Executive Summary and Investment Thesis
Abeona Therapeutics (NASDAQ:ABEO) is a biotechnology company with a number of early- and late-stage assets. This week the company will present initial data on its lead compound, ABO-102 for Sanfilippo syndrome at the WORLDSymposium (We're Organizing Research on Lysosomal Diseases) in San Diego (Feb. 13-17).
The validation of a biotechnology company's key technology is usually a defining moment in the evolution of that company and usually creates significant value for the company's shareholders. Like other gene therapy companies with unproven technology, Abeona trades at a substantial discount to those companies with validated technology, and I believe that validation of Abeona's lead technology could drive the company's share price toward $30/share, which would put its valuation in line with those companies that have validated technology - e.g., AveXis (NASDAQ:AVXS), Spark Therapeutics (NASDAQ:ONCE), and Ultragenyx Pharmaceutical (NASDAQ:RARE).
Longer term, I believe ABEO has the technologies that could be worth more than $100 per share, but the purpose of this report is to preview the imminent clinical catalyst, and readers should refer to previous work for the full company background and the longer-term investment thesis.
I note that a number of highly sophisticated biotech hedge funds have recently appeared on the shareholder register, there has been considerable insider buying over the past 18 months, and in recent weeks, a substantial number of March call options have traded.
As with all biotech investments, the outcome of an investment is very dependent on clinical trial outcomes, which is speculative, and the key company specific risk in Abeona is the failure to commercialize any of its clinical programs. Cash on balance sheet of $72m will provide support for the shares at $1.75 if clinical trials fail. That said, this company is investigating drugs for ultra-rare terminal diseases where there is currently no approved agent, and this means the hurdle for approval at the FDA is very low, and the clinical trial data to date is highly encouraging. With cash burn of about $1m per month, the company has over three years of cash burn on balance sheet, so there is no need for near-term funding.
Clinical Validation of a Biotech Company's Lead Product Typically Creates Substantial Upside for Shareholders
One of the most defining points of time in a biotechnology company's evolution is the moment that proof of concept is demonstrated with clinical trial data in humans. As shown in the exhibit below, gene therapy companies with clinical candidates that have been validated with human clinical trial data trade at market caps seven-fold and enterprise values eleven-fold greater than those companies that have yet to validate their technology in humans. Indeed, AveXis has seen its share price increase 230% during the past 12 months as it has generated additional human data that supports the company's technology. I would argue that Abeona is perhaps 6-12 months behind AveXis in its evolution, and it is therefore likely that the presentation of human clinical trial data for ABO-102 at the WORLDSymposium on Thursday, Feb. 16 (at 4:30 EST), will likely be a transforming event for Abeona and its shareholders.
AVXS: + 230% Since IPO, Now a $1.6B Market Cap Company
In this report, I will review the science, suggest what I believe will be required for the FDA approval, and successful commercialization and the approximate valuation for ABO-102.
ABO-102 is a next-generation adeno-associated viral (AAV)-based gene therapy for MPS III (Sanfilippo syndrome), which involves a one-time delivery of a normal copy of the defective gene to cells of the central nervous system with the aim of reversing the effects of the genetic errors that cause the disease.
Adeno-associated virus is a non-enveloped virus with a single-strand linear DNA genome (4.7kb), which has not been associated with any human/animal diseases (reviewed in Samulski et al, Annu Rev Virol, 2014, 1:427). Transgenes delivered by AAV were able to achieve long-term in-vivo expression in multiple organs (eye, brain, spinal cord, muscle and liver) in preclinical models. AAV enters cells through a series of steps including cell surface binding, intracellular trafficking, second-strand synthesis and persistence as episomes.
Source: Company data
There is a great video here to help explain how the technology works.
