ISIS Pharmaceutical Reveals Dynamic PolyConjugate Efforts

When Arrowhead Research made a splash 1 ½ years ago at the Boston OTS meeting in late October 2012 with impressive subQ DPC knockdown efficacy in monkeys, there was a large number of meeting participants crowding the speaker after his presentation.  Although I sat a few rows away, the most eager questioner was an employee from ISIS Pharmaceuticals who managed to beat the Merck representative to the podium!

There were times when I believed that the interest of large companies in the technology of small companies boded well for partnership potential.  It has become clear to me, however, that while this may be true to some extent, the first instinct by the large guys is to get close to the innovators so that they can appropriate as much of the technology as possible without paying a dime.

While I have long shelved Merck into that category and DPCs will now be a research priority at Alnylam following their acquisition of Sirna/Merck, a now published patent application by ISIS Pharmaceuticals on melittin-based, GalNAc-targeted single-stranded antisense delivery (ssRNAi and conventional ASO) demonstrates that ISIS ticks no different.

With a priority date of WO 2014/089146 A1 of December 4, 2012, this poor employee and collaborating patent agent had to stitch together a patent application in just a month after returning home from the conference.  Unsurprisingly, the patent application lacks any actual experimental data to support the 'ínvention'.

I don’t want to be too cynical about this as this is how the industry and the patent system function.  It is an important lesson for small companies though that they should not get overly excited about interest they generate for their technology and they should only engage in deeper relationships once they have established that the larger party is not just trying to steal their technology by pretending to collaborate with them.

It is this unwillingness to share and attendant mistrust that in my experience impedes much business development that makes great sense on paper.  This ultimately delays timelines, leads to litigation, and makes the pie smaller for everybody.

And for the future of gene knockdown in the liver…if ~5mg per week may be possible for GalNAc-targeted gen 2.5 RNaseH ASOs, one can only wonder what a melittin-type escape mechanism would add to its potency, and safety. 

Comment on Seeking Alpha bear article on ISIS Pharmaceuticals

Last night, a detailed bear article on ISIS came out that shook the market in after-hours trading.  In summary, the thesis rests on the poor safety profile, especially immunogenicity of gen 2.0-based KYNAMRO (2’MOE gapmer) and that this is likely to translate to all other gen 2.0 drugs based on shared chemical composition and the (in my opinion idiotic) claim by the CEO of ISIS that KYNAMRO is a success and is representative for gen 2.0.  Talk about shooting yourself in the foot.

While I agree that the KYNAMRO data have raised legitimate questions around the safety of gen 2.0 phosphorothioate oligonucleotides, the bear article conveniently ignores the abundant clinical data that have emerged since for a number of other antisense drugs based on gen 2.0 chemistry.  These support the claim by ISIS that through improved screening, they are now able to better weed out the sequences that will likely prove immunogenic in the clinic. 

Importantly, in oligonucleotide therapeutics in general, while nucleic acid chemistry has a great influence on whether a molecule is immunogenic, it is the exact sequence composition that ultimately decides whether this is actually the case.

This is illustrated by the recent phase II clinical results for ApoCIII (no discontinuations noted), Factor XI, and GCGR.  Not only were robust gene knockdowns achieved, in sharp contrast to KYNAMRO, the company claimed that they were no flu-like symptoms, chills and other symptoms indicative of the immunostimulatory potential of oligonucleotides.  By contrast, about 1/3 in the KYNAMRO studies exhibited such events, with even higher numbers in the open-label extension phase.

Granted, given that these were 13-week studies, it is impossible to disprove the thesis that things are bound to get worse over the long-run.  However, no flu-like symptoms versus 1/3 of patients exhibiting flu-like symptoms in studies of comparable duration is a dramatic difference and allows one to extrapolate that the safety of the follow-on drugs will similarly be greatly superior to KYNAMRO over time.

While I had been tempted to let the Seeking Alpha article by Dr. Anonymous pass as raising legitimate concerns, the blatant failure to mention the more recent experience with gen 2.0 and the after-hours action last night makes this article suspect and outright useless given the lack of new information or insight.

