Description

Summary

RegenxBio is a unique situation within the biotech space such that it has around 75% of its market cap in cash plus the NPV of a royalty on a drug soon to be approved & commercialized — the rest of what you are paying (~$400-500mm) buys you a healthy internal gene therapy pipeline, royalties on 29 other partnered programs and option value for additional patterned programs as gene therapy research flourishes. The opportunity exists due to 1) mis-modeling of the royalty on AVXS-101, 2) likely unfounded concerns over IP, and 3) beta from the overall sector.

Please note this investment is truly what I would describe as a “cheap option/special situation” and there is above average risk involved. There is opportunity for a 5 bagger embedded here (with the downside case not draconian) and the skew is what attracts me rather than certainty of return on capital.

Idea

RegenxBio (“RGNX” or the “Company”) is the backbone to many developing gene therapy innovations. RGNX focuses on developing the delivery mechanism (vectors) that supplant “drugs” (genes) within patients. The Company leverages a lean internal pipeline across a vast 30 drug install base of partnered programs (including a royalty on a soon-to-be approved blockbuster drug) which partially de-risks the Company to individual binary events.

RGNX owns the rights to a proprietary suite of adeno-associated virus (“AAV”) vectors developed at the University of Pennsylvania, the most notable of which are dubbed AAV7, AAV8, and AAV9. The Company calls this suite of next-gen AAV vectors, collectively, as their “NAV Tech.” The gene therapy researchers at U Penn are largely regarded as the forefathers of gene therapy; as such, RGNX’s ownership of the IP developed there, has resulted in the Company becoming the leader in AAV gene therapy.

These exclusive rights place RGNX in a very unique position within the industry today as they are for the most part the best in class delivery mechanisms (“vectors”) currently being used for viral delivery of genes. Compared to earlier generations of AAV, AAV7+ vectors have better gene expression, generate a more muted immune response, and have improved manufacturability. As such, owning RGNX is akin to owning a suite of powerful (and patented) syringes, without which the delivery of a given drug would be nearly impossible.

Why own it today?

I think an inflection in the adoption of AAV based gene therapy (and as a result RGNX’s tech) in the medical community will soon come about with the first AAV based gene therapy approved a year ago  with many more to come as Phase I/II data read-outs are expected for partners & RGNX in 2019 with nearly 500 new cell & gene therapy INDs expected just in 2019 (up from 250 in 2018).

Furthermore, the stock has traded down around 35% since mid last year on 1) a significant drawdown in the broader biotech sector (the XBI ETF is down around 20% over the same time-frame) and 2) slight concerns over the validity of RGNX’s IP as it relates to new licenses. This is despite cash comprising 1/3 of the market cap and another 50% from the royalty NPV of a highly probable partnered drug.

Resultantly, you can purchase the Company today for just about its net cash position plus its royalty on AVXS-101 (the gene therapy drug for Spinal Muscular Atrophy now owned by Novartis through its recent $8.7bn acquisition of Avexis) giving little credit for its internal pipeline or vast portfolio of other partner companies. This setup essentially creates ReGenX as a call option on the proliferation of gene therapy — you are paying a stub value of $400-500mm (less cash and AVXS-101 royalty NPV) for $10b+ of potential value.

At a high level, I think of the opportunity here more simply in terms of the total market size of gene therapy as a class of treatment. As this market expands, with RGNX currently the owner of the best-in-class delivery mechanisms for in-vivo treatment, the value of RGNX’s suite of vectors will increase consummately.

Business Explained

I view the main business activity of RegenX as the licensing of its AAV portfolio. Licensing is done on a case-by-case basis, e.g. one license for using the AAV8 vector to deliver X gene for patients with Y indication. Such licensing contributes a high single digit to low double digit top-line royalty on sales of the resulting drugs using RGNX’s AAV technology. RGNX owes U Penn a low to MSD royalty on internal pipeline sales (I assume 4%) and a low double digit royalty on license revenue (I assume 11%).

Each additional license requires no additional capital on RGNX’s part, rather the capital burden is solely placed on licensees of the technology. This is the dynamic that truly excites me, given the proliferation of gene therapy as a treatment class comes about, RGNX will gain a meaningful fraction of the economics of all AAV based gene therapy treatments, without really any additional capital required.

