Isis Pharmaceuticals and Roche/Santaris About to Settle Patent Dispute

According to court documents dated March 20, 2015, it looks like Isis and Roche, the new owner of the original defendant Santaris, are about to settle the RNaseH antisense patent infringement suit brought by Isis.  A settlement would have important implications for the future of Antisense Therapeutics. 

To wit, in 2011 Isis sued Santaris for infringing its RNaseH gapmer patents by Santaris signing on Big Pharma partners regarding RNaseH Therapeutics development.  Isis viewed this as a form of monetizing their IP and consequently sued.  If decided in Isis’ favor, the case would have been a notable departure from the long-held practice of shielding preclinical business and drug development under the safe harbor of the Research Exemption. 

As such, the case could have had a chilling effect on particularly small innovative biotech companies seeking to improve upon existing technologies, but by this using aspects of those technologies.  Big Pharma, after all, do not rely on partnerships to finance technology development and can thus go on using and improving the IP of other companies in their own labs impugned.

   

Although one could have thus taken the view that losing the case would have been in Roche’s interest in a perverted sort of way, it could also have more immediately jeopardized the value of their acquisition of Santaris (USD 250M) in addition to payable damages.  For example, Roche might have been ordered to cease any RNaseH work in the US which may be impractical for a global research organization like Roche.

Since I cannot imagine that Isis would tolerate Roche to challenge their control over RNaseH antisense gene knockdown and compete for pretty much the same targets, I expect the settlement to take the form of a significant broadening of the companies’ existing relationship around Huntington’s Disease in the form of additional target picks in exchange for a sizeable upfront fee. 

Unlike Isis’ more recent Big Pharma deals with J&J, GSK, BiogenIdec, and AstraZeneca, however, I expect this to involve less early development work by Isis as some of this would be the obvious job of the former Santaris crew.

Co-delivering Antisense and RNAi for Cancer

The upcoming phase I top-line data for ISIS-STAT3Rx in liver cancer (HCC) to be presented at the upcoming EORTC-NCI-AACR triple meeting in Barcelona (Nov 18-21) will be an important test of the potential utility of RNAseH antisense oligonucleotides (ASOs) incorporating high-affinity chemistry in oncology.  

Based on the body language by ISIS Pharmaceuticals* and last week's $7.5M milestone payment from partner AstraZeneca for progress on ISIS-STAT3Rx (aka AZD9150) , I am tempted to speculate on more than just ‘encouraging’ results.  On the other hand, Regulus Therapeutics partner Sanofi at the Canton Nucleic Acid Forum (CNAF) also last week, noted the need for formulating antisense oligonucleotides to get their anti-miR21 oncology candidate into liver cancer tissue. It is likely that they will be using liposomes for that (--> Tekmira?).

* I was surprised that at the CNAF in Guangzhou, China, Brett Monia from ISIS mentioned STAT3Rx and cancer right after gene silencing in the liver as the next interesting application for ASOs- that is ahead of even the exciting CNS opportunity.

The discrepancy in body language may be explained by just cultural differences (conservative, blasé Big Pharma versus risk-taking, enthusiastic biotech); it may also be a reflection of different requirements for effective tissue concentrations with RNaseH versus anti-microRNA modalities or different target requirements.  Whatever the reason, the Sanofi comments clearly support the notion that getting naked, even phosphorothioated oligonucleotides into cancer tissues is not as robust as with other tissues such as the liver and kidney.

I am therefore pleased that it is a Big Pharma, the last place where I had expected that from, that is connecting the dots and is considering delivery formulations, even the supposedly ‘toxic’ LNPs.  The concept is that the nanoparticle would facilitate a higher tumor concentration of the oligonucleotide, and once in the tumor interstitial space, cellular delivery may be facilitated via two routes.  Firstly, it may be traditional liposome-dependent cell uptake and cytosolic release.  Alternatively, those LNPs that get stuck in the interstitial space would spill the phosphorothioate oligo which may then diffuse further and get into the target cell by self-delivery. 

The same concept, of course, applies not only to phosphorothioate ASOs, but also to self-delivering RNAi triggers (+/- conjugation).

