FDA Cautious on Triglyceride-Lowering Antisense Drug WAYLIVRA

Today, the FDA released review material for the Advisory Committee Meeting on Volanesorsen (trade name WAYLIVRA) this Thursday.  WAYLIVRA has been developed for the lowering of serum triglyceride level in patients with a genetic form of severely elevated triglycerides, familial chylomicronemia syndrome (FCS).  Briefing Docs provide valuable insights as to the safety of drug candidates which are often glossed over by sponsor companies leading up to such regulatory events.

Echoes of KYNAMRO

The last time we had this opportunity in the phosphorothioate antisense space was in 2012 when the subject was the ApoB-lowering KYNAMRO.  Like KYNAMRO, WAYLIVRA is a second-generation 2’-MOE phosphorothioate-backbone oligonucleotide that was given at high doses in the registrational trials (200mg for KYNAMRO, 300mg for WAYLIVRA). 

It should therefore not come as a surprise that the safety and tolerability of WAYLIVRA was quite poor. In particular, not only did the at times drug-induced severe thrombocytopenia cause considerable bleeding concerns, the sponsor Akcea and parent company Ionis Pharmaceuticals have no idea about how to mitigate such risk.  Dose adjustments and increased platelet monitoring apparently did not change the thrombocytopenia risk substantially and apparently are still being made to this date without much rhyme or reason.   

In addition, there were the flu-like symptoms, liver enzyme elevations, renal tox, even a case of anaphylaxis causing around half of what should be highly motivated patients with FCS to drop out from the clinical studies.

This time it’s potent though

Fortunately, unlike KYNAMRO, WAYLIVRA has robust potency.  For those that managed to stay on the 300mg weekly dose, serum triglyceride lowering in excess of 70% was seen (reflecting ~90% target ApoCIII-lowering; cf. ~20% target-gene lowering with KYNAMRO).  Even the FDA had to acknowledge here that this is an unparalleled achievement.

But the positives end here already, with the FDA noting that in contrast to Akcea’s/Ionis’ claims there is no reliable reduction in measures of actual morbidity of FCS patients such as pancreatitis attacks, abdominal pain, and general well-being.  In fact, on most measures there was not even a numerical benefit favoring WAYLIVRA.  This, the FDA agrees, is also a function of the low number of patients available for such ultra-orphan disease studies.

The FDA admits that lowering of the biomarker serum triglyceride, even to much lesser extents than WAYLIVRA, has formed the basis for approving other drugs for hypertriglyceridemia-related disease.  In light of this, I expect WAYLIVRA to receive a favorable panel vote on Thursday on the condition that access to WAYLIVRA is strictly limited and safety tightly monitored (à REMS).  It apparently scares the FDA that while FCS has an estimated prevalence of 1 to 2 in a million, up to 1 in 600 have very high triglycerides resembling FCS.

Having said this, I am not quite clear why there should be different risk/benefit threshold with regard to the triglyceride-related morbidity for FCS and conditions resembling FCS as long as these other patients have exhausted options to treat the underlying causes for their particular hypertrigliceridemia.  As such, I believe the market potential estimates ought to be dramatically increased for WAYLIVRA.

WAYLIVRA competition

It is already clear, however, that WAYLIVRA will have a limited shelf-life.  Hot on its heels is a GalNAc-conjugated in-house competitor (ApoCIII-LRx) which has shown triglyceride lowering approaching that of WAYLIVRA, but at 5-10x lower dose levels, less frequent dosing and, crucially, with supposedly none of the worrisome safety issues of WAYLIVRA.  A phase III study of that compound is planned for 2019.  By that time, an RNAi GalNAc compound by Arrowhead should also be well in the clinic setting up ApoCIII-lowering to become a substantial market for oligonucleotide therapeutics by expanding the market well beyond the FCS population.   

Discrepancy in Claimed Injection Site Reaction Frequency with ISIS-ApoCIII

My fundamental believe as to why RNAi is fundamentally preferable over phosphorothioate (PS) antisense for therapeutic gene knockdown, is that PS antisense involves the saturation of various tissues in the body with a sticky chemistry that is associated with inflammation.  By contrast, with RNAi Therapeutics, you only need to transiently achieve relatively high tissue levels so that enough of the RNAi silencing machinery can be fed with the RNAi triggers.  Unincorporated oligonucleotides can, and usually are, then washed out without much or any loss of efficacy.