Furthermore, as AAV based gene therapy becomes more ubiquitous, with the medical community (both physicians and insurers) more accepting of these radically innovative therapies, the value of both existing and future licenses will increase meaningfully – so both quantity and value of licenses of RGNX’s NAV tech will increase. As AAV gene therapy proliferates, expect RGNX to take a more meaningful fraction of the economics upfront (partially circumventing their mid to late 2020’s AAV patent expiries).

Gene Therapy Overview & Context

Understanding the situation requires some context around gene therapy in general. Gene therapy, at its core, is in many ways the pinnacle of modern medical science. The idea is simple at its core, a patient born of a genetic disorder can have their defective gene transplanted with a working copy – effectively curing the patient. The difficulty comes in gene targeting (i.e. gene x controls protein y expression) and delivery of the targeted gene within the patient (supplanting the working copy of the gene within the patient and fostering meaningful expression). While gene targeting will always be an area of development for the medical community, gene delivery has been improved vastly over time; today, I believe RGNX owns the best in-class delivery mechanisms (Appendix 1).

The first large step forward in gene delivery came with the idea of using viruses to deliver new copies of genes within patients – these worked well in terms of targeting because viruses are naturally optimized to deliver their own genes within cells (i.e. the copy and spread of the virus is a natural event). Experimentation began in the early ‘90s but was for the most part put on hold after the death of Jesse Gelsinger, who died from an immune response to the adenovirus vector (not to be confused with adeno-associated virus) that was used in the trial. Indeed, the largest issue with the idea of using viruses to deliver such genetic code is that of safety – with the body’s own immune system reacting too strongly to the injected virus.

Trials and tribulations were experienced by the medical community throughout the 2000’s with various other vector designs that either did not have efficacy or produced too strong of an immune response – rendering some technologies too harmful for experimentation, let alone FDA approval. Other viral vectors used consisted of adenovirus, herpes virus, and retrovirus – all of which did not fit the bill until the adeno-associated virus came along.

The use of adeno-associated viruses for use as vectors first began in earnest after the release of phase 1 clinical trial data in 2009 from experimentation on patients with Leber’s Congenital Amaurosis that showcased an excellent efficacy & safety profile of the AAV vectors. At the same time, RGNX was effectively formed out of a collaboration between the University of Pennsylvania and James Wilson.  Since then, about 70% of all AAV gene therapy trials have used RGNX’s “NAV” vectors (data from ’12 – ’14 and taken from U.S. gov clinical trial data base for new treatment INDs).

After years of clinical development, we are just now entering an inflection point in the success of AAV based gene therapy, as the aforementioned Leber’s Congenital Amaurosis research led to further development with Spark Therapeutics’ Luxturna for the disease just recently garnering approval on December 19, 2017 – the first gene therapy technology in the U.S. using AAV. Since then two additional gene therapies have been approved. This class of therapies now fully has the FDA’s support with Commissioner Gottlieb recently stating he expects the FDA to approve 40 gene therapies by 2022.

ReGenXBio NAV Technology

Without getting bogged down in the particulars into exactly why RGNX’s portfolio of AAV vectors are superior (of which researchers still do not have a great grasp of), it is a simple enough exercise to observe some of the magnificent data generated to date with these vectors. Observe the difference in the value of the right-most column “max expression” between AAV8 and the older gen AAV2 below (i.e. efficacy):

Additionally, a study comparing immune response (measured by the production of neutralizing antibodies, “NAbs”) in hemophilia patients in 2012 shows a clear standout improvement of AAV8 over prior generations (i.e. safety):

source: http://www.nature.com/articles/gt201190#f1

Prior to the research of AAV vectors and the development of the NAV technology by U Penn, most viral vectors presented very poor safety profiles for in-vivo treatment within humans as the viruses used as vectors were derived from

1. pathogenic viruses (meaning the types of viruses that cause disease in humans)
2. immunogenic viruses (cause a very significant immune response)
3. or viruses that otherwise promoted genomic toxicity (gene delivery takes place in such a way that it interrupts otherwise normal function of the cell)

Research eventually began on the first AAV vectors in the mid-1990s, but at the time only a few AAV viruses were known to exist. These early generation AAV vectors were largely limited in application for gene therapy due to:

1. low gene expression (the delivered genes were producing very little or no protein)
2. short term gene expression
3. poor tissue target-ability (gene expression not seen in the targeted organ or region, i.e. CNS or liver) very high levels of immune response (the subjects’ body readily recognized the vector being used for gene delivery due to pre-existing antibodies that serve to inhibit the therapeutic effect of delivered vectors).
   