ApoCIII Genetic Studies Suggest Oligonucleotide Therapeutics to Generate Next Statin

With ISIS Pharmaceuticals having presented strong phase II data showing that ISIS-ApoCIIIRx leads to ~65% reductions in serum triglycerides, it is now in the lead position to exploit the next statin-like opportunity in cardiovascular disease: ApoCIII inhibition.

Large population studies published yesterday in the New England Journal of Medicine (here and here) have now confirmed that not only very high triglyceride levels are problematic for health (e.g. pancreatitis), but that even normal triglyceride levels may not be optimal.  This is because in those people that harbor a mutation that reduces the activity of the ApoCIII gene by just 50% have roughly half the risk of those with ‘normal genes’ of suffering from a cardiovascular event.  As such, (ApoCIII-related) triglyceride levels have become the second important independent risk factor for cardiovascular disease.

ApoCIII inhibitors like ISIS-ApoCIIIRx are particularly valuable for the industry to develop and commercialize for at least 2 reasons. 

For one, the new studies provide human proof-of-concept that triglyceride lowering via the ApoCIIII route should be beneficial and therefore dramatically lower the risk and cost of outcomes studies.  There may be other ways of lowering triglycerides, but from the experience in lipoproteins, it may not just be a matter of down-regulating triglycerides per se, but also the mechanism of achieving how that goal is achieved.  This could e.g. have critical impact on which forms of triglycerides are affected.

The other reason is that unlike the fish oils (20-30% lowering) and niacins (35-50% lowering with significant side effects) which have achieved only moderate triglyceride reductions, inhibiting the ‘undruggable’ ApoCIII with oligonucleotides has been shown by ISIS to have a much more profound impact on triglyceride levels with additional benefits such as ~40-50% increases in the ‘good’ HDL cholesterol.

As a result, ApoCIII inhibitors could in fact be the real ‘next statin’ for the industry, not the PCSK9 inhibitors which address the same risk factor as statins (LDL cholesterol) and may be largely relegated to the millions that do not tolerate statins.

ISIS Path Forward

Being ahead of the competition by 2 clinical phases for such an exciting target is highly valuable, but also represents challenges to ISIS Pharmaceuticals as the race to competing ApoCIII inhibitors is now officially on.  It is e.g. almost certain that Alnylam will develop an RNAi-based inhibitor, and I would expect the same from Tekmira.

Given the impressive lipid profile of ISIS-ApoCIIIRx and the well-established link between very high triglyceride levels and disease, the subcutaneous ISIS-ApoCIIIRx had been expected to readily obtain marketing approvals for those patient populations (e.g. >500mg/dL) without outcome studies.  Even fish-oil competitor Amarin with much less triglyceride lowering was able to do so.  

However, given the new population genetic evidence and the fact that ISIS-ApoCIIIRx is much more effective in lowering triglycerides, I am wondering whether the threshold for approval without outcomes studies will be somewhat lowered and the initial patient populations therefore widened, including in patients at high-risk of suffering (another) cardiovascular event, but that do not necessarily have very high triglyceride levels.

The next step after that would be outcomes studies in broader populations.  This could be done with ISIS-ApoCIIIRx.  A better idea, however, might be to develop a GalNAc-targeted gen2.5 ASO which would be orally available.  Such a differentiation strategy would also facilitate partnering strategies that the company would pursue and I am almost certain that ISIS-ApoCIIIRx2 will be ISIS' first oral antisense drug.  

And after ApoCIII...how about Apo 'little A' for which ISIS also has clinical studies under way already?

ISIS-GCGR on Path to Become Important Diabetes Drug

At this year’s ADA diabetes meeting, ISIS presented the full dataset for ISIS-GCGR for the treatment of advanced diabetes patients having failed oral options and who are on the way to taking insulin.  After having presented top-line data a few weeks earlier claiming more than 2 percentage point reductions (on top of metformin) in HbA1c, enthusiasm around the results was limited given concerns about liver enzyme elevations that were not much further characterized and ISIS' history of being parsimonious when it comes to sharing safety data. 

Data indicate that good risk:reward will be found

A placebo-adjusted 1 percentage reduction of HbA1c reduction, the gold standard for measuring glucose control, is generally considered to be quite meaningful.  HbA1c stands for glycated hemoglobin and essentially integrates blood sugar levels over time. 

