Traditional Small Molecule R&D Market Reclaiming Investor Interest

Acknowledging my entire audience may not be well versed in the biotechnology sector, I will briefly frame the R&D landscape. Conventional small molecule drugs have been the mainstay of the pharmaceutical industry for almost 100 years. A small molecule by definition is an organic compound with a low molecular weight. Small molecule drugs have conventionally targeted proteins and have distinct advantages relative to other therapeutic classes:

  • Primarily administered orally
  • Ability to pass through cell membranes to reach intracellular targets
  • Varying degrees of systemic exposure (e.g. with or without brain penetration)
  • Insurance and reimbursement pathway is well understood in core markets
  • Easier to manufacture small molecule drugs at scale relative to other modalities

The human genome, which is made up of ~3.1 billion base pairs coiled within 46 chromosomes, is transcribed into RNA (mRNA, pre-mRNA, tRNA, rRNA, miRNA, and long noncoding RNA). These transient products are then translated into protein. This process is otherwise known as “the central dogma of molecular biology”. Roughly 75% of the human genome is transcribed into RNA. That fraction is astronomical compared to the mere 1.5% of the genome which is translated into proteins. This delta is becoming increasingly meaningful from a R&D perspective.

Figure 1. Human RNA regulates key biological processes

For many years small molecule drugs have only attempted to target proteins. Roughly ~20,000 proteins in the human proteome are assembled by one of the roughly ~20,000 genes which are associated with these proteins. Protein targets come from various families including GPCRs, ion channels, nuclear hormone receptors, structural proteins, membrane transport proteins and enzymes. In 2017 the National Institutes of Health (NIH) funded a project which led to the development of the Pharos Database, with the goal of Illuminating the Druggable Genome (IDG). In 2017 it was cited that as many as 4,000 proteins were susceptible to targeting by pharmaceuticals (aka: druggable). However, only 5-10% of proteins considered druggable were targeted by FDA approved drugs.

Three protein families: ion channels, GPCRs, and protein kinases were identified as families containing roughly ~100 promising understudied proteins. These three protein families were also identified to be “well-established druggable families with high potential to impact human health once disease associations were made.”

Figure 2. How The IDG Research Study Was Logically Laid Out

The major issue is several of these ~100 identified proteins of interest work at the intersection of numerous molecular pathways. Modulating these proteins often leads to toxicity and unpredictable downstream effects, and thus a lower likelihood of success. This is known as the “dirty drug” problem. Some examples of dirty drugs include tramadol, chlorpromazine, olanzapine, and dextromethorphan… all of which bind to multiple receptors or influence multiple receptor systems. These drugs all have very broad clinical side-effect profiles.

While small molecules remain a mainstay therapy for the majority of indications… there is growing speculation that we are entering the golden era for small molecule drug development. This optimism has been driven by the rise of AI/ML and the growing capability of small molecules to modulate biological disease driving pathways at the RNA level. This has created an exciting opportunity to reach once “undruggable targets”.

The idea that RNA can be modulated is not completely new, antisense therapy is a form of treatment that uses antisense oligonucleotides (ASOs) to target mRNA. ASOs are capable of altering mRNA expression through various pathways. Some ASOs (phosphorothioate ASOs) can be delivered to cells without a delivery vehicle. This is a big deal. Unforunately, ASOs typically don’t cross the blood brain barrier… but they can if injected into CSF via intrathecal injection.

If your brain likes categorizing things like mine does… ASOs are not considered “gene therapy” as they only make contact with RNA, not DNA. There has been several ASO success stories, particularily within neurodegenerative indications such as Amyotrophic Lateral Sclerosis (ALS). This year Biogen got an accelerated approval of Tofersen (Qalsody) for treatment of SOD1-ALS.

Figure 3. Biogen Turned Their Focus to Neuro Biomarkers Years Ago and It Paid Off

I was somewhat surprised by this, but Biogen emerged years ago as a leader in neurology biomarker research. It is interesting that they were able to enable a pre-symptomatic trial in SOD1-ALS following their discovery of neurofilament elevations preceding symptoms. This was an interesting example of how one biomarker can crack open an indication class and lead to financing avalanches in small pockets of the R&D market. It seems like the FDA took the neurofilament arguments seriously because of the strength of the prognostication evidence. Three studies were completed with n>100, and all three showed, with very strong significance, lowest tertile neurofilament levels being associated with better survival outcomes.