At U Penn, researchers began in the early 2000s to find other AAV designs for potential use as vectors. This was done by a broad search within humans and non-human primates for existing in-tact AAV’s within cells. This search led to the discovery of over 100 new AAV vectors for potential use as gene therapy delivery mechanisms. The first of which were dubbed “AAV7,” “AAV8,” and “AAV9.” These designs were subsequently patented and initial clinical trials utilizing these “next generation” vectors began in earnest in 2010 (the rights of which are effectively owned by RGNX).

In addition to better specific targeting (e.g. AAV8 is extremely effective in transducing the liver and retina, AAV9 the CNS), these newly discovered vectors would be observed to possess much higher gene expression, induce a muted immune response, and generally have a higher ease of manufacturability.

Use of these new vectors, dubbed collectively “NAV technology,” progressed in the clinic as follows:

IP

RGNX owns a vast patent portfolio on its AAV vectors. The key investor debate currently is whether the exact specification around “sequence identity” claims will be upheld. As a brief explanation, the way RGNX effectively patents its vectors is by their DNA structure – how then would courts approach another company using a substantially similar (but not exactly the same) vector? As the US patents lay out currently, another vector would be seen as violating RGNX’s patents if it shares >95% of its sequence identity.

This is the language from RGNX’s U.S. AAV8 patent:

The term “substantial homology” or “substantial similarity,” when referring to a nucleic acid, or fragment thereof, indicates that, when optimally aligned with appropriate nucleotide insertions or deletions with another nucleic acid (or its complementary strand), there is nucleotide sequence identity in at least about 95 to 99% of the aligned sequences.

https://patents.google.com/patent/US9587250B2/en

There have been recent concerns over RGNX’s IP given its EU patent of AAV8 is undergoing a challenge – which has been initially successful in narrowing the scope of sequence identity to 99% rather than 95%. The ruling is under appeal – I don’t have a view here neither do I think this to be particularly material (in the event such a ruling does not set precedence for the US); I note an appeal should still leave a 3 year run-way or so of the claim still lasting as 95% sequence identity.

It’s possible this claim gets narrowed in a review of the patent in the US though I’d suspect there to be little risk to a near 100% sequence claim (which should protect their existing licensing arrangements and likely still attract new licensees).

This is how RGNX’s CEO eloquently describes this particular issue:

“So in the IP space, the way that we think about it is, plus or minus 5% of 750 is up to about 75 amino acid residues. That if there were changes to those residues, or that number of residues up to that, in the development of a new capsid, let's say a derivative capsid, it would fall under the umbrella of the intellectual property. So it's a long way of saying a few minor modifications to something like AAV9 or AAV8 is still going to fall within the capsid IP universe of what we licensed from GSK.”

So in the absence of meaningful risk to their IP, what is all of this worth?

RGNX Pipeline

RegenX is developing a few interesting projects in-house. I will briefly go through each in very limited detail below (please see Appendix 2 for valuation tables on each drug). I’m assuming an across-the-board 15% probability of approval for each project and will update as more data comes out.

