmira’s RNAi Therapeutics candidate SNALP-ApoB which targets the same gene as mipomersen for the treatment of hypercholesterolemia. Assuming that liver enzyme and particularly fat elevations are indeed a function of ApoB knockdown, then some of the same strategies such as gradual dose adjustments may be applicable. In general, as the runner-up, Tekmira has a lot of valuable lessons to learn from mipomersen especially with regard to its regulatory path.

At the same time, to differentiate itself in the market, Tekmira may also want to take advantage of some of the challenges that mipomersen is encountering when it comes to drug tolerability, particularly the injection site reactions and flu-like symptoms that appear to be responsible for quite a few of the study drop-outs and where I believe that SNALP-ApoB could do better. Although I had never thought that being intravenously administered as opposed to the subcutaneous administration of mipomersen would be a major issue, competitive or otherwise, for the high-risk patient populations we are talking about, the fact that ISIS and Genzyme are now considering daily injections as an option to address the injection site reactions even makes intravenous infusion every 2-4 weeks look like a competitive advantage now.

Meanwhile, due to the satisfactory efficacy of mipomersen and the fact that you probably don’t want to overdo it with lowering ApoB as long as it is in the 32-40% range, efficacy should not be too much of a differentiating factor.

All of this is in no way to suggest that the current value of SNALP-ApoB is comparable to ISIS’ mipomersen. Tekmira has yet to enter the clinic with its improved SNALP-ApoB candidate and we all know that clinical development is fraught with surprises. In any case, it is quite satisfying to see oligonucleotide therapeutics progress in the clinic on so many fronts and I wish ISIS and Genzyme well for the commercialization of mipomersen as it would also reflect well on the commercial value of the entire space in general.

For Mipomersen, 10% a Big Difference Does Make

I must admit that I have been very wrong about how clinical results from the ApoB-lowering drug candidates by Tekmira (ApoB-SNALP RNAi, phase I) and ISIS Pharmaceuticals (mipomersen antisense, phase III heterozygous FH) would turn around the sentiment for RNA Therapeutics. In both cases, investors sold off the stocks on results that I believe bring both approaches closer to commercialization. It is almost as if the market is one in which home-run clinical results get rewarded with a doubling in share price, while those in-line with expectations are sold off by about 20%.

Compounding the problem for mipomersen is that investors were secretly hoping for slightly better LDL-cholesterol reductions. This was particularly the case because the modest 25% reduction observed in the previous homozygous FH trial had always been down-played by management as being attributable to a challenging patient population, and that we should expect much more from the subsequent phase III trials. Well, technically, the reduction was slightly improved, but with 28% not too much to get anybody excited.

While analysts were largely focused on potential safety problems, particularly questions about the seriousness of observed liver enzyme elevations, I believe that this attention was very much caused by the disappointment about the degree of knockdown. This is because in order for mipomersen to become a commercially successful drug, it needs to tap the market outside the ultra-orphan homozygous FH population, especially the severe hypercholesterolemic patient population that does not respond sufficiently to current maximal drug therapy and are either on LDL apheresis or eligible for it. Tapping this market or not can make all the difference, the difference between $10M and $500M-$1B in annual sales.

The reason why the LDL apheresis population is so attractive is because mipomersen would aim to replace another procedure that specifically aims at lowering LDL-cholesterol, and demonstrating a comparable knockdown should be sufficient to get approval without time-consuming and costly outcome studies. It could also lead to adoption of mipomersen by some of the other high-risk populations, but this would only be gravy.

LDL apheresis is somewhat unpleasant and, depending on the particular apheresis technique applied, removes good stuff like coagulation factors and HDL cholesterol, too. It does, however, do a fairly good job in removing LDL-cholesterol, by about 30-55%, and in order for mipomersen to be considered an alternative, it needs to get into that range. At the moment, mipomersen is on the lower end of this spectrum where safety could tip the balance, and with lots of questions regarding the safety being left unanswered, this can explain the sell-off we have witnessed.

On a positive note, the 28% LDL-cholesterol reduction may be a very conservative number. This is because it does not take into account the +5% increase in the control group and the fact that the number was on an intent-to-treat basis, i.e. they included also those patients in the drug-treated group that dropped out early, so that the adjusted number could well be in the -36% range. Moreover, the natural person-to-person variation could mean that about half of those that stayed on the drug achieved very meaningful LDL reductions which could prove crucial during the regulatory application process. The roll-over rate into the open-label, uncontrolled phase of the study was also said to be ‘good’ and would be quite meaningful as this provides a good measure about the adoption of mipomersen in real life. However, given a slight tendency by ISIS to over-promise, we really have to await the precise numbers that are to be presented at an upcoming scientific conference.