But why stop there? I propose that for cancer delivery, one should strongly consider and co-formulate RNAi triggers and ASOs into a shared nanoparticle.  They could target either the same gene, or they could target different genes thus taking into account the desire for a multi-pronged attack on cancer   (-> resistance).  In that scenario you would benefit from the superior gene silencing efficacy of RNAi triggers in those cells that they were able to reach, but then extent your reach with the help of the more agile, penetrative single-stranded antisense molecules. 

As such, PS-ASOs have an advantage in addressing intra-tumor heterogeneity of the EPR effect which is a well-recognized problem of nanoparticle delivery for cancer.

Another benefit of combining RNAi triggers with RNaseH ASOs is that you could achieve additive gene silencing activity when going after a shared target.  For example, if the RNaseH ASO and the RNAi trigger had both say a 70% knockdown activity on their own in the nucleus and cytoplasm, respectively, the combined activity would likely be ~90% which genetically could make a huge difference.

There is also a potency benefit, although more minor, when going after different targets because at least in RNAi, the best you can hope for when combining RNAi triggers against different targets is that they do not interfere/compete with each other.

With solid cancer data from both Tekmira (RNAi) and ISIS/AZ out over the coming weeks, we will soon get a sense of whether the field has moved forward in oncology and what the next steps ought to be.

GalNAc 2.0 with Greatly Improved Single-dose Efficacy and Duration

Last night, Alnylam kicked off a week of what promises to be exciting disclosures about continued progress in therapeutic gene silencing of genes expressed in the liver.  In a presentation by oligonucleotide star chemist Mutiah Manoharan at TIDES, the company provided a more thorough chemistry and pharmacology background behind the apparent improvements of the GalNAc delivery platform (Enhanced Stabilization Chemistry).

Journey along a hostile environment

The improvements are based on the observation that the 5’ ends of both the guide and passenger strands are subject to degradation by 5’-3’ exonucleases.  These may act at various stages during the relatively long journey of a subcutaneously administered RNAi trigger-conjugate: in the subcutaneous space, the circulation and lymphatics, and finally along the nuclease-rich endosomal/lysosomal uptake pathway in the target cell itself. 

Accordingly, by adding undisclosed chemical modifications to the 5’ (but also 3’) termini of the RNAi trigger strands, ~5 times the amount of RNAi trigger reaches the liver, and 10-100x RNAi trigger is found in the liver over time compared to first-generation chemistry as exemplified by ALN-TTRsc.  This means that single digit microgram per gram liver tissue can now be achieved at steady-state.  For comparison, gen 2.0 and 2.5 RNaseH ASOs (--> ISIS) depend for activity on ~100-300 microgram per gram liver tissue steady-state concentrations of phosphorothioated oligonucleotides. 

There obviously is a balance between maintaining high drug concentrations for efficacy and avoiding excessive concentrations for fear of causing inflammation and subsequent tissue scarring.  In that regard, Alnylam reports a wide therapeutic index, including in non-human primates which, laudably, were generally extensively used in these studies.

Great benefit for single-dose efficacy and duration

The new pharmacological profile is somewhat counter to a critical advantage of the RNAi platform over single-strand RNaseH technology: achieving great and sustained efficacy with minimal tissue exposure.  

Mechanistically, this fundamental capacity is explained by the fact that once loaded onto the RNAi effector complex, RISC, the duration of RNAi trigger activity in non- or very slowly dividing tissues such as the liver is largely limited by the slow (weeks) natural turnover of RISC.  By contrast, although RNaseH is a catalytic mechanism, too, no such sustained holding on to the antisense oligonucleotide is known for RNaseH such that the guide oligonucleotide has to be constantly available.

According to this model, an important determinant for the efficacy and very feasibility of traditional RNAi approaches is the size of the unloaded pool of RISC during the short period of time that an otherwise unstable RNAi trigger is available.  By contrast, unstable RNAi triggers are ill suited to take advantage of newly synthesized RISC complexes as part of natural RISC protein turnover.   

This is where GalNAc2.0 comes in: by extending the presence of the RNAi trigger, RNAi triggers can now also be loaded into newly synthesized RISC, thus extending the duration of gene silencing by replenishing the pool of RISC that gets lost during its turnover.  As discussed last week, in the case of ALN-PCSsc for the treatment of hypercholesterolemia, GalNAC2.0 can achieve sustained potent gene silencing of PCSK9 for 2-3 months following a single dose compared to only days/weeks with the old chemistry.  Moreover, when it comes to single-dosing schedules, GalNAc2.0 is also vastly (~10x)  in terms of maximal knockdown potency compared to GalNAc1.0 which relies on a loading dose schedule (5x daily injections) for efficient loading of free RISC.

