Preclinical Data Do Not Instill Confidence in iCo Therapeutics Trial Outcome

Oligonucleotide Therapeutics investors keen on playing binary events such as clinical trial read-outs should have the phase II iDEAL study of RNaseH antisense compound iCo-007 in diabetic macular edema (DME) on their radars.  This 2^nd generation compound targeting c-Raf had been discovered by ISIS Pharmaceuticals and was then licensed to iCo Therapeutics.

The outcome of that trial will also serve as a proxy for the utility of RNaseH antisense for ocular indications, although higher-affinity chemistries might be expected to achieve superior results. In January, Roche gave their nod of approval to this approach in signing a partnership with Danish LNA company Santaris.

In trying to predict the outcome of the iDEAL study, I focused on the dose regimen in the trial in light of the preclinical data.  Once again, given that Oligonucleotide Therapeutics enjoy virtually unlimited target space, I typically give the target the benefit of the doubt.  To be successful, however, you need both a good target and sufficient targeting.

Preclinical studies

Similar to OGX-011 discussed recently, the history of iCo-007 goes back to the time when ISIS transitioned to second-generation 2’MOE gapmer chemistry.  Following failure in clinical studies for cancer indications with first-generation chemistry, the antisense compound was converted into 2^nd generation chemistry with the same sequence and re-positioned for ocular indications.

In a seminal study by Danis et al. from 2003, a pig-specific anti-c-Raf sequence (ISIS 107189) was tested in pigs for c-Raf knockdown and a retinal vein occlusion model of neovascularization (c-Raf is thought to play a role in the neovascularization aspect of DME, e.g. downstream of VEGF signaling).

Disappointingly, regardless of whether 34ug or 180ug of antisense were injected into the pig eyes, there mere 40% target gene knockdowns were observed.

Of course, it would have been more insightful if the human sequence, i.e. iCo-007, had been evaluated in an animal efficacy model. Although I looked in vain for such data, a 2009 analyst note by Versant noted that in a mouse model this had indeed been done.  Unfortunately, the knockdown, again, was merely 40%:

iCo-007 preclinical data was encouraging and was generated by partner Isis prior

to out-licensing the drug to iCo. In mice experiencing new ocular blood vessel

growth, Isis administered iCo-007 (then called ISIS 13650) to one eye and saline

to the other in two separate injections, one administered seven days before

blood vessel growth commenced, and then at onset of growth. Three days after

antisense dosing, c-raf kinase mRNA levels in the eye were reduced by 40% but

were unaffected in the saline-treated eye, and after seven days, c-raf kinase

mRNA levels were still reduced from baseline by 30%.

Question marks over phase II study design

The phase II DME study is testing 350ug and 700ug iCo-007 either as monotherapy or in combination with anti-VEGF antibody Lucentis or laser photocoagulation.   

Although 350ug and 700ug in humans appear to be higher relative doses than 34ug and 180ug in pigs, given that there was no dose response in pigs I am not sure why we should expect much increased c-Raf knockdown potencies in the phase II study.

Another cause for concern is the fact that the 208 subject study appears to lack a control group according to the clinicaltrials.gov entry.  So be prepared for positive headlines and fanciful comparisons to historical controls, similar to the ‘positive’ results from the uncontrolled phase I study.  

Altogether, given the weak preclinical data and the questionable phase II trial design, I remain highly skeptical that truly medically positive results, barring non-specific anti-angiogenic effects due to immunostimulation of the antisense molecule, will be forthcoming.   

Imetelstat Off-Target Mechanism Might Be Its Therapeutic Mechanism of Action

This morning’s biotech news featured a clinical hold on Geron’s imetelstat, a 13-mer lipid-conjugated N3’-P5’ thio-phosphoroamidate oligonucleotide telomerase inhibitor.  Since maintaining chromosome ends by telomerase is important for repeated cell replication, it is not surprising that the company is aiming this compound at proliferative disorders such as cancers and essential thrombocythemia (ET).  The latter is a condition in which there are too many platelets.

The clinical hold imposed by the FDA relates to apparent liver toxicity.  Liver toxicity as evidenced by elevated liver enzymes has been a known side effect of imetelstat with 90% of subjects in the phase II trial of imetelstat exhibiting low-grade elevated liver enzymes.  What is more, about 30% of the subjects had concurrent increases in bilirubin.  Taken together this strongly smells like cases of Hy’s Law, the nuclear liver tox bomb of drug development.

