Pullan's Pieces #164  
October 2020
 
 
 
 
www.PullanConsulting.com
linda@pullanconsulting.com
1-805-558-0361
 
 
 
 
 


Dear Reader,


Ready for BioEurope?  It is looking like a very productive partnering forum, even if only digital.  And looking beyond that, I'm hoping that the Biotech Showcase can create the great hub during what has been JPM week to start off January.   


If you did not catch the great webinar panel on New Paradigms in Oncology Oct. 8th, you can listen to the recording.  There were some great nuggets on trends and what may be coming next.


Free.  And if you sign up, you can later get the white paper (when we get it put together).    

https://resources.sharevault.com/webpanel_paradigm-changing-technologies-in-oncology?utm_source=pullan&utm_medium=email&utm_campaign=WBR-78



Thanks,

Linda

 
 
 
 
In This Issue

  • Linda:  The clear, crisp story versus the real-world of translation
  • Linda:   Musculoskeletal Disorders, a snapshot
  • Jessica:  What was that Dyno-mite deal?
  • Trevor:  Originators of 2019 Approvals
 
 
 
 
 
Linda:  The Clear, Crisp Story versus the Real-World of Translation
 
 
 
 


Partners and investors love a clear story where the translation from target to drug impact is clear.  This chain of logic is something like:

  • The target is important in human disease
  • Our drug binds the target selectively
  • And inhibits its function in vitro and in vivo
  • With a practical half-life and biodistribution  
  • To produce a benefit in a disease model
  • With an acceptable safety profile
  • With a biomarker to select the dose in humans
  • And a good patient selection strategy to maximize our chance of success.
 
 
 
 
It is somewhat analogous to moving a boat uphill thru locks.  When you build up the evidence at the first step, you can move up to the next, creating a process to climb up to a clinical trial.  
 
 
 
 
But in the real world, that can quickly become more complex.

I love the article on target validation challenges by William G. Kaelin Jr, in Nature Reviews 17: 441 (2017).  

Some of the real-world questions are:  
  • Is the target validation causation or association?
  • Is the target suitable for intervention, is it an ongoing driver of disease, or is it merely the “on switch”?
  • Is intervention sufficient or just necessary (needing polypharmacy)?
  • Does the target have pleiotropic effects or compensatory regulation or compartment translocation?
  • Do you hit other targets that you have not explored? 
  • Is the assay an up assay, less easy to have artificial interference in than a down assay?
  • Are the effects reproducible (under the same conditions)?
  • Are they robust (working under different conditions) or context-specific?
  • Are the animal models really representative of the disease?

 
 
 
 
So the translation of target to trials is a process of gathering data across many questions with more breadth and depth than the clear simple logic of target, drug, and effect.  With a leaky bucket of evidence, it can be tough to fill up the evidence for each gate, to move up the series of logical lock gates (working my analogy hard).  The chasing of all the leaks complicates the telling of the drug’s story in a clear crisp manner and complicates the decision-making for the potential partners as they worry about the unknowns. 

Perhaps the key is to try to keep returning the clarity of the simple logic chain at the core of your story, after each necessary exploration of complexity.  
 
 
 
 
 
 
Infographic
 
 
 
 
 
Jessica:   What was that Dyno-mite deal?
 
 
 
 


In case you missed it, amidst the hustle of a fully virtual Cell & Gene Meeting on the Mesa earlier this month, news of a big deal dropped on the 14th of October.  Roche is partnering with a fledgling start-up for “up to $1.8B”.  It’s a discovery deal with Dyno Therapeutics, just 5 months after they emerged from stealth mode with a launch pad of $9M.  Whoa!  What technology is so valuable so early?  AI-enabled discovery of next-generation AAVs.  Huh?  In other words, using artificial intelligence (AI) to identify newer, better Adeno-Associated Viral Vectors (AAVs).  Still perplexed, read on:


AAV

Let’s start with the gene therapy piece first, the AAVs.  Adeno-Associated Virus is so-called because of when it was first isolated in the 1960’s it was believed to be a contaminant of an Adenovirus preparation, true story.  AAV-based gene therapies have been gaining in popularity over the last decade as the therapeutic modality of choice for the treatment of monogenic diseases – diseases for which the modification of a single gene can be therapeutically beneficial.