ESC less transformational in multi-dose regimens

Somewhat lost in Alnylam’s press release was the fact that for multi-dosing, the benefit of GalNAc2.0 is less dramatic in terms of the amount of RNAi trigger required to achieve say a 80% target gene knockdown.  For example, for TTR, the ED80 with weekly GalNAc1.0 in non-human primates was ~2.5mg/kg, the same as that now reported for the ED80 with a GalNAc2.0-chemistry improved version in rodents.

This confusion was not helped by the fact that direct comparisons between GalNAc1.0 and 2.0 were only shown for single-dose studies or by the fact the efficacy summary slide compares GalNAc1.0 for TTR with GalNAc2.0 for PCSK9.

Therefore, when the goal is to enhance the target product profile of your RNAi therapeutic by minimizing the frequency of subcutaneous administration (e.g. PCSK9 in light of the monoclonal antibody competition), then GalNAc2.0 certainly represents a very valuable advance, albeit at the cost of (still) relatively large injection volumes (10mg/kgà 4ml).  However, when it comes to the maximal potency against a given target gene, similar results may be obtained with GalNAc1.0 with possibly an improved safety profile.

Of course, more potent and at least equally sustained efficacy following subcutaneous administration may be achieved by Arrowhead's single-molecule DPCs.  If and when they can finally be translated into the clinic, is an important and open question.  I hope we see more data on that this week, also from the TIDES.

ISIS-CRP Phase I Results: The Decreasing Competitiveness of RNaseH Antisense for Liver Gene Knockdown

To maintain is positive trajectory of the recent comeback, it is important for RNAi Therapeutics to firmly claim gene knockdown in the liver for itself.  This is because liver gene knockdown is the lowest hanging fruit for the technology, yet RNaseH antisense has proven to be capable of gene knockdown there as well.   If RNaseH were to crystallize as the preferred technology, it would jeopardize a considerable portion of the near to midterm market potential of RNAi Therapeutics.  Just think of Alnylam's 5x15(TM) portfolio of drug candidates.

As I have explained before, dose/tissue oligonucleotide concentrations and, related to that, safety/tolerability will determine which technology will win the competition.  Although the odds seem in favor of RNAi Therapeutics, ISIS Pharmaceuticals keeps promising that improved chemistry and screening methods will improve upon the modest potency (~30-35% ApoB knockdown) of and the safety concerns with mipomersen.  Due to the importance of liver gene knockdown to RNAi Therapeutics, this blog will continue to comment on the respective data points as they come in.

RNaseH Antisense CRP Phase I Data

Today, ISIS Pharmaceuticals announced phase I knockdown data from its c-reactive protein (CRP) program, ISIS-CRPRx.   CRP, a gene expressed by the liver, is believed (by some, not all) to act as a central player in a host of inflammatory diseases such as atrial fibrillation and rheumatoid arthritis.  Phase II studies for these indications are ongoing.

The phase I study tested whether ISIS-CRPRx can blunt CRP induction upon inflammatory stress.  Accordingly, healthy volunteers were dosed 6 times over 3 weeks with either 400mg or 600mg of the phosphorothioate antisense molecule against CRP.  Following that, subjects received endotoxin, a common laboratory reagent that is interpreted by the body as an infection.

Subjects treated with placebo saw an approximately 30-fold increase in CRP levels due to endotoxin.  When pre-treated with 400mg and 600mg, however, the induction was reduced by 36% and 63%, respectively.

 

Although the study succeeded in showing that ISIS-CRPRx could blunt CRP induction, it is much less clear whether ISIS-CRPRx makes for a viable therapeutic.  A 63% knockdown knockdown may be OK for some indications, but intuitively one has to wonder whether to stop an acute inflammation in its tracks you need to step much harder on brakes like CRP.  Indeed, the press release proclaimed, as a testament to the specificity of the drug candidate, that ‘other important immune modulators’ were not changed- somewhat puzzling given that CRP is thought to play such a central role in inflammation.