In summary, the heterozygous FH results were very likely not as bad as the market reaction would suggest and eyes are now on the full data presentation and, probably even more importantly, results from the phase III study in the critical severe hypercholesterolemic patient population. The reaction highlights, however, that slight differences in knockdown potencies can make a huge difference for RNA Therapeutics. The upcoming dose-fractionation study by ISIS and Genzyme, while labeled as being for the ‘convenience’ of patients, may actually be driven by the hope it might provide for improved knockdown due to different pharmacology. And finally, for RNAi Therapeutics, and Tekmira in particular, the good news is that the pioneering work by ISIS provides them with a very good idea of what their ApoB candidates will have to achieve in order to be commercially successful. A modest 35-40% knockdown should be well within the grasp of current SNALP-siRNA delivery technology.

Tekmira Reports First Knockdown in Man Following Systemic RNAi Delivery

Tekmira today reported very encouraging results from their first clinical study with SNALP siRNA delivery targeting apolipoprotein B for the treatment of hypercholesterolemia (PRO-040201, aka SNALP-ApoB). While the ~20% ApoB/LDL-c knockdown following single dose administration and evidence for flu-like toxicities at the highest tested dose level indicate that the present formulation is not suitable as a therapeutic, the company is rightly optimistic that based on these results, follow-up formulations that it intends to enter into the clinic later this year will be able to provide significant improvements in the absolute dose and therapeutic index. This is because the new knowledge can now be applied to the rapid progress that has been made in finding much more potent SNALP/LNP formulations (see recent PNAS paper and ILS2009 presentation) since the present formulation was locked down 2 years ago. In addition, much has been learned since on how to predict and mitigate SNALP siRNA-triggered immunostimulation in humans (extensive review). I therefore agree that it was probably the scientifically and financially best decision by the company to stop this trial at this stage and focus their resources on the next formulation.

When I first adopted SNALP/liposomal delivery as my pet systemic siRNA delivery approach 2-3 years ago, my main outstanding nightmare stemmed from an unexpected early-stage clinical disaster during a phase I trial by the same company (Protiva branch) almost a decade ago employing related stabilized cationic liposomes for the delivery of plasmids (SPLPs). In that trial, severe flu-like symptoms were observed at the very early and low dosages and the was trial terminated and the drug never heard of again. Instead of giving up, this has stimulated them to undertake almost heroic and industry-leading work aimed at mitigating immunostimulatory effects of liposomal nucleic acid delivery. It is thus quite remarkable that after just seven years after stumbling on siRNAs as more promising payloads for the technology, Tekmira has now demonstrated unambiguously functional liposomal siRNA delivery following systemic (intravenous) administration, with first, transient flu-like symptoms observed at what should be much higher dosages compared to the SPLP trial.

It is worth bearing in mind that the decision to stop the trial was not because the one case of flu-like adverse event was deemed a serious one, but because they felt that at this point they had already learnt what they wanted to learn and given the superior performance of the next-generation formulations it would therefore not have been ethical to further expose patients to a drug that will not be developed for commercialization (all my interpretation). It also might have risked tainting the asset with little to gain. To put the toxicity into context: in the most recent published clinical results on ISIS’ mipomersen targeting the same gene, 70% of patients receiving drug exhibited flu-like symptoms at similar 20% ApoB/LDLc knockdowns.

Equally encouraging is the fact that the knockdown, which based on the natural intra-person lipid variability and the pharmacokinetics of the reported knockdown (fast onset as expected for an RNAi drug) was almost certainly due to drug, occurred at dosages where the company expected to start seeing efficacy based on pre-clinical animal studies. One cannot underestimate the importance of being able to predict efficacious dose (and toxicity) based on these models, particularly at the early stages of evaluating such a novel technology where very little human pharmacological experience has been obtained.

The trial design involved cohorts of 4 patients for each dose level: one treated with placebo and three with study drug. While 8 cohorts had originally been planned, the trial was stopped in approximately the 6th or 7th cohort. According to the company, at this point the dose had been escalated to somewhere in the 0.6mg/kg ballpark and is thus consistent with IC50s for SNALP-siRNAs in non-human primates of ~1mg/kg at that time. Considering that according to a presentation at the International Liposome Society a month ago by Pieter Cullis, scientific founder of Tekmira, SNALP-like formulations have now reached IC50 potencies in non-human primates of ~0.03mg/kg, i.e. around 1.5 magnitudes better, there is every reason to believe that significant improvements in absolute dose and therapeutic index can be achieved with the follow-on SNALP-ApoB candidate. Strengthening this claim, similarly potent formulations have just recently surfaced from a study supposedly (Alnylam IR: I still cannot find the paper) published in PNAS by Alnylam and collaborators at the MIT on lipidoid-containing SNALP-like particles (allow a delay of at least a year from experimental finding to publication).