As the first step of the cell entry, AAV binds to cell surface sugars (such as heparin sulfate, sialic acid and galactose) on proteoglycans and cell surface receptors (such as fibroblast growth factor receptor, integrin and hepatocyte growth factor receptor). It has been shown that the use of different cell surface sugars/receptors could lead to different tropism. The best-studied serotype AAV2 binds to heparin sulfate and fibroblast growth factor receptor 1 (FGFR1) and demonstrated a broad tropism with efficient transduction in liver, kidney, muscle and central nervous system. In comparison, AAV4/5/6 bind to sialic acid and co-receptors other than FGFR1 and showed tropism in tissues including retina, lung and heart (Nonnenmacher et al, Gene Ther 2012, 19:649).
Upon binding to cell surface sugars/receptors, AAV is internalized through endocytosis of clathrin-coated vesicles and disseminated to nearly every cytoplasmic compartment. In two hrs post-infection, most of the AAVs localize around the nucleus, where they escape from the endosomal compartments in response to the acidic environment. The exposure to the acidic environment was shown to be essential for subsequent AAV transduction (at least partly) through the activation of the phospholipase A2 activity and nuclear localization signals of the minor capsid protein VP1 (N-terminal end). The AAV capsids eventually enter the nucleus and release its DNA through unknown mechanism.
As a double-stranded DNA is required for transcription and gene expression of AAV vectors, the second-strand synthesis is essential to enable effective gene expression (the hypothesis of annealing positive and negative strands was invalidated as AAV particles containing only one strand is equally infectious). It has been shown that in-vivo AAV transgene expression took weeks to months after infection to reach the highest level (Flotte et al, Hum Gene Ther 2011, 22:1239).
With the inactivation of one of the AAV inverted terminal repeats (ITRs), the AAV genome could be replicated as inverted dimer and self-anneal after the release of the DNA in the nucleus. The self-complementary AAV vector has been applied successfully in clinical studies including scAAV2/8-LP1-hFIXco in hemophilia B, which showed immediate transgene expression (Nathwani et al, N Engl J Med 2011, 365:2357). The self-complementary AAV vector, however, has only half of the packaging capacity.
Preclinical studies have demonstrated that most AAV vectors are converted into circles to form episomes (DNA that can be replicated independently of the host), which contain several copies of the transgene (ligated in the order of head to tail). There is a low frequency (up to 1% in liver, lower in brain/muscle) that the AAV vectors can integrate into the host genome. The transgene expression is maintained mainly through the persistence of episomes, therefore AAV vectors are particularly suitable for gene delivery in non-dividing somatic tissues such as eye, brain, muscle and liver (adult).
Abeona is Using AAV9 (Double Stranded) for its Vector
Sanfilippo syndrome has a significant CNS component to the disease, and it is imperative that the virus is able to penetrate the blood brain barrier, persist in the central nervous system and deliver its genetic payload to the patients' cells. As shown below, AAV9 appears to have the greatest ability to enter and persist in the central nervous system. The diagram below shows the results of a brain-directed injection of AAV vectors encoding green fluorescent protein gene (AAV/GFP). In this particular experiment, approximately 2.0x10 ^10 vg of AAV/GFP vectors (serotypes 1,2,5,8,9 and 10) were injected into the right striatum over a period of five mins and expression of GFP was analyzed using fluorescent microscopy at two weeks.
Source: Company data
To further enhance the potency of its vector, Abeona is using a double-stranded (self-complementary) AAV as opposed to a single-stranded AAV. Self-complementary vectors can facilitate gene transcription and translation without relying on the infected host's replication machinery, which could hinder therapeutic efficacy. The prime limitation of scAAV vectors is their size. They can carry DNA pieces up to ~2.4kb^9. scAAV vectors are 10-100 fold more efficient than traditional single-stranded (ss) AAV vectors, as demonstrated in the pictures below:
Source: Company data
Preclinical Trial data Supports Mechanism and Validates the Science
There is a substantial amount of preclinical trial data on Abeona's AAV9 gene therapy approach to treating MPS IIIA and B. As shown below and originally recorded in Fu H. et al., Molecular Therapy (2011) 19(6):1025-1033, untreated MPS IIIB mice have a limited 8-12 month life span. Treatment of four- to six-week-old animals with either low (AAV9-L) or high dose (AAV9-H) gene therapy resulted in 16-28 months of survival. Wild type mice live on average 24 months.
