Strong Presumption Moderna Covid Vaccine Infringes on Arbutus IP

News last Thursday that the US Patent Trial and Appeal Board (PTAB) upheld a broad liposomal nanoparticle (LNP)-related patent added an interesting wrinkle to the race to develop a vaccine against covid19.  Given the way that LNP delivery of mRNA is practiced today (see also below), the patent owned by Arbutus Biopharma greatly impacts on the development of some of the most prominent covid19 vaccine candidates, most notably by Moderna, but also BioNTech, CureVac and others.

This blog entry will provide some background on the history of the science and business of LNPs and why there is a strong presumption that the mRNA-1273 formulation by Moderna which is about to enter phase 3 development infringes on the US patent 8,058,069 ('069) in question.

Disclosure: while anybody who has been following RNAi from the early days of IP battles should be able to call themselves an IP expert, I do not have a corresponding certificate hanging on my walls.  Moreover, since I am long Arbutus Biopharma (ticker: ABUS), I have a vested interest in Arbutus prevailing in the present case, but hear me out anyway...

A history of trade secret theft and IP misappropriation

In 2005-6, Protiva (a predecessor of Arbutus) emerged as the leader in the delivery of RNAi Therapeutics by reporting the first clinically relevant success in taking RNAi triggers into cells in the human body.  The delivery formulation, then called SNALP, involved a mixture of lipids that has essentially remained the same until today:

•         the nucleic acid payload;
• an ionizable cationic lipid for formulation and cell penetration;
• a (PEG-)conjugated lipid for stability;
• cholesterol;
• neutral phospholipids.

-           

What has mostly changed since is the nature of the cationic, typically ionizable lipid, but the 4 lipid-component-system has stayed the same despite at times frantic IP workaround attempts like using just 3 lipid components.

Due to the central importance of these LNPs to RNAi Therapeutics, large players (Sirna Therapeutics/Merck, Alnylam, Roche) soon came to collaborate, then misappropriate related IP from the small Canadian biotech company.  A bitter divorce of Protiva and scientists related to Pieter Cullis from the University of British Columbia certainly aided that goal by facilitating the transfer of know-how and trade secrets for peanuts. Divide and conquer.

Despite of what appeared to be a hopeless battle between David and Goliath, a settlement was reached in 2012 with Alnylam and 'Cullis' in which Alnylam got access to (now) Tekmira’s IP mainly for covering ONPATTRO for the treatment of TTR amyloidosis (now approved) in return for allowing Tekmira survive a system that greatly favors the guy with the most financial wherewithal (disclosure: I was an expert witness called on by Tekmira in that litigation).  The settlement also provided for the limited use of certain IP by 'Alnylam Canada' (AlCana, now Acuitas).

LNP going out of favor in RNAi as mRNA gains traction

The fact that Alnylam started to see success with a less invasive conjugate delivery strategy (GalNAc) for RNAi around that time, certainly helped with settling the dispute.  In ~2015 then conjugate technology in the form of second-generation GalNAc technology started to demonstrate superiority over LNPs in RNAi delivery and Tekmira faced the decision of where to take their company.

In my mind, the decision was obvious: leverage the LNP know-how and IP in order to cement their position in messenger RNA (mRNA) Therapeutics, a hot new field in biotech where Tekmira had been generating industry-leading data.

Instead, in what must be one of the most catastrophic business decisions that to this day I fail to grasp, Tekmira in 2015 gave up half of the company in a merger with a paper company called OnCore Biopharma in establishing an HBV therapeutics solutions company.  This shell contained nothing more but a list of untested chemical structures written on the back of an envelope by new biotech Wunderkind Vivek Ramaswamy and his scientific lieutenant Michael Sofia.

To wit, Vivek Ramaswamy is famous for dumpster diving and buying rights to a failed Alzheimer’s asset from GSK for $5M to then sell it to the public without much additional development for $1.5B soon thereafter.  It then only took one clinical trial to incinerate that value to essentially zero...

Arbutus Biopharma was born.