No wonder the alarm bells at the FDA are ringing.  According to the press release by Geron this morning, the agency is concerned about the reversibility of these elevations.  If not reversible, chronic liver disease, if not failure might ensue.  

I am not too surprised by these developments.  What surprises me is that the stock market apparently has not picked up on the concurrent liver enzyme/bilirubin increases since these had already been known.

The reason why I am not surprised is that phosphorothioate oligonucleotides can be expected to result in liver injury at the ~10mg/kg very high dosages given in the ET trial (dosage similar to the 640mg phase III prostate cancer trial of OncoGeneX discussed yesterday).  Note that ISIS Pharmaceuticals generally settles with 200mg, at most 300mg of systemically administered phosphorothioates and Santaris had to terminate two candidates targeting genes expressed in the liver due to liver tox.

It should be added here that the Geron, ISIS, and Santaris chemistries are slightly different, but share the phosphorothioate modification which in my opinion is what is causing these toxicities.

Given that the half-lives of phosphorothioate oligonucleotides in the liver are about 1 month, one would expect the low-grade liver enzyme elevations to go away with time and they might not be a show-stopper for non-chronic applications of imetelstat.   

That’s the somewhat good news.

The bad news: applying a phosphorothioate oligonucleotide-based telomerase inhibitor for the treatment of cancer and increases in platelet counts sounds like a bad joke.  Thrombocytopenia (decreases in platelet counts) and anti-proliferative immunostimulation are well known side effects of large doses of phosphorothioate oligonucleotides.  In light of that, claiming that imetelstat works via telomerase inhibition seems a bit optimistic to put it kindly.

Don’t Bank on OncoGeneX Phase III Data

As much as RNAi Therapeutics is contributing to and benefiting from the fact that RNA Therapeutics as a whole is rapidly materializing as the 3^rd major drug discovery platform, this relationship could also backfire to some extent.  I am particularly thinking here about the upcoming OncoGeneX 2nd gen RNaseH-antisense (ASO) phase III results in prostate cancer (partnered with TEVA).

Obviously, a successful outcome of the SYNERGY trial which is analyzing the ASO compound custirsen (aka OGX-011) in combination with chemotherapy would be a big win for the field and represent the first more significant commercial success.  In February, the company announced that the pre-specified number of ‘events’ (i.e. deaths) had been reached in this ~1000-patient study, and depending on the robustness of the results (good or bad) we could be hearing any day now of the top-line results.

Reasons to be skeptical

In evaluating the prospects of an Oligonucleotide Therapeutics compound, I first ask whether the delivery has been firmly established and how good the target is. 

Usually, since Oligonucleotide Therapeutics enjoy the benefit of being able to pursue virtually any target, target risk should be minimal.

Unfortunately, in this case, there are considerable risks in both the delivery and the target.

Custirsen targets clusterin expression.  Clusterin has chaperone activity and is thought to protect cancer cells from stress, including stress arising from cancer therapies such as radiation and chemo.  Accordingly, limiting clusterin activity by knocking it down would be predicted to improve the effectiveness of conventional cancer therapies.

The most important validating preclinical study in that regard comes from a publication by Zellweger and colleagues…from the year 2001 (which got me wondering about the patent clock).  While some of the predicted effects were seen such as increased sensitivity to chemotherapy agents, the effect sizes were rather modest and their correlation with ASO potencies poor. 

In general, in reviewing even more recent ASO chemistries such as cET2.5 for cancer (e.g. STAT3 by ISIS/AstraZeneca), I was struck by the poor correlation often seen between knockdown potency and therapeutic effect (e.g. much more potent anti-cancer effects with an ASO that had 20% knockdown activity compared to one that had 60% knockdown activity) and sometimes also between mRNA and protein levels (mRNA e.g. reduced by 50%, protein essentially disappearing; patent application WO2012/135736).

This raises the prospect that some of the apparent anti-cancer activities were not due to on-target efficacy, but due to some unintended side effects such as immune stimulation.

Promising phase I PK-PD data at very high dose of 640mg

Consistent with the concern of unintended/undesirable immune stimulation, the apparent increase in overall survival observed in the phase II prostate cancer study of custirsen was seen at the dose of 640mg, a dose at the very upper limit of phosphorothioate chemistry (Chi et al. 2010).  This study in ~80 prostate cancer patients provided the rationale for investing in the ongoing phase III trial.