While all this sounds good, a critical question to ask is what gives them also the confidence that the new formulations (= more potent liposomes + less immunologically active siRNAs through modification) will not perform worse immunologically than the first one, thereby eliminating some of the benefit from the improvements in SNALP potency. The ability to answer this question would hinge on the ability to sensitively detect immunostimulation and translate such findings pre-clinical models into humans. The JCI paper last year by Tekmira scientists on the application of SNALP for distant tumors in mice showed that Tekmira indeed understands how to detect such responses even in the absence of cytokines in the serum (tissue-based PCR assay for IFIT1 mRNA). And from all the presentations I have seen, another important piece of the puzzle in being able to translate from animals into humans is the good correlation between immune responses seen in in vitro human PBMC immuno-assays and responses in animals. In combination, these elements allow for efficiently modifying siRNAs until predicted immune-related tox in humans have been minimized. Through such knowledge, Tekmira was able to provide a good explanation for why some non-human primates exhibited liver toxicity in the 2006 Nature study (the first demonstration of systemic RNAi in non-human primates) even in the absence of outward immune activation and remedy it through siRNA modification. It is interesting, however, that in the present clinical trial the presumed immune stimulation appeared to have (thankfully) been uncoupled from liver enzyme elevations as an absence of liver toxicities was noted. Although the reasons behind these are unclear, this uncoupling lends further credence to the notion that also efficacy and liver toxicity can be uncoupled: the mechanism that allows for functional entry of the liposome itself does not cause hepato-cytotoxicity.

According to a conversation with the company that I had following yesterday’s release of top-line results, it also appears that they have developed what appears to be a separate assay that they have not disclosed yet, and that should be even better in predicting immune responses in humans. Although, in the absence of seeing the data, I will have to take the company by its word here, as a proven leader in liposomal siRNA delivery, I’d like to give them the benefit of the doubt.

Looking ahead

The company indicated that the new formulation will go into the clinic later this year, although discussions with the FDA have yet to determine the exact regulatory requirements for this (e.g. an abbreviated IND package would make sense IMO). Like the first trial, the second one is likely to be a single-dose one, and a multi-dose trial in humans would similarly require additional pre-clinical data. All this makes it clear that the main value of this program to the company is that ApoB is an exceptionally good target to most rapidly develop insights into the entire SNALP-siRNA delivery platform. The market reaction tomorrow, however, will show whether this is shared by investors who may be disappointed by the delay on ApoB and may have also hoped that SNALP-ApoB could find a development partner soon. However, I would think that most invested in Tekmira as a leader in systemic RNAi delivery and are equally encouraged that the company is now able to build real data which have for the first time demonstrated unambiguous systemic RNAi gene silencing in Man. In addition to investor perceptions, the financial repercussions of these results also depend on what value such pioneering work will fetch from potential partners that are interested in the broad development of the RNAi Therapeutics platform.

Whether Tekmira’s enthusiasm is shared by current partners will be seen by their clinical actions. As such, Alnylam until recently reiterated that it was still on track to file an IND for the liver-directed SNALP-TTR. Since Alnylam apparently was aware of the status of SNALP-ApoB in November already (see RNAi Clinical Experience chart in this slide presentation), Alnylam also seems to be comfortable that the more potent SNALP formulations will have clinical applicability given Tekmira's results. I look forward to learning about the TTR trial design, including whether multi-dose will be attempted, and there should be results by the end of this year. In addition, the fact that we have just learned that Roche entered IND-enabling studies, most likely with a SNALP product, further validates Tekmira’s decision. The fact, however, that the IND submission of SNALP-PLK1 has been moved into the second half of this year represents a slight delay, and it remains unclear whether this might be related to today’s news.

In any case, today makes me feel like we have witnessed possibly the beginning of the most important line of RNAi Therapeutics clinical research thus far. Many more data-points on clinical SNALP-siRNA delivery should accumulate over the next two years.

Disclosure: I have financial interests in the company, largely by holding TKM shares.

Note added in proof: Scientific papers by Tekmira and Alnylam on the development of significantly improved SNALP formulations:

1) By combinatorial chemistry screen (December 2009 PNAS paper)

2) By rational design of lipids (January 2010 Nature Biotech paper)