Unsurprisingly, in what must be one if not the longest string of failures in biotech history, small molecule after small molecule crashed and burned, mostly due to preclinical and early clinical tox issues.

Realizing that their legacy IP and know-how could be quite valuable for financing their string of failures, Vivek took advantage of Arbutus and its shareholders yet another time by spinning out Arbutus' mRNA assets into Genevant (January 2018).  

Arbutus retains a 40% stake in Genevant, a number that Arbutus said needs to be adjusted for 'significant' dilution due to the convertibles that Genevant has issued since.   Possibly equally if not more importantly, Arbutus is eligible to 20% of the revenues (e.g. from damages and royalties payable to Genevant) from the sublicensing of LNP IP by Genevant, for example as the result of a settlement with Moderna, but also other prominent covid vaccine players like BioNTech and CureVac.

BioNTech and Moderna acknowledge fundamental importance of IP controlled by Arbutus

The Genevant creation apparently hinged on a settlement with the rival Vancouver group (now 'Acuitas') announced the following month (February 2018) where it was determined that Acuitas could not sublicense certain fundamental LNP IP under the grand 2012 Alnylam settlement.

Not long thereafter, covid vaccine player BioNTech which until then had worked with Acuitas on mRNA LNP delivery took a license to Genevant's LNP IP in an obvious acknowledgement of the new settlement (note: there has been no explicit disclosure of whether the original license agreement between Genevant and BioNTech has been extended to cover the lead covid vaccine candidates by BioNTech and partners Pfizer and Fosun Pharma; potentially another important puzzle piece of the intrigue).

Importantly, the settlement also provided that Moderna, which similarly had been working with Acuitas, would only retain legacy rights to such IP for 5 viral targets that had been selected by then, well before SARS-CoV-2 was on anybody’s radar.

Unhappy and feeling vulnerable- not surprising since the '069 and other IP may cover most if not all of Moderna's current pipeline- Moderna set out to challenge the validity of Arbutus patents in front of the patent courts.

In a great setback to those efforts, by upholding the ‘069 patent last week, the PTAB not only confirmed the validity of fundamental LNP claims, but strengthened them enormously to the point that Arbutus (the patent owner) would run little risk having its IP found unpatentable (not unusual in IPR proceedings like the one concluded last week) during an infringement lawsuit.

The ‘069 patent

In order for a biotechnology/product to infringe on a patent, it is sufficient that a single claim applies to the technology/product that is being monetized by the infringer.  Receiving government money specifically to develop and manufacture such product like Moderna did in April could be interpreted as such monetization.  More typically, however, it is the actual sale of pharmaceutical products that is viewed as an act of infringement.

The critical claim in the present controversy is as follows (color highlights are mine):

1. A nucleic acid-lipid particle comprising:

(a) a nucleic acid;

(b) a cationic lipid comprising from 50 mol % to 65 mol % of the total lipid present in the particle;

(c) a non-cationic lipid comprising a mixture of a phospholipid and cholesterol or a derivative thereof, wherein the phospholipid comprises from 4 mol % to 10 mol % of the total lipid present in the particle and the cholesterol or derivative thereof comprises from 30 mol % to 40 mol % of the total lipid present in the particle; and

(d) a conjugated lipid that inhibits aggregation of particles comprising from 0.5 mol % to 2 mol % of the total lipid present in the particle.

In order to infringe a ‘comprising’ claim, all individual elements need to be present in a covered product.  As detailed in Moderna’s recent publication in the New England Journal of Medicine on their phase I results with mRNA-1273 (Jackson et al), there is no controversy of whether mRNA-1273 contains all elements covered by the claim:

Messenger RNA -->  nucleic acid

Ionizable lipid --> cationic lipid

DSPC --> phospholipid

cholesterol

PEG2000-DMG --> conjugated anti-aggregation lipid

What then becomes critical is whether the percentages in the mRNA-1273 formulation fall within the ranges specified in the claim.  Curiously, this information is lacking in the publication, in sharp contrast to previous publications. Interesting!