Figure 4. Strong Independent Biomarker Prognostication Evidence Can Be Indicative of Future FDA Outcomes

The Biogen success story led me down a personal research journey. I believe analyzing the strength of evidence for a biomarker has a higher predictive capacity than the actual drug efficacy when analyzing FDA decisions. I will save this discussion for a later post however.

Regardless of ASOs, the space that is now accessible to a small molecule has exploded and the ability to modulate RNA is being explored. Novelly deployed AI/ML and biomarker discovery is perpetually refreshing interest in the small molecule space. As referenced in the diagram above, there are roughly ~50,000 targets in the genome which are associated with RNA products. That is a lot more than what has been accessible in the past. The takeaway… the solution space for conventional drugs is rapidly expanding.

New Modalities Will Face Financing Headwinds

While it is evident small molecules are not going anywhere, when the market was frothy in 2020 and 2021 there was an aggressive trend towards the development of alternative therapeutic modalities. This was partially because the COVID-19 mRNA vaccine race resulted in billions of dollars of targeted public research money, which subsequently spurred private investor interest in a once hidden corner of the market. Roughly speaking, there are six categories which encompass 18 seperate new modalities. Of course, these could be further subcategorized, but I spare this triviality for market framing purposes. Outside of small molecules, this is really what the landscape looks like:

Figure 5: The New Modality Market From 10,000 Feet

One of the core themes of the last three years in the industry has been the rise of new modalities. New modalties are on track to outpace conventional therapies in terms of aggregate marketed drug revenue generation. The relative projections presented below, which are from a BCG internal report from 2023, are quite aggressive. Monoclonal antibodies were also included as a new modality in this report… and the argument could be made that monoclonal antibodies shouldn’t be classified as so.

Figure 6: New Modalities Now Outpace Conventional Modalities in Revenue Projections

Regardless, new modalities are on pace to make up roughly half of aggregate big pharma revenue generation. Of this revenue share, not suprisingly, the vast majority has been produced by monoclonal antibodies. As you can see below, monoclonal antibodies still dominate the market from a pipeline perspective and this trend likely won’t be reversing anytime soon.

Figure 7: Pipeline Snapshot by Modality

When interest rates bottomed out and private financing was frothy in late 2020 through 2021 there was expanding early stage private interest in cell therapy, as CAR-T had a fresh batch of FDA approvals in 2020 and 2021.

Figure 8: Approved CAR-T In the United States

The downfall of commercial CAR-T products is that they are currently “centrally manufactured” (meaning patient products need to be cryopreserved, shipped to a factory, processed, shipped back to the patient site, etc.). Access is grossly limited by the current CAR-T manufacturing capacity. There is a 1-2 month wait for a “scheduled manufacturing date”… from there the manufacturing time often takes an additional 4 weeks. This gives the manufacturing lab time to process the autologous apheresis products. During that time samples undergo a transduction and expansion of T-cells. While the technology is extremely promising… the “vein-to-vein” time is simply far too long. Lastly, commercially manufactured CAR T products cost around $500,000 USD. I won’t get into resource and insurance constraints, but this is obviously burdensome.

There was a period a few years ago where there was promise and investment interest in “off-the-shelf allogeneic CAR T”, but subsequent trials have shown efficacy and persistence to be limited by rejection. There is also concerns regarding donor variability. I personally don’t see off-the-shelf CAR T as being a successful market niche moving forward, and investment interest here is currently falling away. There is some renewed interest in decentralized manufacturing, but those discussions fall outside the scope of this report and outside the investment mandates of R&D stage funds.

