What a year 2023 has been for tech advances in bio!
Ironically, the elation from a kaleidoscope of technological and scientific advances is in stark contrast with the overall industry’s economic performance—it was a rough year for biopharma overall. There are signs 2024 might be better for the biopharma industry.
Anyway. it is a good time to take a look back and reflect on the things that have impacted life sciences and changed our perception of what may be possible soon and what might need more time to mature into real-world applications.
Organoid Intelligence
In February, scientists founded a new field: “organoid intelligence” (OI), which I consider one of the potentially most impactful ideas in biological sciences —for the better of worse.
Led by Dr. Thomas Hartung in the U.S., are developing biocomputers using brain organoids—lab-grown tissues mimicking organ functions—from human stem cells.
These brain organoids, though not structurally identical to human brains, exhibit neuron-like functions and are envisioned to surpass the computational efficiency of supercomputers, offering novel approaches in pharmaceutical testing and insights into brain functioning.
The field confronts technological challenges like scaling up organoids, developing brain-computer interfaces for data exchange. It also confronts ethical considerations regarding the potential consciousness and rights of these organoids, necessitating a rigorous and inclusive ethical framework for development.
We are in the early days. Thankfully! Why? While I do not believe LLMs or other in silico artificial intelligence can become dangerous AGI any time soon, I am certainly less sure about that when we talk about actual biology-based systems combined with hardware, digital interfaces and data. These little cyborgs are frightening and exciting at the same time!
A mixed year for AI in drug discovery
2023 has been a pivotal year for reconsidering the role of AI in early-stage and preclinical drug discovery.
On the one hand, we have distinct success stories, like those of Insilico Medicine, a company that managed to build a diversified clinical pipeline, including a recent start of Phase 2 for a drug candidate to treat idiopathic pulmonary fibrosis (IPF), five phase 1 candidates for various indications, including kidney fibrosis, inflammatory bowl disease (IBD), immuno-oncology, and COVID-19, and around a dozen preclinical programs in late stages of development — all within 3-4 years since the start of internal pipeline. What is more striking is that the majority of programs are based on actually novel targets, discovered by company’s PandaOmics system, which is a multimodal AI platform for target discovery.
Another notable example which I would count as a 2023’s AI success is that Verge Genomics got positive safety and tolerability data from the Phase 1 clinical trial for its leading candidate VRG50635, a potential best-in-class therapeutic for all forms of ALS. Verge Genomics used CONVERGE™, the company’s all-in-human, AI-powered platform to develop its drug discovery program.
Also, the FDA’s clearance of A2A Pharma’s Investigational New Drug (IND) application for A2A-252, a TACC3 protein-protein interaction (PPI) inhibitor, showcases the potential of AI in accelerating drug development. Utilizing its AI-driven SCULPT computational platform, A2A Pharma, with a lean team of four and limited funding, managed to advance two clinical stage programs, including A2A-252.
Finally, there were successes in drug repurposing and indication expansion, for instance, by Dallas-based clinical stage company Lantern Pharma, a developer of RADR®, or “Response Algorithm for Drug Positioning & Rescue”, AI platform. The company is applying its biomarker discovery and drug design platform to finding novel indications, drug combinations, and likely-responder patient groups among broader patient populations, allowing for more cost-efficient, and robust clinical trials, as well as personalized therapies. The company’s pipeline includes two Phase 2, five Phase 1 candidates and a number of preclinical programs in oncology and CNS diseases. Notably, using their novel highly accurate AI algorithm, Lantern Pharma managed to predict blood brain barrier (BBB) permeability with an impressive 89-92% accuracy, offering a rapid, cost-effective way to screen drugs or compounds to determine their potential to cross the BBB.
On the other hand, in 2023 we have witnessed a number of clinical trial setbacks for some AI-designed drug candidates, including Exscientia’s cancer drug candidate EXS-21546 which did not reach target results in a Phase 1/2.
An AI-inspired schizophrenia drug candidate from partners Sumitomo Pharma and Otsuka Pharmaceutical failed to outperform a placebo in two Phase 3 studies. Sunovion, a subsidiary of Sumitomo Pharma, brought compounds to the alliance that were then screened using PsychoGenics’s SmartCube technology, which employs computer vision to analyze and define behaviors of mice treated with a potential drug.
Adding to negative statistics, BenevolentAI’s lead drug BEN-2293 failed to beat a placebo in a Phase 2a atopic dermatitis study, leading to cutting up to 180 jobs and reorganizing its pipeline to conserve cash.