But since Moderna is using a platform approach to developing mRNA therapeutics and vaccines and has stated that the development risk of mRNA-1273 is greatly reduced because it relies on already clinically tested LNP formulations, one could simply look up and compare the percentages used in other mRNA vaccine candidates currently being developed by Moderna.  This is just what I did by looking up the hitherto 3 most recent freely accessible mRNA vaccine publications by Moderna as listed on their website and where the ratios were explicitly detailed.

1)      HIV (Moyoet al, 2020)

Ionizable lipid: 50 mol % (i.e. within 50-65% stipulated in claim)

DSPC: 10 mol % (cf 4-10%)

Cholesterol: 30.5 mol % (cf 30-40%)

PEG-lipid: 1.5 mol % (cf 0.5-2%)

è The HIV formulation infringes on the ‘069 patent.

2)      RSV (Espeseth et al, 2020)

Ionizable lipid: 58 mol % (cf 50-65%)

DSPC: 10 mol % (cf 4-10%)

Cholesterol: 30 mol % (cf 30-40%)

PEG-lipid: 2 mol % (cf 0.5-2%)

è The RSV formulation infringes on the ‘069 patent.

3)      Chikungunya (Kose et al, 2019)

Ionizable lipid: 50 mol % (cf 50-65%)

DSPC: 10 mol % (cf 4-10%)

Cholesterol: 38.5 mol % (cf 30-40%)

PEG-lipid: 1.5 mol % (cf 0.5-2%)

è The Chikungunya formulation infringes on the ‘069 patent.

Similarly, since Moderna has referred to the clinical experience with their 1273 formulation, I finally checked on their latest clinical research paper.

4)      Flu (Feldman et al, 2019)

Paper references Richner et al 2017 paper for formulation details.

Ionizable lipid: 50 mol % (cf 50-65%)

DSPC: 10 mol % (cf 4-10%)

Cholesterol: 38.5 mol % (cf 30-40%)

PEG-lipid: 1.5 mol % (cf 0.5-2%)

è The flu formulation infringes on the ‘069

As you can see, the preponderance of evidence points to the fact that the covid vaccine candidate by Moderna infringes on ‘069.  Issuing a PR, as Moderna did on Friday, that it is not aware of an IP problem, of course is par for the biotech IP game, not only for public posture, but in particular to downplay the view that Moderna is willfully using somebody else’s IP.  If found to have done so after starting to commercialize the vaccine, this could lead to up to triple the amount the damages awarded to Genevant and Arbutus.  But then again, the motivation behind attempting to invalidate the patent right after it loses access to it following the Genevant-Acuitas settlement and hiding the lipid ratios in the NEJM paper will be obvious to any judge and should lead to the presumption of willful infringement.

What’s next?

In the typical biotech game, what would follow now is a last-ditch attempt by Moderna to still invalidate the patent by appealing the ruling.  Odds, however, are now strongly against Moderna that they will be able to reverse last week’s ruling.  In fact, the ‘069 is now stronger than ever and the above evidence will give them sufficient ammunition to sue Moderna on the presumption of infringement.  Appealing the decision would, however, buy Moderna some time trying to make their smaller adversary willing to settle for more favorable terms (following the Alnylam playbook).

This, however, is happening during a pandemic and IP-related tactical games may not be viewed kindly.  This also means that Arbutus would be well advised not to make a big public fuss out of what could be very valuable to them financially and continue with their low-key, matter-of-fact approach to the issue.

So in short, I don’t know when the issue will be resolved, but it certainly won’t be next week or month, but more likely at least after a first read-out of the phase 3 results before we hear about any resolution to the matter- most likely a settlement.

How much all of this is worth to Arbutus stock is anybody’s guess, too, depending, of course, mostly on the performance of mRNA-1273 in the clinic, the price Moderna could charge for its vaccine (note: Moderna is a proponent of whatever-the-market-will-bear) and whether covid19 vaccines will become an annual re-administration market.  Personally, the current share price of $5 is already justified by the promising HBV-RNAi phase I results disclosed in May (update pending soon) and the optionality from the fact that Michael Sofia from Arbutus, the inventor of the most impactful HCV medicine, polymerase inhibitor sofosbuvir, has now set his sights on inhibiting the SARS-CoV-2 and other coronavirus polymerases.