Especially in chronic settings and systemic administration, this persistent inflammatory herd raises the specter of multi-organ fibroses and other adverse immune effects, including vasculitis.  The mipomersen briefing docs showed that such concern is warranted, especially with the findings of vasculitis in non-human primates and frequent injection site reactions which, in contrast to ISIS management which routinely characterizes them as ‘cosmetic’ and ‘nuisance’ side effects, should be interpreted as indicators of what might be happening more systemically.

As you also know, I am equally wary of the reliability of ISIS’ forward and backward looking statements.

Obviously, when ISIS reported phase II data from theISIS-ApoCIII study showing a deep 88% knockdown of the target gene and impressive >70% reductions in serum triglycerides, the question of whether this is a real drug or another KYNAMRO/mipo-like dud hinges on its safety profile.  Unfortunately, no hard numbers were provided on this front in the initial press release.  Instead, there was much beating around the bush characterizing them as being very infrequent and mild in nature. 

However, in the conference call, one analyst apparently wasn’t satisfied with this and asked for a straightforward quantification of the injection site reactions (the following is an adaptation of the telcotranscript on Seeking Alpha):

Jim Birchenough – BMO Capital Markets

Yeah. Hi, guys. Congratulations on the data. A few questions…And then just final [third] question, if you could quantify injection site reaction profile and contrast it with KYNAMRO, I think that would be helpful…

Richard S Geary (ISIS Pharmaceuticals)

Good. This is Richard, and thank you for those questions...The third question, I don’t think I got.

Stanley T Crooke

Reactions compare and contrast to KYNAMRO.

Richard S Geary

So you will remember that in our Phase 1 experience we actually compared directly with three weeks study in normals and we saw almost a 90% reduction in the frequency of ISRs.

Stanley T Crooke

Compared to KYNAMRO.

Richard S Geary

Compared to KYNAMRO, and I would say that looking at these Phase 2 studies, this is being replicated very low frequency and much less severity. These are very mild and it’s almost all erythema that resolves very quickly.

Stanley T Crooke

So we’re seeing much more mild and much less frequent injection site reactions in this study as we saw in earlier work. Is that a fair way to say it, Richard?

Richard S Geary

Yes.

A 5-fold exaggeration of the reduction in injection site reactions

So while ISIS did not add much regarding the frequency of injection site reactions in the phase II study, at least a relative number was provided for ISIS-ApoCIIIRx.  Going back to the phase I trial for which the results were recently published (Graham et al. 2013), the following was said about the frequency of injection site reactions:

‘The most common adverse event was mild injection site reaction, a typical response to subcutaneously administered drugs.52 No subject dosed with placebo complained of injection site reactions, al- though 13 of 25 (52%) subjects dosed with ISIS 304801 experienced at least one. Approximately 1 of 6 injections (median, 17%) led to an injection site reaction, the majority of which resolved within an hour.’

To test the veracity of ISIS’ comments, you would then have to dig up the numbers for mipomersen.  In the FDA briefing docs, the following can be found about the frequency of injection site reactions:

‘ISRs were the most commonly reported AE in the clinical development program. In the pooled Phase 3 trials, 84.3% (220/261) of mipomersen-treated individuals experienced 3,683 ISR events and 33.3% (43/129) of individuals in the placebo group experienced 139 ISR events.’

Assuming for simplicity that the 261 subjects in the phase III studies received 52 injections minus X due to drop outs (note: this fudge factor actually favors ISIS’ numbers as it increases the injection site rate for mipo) we arrive at around 3683 injection site reactions from 10,000 injections, i.e. a rate of around 37%.  If ISIS’ comments on the 90% drop in injection site reaction frequency were then true, the rate for ApoCIII should be 3-4% and not 17% (according to ISIS the mipo injection site reaction rate would have to be 170% for the numbers to match up).  In a 13-week trial, a 17% injection site reaction rate also means that the majority of subjects will experience such reactions, some more often and severe than others.

Of course, I am just a molecular biologist and know little math. I therefore look forward to the ISIS bulls or even the company itself to point out where I went wrong in my calculations.

First Antisense Oligo Approved for Gene Knockdown in Liver

In a milestone for antisense oligonucleotide technology, Kynamro (aka mipomersen) yesterday received US regulatory approval for the treatment of homozygous familial hypercholesterolemia (hoFH).  It is the first regulatory approval for a systemically delivered oligonucleotide-based drug addressing a target in the liver, following two locally applied oligonucleotide drugs (aptamer Macugen and antisense oligo Vitravene) that have been approved for ocular diseases.  Considering that next month a TLR oligo-enhanced HepB vaccine by Dynavax could join the ranks of oligonucleotides licensed for medical use, oligo drug technology is rapidly shedding its image as an oddball technology and going mainstream.