Source: Company data
The results below are for mice with MPS IIIA.
Source: Company data
Abeona has already presented very preliminary data in October 2016. Low-dose ABO-102 (5E+12vg/kg) led to 58/26% reductions in urine/CSF GAG levels and 18% reductions in liver/spleen volumes at 30-day follow-up. Of note, the urinary and CSF GAG reductions appear comparable to other ERT data (57-75% urinary GAG reduction and 26% CSF GAG reduction). Since liver/spleen volumes remained 2x normal size over 1yr in natural history studies, the 18% reduction observed in ABO-102 Ph1/2 study looks encouraging. The safety profile appears good with no treatment-related AEs or SAEs (all 3pts received 3mon prophylactic tapering-dose steroid), and the Data Safety Monitoring Board has allowed enrollment of the high-dose cohort. The table below shows the magnitude of GAG changes seen in currently approved drugs for other MPS diseases that are currently marketed by Shire (NASDAQ:SHPG), Sanofi (NYSE:SNY), BioMarin (NASDAQ:BMRN) and Ultragenyx.
Source: Company data
What Data Might We Expect on Thursday Feb. 16?
The presenter is likely to present three-month follow-up data from the three initial subjects and six-month follow-up data from the first two subjects. This will allow investors to gauge the durability of the response seen at one month and to gauge the clinical outcomes at six months.
During the presentation of the initial two subjects at six months, it is likely that the presenter will report neurocognitive data which is likely to be the primary clinical endpoint upon which the FDA will review the BLA (that said, changes in GAG, and liver and spleen size are also clinically relevant). It may be possible to compare this data to the results of the natural history study.
The principal investigator has conducted a natural history study (see K.V. Truxal, et al., A prospective one-year natural history study of mucopolysaccharidosis types IIIA and IIIB: Implications for clinical trial design, Mol. Genet. Metab. (2016)) so it will be possible to compare the neurocognitive data to the untreated population in the natural history study. It is likely that there will be three main neurocognitive tests analyzed.
Leiter - 3 International Performance Scale
For cognitive assessment, the Leiter-3 has several features that are useful in this population. As it measures nonverbal IQ, it does not require participants to comprehend auditory instructions or provide any verbal responses - critical test characteristics for assessing this population. In the natural history study, two subjects between two and three years of age began the study with a Leiter-3 (or converted Leiter-R) NVIQ score N85, which is within the normal range. Consistent with the understanding of disease pathogenesis, this observation suggests that an earlier successful treatment could result in preservation of normal cognition. However, with the Leiter, subjects tended to score at the floor of the measure starting around six to eight years, which limits the utility of this measure to detect further decline at this stage of disease progression. In addition, and particularly near the test floor, fluctuations in attention and cooperation affected scores if a subject was unable on that particular day to attend to the test long enough to receive a score.
In the natural history study, a majority of subjects (N=16) were able to receive a NVIQ standard score at two consecutive visits, allowing for assessment of change over a six-month interval. For statistical analysis, mean change was assessed in these 16 subjects using the first six-month interval for which a subject was able to be scored at both time points (i.e., either baseline to Month 6 or Month 6 to Month 12). The mean decline in Leiter-3 NVIQ score over six months was 8.625 points (N=16, p=0.002). The mean decline in NVIQ score over six months of 8.625 points is considered a reliable change beyond what can be explained by measurement error alone, as the Reliable Change Index in the normative population is ≥6.43 for children three to six years old or ≥6.76 for children 7 to 11 years old, using a significance level of pb 0.05
Intra-subject change in Leiter-3 nonverbal IQ over six-month intervals (N=21). Normal range (shaded, partially shown on graph) is 85 to 115. The lowest score possible is 30. Blue circle(s) connected by line(s) = subject with MPS IIIA, red square(s) connected by line(s) = subject with MPS IIIB.
Source: Company data
Other Neurocognitive Tests
In the natural history study, the Mullen Scales of Early Learning also demonstrated significant change in age-equivalence scores over the course of the study in three of five domains (Visual Reception, Fine Motor, and Expressive Language), and the decline neared significance in the Receptive Language domain.