Unsurprisingly, immunostimulatory side effects in the form of chills and fevers were seen in the majority of subjects taking custirsen in combination with chemo, but not those on chemo alone (plus the liver and kidney tox you’d expect from such a high PS-ASO dose).  Moreover, the apparent clusterin knockdown, as assessed on the basis of the protein found in serum (not in cancer tissue directly), was a very modest ~30%.

This unsatisfactory degree of knockdown stands in stark contrast to a promising phase I PK-PD study which showed >90% knockdowns in prostate/tumor biopsies (Chi et al. 2005).  This was accompanied by mid- to high single-digit microgram of oligonucleotide concentrations per gram tumor tissue.  Such concentrations do not necessarily predict (normal) mRNA knockdown efficacy, but do not exclude such activity either. 

One explanation for the apparent discrepancy could be that the phosphorothioate oligo preferentially localizes to solid tumors and that solid tumors are not a significant source of serum clusterin. Possible, but...  

Under the radar for a reason

A sub $200M market cap for a company with imminent pivotal phase III data in prostate cancer and low trading volumes indicate that the market has not been paying all that much attention to Oncogenex. 

In this case, I would like to think for a good reason, and while I do not exclude ‘positive’ results (at the cost of considerable side effects and an uncertain mechanism of action), I hope that the public won’t read too much into them regarding the ongoing RNA Therapeutics revolution.  Every mechanism and chemistry has its place. PS-ASOs knockdown of (cytoplasmic) mRNA for cancer is not it. 

For cancer, RNAi and microRNA Therapeutics have the edge.

An RNAi Therapeutic Could Be Last Man Standing in PCSK9

PCSK9 has arguably become biotech’s hottest drug target as it promises to unlock much of the remaining unmet need in lowering ‘bad’ LDL cholesterol.  While monoclonal antibodies have been the predominant and most advanced modality to address PCSK9 (e.g  alirocumab Regeneron/Sanofi-Aventis and AMG145 by Amgen), RNAi Therapeutics have made tremendous, yet widely underappreciated progress in this area with predicted potencies equivalent to if not superior to monoclonal antibodies.

Alnylam’s ALN-PCS leads in this effort (preclinical 67% LDLc lowering without statins) and should have entered clinical development by the beginning of 2015.  It is partnered with The Medicines Company.

The news today that the FDA has become concerned about neurocognitive adverse events in the ‘PCSK9 class’ could mean that RNAi Therapeutics, despite their development delays, will carve out a nice junk of the PCSK9 market which is widely estimated to be a multi-billion dollar market in terms of annual sales.

But if it’s a ‘PCSK9 class’ issue, so how could this be positive for RNAi Therapeutics?  Wouldn’t this concern also apply to ALN-PCS?

Not necessarily.  Since by far the most clinical experience with PCSK9 has been obtained with monoclonal antibodies, the FDA may be inappropriately lumping all PCSK9 inhibitor agents in one bucket confusing a short-coming of a therapeutic target with a short-coming of monoclonal antibodies.

The reason why I have a good feeling that this turn of events could be very positive for RNAi Therapeutics is that RNAi is a genetic tool that reduces PCSK9 expression before any protein is made.  The approach therefore resembles populations that under-express PCSK9 for genetic reasons and which enjoy improved cardiovascular health compared to the general population without any apparent negative consequences of their PCSK9 deficiencies.  It is for this very observation that PCSK9 has become such a sought-after target and RNAi, a gene knockdown approach, best mimics human genetics.

Monoclonal antibodies don’t.  One mechanism by which antibody-specific toxicity might arise from is the fact that they form complexes with their target.  Such complexes could elicit adverse immune reactions in organs where they form and accumulate.  While I am not familiar with the binding sites of the particular Regeneron and Amgen antibodies to PCSK9, it might also be possible that the immune complexes form directly on neuronal cells expressing LDL-receptor-like proteins (note: PCSK9 binds the LDL-receptor).

Beyond PCSK9, today’s events emphasize yet another hitherto under-appreciated value proposition of RNAi Therapeutics, namely providing differentiation value when going head-to-head with the small molecules and monoclonal drug establishments.

With Success, Pure-Play RNAi Therapeutics No More

As RNAi Therapeutics are charging forward towards first marketing approvals, it is time to consider how this will change the face of some of the pure-play companies involved.  Alnylam is set to become the dominant player in TTR amyloidosis, Arrowhead Research and Tekmira are vying to become just that in Hepatitis B, and Tekmira is increasingly isolating their biodefense division involving the formation of a filovirus alliance with non-RNAi modalities (prophylactic and therapeutic vaccines) as revealed yesterday.