Commercial Prospects

The hope, of course, is that Kynamro will also be the first commercially meaningful oligonucleotide considering lackluster sales histories by Macugen and Vitravene.  Favoring such success is the strong pricing power with drugs for orphan indications such as hoFH which is estimated to affect 3000 patients in the US and Europe each.  Small molecule drug Juxtapid by Aegerion for example received approval for the same indication last month and will price at $250.000 per annum.  With 1000 patients on the drug, this would mean sales of $250M already. And with the fudgible definition of symptomatic, instead of strictly genetically defined homoFH and the aggressive ‘friending’ of orphan drug salesforces with disease groups, I expect quite a bit of ‘label creep’ (Aegerion's CEO has indicated that he expects hoFH patients to somewhat respond to PCSK9 inhibitors which I can only explain if the patient population he considers homoFH includes quite a few non-hoFH when defined genetically).

With an expected price for Kynamro of $100k per annum, I optimistically project around $100M in annual sales assuming marketing partner Genzyme can get 1000 patients on drug in the US. With a ~35% profit share, a 50% profit margin and 50% attrition, ISIS Pharmaceuticals would pocket around $10M in annual profit share from this approval in a few years.

Update 30Jan13: It appears that according to a WSJ article (to which I cannot get access), Kynamro will be priced at $176k per year.  This, of course, would significantly improve ISIS profit share assuming same patient numbers treated (possibly 70% profit margin, slightly higher royalty rate in addition to higher sales --> $20-25M p.a.).

   

RNAi versus Antisense for gene knockdown in the liver

To maintain such sales numbers, however, the salesforce will have to fight hard to keep patients on the drug.  During the extension of the phase III study in homoFH, over half of patients  discontinued due to safety and tolerability issues, injection site reactions and liver safety (prompting a Black Box warning) being the main culprits. Having said that, the label for Juxtapid does not look any better, in fact a bit worse with all the intestinal side effects and drug-drug interactions, offset, however, by the increased potency in LDL-c lowering.

As some of the side effects are not target-specific (e.g. the injection site reactions and flu-like symptoms and also some of the liver enzyme elevations which are not fully accounted for by the liver fat elevations), the modest tolerability of and safety concerns with Kynamro (including vasculitis as well as liver and skin cancers especially when considering the preclinical data) raises the question whether systemically delivered phosphorothioate-based antisense technology has much of a future for gene knockdown in the liver also in light of increasing competition from RNAi delivery technologies.  You will not be surprised that I subscribe to the view that among Tekmira’s SNALP and the subcutaneous alternatives by Arrowhead (DPC) and Alnylam (GalNAc), RNAi will become the preferred technology for gene knockdown in the liver.

Whether RNAi or antisense, oligonucleotide therapeutics are here to stay. 

The Mipomersen Briefing Docs: Gymnotic Delivery Revealed

After Alnylam’s CEO John Maraganore famously failed to deliver on his Big Pharma platform partnership promise in 2009 (aka ‘SRTL’), and ISIS Pharmaceuticals announced a preferred partnership with GSK in March 2010, it seemed that Dr. Maraganore salesmanship was outdone by one of the best biotech salesmen ever: Dr. Stanley Crooke. 

In addition to likely having deprived Alnylam of a deal, what is this salesmanship claim based on?  It is the fact that not all that long ago, antisense technology had fallen out of favor in Big Pharma.  This was because it required large amounts of oligonucleotides and, partly as a result of that, was associated with non-specific immunostimulatory effects.  And this is where the more elegant RNAi Therapeutics was supposed to take over. Despite this poor track record, the technology managed a comeback, ironically partly on the back of RNAi Therapeutics which was seen stumbling at the ‘delivery’ stage.  In some cases out of the frustration of making RNAi delivery work, the 'naked' delivery approach of antisense would be embraced by the likes of GSK, BiogenIdec, Pfizer, Genzyme, and VCs making the application of Santaris' LNA technology the basis of start-up biotechs targeting emerging classes of non-coding RNAs such as microRNAs or long non-coding RNAs.

The reason, however, why I had always favored RNAi as the more ‘natural’ gene silencing modality over antisense is that antisense, to this day, relies on achieving extremely high steady-state tissue concentrations (>100-300mg/kg of liver and kidney tissue e.g., see ISIS TTR patent application US2011/0294868) either by using reactive/sticky chemistries such as the phosphorotioate backbone or more gentle backbones such as the morpholino chemistry that, however, require even larger amounts of oligonucleotides.  