For behavioral assessment, the investigators of the natural history study found the Vineland Adaptive Behavior Scale to be measurable across the entire subject population, as expected from a parental interview that measures adaptive living skills. A decline in the Vineland composite standard score correlated strongly with chronological age in this entire cohort, with a significant mean decline of 5.09 points over a six-month interval (p=0.0004) that notably does not differ for those below versus above seven years old in contrast to the Leiter NVIQ standard score.
Below are the results of the Intra-subject change in Vineland composite standard score over six-month intervals (N=25) found during the natural history study. Due to late addition of the test and subject withdrawals, three subjects were assessed at one, 16 subjects at two, and six subjects at three time points. Normal range (shaded, partially shown on graph) is 85 to 115. The lowest score possible is 20. Blue circle(s) connected by line(s) = subject with MPS IIIA, red square(s) connected by line(s) = subject with MPS IIIB.
Source: Company data
Anecdotal Evidence Suggests that ABO-102 May Have an Effect on Neurocognitive Function but We Await Data on Feb. 16
As I mentioned in a previous report, it is possible to track the first patient in real time. The family keep a Facebook page (Saving Eliza O'Neill), and the family have appeared on numerous television shows and in various newspapers and journals. Eliza's story is pretty incredible. In order to ensure that Eliza was eligible for enrollment into the clinical trial, the entire family had to isolate themselves for 726 days (they could not allow Eliza to come into contact with the virus that is being used to deliver the gene). Her nine-year old brother, Beckham had to be home schooled so that he wouldn't come into contact with other people.
Eliza received the first treatment in the clinical trial in May 2016 and has just passed her six-month follow-up, and we hope to hear about six-month results in 1Q 2017. Anecdotal evidence from the Facebook page suggests that she is doing very well. In fact, Eliza had lost the ability to speak in April 2016, but now she appears to be able to say basic words. The family have also been on numerous television shows and in magazines and journals - it really is an amazing story. She appears to have had a great seventh birthday, is enjoying her first year at school (she is attending a special school for autistic children), and Beckham (who should be awarded "the best big brother in the world" award) spends much time talking about how Eliza has changed following treatment and how she no longer grabs his iPad from him! Based on this anecdotal evidence, perhaps ABO-102 is having an effect on Eliza's neurocognitive function.
The Path Forwards and the Approval Process
The current clinical trial has already enrolled three patients that have received the low dose formulation. An additional six patients will receive the higher dose with the first patient receiving treatment on or around Jan. 24, 2017. This will therefore be viewed as a multicenter clinical trial.
It is likely that the company will have >20 patients of data with six-month follow-up by the end of 2017. At this time point, there will also be three patients with >12 months' follow-up. As shown in a previous exhibit, most BLA submissions for other MPS diseases have involved trial populations of around 20 patients, and we would be surprised if the FDA did not accept such a data package for regulatory review and subsequent approval. It therefore appears likely that Abeona will receive approval around mid-year 2018 for ABO-102.
There have been two gene therapy products approved in Europe.
Strimvelis is the first ex-vivo stem cell gene therapy to treat patients with a very rare disease called ADA-SCID (Severe Combined Immunodeficiency) due to adenosine deaminase deficiency, a rare disorder caused by the absence of an essential protein called adenosine deaminase (ADA), which is required for the production of lymphocytes. Children born with ADA-SCID do not develop a healthy immune system so cannot fight off everyday infections, which results in severe and life-threatening illness. Without prompt treatment, the disorder often proves fatal within the child's first year of life. ADA-SCID is estimated to occur in approximately 15 patients per year in Europe. Strimvelis was approved in 1H 2016 following a clinical trial involving 22 children.