In type II diabetes, blood sugar is elevated either due to insufficient insulin-mediated uptake of sugar (glucose) into liver, muscle, and adipose tissue and/or due to excessive release of glucose from the liver stores due to the action of glucagon.

Due to the time-dependent nature of hemoglobin glycosylation, it takes at least 2-3 months for HbA1c to reflect the full impact of an antidiabetic medication on blood glucose levels.  Because it takes another 1-1.5 month for a PS-based ASO like ISIS-GCGR to promote full target knockdown and in the case of GCGR work its way downstream biologically, it will be about 4-5 months before the full extent of HbA1c lowering will be seen with ISIS-GCGR.

This is borne out by the HbA1c curves (slide 37) which appeared to be still declining after the 3 months of dosing in the study. Importantly, coincident with plateauing fasting plasma glucose levels, a more proximate measure of the drug effect, liver enzyme elevations, markers of liver injury, plateaued after ~1.5 months. 

This not only further supports that the liver enzyme elevations, the main safety concern with ISIS-GCGR, were due to on-target pharmacology rather than a flaw of the ASO platform, it also suggests that HbA1c lowering on the order of 1.5 percentage points or greater can be achieved without triggering liver safety concerns.

At the 200mg dose where more than 2 percentage points reductions were reported, liver enzyme elevations as measured by ALT were 2.7 times the upper limit of normal (ULN).  Although the variability of these ALT elevations was remarkably small, consistent with on-target toxicity, a 2.7 ULN is a non-starter for a drug in this population when in the drug world instances of 3x ULN are often a show stopper.

At the 100mg dose things become much more tolerable with 1.6 times ALT elevations and little variability all the while 1.35 percentage reductions in HbA1c were achieved.  I therefore expect more than 1.5 percentage reductions in HbA1c following more prolonged dosing without a further increase in liver enzyme levels.  Additional dosing optimizations and potentially drug combinations will be the subject of future phase II studies.

Best-in-class profile

Merck and Eli Lilly have also been developing GCGR inhibitors, but using small molecules to inhibit the proteins instead of oligonucleotides to inhibit glucagon receptor synthesis.  Apparently, these drugs have not only seen liver enzyme elevations in humans, again consistent with the on-target toxicity hypothesis of the ALT elevations, but more importantly cardiovascular safety signals such as increases in blood pressure.  Whereas Merck seems to have given up on further development of MK-0893, clinicaltrials.gov shows that development is ongoing for LY2409021, including a safety-related study to characterize its effect on blood pressure.

In data presented at the 2011 ADA meeting, MK-0893 achieved HbA1c reductions of ‘0.6 to 1.5 percentage points’ in a 12-week study, with 0.8 and 0.5 drops observed in the metformin and placebo control groups, respectively.  As presented at the 2011 European diabetes meeting, an up to 1 percentage point reduction was achieved with the Lilly drug LY2409021 in a study of shorter duration.  I am not sure whether results from longer studies are known, but it is probably a good assumption that the reductions will fall in a range similar to the Merck drug.

In sum, the ISIS drug is at least as potent, if not more potent (à on top of metformin) as the small molecule competitors, but without the additional safety issues, particularly the blood pressure increases.  In addition, being an oligonucleotide, it should be much easier to combine them with other antidiabetic medications.  On the competitive downside, the small molecule inhibitors are orally available, whereas ISIS-GCGR is administered subcutaneously.

Given the central importance of glucagon signaling in glucose homeostasis, a GCGR inhibitor will have its place in the enormous and fractured diabetes market.

Alnylam Pulls Chair from Underneath Arrowhead Research

Machiavelli would have blushed. Feeling the heat from both RNAi and antisense competition for targets in the liver, Alnylam has made it clear in its recent McSwiggen patent-related press releases that it is determined to wipe out at least the RNAi competition through patent and contract strategy (see here and here).  

Whereas before it used to call out Tekmira/Marina/Arcturus by claiming that unlocked nucleic acid would fall under their patent estate, today it added Arrowhead Research to the club of companies that entered an RNAi trigger licensing contract with Alnylam, but then find that those contracts were not worth the paper written on: Merck, Tekmira, Arrowhead Research, and others.