Beyound turbulent drug candidate progress dynamics, 2023 was very productive for the AI adoption by pharma industry, including numerous academic breakthroughs, technology launches, like NVIDIA’s introduction of BioNeMo generative AI platform for biology research, and research pilots to implement so called AI foundation models for large scale omics projects, like in the case of Ginkgo Bioworks parntership with Google Cloud.
For a deepdive into AI progress in drug discovery and biotech, as well as 2023 briefings, refer to BiopharmaTrend’s landmark report “The State of AI in the Biopharma Industry.“
A great year for AI in clinical research
In contrast to early drug discovery and preclinical, where data bottlenecks are still a major roadblock for swift AI adoption, the field of clinical trials is advancing much smoother.
Big pharma hints that AI is already impacting clinical research. According to a 2023 Reuters report co-authored by Natalie Grover, Martin Coulter, and Julie Steenhuysen, Amgen’s AI tool, ATOMIC, now scans vast data to rank clinics and doctors based on recruitment history, cutting enrollment time for some mid-stage trials by half. By utilizing ATOMIC, Amgen aims to shorten the typical drug development timeline by two years by 2030.
Novartis also leverages AI to expedite patient enrollment in trials, making the process faster, cheaper, and more efficient. However, AI is only as good as the data it is trained on. With only about 25% of healthcare data available globally for research purposes, there are still limitations.
Bayer utilized AI to decrease participant numbers in a late-stage trial for Asundexian. Specifically, it used AI to bridge mid-stage trial findings with extensive real-world data from millions of patients across Finland and the US, facilitating the forecasting of long-term risks among a population analogous to the trial. Bayer plans to use real-world data for an external control arm in a pediatric study of the same drug.
According to the Reuters report, Dr. Blythe Adamson, a senior principal scientist at Roche subsidiary Flatiron Health, emphasized how AI enables rapid and large-scale analysis of real-world patient data, contrasting it with traditional methods, which could take months to analyze data from 5,000 patients, whereas now millions of patients’ data can be analyzed in just a few days.
IQVIA’s AI platform, using their real-world data assets, has enabled precise patient and healthcare professional (HCP) targeting, doubling the number of identified eligible patients, finding 30% more with uncontrolled symptoms, predicting 81% of patients are likely to quit treatment early, pinpointing events linked to early discontinuation, and boosting treatment transition success by 500% compared to prior methods.
Smaller companies are also applying AI to contribute to the progress in clinical research. In one case, Lantern Pharma’s RADR® was able to predict the drug’s response with an accuracy rate of 88% across all solid tumors being tested for Elraglusib; using those models they were also able to predict sub-populations of melanoma patients that may benefit from Elraglusib.
Another AI platform, the inClinico system by Insilico Medicine, was validated through three methods in a study evaluating AI’s prediction accuracy on Phase II trial success. This transformer-based platform, utilizing generative AI and multimodal data, was trained on 55,600 unique Phase II trials over 7 years. Insilico’s developed model showcased 79% accuracy in predicting real-world trial outcomes in the prospective validation set.
I had an interesting discussion with Fareed Melhem, Senior VP and Head of AI at Medidata, a Dassault Systèmes company and he outlined some interesting insights about AI in clinical research, read it here.
AI foundation models are coming to biotech
In the past three years, AI foundation models have been on fire, swiftly entering various industries. The most prominent examples of foundation models are the GPT-3 and GPT-4 models, which form the basis of ChatGPT.
These are huge models trained on enormous volumes of data, often in a self-supervised or unsupervised manner (without the need for labeled data). Thanks to special model design, including transformer architecture and attention algorithms, foundation models are very generalizable, allowing their adaptation to a diverse array of downstream tasks, unlike traditional AI models that excel in single tasks like, say, predicting molecule-target interaction.
For instance, Deep Genomics unveiled BigRNA, a pioneering AI foundation model for uncovering RNA biology and therapeutics. According to Deep Genomics, it is the first transformer neural network engineered for transcriptomics. BigRNA is informed by nearly two billion adjustable parameters and has been trained on thousands of datasets, totaling over a trillion genomic signals.
A month earlier, Ginkgo Bioworks, Inc. and Google Cloud announced a 5-year partnership where Ginkgo would work to develop new, state-of-the-art large language models (LLMs). The AI foundation model would be focused on genomics, protein function, and synthetic biology and would be running on Google Cloud’s Vertex AI platform. The model is supposed to help Ginkgo’s customers accelerate innovation and discovery in fields as diverse as drug discovery, agriculture, industrial manufacturing, and biosecurity.
Read a case study 7 Companies Building Foundation Models in Biology and Chemical Synthesis for more context.