The first AAV-based medicine to achieve marketing authorization is Glybera® (uniQure), approved in 2012 by the EMA (but withdrawn from the market 2 years later).  The first AAV-based product to be approved by the US FDA is Luxturna® (Spark Therapeutics, now Roche) in late 2017 and the second AAV-based medicine to gain marketing approval from the US FDA is Zolgensma® (Avexis, now Novartis) last year, in 2019. 


With the Roche partnership, the Dyno team will be aligned with the Spark team to develop new AAV vectors while overcoming some of the challenges of the existing vectors.  But there are 2 products on the market currently, how challenging can the development of AAV-based medicines be?  The struggle is real.

 
 
 
 
 
 
 
 


There are scientific and clinical challenges associated with the development of AAV-based gene therapies:


Scientific/manufacturing Challenges


  • Limited carrying capacity – the length of the transgene allowable in an AAV capsid (shell, so to speak) is typically 4-5Kb.  If the transgene of interest is larger than this the developer will need to re-distribute the transgene into different pieces which is generally achieved by 2 overarching approaches (each with unique sub-strategies):

  • Fragment AAV (fAAV) - heterogenous single-stranded genome fragments can be packaged in a single capsid.  As long as host cells are transduced with multiple particles, the appropriate fragments will likely come together in the host cell, relying upon host cell DNA synthesis to construct the full-length transgene. 

  • Split vector AAV - each cell can be c-transduced with vectors containing specifically designed pieces of the transgene, which can be brought together via homologous recombination or non-homologous end-joining in the host cell. 

  • Partial/empty capsids – Remember the “I Love Lucy” episode at the chocolate factory?  Packaging of virions from producer cells – not that different – is highly error-prone, and can result in partially-filled or empty capsids in the viral vector preparation.  The presence of partial or empty virions can be problematic for many reasons including the need with manufacturing via fAAV (see above) for much higher doses of viral particles to ensure enough complete vectors are administered to the patient, and the risk that empty capsids can elicit an immune response to the capsid proteins


Clinical Challenges


  • Immunogenicity – Approximately half of all humans have been exposed to, and therefore have neutralizing antibodies to the most common AAV capsid proteins used for gene therapy.  If they did not already have antibodies, on re-dosing the patient will likely have developed antibodies to the first dose.   Capsid-specific antigen expressed on the surface of transduced cells (“infected” with viral vector) can result in an adaptive immune response (CTL, cytotoxic T-cell).  Cell-free double-stranded RNA (dsRNA) can also elicit innate immune response.

  • Tissue tropism – at least 12 natural serotypes, and over 100 variants (see Nature review link above) have been isolated or studied as vector candidates









So, yes, developing AAV-based medicines is challenging.  Strategies for transgene packaging/distribution, combatting immunogenicity, and optimizing serotypes for the given indication are all significant barriers to overcome in the development of AAV-based therapies.  How can artificial intelligence help?


Artificial Intelligence (AI)

In the field, and in the Dyno deal, AI is really referring to machine learning.  Machine learning is the discipline of teaching computers to identify patterns in data sets and subsequently make predictions.  This is accomplished by feeding the computer training data.  As with manufacturing: the higher quality the inputs (data) going in, the better the product coming out. 


In healthcare right now, there are concerted efforts to use machine learning to predict disease diagnoses.  For example, the training of computer programs to recognize concerning regions in radiology images in order to direct the attention of radiologists.  Importantly, the goal is not to eliminate the role of physicians, but rather to improve the “signal-to-noise ratio” or, in other words, to reduce the error rate.  The AI in healthcare can also be directly applied with the patient.  There are many devices and apps that can collect data (usually from smart devices) which can be used to monitor and signal patients, and/or their physicians, in order to control chronic conditions (eg, Diabetes) or identify signs of impending high-risk events (eg, heart attack). 


In academic research and drug development, AI is augmenting the power of bioinformatics (collections of complex biological data).  Applying machine learning to new or existing data sets has been very powerful for identifying new drug targets and/or product candidates.  It is also being used to predict outcomes as a method to test the suitability of a drug candidate “in silico” or in the computer.  The benefit here is to narrow the list of candidates for advancement to the more expensive studies in the lab and/or with animal models.  There is even a name for this emerging field, Computer Aided Drug Design (CADD).  The majority of these approaches are addressing the development of novel small molecule compounds.  In a recent Global Data search to identify deals involving the keyword “intelligence,” of the 195 strategic alliances identified, 46 involved small molecule drug discovery (compared to 11 for antibodies, 6 for gene therapy, and 3 for cell therapy).  Also, not surprising, is the upward trend of deals involving the use of AI in the drug discovery process. 