Be that as it may, what the press release did not mention was the side effect profile seen in the ISIS-CRPRx phase I study; to wit, safety and tolerability is the primary objective of any phase I study.  What concerns me is that mipomersen is 200mg weekly and causes injection side reactions, flu-like symptoms, and indications of chronic inflammation which are likely the result of the elevated steady-state tissue concentrations required for RNaseH antisense efficacy.  In the CRP study, 2- to 3-times more phosphorothioate antisense molecules were given.  Although the side effect profile of an RNaseH antisense molecule is also sequence-dependent, doses of 400mg and higher have historically been associated with toxicity/tolerability considered unacceptable outside of cancer.   

My prediction is that due to the poor efficacy-dose relationship, ISIS-CRPRx will not go into phase III.  We have to see whether the gen2.5 cET chemistry brings any significant increases in clinical potencies as inherent molecular potencies as measured in cell culture studies may not take into consideration other pharmacological rate-limiting factors.  If gen2.5 can lower liver tissue concentrations required to obtain 50% gene silencing into the low single-digit microgram/g range, RNaseH would become more interesting again for the liver.  Otherwise, it seems that more fertile hunting grounds for antisense technologies may be found in the areas of noncoding RNAs (e.g. microRNAs and lincRNAs) and therapeutic splice regulation, especially when the target molecules have a long nuclear residence time.  With ISIS about to present phase I data for its spinal muscular atrophy splice modulation drug candidate at next week’s AAN meeting, a next step into that strategic re-direction may be taken.

ISIS TTR Results Show 1st Gen SNALP Already on Par with RNase H Antisense

ISIS today reported knockdown results from a phase I trial with its RNase H antisense compound ISIS-TTRRx against the transthyretin gene for the treatment of TTR amyloidosis. The results show that while high 81% knockdowns can be achieved, very large amounts of oligonucleotides (400mg weekly for 4 weeks) had to be given. When clinically more relevant doses were given (200mg as in mipomersen and the next lower dose from 400mg), a mean knockdown of 44% was achieved.

As you will remember (SNALP Works!), the RNAi Therapeutics candidate ALN-TTR01 last November showed a mean reduction of 41% of serum TTR from baseline following a single 1.0mg/kg infusion. In light of yesterday’s results with ALN-PCS02 where a supposedly less potent RNAi trigger was employed and a ~60% knockdown was achieved with 0.25mg/kg, it seems likely that the Tekmira SNALP-enabled ALN-TTR02 which is about to enter clinical development and which employs a formulation similar to PCS02 will show the type of 70-80% knockdown at 0.25mg/kg.

We have yet to learn more about the safety profile of ISIS-TTRRx which is being developed under an option agreement with GSK. ISIS stated in their press release that they were planning to enter pivotal studies with this compound soon. If it is with the 400mg dosage, they may be tempting fate. It might be interesting to try 300mg.

As you know, these days I’m strongly advocating RNAi Therapeutics as the technology is currently being sold at much below the worth of the science which has progressed steadily over the years despite the up and downs in public perceptions. Today’s results by ISIS show that the new darling in nucleic acid therapeutics can certainly knock down genes in the liver (and some other tissues). They also show, however, that even the initial SNALP formulations are already on par with it in terms of efficacy at the higher end of tolerability (not talking about other aspects such as absolute differences in dosage or route of administration).

SNALP RNAi versus RNaseH Antisense for Gene Knockdown in the Liver

Following recent phase I results from ISIS Pharmaceutical’s Factor XI (ASH abstract 12999; addendum: PR on phase I data reported on December 12) and Apo C III programs, there is little doubt left that phoshorothioate-based RNaseH antisense as developed by this company and Santaris can mediate target-specific gene knockdown in the liver in Man. These results confirm the clinical experience with the registrational hypercholesterolemia candidate mipomersen and are corroborated by the impressive HCV results obtained by Santaris’ with its anti-miR122 HCV candidate. Beyond the liver, RNaseH efficacy has been demonstrated for solid cancer (custirsen) and possibly Excaliard’s (now Pfizer’s) anti-scarring candidate. On the other hand, the recent trial termination(s) by Santaris, and the safety profiles of mipomersen and OncogeneX' custirsen highlight some of the challenges facing phosphorothioate antisense technology.

Do you believe that your writing style or English language skills put you at a disadvantage in publishing your research? OxTERMS can help you.