In brief, the McSwiggen patents that Alnylam bought from Merck claims RNAi triggers that satisfy the following conditions:

1)      RNAi trigger with strands that have a length that fall within the classical range for siRNAs;

2)      At least 10 or more bases (often pyrimidines) of either or both strands have to be one of the following modifications: deoxy, 2’F, or 2’0-methyl.

Depending on the patents there are slight variations as to the strand lengths covered, additional requirements for phosphorothioate linkages etc.

Given the above limitations, it is strange that coincident with the acquisition of the McSwiggen patent estate, Alnylam repeatedly adds in the related press releases that it owns RNAi triggers, including those modified with acyclic nucleotide analogues (--> usiRNAs).  By contrast, it would appear that replacing any of those modifications with the very unlocked nucleic acid analogue would be a simple way of circumventing McSwiggen without any destabilizing effect on the RNAi trigger. 

Isn’t that odd?  A company in a position of strength would not do that.  But we all knew that Alnylam still wishes Tekmira to disappear from the face of the earth, so that’s not really new.

The real shocker may be to investors of Arrowhead Research (from Alnylam’s press release):

‘we intend to maximize the value of this newly issued IP solely through the advancement of ALN-HBV - our GalNAc-conjugated siRNA targeting the HBV genome for the treatment of HBV infection’

In other words, Alnylam says that the HBV target-specific license it gave to Arrowhead just 2 years ago does not include rights to the newly acquired McSwiggen patent estate.  And indeed, ARC520 by Arrowhead infringes the newly issued patent subject of today’s press release (US 8,618,277).  I will just use si-74 (one of the 2 siRNAs in ARC520) to illustrate why.

chol-siHBV-74 sense: chol-uAuCfuGfuAfgGfcAfuAfaAfuUfgGfuAf(invdT) 23 bases,

chol-siHBV-74 antisense: dTAfcCfaAfuUfuAfuGfcCfuAfcAfgdTsdT 21 bases.

I’ve highlighted the pyrimidine bases in red.  ‘f’ after a symbol stands for 2’-fluoro, small case is 2’-o-methyl, dT is deoxy-T.

1.       ‘277 claims an HBV-targeted RNAi trigger which meets the following requirements:

a)      Each strand 18-24 bases (applies to si-74);

b)      the sense strand comprises 10or more 2'-deoxy, 2'-0-methyl, 2'-deoxy-2'-fluoro, or universal base modified nucleotides (applies to si-74 where 21 nucleotides are accordingly modified in si-74 sense);

c)       the antisense strand comprises 10or more 2'-deoxy, 2'-0-methyl, 2'-deoxy-2'-fluoro, or universal base modified nucleotides (applies to si-74 where 21 nucleotides are accordingly modified in si-74 antisen);

d)      10 or more pyrimidines (i.e. U or C) of either strand modified by 2'-deoxy, 2'-0-methyl, 2'-deoxy-2'-fluoro nucleotides (applies to si-74 as each strand has at least 10 of those modifications).

Genius, by acquiring the Merck patents in January, Alnylam not only gave the withering Kreutzer-Limmer patents a new life given that Merck was in some cases the only party fighting it.  No, Alnylam thereby also pulled the chair from underneath Arrowhead given the reliance of that company on ARC520 for its valuation.  Of course, a crumbling valuation of Arrowhead could really help Alnylam as it tries to catch up and develop its own version of DPCs based on what Merck has been able to copy from Arrowhead.  Given that ‘277 will only expire in 2022/3, just waiting it out may not be an option for Arrowhead.

Sorry, Arrowhead, just because Alnylam really, really ‘dislike’ Tekmira and is calling you ‘a friend’ does not mean that you can trust them any more than Tekmira or that you are any more savvy in your drafting contracts.

Game-Changing Potency Improvement for 2-Molecule DPC by Arrowhead

At the TIDES meeting last month, Arrowhead Research presented advancements with its intravenous 2-molecule DPC delivery approach for gene silencing in the liver. By adding a protease-sensitive hydrophilic extension to the endosomolytic peptide (MLP), the company has improved both the safety of the delivery approach, and even more importantly so, the potency of 2-molecule DPC. 