AI-driven protein design is a big deal
AlphaFold, a product of Google DeepMind, revolutionized predicting protein structures using amino-acid sequences, which was a truly extraordinary achievement.
Recently, a wave of AI tools have been introduced to do something potentially even more useful: design novel proteins with desired structures and, hence, biologic functions. Even those previously unseen in nature.
In his recent Nature paper, Ewen Callaway delves into this exciting field and explains emerging AI tools for protein design, like RFdiffusion and other similar AIs, inspired by neural networks like DALL-E. Such tools offer a promising solution, working on ‘denoising’ data for protein creation. Trained on the vast Protein Data Bank, RFdiffusion, for example, can generate novel proteins or be conditioned for specific designs, like emulating the English alphabet or Arabic numerals.
Some comanies that made headlines in protein design space and raised money in 2023 include Generate:Biomedicines, and Cradle Bio, among others.
Generate:Biomedicines’ recent Series C funding round, totaling $273 million, marks the largest in 2023 for a biotech company, surpassing Apollo Therapeutics’ $226.5 million. This milestone elevates Generate:Biomedicines’ total equity financing since 2020 to approximately $700 million. The substantial funding underscores the significant investor confidence in Generate:Biomedicines’ innovative approaches in the biotechnology sector.
Cradle, a biotech and AI startup specializing in protein design using generative AI, has secured $24 million in a Series A funding round, following a $5.5 million seed round last year.
To be fair, the field is still to overcome many challenges, some of which are outlined by Derek Lowe in his commentary Protein Design the AI Way.
A bang year for genome editing
The year marked a series of landmark achievements, most notably the first clinical approval of CRISPR-based therapy for sickle cell anemia and beta-thalassemia in the United Kingdom, a decision soon echoed by the U.S. Food and Drug Administration and the European Medicines Agency.
The highlight of CRISPR’s success was the emergence of Casgevy, a novel gene editor. This therapy, which corrects genetic errors in stem cells harvested from patients’ bone marrow, promises to alleviate the debilitating effects of blood disorders.
Once reintroduced into the body, these modified stem cells are capable of producing healthy blood cells, offering a beacon of hope for those afflicted.
However, despite its potential, CRISPR-Cas9 isn’t without its challenges. The technology’s approach to cutting both strands of DNA poses risks, including the unintended activation of cancer-triggering genes or other unwanted genetic alterations.
In 2023, there were a number of successes where AI algorithms have not only enhanced the precision of CRISPR but also broadened its scope. Machine learning techniques have been instrumental in predicting off-target effects, particularly in CRISPR tools that target RNA instead of DNA.
This expansion in capability is largely attributed to AlphaFold’s algorithm, which has honed the art of identifying smaller, more precise CRISPR “scalpels” for targeted genetic alterations. These refined tools promise greater accuracy and ease of delivery to their intended genomic destinations.
Also, a team led by Feng Zhang at the Broad Institute of MIT and Harvard has discovered a new method for genome editing. They have found a system, based on a protein called Fanzor, that uses RNA to guide DNA changes. This is the first such system found in eukaryotes—organisms like animals, plants, and fungi.
The Fanzor system could potentially be more precise than existing CRISPR/Cas methods. The team has made the system more efficient, suggesting that Fanzor could become a new tool for genome editing. Importantly, Fanzor does not cause unnecessary damage to nearby DNA or RNA.
There is a wonderful article by Shelly Fan, published in in Singularity Hub, which nicely summarizes the major gene editing developments in 2023 — read it here.
RNA modalities keep growing, delivery is key
More than 30% of R&D investment over the next 10 years will be in RNA medicines (siRNA, ASO, mRNA, etc.), according to John Maraganore, former founding CEO of Alnylam Pharmaceuticals, and pioneer of RNAi therapeutics as a whole new class of medicines for patients.
Alnylam is often recognized as a pioneer in the field of RNA interference (RNAi) therapeutics. One of the notable milestones for Alnylam and the RNAi field was the FDA approval of patisiran (Onpattro) in 2018, which is an RNAi therapeutic developed by Alnylam for the treatment of hereditary transthyretin-mediated (hATTR) amyloidosis. This approval marked the first-ever RNAi therapeutic to receive approval from a major regulatory agency, affirming Alnylam’s pioneering role in advancing RNAi as a new class of medicines.
But RNAi is not the only promising class of drugs in the growing arsenal of RNA-focused therapeutics. For instance, San Francisco-based Atomic AI is focused on RNA-targeting small molecules.
Atomic AI’s proprietary AI-driven 3D RNA structure engine, known as PARSE, generates RNA structural datasets, integrating machine learning foundation models with large-scale, in-house experimental wet-lab biology to unveil functional binders to RNA targets.