Therefore, as RNAi Therapeutics have progressed from the basic discovery of its mechanism in mammals 10 years ago to solid proof-of-concept gene knockdown in the form of the ALN-TTR01 results 2 weeks ago, it may be a good time to compare and contrast these two technologies also in light of the fact that, after the RNAi Therapeutics backlash, there is a clear trend towards Big Pharma (and other pharmaceutical companies) opening themselves up again towards antisense, meaning that the 2 technologies are competing for precious non-dilutive funding. For this purpose, I will focus on the liver as the best developed target organ for both these technologies.

20 to 100-fold more antisense required

An obvious advantage of antisense, with about 3x the age of RNAi Therapeutics, is that more is known about its clinical pharmacology. As such, there is good visibility as to how much antisense will be needed to achieve the kind of 50-75% knockdown that will be required for therapeutic outcomes in most cases. Dose has important implications particularly in terms of safety and cost.

Current ‘2^nd gen’ molecules (fully phosphorothioated gapmers) like mipomersen require 200mg oligonucleotide per week. Actually, if 300-400mg would have been better tolerated, the higher dosages would have enhanced the commercial profile of mipomersen considerably. Let’s therefore say 1000mg per month for 2^nd gen RNaseH antisense.

With the higher-affinity ‘2.5 gen’ technologies that are starting to move into the clinic, best exemplified by Santaris’ LNAs which in fact may also symbolize the most potent version, it is expected that clinical dosages can be further reduced. Based on the non-human primate data and clinical dosage regimes, I expect that dosages of around 100mg/week or 500mg per month are feasible in the foreseeable future. Also because of the modifications involved, I would be therefore very surprised if the cost of oligonucleotides for treating a patient over a year would be below $15,000 even at commercial scale.

By contrast, it can be expected that it will take about 0.15mg/kg of siRNA formulated in the ‘2^nd gen’ SNALPs that are now moving into the clinic to achieve the type of once-a-month pharmacology that Tekmira and its licensees are aiming for. If you do the math, that translates into about 10mg per month of siRNA oligonucleotides. Give and take the added costs of the lipids and formulation process, but cheaper nucleotide chemistries involved, this translates into a maybe 50-fold cost of goods difference alone. For some diseases and in some countries, this may be less of an issue, but it will be a factor for others.

Safety

The even larger implications of dosage is the related safety. Although clinical repeat-administration studies with SNALP have yet to be conducted, it seems that with the 2^nd gen SNALP formulations, the main safety challenge with SNALP will be in managing acute hypersensitivity reactions around the time of drug administration. Based on similar issues with intravenously administered biologics such as monoclonal antibodies where e.g. transient immune suppression with steroids is routine (and widely accepted), I believe that infusion-related acute toxicities will be manageable.

What is nice with the pharmacology of SNALP RNAi Therapeutics is that the bulk of the drug that does not hit the target, i.e. gets incorporated into the RISC silencing complex, is rapidly turned over by the body, meaning that drug exposure levels between drug administrations will be extremely low. It is because of this that I am hopeful that the risk of causing liver toxicity, long believed to be the main toxicity challenge for SNALP, is quite limited at dosages of 0.15mg/kg/month.

By contrast, RNaseH antisense do not harness a natural gene silencing mechanism and, in the case of the phosphorothioate-based gen 2 and gen 2.5 antisense, work by saturating the target (and off-target) organs with high levels of the ‘sticky’ phosphorothioated oligonucleotides so that mass action carries enough of them into the cells. Consequently, the exposure of the body to the antisense drug is significantly higher compared to SNALP-delivered siRNAs. Assuming ~100-300mg of antisense oligonucleotide per kg of liver or kidney tissue (e.g. ISIS TTR patent application US 2011/0294868) and single-digit microgram siRNA oligonucleotide per kg of liver tissue (e.g. Landesman et al. 2010), you could argue that the real difference in bioburden between antisense and SNALP RNAi is about ten thousand fold. It also does not take into account that it is the sticky phosphorothioate chemistry that is thought to be responsible for much of the toxicity (interactions and turnover) whereas RNAi triggers employ more natural chemistries. On the other hand, double-strand RNAs are recognized by more innate immune receptors than highly modified, single-stranded oligonucleotides.