• RGX-314 Wet-AMD
• Wet-AMD is a disease of the retina that causes vision characterized by abnormal blood cell growth within the retina. Such abnormality causes fluid build-up that leads to vision loss. The cause of the blood cell growth is the presence of a protein known as “vascular endothelial growth factor” (VEGF). There currently exist a few treatments (Lucentis and Eylea) that take the shape of VEGF inhibitors (and are administered monthly directly into the retina). There are close to 150k new cases in the U.S. alone per year (I model a current base of about 565k patients for potential treatment). Given the extent of debilitation this disease causes (vision loss) and level of new cases per year, Wet-AMD is as a result, a very large market at ~$7bn total. A base of patients requiring such frequent dosing as shown below (data from the HARBOR study on Lucentis) is clearly ripe for innovation.
• 
• Given such a large market size, limited data to date, and little disclosure on mechanism of action, I am fairly conservative in my modeling, and assume a very small ramp to a small mid-single digit penetration rate at a 15% probability of approval.
• RGX-501 HoFH
• Those afflicted with HoFH have a defective copy of the LDL receptor gene (“LDLR”). These receptors detect LDL cholesterol (“bad” cholesterol) and remove it from the body. Patients today, depending on severity, are treated with PCSK9 inhibitors or Novelion’s Juxtapid – older treatments consist of Statins, LDL Apheresis, and other concoctions which are very cumbersome processes and have very poor to modest results. Juxtapid is probably the best-in-class treatment here but is exorbitantly expensive ($295k) which has led to insurers mandating the trial of PCSK9 inhibitors (cost around $15k) and subsequent failure before reimbursing Juxtapid. PCSK9 inhibitors only work if the patient has some amount of function in their LDL receptors however – some have no function (about 10-20% of the HoFH population) and others would experience sub-par LDL reduction on just PCSK9 inhibitor treatment. A one-time gene therapy treatment is a very attractive alternative compared to the current treatments on the market.
• We will get more data this year, but the preliminary data is marginally positive, with 80% reduction in serum cholesterol in mouse models:
• 
• RGX-111 MPS Type 1 & 2
• For purposes of valuation, I am only including MPS Type 1 as I think it may be too aggressive (even using a probability adjusted NPV) to include both indications (type 1 & 2) for now given limited data — though Regenx is tracking type 2 studies a bit faster than type 1.
• MPS disease covers a very broad range of particular disorders (labeled MPS type 1 through type 9). MPS type 1 and 2 are the most prevalent (about 1/100,000 and 1/200,000 respectively). Even within each type, there are further classifications; the most severe form of type 1 MPS is called “Hurler” syndrome. Type 1 MPS patients are largely treated with Aldurazyme (collab between BioMarin and Genzyme) – but is currently only offered as an IV treatment and is administered weekly with rare but deadly side effects (risk of anaphylaxis (life-threatening allergic reaction) is on the label. Given the data is so early, I will not review it, but this indication is clearly a prime candidate for gene therapy innovation.

Collectively, RGNX’s internal pipeline amounts to ~$576mm PNPV, or just 33% of the market cap (detailed in Appendix 2 – I only value the most advanced pipeline candidates).

Partnered Programs

With 30 total partnered programs and 13 in the clinic (the most advanced of which is AVXS-101), RGNX possesses a significant amount of optionality. The TAM of the 8 most advanced programs in the clinic is roughly $14.4bn:

Going forward, I expect a significant increase in licensees of RGNX’s NAV tech as the existing partners steadily progress through the clinic. As a rough guide for valuing the partnered programs, I take the TAM less what is attributable to SMA (as I value this separately through Avexis) with similar math to the NAV Tech platform (40% peak penetration, 11% take rate, but a 15% chance of approval since the starting point is Phase I/II). This gets to around $340mm of value:

Platform Value

Valuing the potential for new licensees is inherently difficult. As a back of the envelope guide, the global TAM for genetic diseases is about $28bn, with about $13.7bn of this not already a part of RGNX’s partnered programs. I assume an average 40% peak penetration rate for gene therapy in their respective indications and a 7% probability of approval (from preclinical to approval) with RGNX’s NAV Tech capturing 75% of this at an 11% take-rate for roughly $158mm in platform value:

it is  difficult to ascertain the future economics of royalties as I would expect the company to steadily transition to an upfront fee model as they approach their mid-2020’s patent expiries. This estimate is therefore likely conservative

Novartis (Avexis)

Since the Avexis drug is currently the most advanced gene therapy program within RGNX’s partnered portfolio and the market ascribes such a high probability of success to it currently, I will discuss this drug & company in particular in Appendix 3. Important to note here is I am assuming there is a patent term extension for AAV9 (this covers the Avexis royalty) such that RGNX will retain royalties for a few years post 2026 (the expiry of the patent) – this is a function of awarding patent extension to account for time spent in the clinical trial process (you get 50% of the time between IND and NDA filing plus the FDA review time, or typically around 3.5 years for most drugs). This is a key driver of additional value vs. what most people model.

Valuation

Below is how it all stacks up, I estimate 31% upside giving effect to probability adjustments:

Note: I have not incorporated milestones that RGNX is due to receive upon certain clinical, regulatory, and commercial events (other than $80mm due from Avexis) due to lack of specific disclosure by the company. In aggregate, the Company is due roughly $300mm in clinical, regulatory, and commercial milestones.  

Risks

• All values are probability adjusted; naturally there is significant clinical risk and any unforeseen challenges with RGNX’s AAV tech or gene therapy as a whole could impair value permanently across the portfolio.
• Avesis regulatory / commercial outcomes  
• IP challenges
• Manufacturing issues with the AAV vectors.