Mipomersen Briefing Docs Surprises

Unfortunately, the publication last week of the FDA Briefing Document related to the new drug application (NDA) for Mipomersen (tradename: KYNAMRO) by ISIS and partner Genzyme, should serve as a reminder that giving an old technology new names such as ‘gymnotic’ or saying that the algorithms now allow for the discovery of much more potent antisense sequences without any new chemistry (despite the fact that narrow tiling had always been practiced) are telltale signs that the biggest change that has occurred was in the marketing department.

The selective disclosure strategy by ISIS Pharmaceuticals on mipomersen has certainly contributed to the impression that antisense technology had indeed made important progress.  Dr. Crooke claimed in a recent interview on Mad Money with Jim Cramer that ISIS had been very transparent with mipomersen.  I could agree that there has been a lot of scientific data released on mipomersen by ISIS and their clinical investigators, but in hindsight this was arguably not for the claimed sake of transparency, but for providing the impression that mipomersen was a safer drug than it probably is. 

Here are a few examples of the ‘surprises’ that were revealed in the Briefing Docs:  

1) Liver Fat: More and Persistent

While the significance of the non-alcohol-related accumulation of fats in the liver in predisposing towards liver fibrosis is still hotly debated among experts, ISIS tried to dispel the still understandable fears about ApoB-knockdown-related fat accumulation in the liver by claiming a) that it was only modest, and b) that feedback mechanisms allowed the liver to adjust and that liver fat would decrease over time.  Supporting the impression that ISIS was concerned about transparency was the publication (Visser et al. 2010) of liver fat data from a 13-week study in a small, 21-patient study population which only showed a trend of such accumulation (median control-adjusted increase from baseline to day 99 of < 1%).  

Compare that to the actual results (as revealed in the Briefing Docs): In study ISIS301012-CS7 and CS12, 61.8% (63/102) in the mipomersen group had a > 5% increase in hepatic fat content.  This already shows that the median increase is definitely higher than 5%, not < 1%. Moreover, in study CS6, 16% had average liver fat of > 20% measured at least once.  30-40% liver fat contents also were not uncommon.  Such values were practically not observed at baseline or in control subjects despite ISIS’ repeated claims that such values are of no concern since NAFLD is so common anyway.

Furthermore, the data such as capture in Figure 10 of the Briefing Document show that there is no basis for claiming that the liver adapts and liver fat contents come down over time.  They only come down after you discontinue treatment.

Suggesting that the fat accumulations could be of clinical significance, a fibrosis-related biomarker score (ELF) was elevated in the mipomersen population.

To be clear, at least the liver fat accumulations (not sure about  the ELF score) is very likely a target-related side effect and cannot be attributed to the phosphorothioate antisense platform per se, but it is an important issue for mipomersen and raises suspicions about the way ISIS Pharmaceuticals has dealt publicly with platform-related safety issues.

2) Partial Clinical Hold: Vasculitis

Were you surprised by the fact that, according to the Briefing Doc, the FDA issued a Partial Clinical Hold for the non-severe LDL patient population in January 2008?  I certainly was taken by surprise and would also think that this should have been disclosed in an SEC filing as a Partial Clinical Hold for your lead program should be considered a material event.  Not only can shareholders be pitied, but also Genzyme which less than a month before invested more than $300M in mipomersen! I cannot imagine that they would have closed such a deal had it known that there was data that would trigger such a Hold (I do not exclude that the data was there under their nose, but in that case it was probably either buried under a pile of paper, or Genzyme just did not do their job well).

Apparently (à Briefing Doc), the ‘final interim report’ from the 52-week monkey tox study that was submitted in June 2007 showed vasculitis (inflammation of the vasculature which can impair blood supply and therefore organ function, possibly leading to organ failure) at clinically relevant doses of 3mg/kg and up, mainly, but not exclusively, in the gastrointestinal tract. Although the Hold was lifted after the actual clinical safety data did not indicate vasculitis to be an issue, subsequent skin biopsies around the injection sites showed an accumulation of inflammatory cells around the vasculature.

Although similar to the liver fat issue, the clinical significance and degree of systemic vasculitis due to phosphorothioate oligonucleotides remains unknown in Man, the potential for sustained inflammatory processes in the entire vasculature is a serious safety concern that needs to be studied in more detail before the technology can be applied to less than the most severe, 'orphan' patient populations- if at all.