The new formulation can be expected to mediate deep knockdown at the 2mg/kg endosomolytic peptide dosage at which Arrowhead Research stopped their dose-escalation for the HBV product candidate ARC520 which is based on the original 2-molecule DPC chemistry.  The reason why I have come to be on the fence regarding the future of ARC520, especially in a competitive environment, is that I consider 2mg/kg insufficient to mediate the multi-log HbsAg knockdowns that might be desirable.  If they had increased the dose to 3mg/kg, or better 4mg/kg, I would have a much more positive outlook on ARC520 (maybe this is still in the cards pending the outcome of the phase IIa result in Q3).

It is important to keep in mind that with DPC, it is the amount of endosomolytic escape agent that is rate-limiting for knockdown potency and dose-limiting for safety. The amount of (otherwise very safe) RNAi trigger by contrast is in vast excess, just to make sure that no potency is wasted. With first-generation DPC 2.0 (2-molecule), it was from around 2mg/kg for both Factor VII and HBV mRNAs that target knockdown started to take off. The fact that the same can be observed for various target genes supports the notion that as long as you have a fairly potent RNAi trigger, it is the amount of endosomolytic escape agent that determines that depth of the knockdown.

There was no reason for Arrowhead Research not to explore higher doses if they had no trepidations about the safety profile. Somewhat contrary to this notion, the company had claimed seeing no preclinical tox at up to 10mg/kg.

Nevertheless, the thought process behind the improved 2-molecule DPC version leads me to believe that safety has been a concern. This is because the reason for adding the hydrophilic extension to the endosomolytic backbone was to limit non-specific, potentially toxic interactions with membranes outside the endosomes. This minimization of non-specific membrane interactions was confirmed in a test tube membrane interaction assay.

In a second step, Arrowhead scientists sought to get rid of the hydrophilic extension in the endosome so that it would not inhibit membrane disruption there. For this, the hydrophilic extension was rendered susceptible to cleavage by proteases that are found only in endosomes. With this trick, they regained the potency of the original DPC 2.0 version, but with the improved safety feature.

Going one step further and truly making lemonade out of lemons, they replaced PEG as the hydrophilic extension with more of GalNAc targeting ligand: whereas DPC 2.0 exhibited 40% and >90% gene silencing at 1 and 3mg/kg endosomolytic agent, respectively, the improved version DPC 2.1 achieved a >90% silencing at 1mg/kg already.

Given the increase in potency, even if the safety profile was not enhanced with DPC2.1, a conservative assumption, a 2mg/kg dose should now facilitate very potent gene silencing in the liver. This bodes particularly well for the 2^nd development program that Arrowhead will disclose this month. As a follow-up to ARC520 in HBV...?  I believe the company would be better advised to wait for a single molecule, subcutaneous version, the ultimate future of DPC.

Question of the day: why does adding more GalNAc to an already GalNAc-targeted molecule enhance gene silencing?  Could it be that they added a trivalent GalNAc (like Alnylam), and not more monovalent GalNAc as on the endosomolytic backbone?

Oligonucleotide Therapeutics Need to Embrace New Genomic Era in Cancer Medicine

With ASCO 2014 behind us, I am left with the impression that oligonucleotide therapeutics have failed to keep pace with important developments in cancer drug development.  Most importantly, cancer oligonucleotide therapeutics need to take advantage of the latest genomic tools and insights in cancer biology to select the patient populations most likely to respond treatment to both increase success rates and to save development dollars.  After all, this is what you would expect from a platform most suited to personalized medicine.

In some cases, drugs are being developed a priori with specific mutations in mind such as the selective EGFR-mutant inhibitors by Clovis Oncology and AstraZeneca which have caused quite a stir amongst clinicians and investors at ASCO.  In other cases, and this is something spearheaded by companies like Foundation Medicine and their pharmaceutical partners, comprehensive next-gen sequencing is employed as trials progress to identify often complex signatures that render a cancer susceptible to a treatment. 