Glybera is a gene therapy that is designed to restore the LPL enzyme activity required to enable the processing, or clearance, of fat-carrying chylomicron particles formed in the intestine after a fat-containing meal. The product consists of an engineered copy of the human LPL gene packaged with a tissue-specific promoter in a non-replicating AAV1 vector, which has a particular affinity for muscle cells. Ahead of EMEA approval, Glybera was studied in 27 patients with lipoprotein lipase deficiency on a low-fat diet. The majority of patients who received Glybera also received immunosuppressive treatment. The main measures of effectiveness were the reduction in blood fat levels after meals and the reduction in the number of pancreatitis attacks.
The Commercial Opportunity
The incidence of MPS III is estimated to be ~1/70,000 births (Meikle et al, JAMA 1999, 281:249; Heron et al, Am J Med Genet 2010, 155:58). It is also estimated that there are approximately 2,000 Sanfilippo type A patients in the world with approximately 60 births in the US each year and about 100 in the developed rest of world (Source: Wikipedia and industry websites). Unlike primary care illnesses such as cardiovascular illnesses, these rare genetic diseases have very high levels of penetration and reimbursement, and it is quite likely that c. 80% of patients will receive treatment. The company has not disclosed the price it intends to charge, however, other gene therapy companies such as BioMarin have suggested a gene therapy treatment cost of $1.5m per treatment is where they intend to price their gene therapy products for similar diseases and this appears to be a sensible assumption, in my view. This is also considerably lower than lifetime treatment costs for diseases such as Gaucher's disease (c. $5m), Hemophilia (c. $3m) and Hunter syndrome (c. $5m). Glybera (the first ever gene therapy product approved) was launched in Europe at $1.4m per patient.
Based on these assumptions, this product could potentially generate $4.4bn of revenues in its first 10 years. On its recent Q3 conference call, the company suggested that the cost of each dose will be approximately $25k, and it is in the process of constructing a manufacturing facility in Cleveland. Abeona will also need to finalize the financial terms with REGENXBIO (NASDAQ:RGNX) which owns the rights to the AVV9 vector. REGENXBIO's patent that protects the AAV9 vector expires in the early part of next decade (c. 2023). Looking at precedent transactions suggests that Abeona will have to pay a royalty in the low-double-digit area. Lysogene (private) received exclusive licensing from RGNX for AAVrh10-based gene therapy in MPS IIIA in Dec. '13, with financial terms including 1) $0.5M initial fee, 2) an annual maintenance fee, 3) up to $7.75m milestones per product and 4) mid-single- to high-single-digit % royalty. Esteve (private) received non-exclusive licensing from RGNX for AAV9-based gene therapy in MPS IIIA in March '14, with financial terms including 1) $0.5M initial fee, 2) an annual maintenance fee, 3) up to $8.5M in milestones per product, and 4) mid-single to low-double-digit % royalty.
This suggests that gross profit derived from ABO-102 during its first 10 years will be over $4bn. There are minimal selling and marketing costs associated with this disease because it is so rare and patients are very aware of the disease. If we discount this $4bn of cumulative profit back a few years, then $3bn probably isn't far from the correct number of where fair value will reside, once the product has been approved. Take a 50% haircut for an unapproved product and it isn't unreasonable to think that the mkt cap should reside in the $1.5bn region fairly soon ($30/share). This is also close to the value where other gene therapy companies with validated clinical data trade (Spark Therapeutics, bluebird bio (NASDAQ:BLUE), AveXis, Ultragenyx and Alexion (NASDAQ:ALXN)).
Valuation and Target Price
As with all early-stage biotech companies, it is very easy to get carried away with potential valuation etc. under blue-sky outcomes. However, giving the company $3bn of value for ABO-102 when fully commercialized. In addition, as we have previously written, there are a number of other technologies that may provide upside to >$100 per share over the next several years, however, this report focuses on ABO-102 and Abeona's Sanfilippo gene therapy franchise.
This is an unprofitable biotechnology company with cash on balance sheet of $72m. The key risks to an investment in Abeona is clearly the failure of the company to commercialize any of its clinical programs. Under such a scenario, cash on balance sheet is approx $1.75/share.
Disclosure: I am/we are long ABEO, ALXN, RGNX, AVXS, RARE.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article.