3) Liver Enzyme Increases

That mipomersen increased liver enzymes in the serum, a measure of liver tox, has been widely known.  But ISIS used to claim that in a given individual these were one-off findings that could just as well have come after a night of heavy drinking.  By contrast, it seemed to me new news that there were cases where elevated (>3x ULN) liver enzymes were found on at least two subsequent investigations.  In fact, the FDA feels that the liver enzyme increases warrant them be tightly monitored as part of a REMS program should mipomersen be approved in the homoFH population.

4) Injection-Site Reactions and Discontinuations

One of the major marketing tools, and in fact actual draws, of ‘gymnotic delivery’ has been that it can be subcutaneously administered, while the leading RNAi Therapeutics delivery technologies, foremost Tekmira’s SNALP, still rely on intravenous infusion (for most applications at least).  Turns out that injection site reactions ranging from pain, redness, swelling, to skin discolorations and haematomas were a major factor for study drug discontinuations, especially in the open-label phases of the studies which should be indicative of what will happen in a real-world setting.  Overall, 61% of HoFH patients (23 of 38) discontinued study drug during the open-label phase, and 77 of 141 discontinued in the pooled Phase 3 population (including non-hoFH subjects).  Other adverse events attributable to the inflammatory potential of the technology, such as flu-like symptoms, contributed to the high discontinuation rate.

A curious, hitherto undisclosed finding, were injection site recall reactions with mipomersen (but not control).  This refers to an inflammatory reaction at an old injection site when the drug was administered at a new site.  The molecular basis and importance for such recall reactions seem unknown.  In fact, my literature searches suggest that they have almost exclusively been described for the TNF-alpha blocking antibody Enbrel.  However, to me, they once again indicate the systemic inflammatory potential of phosphorothioate antisense chemistry, likely involving adaptive immunity.  Accordingly, 30/50 patients (60%) in the CS5 and 6 studies exhibited mipomersen antibodies compared to none in the control population- another ‘surprise’ that should be applicable to the phosphorothioate platform in general.

5) Carcinogenicity

While there were ~2-3x more cases of tumors in the mipomersen-treated population compared to control, I agree with the reviewers that they are not the basis for concluding that mipomersen increases the risk of developing cancer.  This is because of the nature of the reported cases and the relatively short treatment duration after which many of them occurred.

However, what would worry me more is that in the preclinical carcinogenicity studies in rodents (mice and rats), mipomersen clearly increased cancer risk- despite ISIS’ claim that they have been ‘clean’.  Among the findings were hepatocellular adenoma and subcutaneous tumors.  Having a report by your CRO stating the findings are likely species-specific issues and not applicable to humans, does not mean that the findings are not material in a regulatory sense, especially as they were made during a period at the FDA when the agency was much more risk-averse.  I expect the clinical and preclinical findings to cause carcinogenicity to be included in a REMS program.    

6) Kidney Damage

The liver, spleen, and kidneys are the major sites of phosphorothioate accumulations.  Consequently, it is not surprising that kidney damage has been an important safety concern of such antisense technology.  Although the clinical safety findings did not demonstrate that mipomersen impaired kidney function, the study drug did increase the amount of protein found in the urine (proteinuria).  This finding supports that kidney function remains a safety concern and such monitoring should be included in a REMS.

In summary, the safety findings were not all that surprising given what we know about phosphorothioate safety in animals and study discontinuations by competitor Santaris, very likely due to safety.  However, if you indeed have taken the reassurances by ISIS CEO Dr. Crooke literally, you may be justified in feeling misled.  Even on that count, however, I am not all that surprised, and his salesmanship may indeed be one of ISIS' great assets.  

Disclosure: The author is 'long' the stock ($ISIS).

Clarification: The term 'gymnotic delivery' in the linked paper refers to a specific protocol for optimizing gene knockdown in tissue culture in the absence of assisted delivery.  However, the term has been more broadly used by Santaris to refer to unassisted delivery in general (including clinical applications), obviously in an effort to express that in contrast to phosphorothioate oligonucleotides, RNAi needed assisted delivery.  I found it suspect, however, that they used a new term to refer to an over decade-old industry practice.  In a further clarification, I have not seen ISIS Pharmaceuticals adopt the term either. But clearly, not using (in most cases) assisted delivery has been a greatly exploited marketing tool of the antisense industry built on the back of RNAi Therapeutics.