Route of administration

Although the subcutaneous administration of SNALPs has been demonstrated (see e.g. Tekmira's ApoB patent) and may become practical with the higher potencies of SNALPs and extracellular matrix-degrading technologies as developed e.g. by Halozyme, antisense is currently more amenable to subcutaneous administration whereas SNALP have to be infused in an institutional setting. This means that, as is the case for essentially all monoclonal antibody drugs, SNALP drugs have to address diseases of considerable unmet medical needs where patients do not perceive a once-a-month trip to the infusion center a huge burden. Maybe Pfizer can't, but I can think of many such diseases. Infusion in an institutional setting also has the advantage that acute toxicities, the main safety challenge for SNALPs, can be well managed through professional supervision, whereas patients that inject themselves with antisense at home may be slightly panicked on seeing redness develop at the injection site or on experiencing ‘flu-like symptoms’ that have been reported to occur at high frequency with antisense (often 1/3 to 1/2 of patients), but has surprisingly been little discussed by ISIS Pharmaceuticals.

Manufacturing

Like route of administration, manufacturing is considered to be a practical advantage of antisense over SNALP RNAi. I agree…in purely practical terms. What is, however, entirely forgotten is that as long as you can deal with the manufacturing complexities, it suddenly gives you an invaluable competitive advantage. How about unlimited market exclusivity? Isn’t one of the lessons that Big Pharma should have learned from the current patent cliff that simple small molecule chemistries are highly vulnerable to generic competition? Isn't this also a major reason for why everybody obsesses about monoclonal antibodies these days, yet is often strangely held against SNALP RNAi? To my knowledge, there are no generics of a nanoparticle-formulated drug.

So in summary, as antisense has reached an inflection point as a slew of clinical data is confirming the early clinical results with mipomersen from 6-7 years ago which demonstrated gene knockdown in the liver, SNALP RNAi is making even faster progress with many of its theorized advantages, especially related to the amount of oligonucleotide required and pharmacology, turning into clinical reality quickly. The race is on. The most likely winners meanwhile are the patients.

RNAi Therapeutics versus Antisense- Where Delivery Makes a Difference

Regulus, the microRNA joint venture between Alnylam and ISIS Pharmaceuticals, is the most visible manifestation of the scientific overlap that exists between antisense and RNAi. The overlap, however, is not just limited to the science, but also extends into the capital markets. To better help the investor differentiate between the two technologies, I’d like to use this blog to provide an overview of some of the fundamental differences underlying the development of RNAi and antisense into therapeutics and their long-term prospects, with an emphasis on delivery. Needless to say, beware this discussion will be heavily biased in favor RNAi, but then again these are the reasons why I’ve been attracted to RNAi in the first place.

Certainly, progress in both areas in the last 3-5 years has mutually benefitted the investment climate for both technologies as it has heightened interest and increased confidence in RNA therapeutics in general. However, the two technologies also compete for investment dollars with many of the same investors, which are typically upbeat about the future of gene-based medicines but unsure where to place their bets, allocating their investments based on where they see most promise. One issue that often comes up in making this decision is the observation that while for systemic applications antisense, as practiced in the most advanced programs today, is typically administered without a particular delivery formulation, the development of specialized delivery technology is frequently cited as the key challenge for RNAi to realize its ultimate therapeutic potential.

Antisense for gene knockdown works largely by two mechanisms: interfering with translational initiation (e.g. AVI Biopharma's morpholinos) or through an RNase H-type mechanism (e.g. Santaris and ISIS Pharmaceuticals). For this, the key factor is to achieve efficient hybridization of a single-stranded oligonucleotide antisense with its target mRNA which either prevents productive ribosome association to the mRNA (inhibition of translation initiation) or may be recognized as a substrate for the RNase H enzyme which may degrade the RNA portion of the mRNA-DNA duplex, but normally functions in the degradation of the RNA primer during DNA replication.

Various oligonucleotide chemistries have been developed to optimize these processes for in vivo applications. Essentially all of these are single-stranded oligos of which the sugar phosphate backbone is heavily modified to a) increase their in vivo stability; b) improve their pharmacokinetics and avoid rapid renal excretion by promoting their association with components of the blood; c) similarly allows them to be retained in tissues; d) facilitate crossing of cell membranes; and finally e) increase their target mRNA binding. By contrast siRNAs, because of their charge and more rigid double-stranded nature and with apparently some exceptions that include mucosal epithelia, do not cross cell membranes efficiently on their own and therefore need to be specially formulated for most applications.