Under normal circumstances, getting drugs approved that revolutionize the treatment of particular diseases provides great incentives to protect the resultant franchises.  Witness BiogenIdec .  BiogenIdec started out as a molecular biology pioneer and serendipitously established itself as the major player in multiple sclerosis, and is now considering all therapeutic modalities, including RNA Therapeutics, to build on that strength and further expand into neurology.  

Although it breaks my scientific heart, I do not expect that to be any different for RNAi Therapeutics companies.  HBV is a great example since an immune de-repressing RNAi Therapeutic alone is unlikely to facilitate the 100% cure rates now seen with combinatorial HCV treatments.  Instead, the addition of a direct immune-activating agent such as a TLR agonist should enhance cure rates and shorten treatment durations.  For an acute fatal infection such as Ebola, you also do not want to limit yourself to one treatment, but recruit multiple mechanisms to save an infected person, or start vaccinating populations around areas of viral outbreaks.

To extract maximal value from such disease indications, it is often best to combine the various options under one roof.  It may be partly this fear of moving away from your roots that is responsible for ISIS Pharmaceuticals giving up marketing rights early on for a few cents on the dollar worth (SMN, myotonic dystrophy, TTR).

Having said that, just like BiogenIdec with recombinant proteins and monoclonal antibodies, most of these RNAi Therapeutics companies will continue to focus their internal R&D efforts on RNAi Therapeutics.  In special cases such as Ebola and other biodefense opportunities, a spin-out may create most shareholder value without damaging the platform technology value of the parent company.

Nothing of what I just said is revolutionary.  It is, however, worth keeping in mind as the first major RNA(i) Therapeutics wave will make landfall in 2-3 years. 

The Ebola Clinical Trial a Real ‘Haertetest’ for SNALP LNP Delivery

In German, a Haertetest is a test that challenges the performance of something under the most exacting conditions.  

It is therefore an apt description for the Ebola clinical trial now underway which is testing the safety and tolerability of SNALP LNP delivery following multiple consecutive daily administrations.  This compares to once-weekly and once-a-monthly regimens tested so far.  If SNALP LNP comes out positive from this, it would provide great comfort around the safety and tolerability of the industry’s most potent delivery technology for gene knockdown in the liver, in addition to paving the way towards more biodefense applications.

Teething problems with 1^st gen SNALP LNP

In January 2010, Tekmira reported clinical results from the phase I study of TKM-ApoB, the first clinical candidate employing SNALP LNP technology.  This trial was terminated prematurely due to immune stimulations observed at 0.6mg/kg, a dose at which only minimal target knockdown was observed at that.

This event triggered a healthy amount of soul searching and meant that subsequent SNALP LNP trials had to employ transient immune suppression.  A number of important practical lessons were learned from this.  

Firstly, it showed that the preclinical immune stimulation tests had been inadequate and this prompted the development and adoption of a more predictive test to better screen out candidates that are immunostimulatory before they enter humans.  Secondly, the phase I results were consistent with the notion that the DLinDMA lipid itself (and not just the RNAi trigger), the basis for the 1^st gen SNALP formulations, was immunostimulatory and that new lipids were needed.  And lastly, it highlighted the need to develop more potent SNALP formulations, including new lipids.

Research rises to the challenge

Three years following these events, all three challenges have been successfully addressed.  Today, the potency of SNALP LNP has increased ~30 fold from 1^st gen to 3^rd gen formulations (based on measuring gene knockdown in the liver), with the 2^nd gen MC3-based SNALP generation already with a confirmed 10-fold improvement in clinical potency based on phase I and II results of ALN-TTR02.

Tekmira not shy to ask the tough questions

Despite the progress, detractors of the technology like to point out that (transient) immune suppressions has still been employed in subsequent clinical trials.  With the recent initiation of the phase I study of TKM-Ebola by Tekmira, this has changed however.  Not only that, transient immune suppression has been dropped even as administration intensity is ratcheted up to 5-7 daily administrations. This compares to once-a-month and once-a-week dosing schedules anticipated for SNALP LNP applications involving gene knockdown in the liver and solid cancers, respectively.

The reason why TKM-Ebola necessitates daily administrations is due to the rapid progression of the otherwise fatal Ebola virus.  Therefore, in order to get it approved under the Animal Rule, Tekmira needs to demonstrate the safety and tolerability of SNALP LNP in human volunteers under conditions where they are able to rescue monkeys from succumbing to otherwise fatal infections.  