Appendix 1: AAV Gene Therapy Graphic

Appendix 2: Pipeline Valuations (Values in USD ‘000s)

Appendix 3: Avexis

Novartis acquired AveXis in May of 2018 for a chunky $8.7bn. This move validates the gene therapy space but also importantly puts a full commercial team behind AVXS-101 which should support an incredible ramp (and resultantly make RGNX’s ~11% royalty on this drug a cash cow). This is my AVXS-101 model as it relates to the economics for RGNX (as a sanity check, I estimate $897mm of value from a low double digit royalty on a single-drug company taken out at $8.7bn):

I could wind up being high on my pricing estimates (Spinraza currently has a net $2.25mm price).

The current leading therapy on the market for SMA is the recently approved Spinraza which is inferior to AVXS (though a nice drug in its own right). I’ll detail the differences in brief below:

Spinraza works differently than AVXS-101, it targets SMN Gene 2 protein production where AVXS-101 attempts to supplant a working copy of SMN Gene 1 into patients. SMN gene 1 produces much more SMN protein, but patients afflicted with Spinal Muscular Atrophy are totally deficient of their SMN 1 protein producing genes. Spinraza fixes this, as you can see in an illustration below by juicing their working (but sub-par in terms of total protein generation) SMN 2 gene.

This means that you would theoretically want to take both drugs, as an SMA patient, to have maximum recoverability from this terrible disease. SMA kills infants; often with less than a 12-month lifespan (for type I). Insurers are as a result in a tough situation having to cover these patients for maximum chance of survivability. My view is that when approved, patients will be given AVXS-101 first and likely taken up on Spinraza over time (i.e. the standard of care will be AVXS-101 supplemented with Spinraza).

AVXS-101 is a great example, so far, of the power of the AAV9 vector and its ability to cross the blood-brain barrier (which is a large item of debate given the size). Instead of the yearly intrathecal (spine) booster injection required of Spinraza, AVXS-101 is a one-time injection.

Mako Research (essentially reposted by Citron on his site) put out a short report on Avexis some time ago mostly predicated on the idea that the lead researcher of the study, Jerry Mendell, has a dubious past and that given the fallibility of the study design (single arm, no placebo) as well as the recent Spinraza approval – that Avexis would have a very tough time further researching AVXS-101 (or worse, it is a total fraud).

I think the fact that Novartis acquired AveXis probably serves largely to dispel this thesis, but simply put the data speaks for itself; without therapy most of the babies (shown as horizontal lines below) would unfortunately not have likely lived beyond 20 months or so. In the data shown below (which is a tad dated now), 9/9 subjects reached 20 months totally event free:

This piece drives the majority of RGNX’s value and I assume 90% probability of approval (for both type 1 and type 2). I note that prior to its acquisition, consensus was AVXS-101 would be approved (and the $8.7bn price tag sets a high probability on this outcome as well). Value would be materially impaired if AVXS-101 is not approved, though I find this outcome unlikely.

Appendix 4: Sarepta DMD litigation option

Sarepta uses the AAVrh74 vector for its DMD gene therapy. Rh74 is apparently 98.9% identical to AAVrh10 which is patented by RGNX (I say apparently as this comes from a patent lawyer focused on AAV that Leerink did a conference call with – I’ve not independently confirmed this claim yet). It’s possible, resultantly, that Sarepta’s DMD gene therapy violates RGNX’s patent. RGNX has not brought about any litigation (or made any indication it will do so) yet – but I note this would likely only happen if Sarepta’s DMD gene therapy is commercialized.

Sellside assigns about $1-2bn of probability adjusted peak sales to Sarepta’s DMD gene therapy, resultantly there is a couple hundred million of option value here – I do not model this.

Disclaimer: This memorandum is for discussion purposes only and is not intended to be, nor should it be construed or used as, financial, legal, tax or investment advice or a general solicitation. This memorandum is as of the date posted, is not complete and is subject to change. The data contained herein are prepared by the author from publicly available sources and the author's independent research and estimates. Certain information has been provided by sources believed to be reliable, but has not been independently verified and its accuracy or completeness cannot be guaranteed and should not be relied upon as such.

Catalyst

• Potential approval of AVXS-101 in May 2019
• Current partner readouts throughout 2019
• Further partnership announcements
• Acquisition by strategic
• WetAMD cohort 4 data in 2Q19
• Internal pipeline advancements
• Other non-valued pipeline asseets advance in the clinic