Scoping in On The Cell & Gene Therapy (CGT) Market

For the non-technical audience, gene therapy can be classified as:

  • ex vivo: modification of cells outside of the body and subsequent transplantation back into patient (e.g. CAR T, CAR NK)
  • in vivo: therapeutic or genetically altered DNA directly into the patient’s body, typically in the form of a plasmid vector and injection (into the blood)
  • in situ: administration of the gene product directly into a specific site (into a specific tissue)

Modalities include:

  • gene augmentation: deliver a functional version of the mutated gene
  • gene editing: use CRISPR/Cas9 or molecular machinery to modify the genetic material of cells

So far, this is what approvals look like in the CGT market, you’ll notice most approvals are in oncology and hematology, but there is growing promise in rare neurology and CNS indications:

Figure 9: Existing Gene Therapies Which have Obtained Approval (2004 - 2020)

Figure 10: Existing Gene Therapies Which have Obtained Approval (2021 - 2023)

I will save comments on pricing and insurance for a later post, but there are concerns regarding the marketability of most of the newer gene therapies.

The Current Cell & Gene Therapy (CGT) Pipeline

Not surprisingly oncology and rare disease remain the top areas for gene therapy development in both preclinical and clinical settings. CAR-T cell therapies make up 47% of the pipeline (preclinical through to pre-registration) of genetically modified cell therapies according to the Q3 2023 report by the American Society of Gene and Cell Therapy. 97% of CAR-T cell pipeline therapies are for oncology indications.

Later stage CGT companies maintained steady deal activity in Q3 2023, but start-ups faced a very challenging financing environment. The most interesting statistic I was able to pull was the number of seed and series A rounds for gene therapies, which decreased by 15% to 17 financings in Q3 from Q2. The aggregate dollar value of seed and series A financing plummeted 73% to 348.2 million USD in Q3. I expect much more activity within optho/sensory in the coming years… I have my eye on Tenpoint Therapeutics after their $70 million USD series A for in vivo reprogramming. They also acquired AAV ex vivo GIRK technology from SparingVision in July 2023 for late-stage retinitis pigmentosa.

Figure 11: The Current Cell and Gene Therapy Pipeline

There are 1023 preclinical to pre-registration gene therapies which are currently in the pipeline for “rare diseases” (around 80% of these are for rare oncology indications). The top five rare diseases for which gene therapies are being developed are; myeloma, non-hodgkin’s lymphoma, acute myelogenous leukemia, b-cell lymphoma and ovarian cancer. According to the American Society of Gene and Cell Therapy “the proportion of gene therapy trials for non-oncology indications has increased by six percentage points since the previous quarter, to 38%, continuing the trend of an increasing proportion of non-oncology gene therapy trials initiating each quarter since Q4 2022.”

Figure 12: Growing Registration of Gene Therapies in Rare Diseases Outside of Oncology

This is interesting but not surprising, from an investment perspective oncology is definitely crowded which is why there is a growing pool of earlier stage investors looking to rare CNS and various neurological indications. For completeness sake, I have included pipeline numbers for targets below, no real surprises here with CD-19, TNF and VEGF along with clotting factors dominating. No real trends to pick up on from the last 12 months.

Figure 13: Gene Therapy Pipeline by Target

GenAI/AI/ML Use Cases for The Cell & Gene Therapy (CGT) Market

I’ll start off with outlining the challenges. Explicitly, the application of GenAI/AI/ML is over-hyped. There is a growing trend with pre-clinical programs where you very much need to have some marketable AI/ML component to your story or you are at a competitive disadvantage. The reality is the potential applications of AI/ML in R&D drug discovery are still relatively nascent. Technical talent is still attracted to alternative sectors (e.g. tech, finance) where there is higher pay and more job stability.

The deployment of AI/ML in the setting of small molecule target discovery is probably more valuable than in the setting of CGT given the smaller solution space (aka: less attributes to model therefore less data required). The fundamental issue with AI/ML driven in silico companies is that even if there is a statistically significant target validation event… that discovery still needs to be experimentally validated in a pre-clinical setting before moving to a clinical setting. You can’t skip steps. So in essence… the AI/ML component can be viewed as a “pre-pre-clinical tool”, with the hopes that the discovery phase of a pipeline can be time-compressed and the IRR can be juiced by that shortened timeframe. That is a narrow window to unlock value given the mathematical hurdles regarding data constraints that must be overcome. I am also a data scientist and happen to believe that only ~75% of the data collected is actually clean enough to be loaded into a data pipeline and eventually a model.