 
 
 
 




















Global Data Query for Deals with keyword* “Intelligence” (21 October 2020)

*Keyword “artificial” yielded deals for artificial therapies (eg artificial skin); Bioinformatics deals included



Dyno Deal

Roche is putting a lot of money on the table in order to leverage the AI technology that Dyno has developed to identify new AAV vectors in order to mitigate some of the challenges associated with the development and/or manufacture of AAVs.  With AAV playing a dominant role in the gene therapy field, and AI being increasingly leveraged in drug design and development the total deal value of $1.8B may be worth every penny.

 
 
 
 
 
Trevor:  Originators of 2019 Approvals
 
 
 
 


Wouldn’t it be nice to have a roadmap to clinical and commercial success?  A guidebook drafted from the review of corporate development trajectories of clinical-stage companies?  Do you have to be kissed by the benevolence of the venture capital gods to even stand a chance?  Does everything have to come out of Bob Langer’s lab with an Arch-led syndicate?


Nothing against those giants but thankfully, no, of course not.  Good ideas can still come from anywhere. 


Last issue, we saw that Big Pharma held almost 50% of the approvals in 2019, but 60% of the approvals originated elsewhere.   

This time, we are looking at Phase III oncology drugs. The charts below are based on GlobalData queries for oncology drugs (geography agnostic) where Phase III is the highest development stage for those assets. 

Big Pharma holds 41%,  and 50% more are in other public companies, with 8% still in the hands of private companies and 1% still in the hands of academic institutions.  
 
 
 
 

 
 
 
 
But where did the drugs come from? 


The charts below are based on GlobalData queries for geography agnostic oncology drugs where Phase III is the highest development stage for those assets.  The data was then cleared of those assets that were in multiple, parallel studies due to regional development or combination trials.  The result was 205 originator Phase III drugs.  As one might expect, the US is home to the most originators with 43% of the total, roughly forty percent more than the next closest territory where 26% of the originators found their genesis on Chinese soil. Japan, Germany and Switzerland round out the Top 5 countries.

 
 
 
 

And what type of companies were the originators?


Reading the data for the types of entities giving birth to these clinical distance-runners (below), one finds a surprisingly high percentage of private companies leading the charge ringing the bell with 62 originators.  And of those, China is at the front of the pack with 22 of the sum, followed by the US with 19.


Another aspect in the data that caught my eye: relative to overall numbers of originators where their percentage is ~26%, Chinese originators are overrepresented in Private, Public Mid-Cap and Public Large-Cap companies.  China leads in all three and particularly in the Mid-Caps where it is over half of the total with a number of newly minted companies sporting valuations in the $2-$5B range. 


As a class, Universities and Research Institutes are right up there near the, well, near the head of the class.  It pays to stay close to the researchers making the new discoveries as almost 20% of the originators were begun in these labs of higher learning.  Indeed, it’s likely that some percentage of the Private entities and the Public companies also got their start on IP that “came out of” a University. 

 
 
 
 
Interestingly, Large Cap and Mega Cap companies (as defined in the graph) with 41 originators equal roughly the same number as Universities.  Personally, I find it encouraging that almost half of the Phase III oncology drugs in this data set were originated in either University settings or incubated in Private companies.  It’s true, not all start-ups start the same… a few of the Private company names in this data set launched with massive financings – but most of them just went check to check, inflection to inflection. 
 
 
 
 
 
 
Pullan Consulting (www.PullanConsulting) provides advice and execution for biotech partnering and fundraising, with outreach to partners and investors, help with the shaping of presentations, evaluations and market analysis, preliminary valuations and deal models, and negotiations from deal prep to term sheets to final agreements.      
 
We have extensive scientific and financial experience, with many deals signed.  Check out the website for more on clients, tasks, CVs, etc.  And don't forget to look at the resources with whitepapers and decks to help. 

Send us an email or set up a call if you want to explore how Pullan Consulting might be of help!     
     
Linda Pullan                     Linda@pullanconsulting.com 
Trevor Thompson             Trevor @pullanconsulting.com 
Jessica Carmen               Jessica@pullanconsulting.com 
 
 
 
 
 
 
        
Footer-logo