SNALP Structure Reconsidered

SNALP delivery technology has not only been the subject of a heated fight over ownership and control, but also continued mechanistic, chemistry, and structural investigations.  

In terms of structure, I used to think of SNALP as simple, unilamellar liposomes with RNAi triggers captured in their aqueous interiors.  A patent application by Tekmira (WO2012/000104A1) and a recent paper by the Cullis/AlCana group (Leung et al., 2012) convincingly challenge this view.  Accordingly, SNALP particles are highly electron-dense entities in which pockets of RNAi triggers surrounded by positively charged lipid micelles lipids fill a lipid-enclosed vesicle.  

This new view should guide the future development of the technology, especially the ratio space of its components, targeted delivery approaches, and cytoplasmic release.

Numbers hinting at need for new model

One metric of Tekmira’s results that has always impressed me were its RNAi trigger formulation efficiencies, often well above 90%.  As Leung and colleagues note in their paper, this is at odds with the old simple aqueous encapsulation model which would predict that the likelihoods of an RNAi trigger to be encapsulated or not are roughly equal.  

It now seems that Tekmira’s manufacturing method of rapidly mixing lipids and nucleic acids is at least one critical factor allowing for this remarkable concentration effect.  Other groups largely failed to reproduce such data at least partly because the published method called for the use of costly amounts of reagents.  This is changing, however, due to the use of microfluidic formulation methods as was also practiced by Leung and colleagues. This method likewise allows lipids and nucleic acids to be mixed vigorously and consequently also generates electro-dense structures, but using much smaller volumes.    

Limit-size particles smaller than anticipated

A consequence of the new model is that SNALP LNPs may be as small as 15nm in diameter, whereas previously I subscribed to the view that 40-50nm was the limit.  While a 15nm spherical particles would still have problems in passively getting out of the vasculature in most tissues, it is in a dimension where it might become of interest for additional tissue targets than those that we have assumed to be suitable SNALP targets. 

While it is nice to think that SNALP LNPs has more tissues that it might be able to address, it seems that the more tangible, near-term value of the new insights is in how the particles can be designed to target the existing low-hanging fruit tissues such as liver, sites of inflammation, phagocytes and other cells in the blood/lymph, solid tumors (by systemic delivery), and local applications such as lung epithelia by inhalation.

You could for example imagine that this model makes the post-insertion method of adding a targeting ligand more attractive compared to a co-formulation one (both of which have been considered). Similarly, the model changes the space of lipid and RNAi trigger ratios that should be explored.  It also requires new models for the cytoplasmic release of the RNAi triggers from the endosomes to be considered.

Who was first?

While I do not wish to carry on with pointing out potential tensions between Tekmira, Alnylam, and AlCana, the fact that both Tekmira and the Cullis laboratory (‘AlCana’) came out with essentially the same discovery does not leave me much choice but to briefly comment on the coincidence (note that this clash was set in motion before the settlement).

In the patent application by Tekmira, the priority date is June 30, 2010 2011.  As this is also the international filing date, the invention/discovery must have occurred at least a few months, if not at least a year earlier.  The Leung et al. paper was received by the the Journal of Physical Chemistry approximately 2 1 year later (April 5, 2012).  This suggests that the Tekmira were the first to have these insights.  This, of course, could also be critical for the patent application.  If Tekmira were able to get patent protection for such ‘non-lamellar’ LNPs, this could be an important one and somewhat replace in importance the Semple/Wheeler patent estate which are about to expire over the next years.

The new findings illustrate that SNALP LNP delivery continues to be an area a rapid progress.  One would make a mistake to assume that the MC3 formulations currently in the clinic are as good as it gets and that other somewhat overlapping, and therefore competing approaches such as Arrowhead’s DPCs are about to catch up.  Here’s hoping that with the leading delivery technologies having so much obvious room to mature, 2013 will be at least as exciting if not more so than 2012 for RNAi Therapeutics.