Figure 14: This is How To Think About ML/AI Model Efficacy in Different Market Settings

Publicly accessible and commercially available experimental data is exceedingly challenging to come by. The CGT subsector remains nascent and somewhat poorly defined. Creating data for these new methods from scratch is costly and time-consuming. Data scarcity poses obstacles for training AI systems. While there are ML approaches that can circumvent data scarcity (e.g. data imputation), AI’s ability to unlock value is rate-limited by data availability. Which brings me to highlighitng the enormus solution space of these novel methods. There is an astronomical number of attributes embedded within FASTA and FASTQ files when doing sequencing runs, which leads to the inevitable curse of dimensionality.

Figure 15: More Attributes Makes the Rate Limiting Step of Data Collecting More Pronounced

There is also a fundamental dislocation between wet-lab and in silico research, in silico drug discovery demands being facile with programming, which is a very different skill set compared to the specialized expertise needed for CGT wet-lab experiments. These teams tend to operate in isolation. Attempts to join in silico and wet-lab teams have had limited success thus far. I’ll give the example of Deep Genomics, a Toronto based company which was founded in 2014 and has raised over $200 million USD. The company went all the way to Series C in 2021 and been backed by high profile institutions like CPPIB without being able to take anything past the pre-clinical stage. This was after Deep Genomics assembled some of the most high profile researchers in the space as well.

While there are headwinds, there are a few cases where AI/ML does make sense. At this stage I think applied AI/ML should be viewed as a potential tool and not an early stage business you’d want exposure to. The highest value add for R&D remains in large-scale in silico target screening to help reduce wet lab discovery time. Until further validation, it is very much still a “discovery tool”.

Figure 16: Summary of Major AI Use Cases Across the Cell and Gene Therapy Value Chain

Private Investment is Cooling & Public Market Opportunity is Developing

Within the private financing realm there is a much higher “conviction threshold” for investment committees to allocate to a any pre-clinical program (conventional or new modality) in this tight financing environment. I’ve identified some core factors currently dampening private pre-clinical investment:

  • Heightened interest rates and macroeconomic volatility in the past 12 months has produced a sector agnostic drift away from non cash flow centric business models.
  • The prolonged period of low interest rates following the pandemic created an environment whereby family offices and institutional LPs became overweight growth equity and thus many of these participants are now looking at “balancing down” and allocating less capital to sector strategic fund-of-funds and growth equity… with REIT dividend yields sitting above 10%, I don’t blame them. Many institutional LPs (particularily those in Canada) were overweight growth equity within the technology sector and have underperformed with those investments, there is little to no impetus to allocate funds to early stage companies currently.
  • Within VC there is a prioritization of follow-ons supporting existing pre-clinical portcos as syndicate fund pools are shallowing.
  • It is nearly impossible to raise private capital right now unless you have an international footprint and are starting a royalty/credit fund.

Despite private deals slowing down, there is a growing window of opportunity for alpha in the public markets. There is a large swath of pubcos trading below cash per share and at negative enterprise values. Some of these names have near term catalysts. Stay tuned as my next market research post is going to be a deep dive on my favourite publicly traded opportunity.

More broadly, the XBI has rallied aggressively since October and there has been a cluster of M&A activity: this past week Bristol Myers agreed to acquire radiopharma RayzeBio for 4.1B USD, while AstraZeneca bid for the Chinese CAR-T firm, Gracell, in a deal valued 1.2B USD. According to Barron’s nine of the twenty biotech acquisitions worth 1.0B USD or more announced so far this year have come since the start of October. Geeze, I wish I was staring down a bonus at Centerview Partners right now. All that M&A tends to really shake up strategic directives within managerial suites, and this often creates opportunities for saavy public managers. Safe to say I am anticipating an extremely eventful JP Morgan Conference this coming January (I really wish I was going).

Conclusions

  • Reinvigoration of small molecule market - growing solution space, derisked, less manufacturing uncertainty
  • Private market will remain quiet in 2024, public deal activity likely to continue
  • Headwinds and tapering of optimism in new modality market despite CRISPR/Cas9 approvals
  • Oncology will continue dominating the CGT market and expect more action in Optho
  • AI/ML synergies in small molecule space are coming to the forefront
  • AI/ML utility will slowly advance into new modalities
Download