The use of unformulated antisense is consistent with their mechanism of action. Since antisense does not harness a naturally existing endogenous gene silencing pathway, it relies on achieving concentrations of oligonucleotides in the target tissue over a prolonged period of time that are high enough such that, as a result of the rules of thermodynamics, a sufficient fraction of target mRNA will be recognized. Similarly, unlike RNAi, the specificity of antisense is largely governed by biophysics and benefits only relatively little from biological proof-reading.

In practice, to achieve the necessary tissue concentrations, patients are typically dosed frequently at the initiation of therapy so that the tissue concentrations reach steady-state therapeutic levels. Targeted delivery of antisense into cells of interest may allow one to achieve a knockdown earlier, but any benefit would only be short-lived as antisense is not retained in specific gene silencing complexes but will soon redistribute according to their partition coefficient throughout the entire tissue so that ultimately similar amounts have to be administered and a formulation would only be a nuisance with little benefit.

By contrast, RNAi harnesses an endogenous and catalytic gene silencing mechanism, which means that once it has been delivered, either by conjugation or in nanoparticles into the cytosol, they are efficiently recognized and stably incorporated into the RiSC silencing complex to achieve prolonged gene silencing. In fact, measurable RNAi-mediated gene silencing can be observed at siRNA concentrations so low that it becomes difficult to detect them (e.g. fluorescently-tagged siRNAs by microscopy). This means that as the majority of siRNAs that do not reach the cytoplasm may disappear quite rapidly, the total exposure of the body to the nucleic acid can be much lower compared to antisense which should be beneficial both in terms of safety and pharmacodynamics (activity profile of drug over time).

This is not to say that chemical modification is not practiced in RNAi. However, unlike in antisense, the purpose of modification in RNAi is mainly to avoid triggering innate immune responses, making the siRNA sufficiently stable so that they survive their journey into their target cells, and also to stabilize them as part of RiSC (Merck has been talking about that concept on several occasions); and as we learn more about the biochemistry of endogenous RNA silencing pathways, modification is also increasingly used to increase the inherent biological specificity of RNAi. Unfortunately, it is surprising to me that compared to RNAi only very little, if at all, is reported about the specificity of antisense and I would be grateful if somebody here could point out pertinent studies that I should be aware of.

Targeted delivery may also avoid unnecessary drug exposure of non-target tissues. For unformulated antisense, no matter what the indication and target tissue, the biodistribution is essentially the same, and toxicities of the liver and kidney due to extended exposure to large amounts of the heavily modified antisense compounds is well known.

Certainly, improving the therapeutic index is an important issue for RNAi Therapeutics, too, but as the many transgenic mouse models which express ample and highly efficient RNAi throughout their life without causing overt toxicity attest, ultimately the improvement in the therapeutic index of RNAi is not limited by its very mechanism of action.

While it is a certainty that antisense companies will come out with 4th and 5th generation antisense technology, advances after decades of antisense research aiming to improve target mRNA recognition will only be marginal and based on trying out yet more nucleic acid modifications, although it appears to be a challenge to improve upon the efficacy of probably the most potent antisense modification that have now been known for a while, namely LNAs and their derivatives.

While RNAi efficacy in animals has already surpassed that of antisense for applications of the liver and lung as well as other tissues, I am confident that future advancements in RNAi will be more than marginal. For example, even as recent liposomal formulations achieve 90% gene knockdown in the liver at 1mg/kg, this still means that only about 1 in 10,000 siRNAs that have reached the liver makes it into the cytoplasm (assuming it takes about 1000 cytosolic siRNAs to achieve that level of knockdown according to a recent presentation by Phil Sharp). Alone a better understanding of the endosomal uptake of these nanoparticles, which is only in its infancy and starting to be explored, should allow for more than incremental improvements in the therapeutic index of RNAi Therapeutics.

And if you are still undecided on where the future is heading, numerous transfection studies in vitro where it can be assumed that equal amounts of antisense and siRNAs are present in cells, have shown that RNAi is quite a bit more potent on a mole-by-mole basis comnpared to antisense.

I am aware that some in the antisense community, including investors, may take offense with this blog, but since I am often asked about this issue, I think a more straightforward approach is better than to keep beating about the bush. And, of course, there is always the comment section.