And particularly relevant for delivery aspects, other companies such as Endocyte are using imaging diagnostics to identify those cancers that either express the right surface receptors for a ligand-targeted agent or those amenable to the EPR effect relevant for nanoparticle-based delivery.

Although I believe that Tekmira has also a delivery-related rationale in selecting adrenocortical carcinoma and neuroendocrine cancers for TKM-PLK1 based on their insights in liposomal delivery, the oligonucleotide field at large is failing to take advantage of delivery-related cancer selection strategies.

In terms of tumor response related to hitting the right targets and pathways, I would like to see the field embrace the personalized genetics approach.  Prospectively targeting KRAS mutant colorectal cancers with a PLK1-RNAi therapeutic would be one example.  Going one step further, a development strategy that looks at the genomic profiles of various cancers and then pull out the formulation with the right RNAi trigger or antisense oligo against critically activated pathways would play into the strength of the technology: every signature is ‘actionable’ as the technology leaves no target behind.

Especially for delivery approaches that can penetrate tumors well, innovative cancer development could unlock the considerable potential that oligonucleotide therapeutics have for oncology.  Beyond TKM-PLK1, the androgen receptor antisense compound by ISIS/AstraZeneca could be an important first test case for this concept in oligonucleotide therapeutics.

The GalNAc Wars

‘All roads lead to GalNAc’ was the thoughtful, almost philosophical response by Alnylam’s CEO John Maraganore when asked about the GalNAc-conjugate challenge mounted by ISIS Pharmaceuticals for gene silencing in hepatocytes. 

With 20mg per week (corresponding to less than 0.1mg/kg per week) and a much improved hepatocyte-specific biodistribution of otherwise more promiscuous phosphorothioate chemistry, GalNAc-antisense has suddenly re-emerged as a significant competitive threat to the hepatocyte-targeted RNAi delivery solutions by Alnylam, Arrowhead Research, and potentially other undisclosed efforts.  By comparison, 1^st gen GalNAc-based ALN-TTRsc by Alnylam required at least 10x more oligonucleotides to effect potent gene silencing in a phase Istudy in TTR amyloidosis. 

2^nd gen ESC GalNAc-conjugates are claimed to be significantly more potent, but I suspect that they won’t be that much more potent than GalNAc-ASOs and will have to win the GalNAc battle based on safety.  Arrowhead’s GalNAc-targeted single-molecule DPCs should still be the most potent of the bunch, but it is about time for them to advance them into the clinic as the first-in-class train is leaving the station for an increased number of targets and it is feeling the most-potent-in-class heat in its lead HBV indication that is based on 2-molecule DPC.

With GalNAcs applied in such a ubiquitous manner, you’d think that the likes of Alnylam and ISIS Pharmaceuticals, not known for sitting on their hands when it comes to claiming ownership over technologies would start flexing their IP muscles.  As I noted in a post from AsiaTIDES earlier this year, the word on the lab floors is that GalNAcs per se are an old, non-patentable free-for-all such that the IP focus should shift to other aspects of the GalNAc-targeting approach such as linker chemistries and economical manufacturing methods.

As much as it is an interesting twist that ISIS would adopt Alnylam technology to become competitive again for liver-directed gene silencing, the IP situation is intriguing given the cross-licensing relationship between the two companies.  The original arrangement going back to over a decade ago gave Alnylam access to IP from ISIS for the use in double-stranded oligonucleotide therapeutics, and in return ISIS got access to IP from Alnylam for use in single-stranded oligonucleotide therapeutics.  Whether this IP exchange includes IP that does not directly impinge on oligonucleotide chemistry, such as aspects of delivery will be a critical element in how easily an outright IP war could be triggered. 

If Regulus Therapeutics which has rights to Alnylam’s GalNAcs as evidenced by its anti-miR122 candidate is a guide, the IP exchange would be broader than just oligonucleotide chemistry.  On the other hand, I would be very surprised if strategy- and IP-savvy Alnylam did not foresee the potential perils of also giving ISIS access to its delivery technologies.

In any case, after the first verbal skirmishes, expect the saber-rattling around GalNAc to increase with IP adding a new dimension to what has largely been a battle based on potency and safety considerations.