Moreover, while 0.2mg/kg already provided good protection in the monkey studies (2/3 survival), it was the 0.5mg/kg dose that complete protections were achieved.  0.5mg/kg also happens to be the ‘magic’ dose for SNALP LNPs following which idiosyncratic immune stimulations and other side effects may be expected.

The reason why I am pretty sure that TKM-EBOLA, despite the daily administrations and up to 0.5mg/kg dose, should not involve immune suppression is because it would be incompatible in a viral infection.  There is also no evidence that such pre-treatments are planned in the clinicaltrials.gov entry nor when Tekmira reported the monkey results.  The same considerations apply to the HBV candidate by Tekmira for which an IND is expected by year-end.

Should Tekmira come out of this trial unscathed, it would further de-risk the entire SNALP LNP platform and solidify its position as the most potent oligonucleotide therapeutics technology for gene knockdown in the liver, for hemorrhagic fever applications, and possibly oncology.

Where some see risk, others see opportunity.    

Oligonucleotide Therapeutics Companies Crowding into GalNAc-Conjugation

Following Alnylam’s achievement of making subcutaneous administration work for RNAi gene knockdown in the liver by conjugating RNAi triggers to the GalNAc sugar, antisense companies have started to copy the approach.  As RNAi Therapeutics have made great progress in targeted delivery, antisense companies are starting to realize that in order to stay competitive and improve the safety of systemic uses of antisense, they need to get away from the notion of 'naked'/unformulated delivery that relies on swamping the body with phosphorothioated oligonucleotides.

This has become apparent at the AsiaTIDES meeting here in Tokyo where both Santaris and ISIS Pharmaceuticals disclosed their great interests in GalNAc conjugation.

In collaboration with Axolabs (the part of Roche RNAi Therapeutics that was not acquired by Arrowhead Research and that had familiarity with GalNAcs), Santaris presented phosphorothioate-LNA Factor VII (liver) knockdown data showing 80% gene knockdowns in mice following a single dose of 0.1mg/kg.  Even more potent knockdowns were seen at 0.25mg/kg and above. 

In another presentation by ISIS Pharmaceuticals on their cardiovascular franchise, the company noted that they will follow up on their initial Apo(A) (‘little A’) candidate with a GalNAc-conjugated version.  With this, they expect to increase potency by up to 10-fold, thus allowing for 10-30mg (~0.15-0.45mg/kg) dosing.

This illustrates the utility of the GalNAc receptor (ASGPR) and how the competing RNAi and antisense technologies are fertilizing each other.

Regulus Therapeutics, of course, is the first antisense (anti-miR) company of sorts that has adopted GalNAc conjugation for their liver-directed programs, most notably anti-miR122 for HCV that is about to enter clinical development.  Regulus obviously has a license to GalNAc from Alnylam.  Whether this also applies to ISIS Pharmaceuticals, remains to be seen.  The word on the lab floors is that GalNAcs per se are not patent-protected, so ISIS may use an entirely different linker strategy to Alnylam just as Arrowhead Research does with its GalNAc-targeted DPCs.

What it means for Tekmira and Marina Biotech

Since some of you are following my investment strategy, here is what I think this means for the liver-directed efforts by Tekmira and Marina Biotech (both of which I own). 

Marina Biotech could obviously follow the same path as ISIS Pharmaceuticals and Santaris in adopting GalNAc conjugation with its CRN technology.  As such, there should be no change in the competitive value of CRN compared to Santaris and ISIS antisense.  It could also attach GalNAcs to their liposome-based SMARTICLES for which delivery to the liver remains to be demonstrated.  To do all this, however, Marina Biotech needs to grow and establish in-house R&D.  

For Tekmira, it means that RNaseH antisense are getting close in potency for gene knockdown compared to the 2^nd gen MC3 SNALP LNPs (85-90% TTR gene knockdown in humans at 0.3mg/kg).  With the 3-fold more potent 3^rd generation SNALP LNPs which should enter the clinic this or early next year (TKM-ALDH2, TKM-HBV), Tekmira should stay well ahead of its competition with the most potent gene knockdown technology for the liver.  This means more addressable diseases and in most cases higher treatment effects as well.   And if it incorporated GalNAc-conjugated lipids into their liposomes, too, maybe that would extend that lead even further.

If gene knockdown in the liver is not a great example for why you need a competitive free market economy, then I don’t know what is. And, of course, there is no better example of why you need a lab ;).