The company’s technology has the ability to predict structured, ligandable RNA motifs at unprecedented speed and accuracy, a key barrier to current approaches to RNA drug discovery.
Atomic AI plans to use its database of discovered and designed 3D RNA structures to develop a pipeline of rationally designed small-molecule drug candidates.
What is interesting, Atomic AI is using so-called geometric deep learning, and can learn from very small RNA data (more on small data vs big data for AI in biology in tomorrow’s issue of ‘Where Technology Meets Biology’ newsletter; make sure to subscribe to get notified).
Recently, I highlighted some other promising RNA-focused startups, including those focusing on mRNA, ASO, RNA editing, and circRNA theraeputics.
Delivery is key in RNA therapeutics field.
In contrast to good old small molecules, RNA can’t be administered without proper delivery systems. One such system is lipid nanoparticles (LNPs).
This US and Israel-based startup, Mana.bio, really caught my attention, as it is focused on the design of novel lipid nanoparticle candidates in silico using its machine learning models that predict desirable properties to select best-in-class cell-specific tissue-targeted LNPs.
Alternatives to animal testing keep advancing
The introduction of FDA Modernization Act 2.0 signed by President Biden at the end of 2022 marked a significant regulatory change for the preclinical research industry. The Act basically allowed to consider alternative drug testing systems, in place of earlier mandatory animal testing. This, of course, was great news for everyone in the business of creating organ-on-a-chip systems, organoids, and other drug testing alternatives.
Here I just mention one recent development by Israeli company Quris AI, that announced an extension of its collaboration with Merck KGaA following a successful preclinical study. It showcased Quris-AI’s ability to accurately predict liver toxicity risks in a selection of drug candidates.
Quris pioneered a high-throughput patient-on-chip platform called BioAI that integrates real-time nanosensing, stem-cell-derived tissues with genomic diversity, and machine learning models to predict drug toxicity.
Quris’s patient-on-chip systems are models where multiple miniaturized organs are interconnected using a system similar to blood circulation. Nano-sensing provides real-time data on metabolites in miniaturized tissues, documenting responses to each drug.
Using stem-cell genomic diversity, drugs can be tested on a wide variety of genetically unique patient models on a chip, simulating real clinical trials. Thanks to recent breakthroughs in stem cell technology, it’s now affordable and quick to produce hundreds of miniaturized organs, like livers and brains, from iPSCs derived from basic blood samples. With this, machine learning can better predict not just a drug’s general safety but also its safety for specific individuals, enhancing personalized medicine and refining clinical trials, drug repurposing, and saving failed drugs.
The Quris’s high-throughput patient-on-chip system generates an enormous amount of real-time experimental data. But Bio-AI doesn’t use this data to directly determine drug safety. Instead, this vast amount of data is used to train machine learning models. After training, the ML predicts drug safety.
The field of organ-on-a-chip systems and organoids is complex, there is some useful follow-up reading:
Radiopharmaceuticals are a growing niche
Drug makers have developed some radiopharmaceuticals over the past several decades. Most of them did not reach commercial goals and were eventually shelved.
The market for radiopharmaceuticals is on the rise now. There are now more than 70 radiopharmaceutical startups in the U.S. alone, approaching a critical mass.
In an impactful deal, Bristol Myers Squibb acquired RayzeBio for $4.1B, along with its lead phase 3 candidate, RYZ101. The 225Ac-bearing molecule RYZ101 is aimed at treating certain cancers; it is designed to target somatostatin receptor 2 (SSTR2), a protein frequently found in high levels in gastroenteropancreatic neuroendocrine tumors (GEP-NETs) and extensive-stage small cell lung cancer (ES-SCLC).
The company has initiated a Phase 3 clinical trial for RYZ101, focusing on patients with SSTR-positive GEP-NETs who have previously undergone treatment with Lutetium-177-based somatostatin therapies.
Currently, there are more than 70 radiopharmaceuticals startups in the US alone, which is close to a critical mass for a field expansion.
A busy year for antibody-drug conjugates (ADCs)
In 2023, there were over 20 deals involving ADCs, surpassing the number of deals in other emerging cancer drug categories. In stark contrast to the general slowdown in dealmaking, marked by higher interest rates and geopolitical tensions, the ADC sector is witnessing a flurry of activity.
Notably, Pfizer has acquired Seagen for $43 billion, a move followed by Seagen’s collaboration with Nurix Therapeutics. Other pharmaceutical giants like Amgen, AstraZeneca, and Merck have also invested heavily in ADCs, with over $60 billion worth of licensing deals signed in the past five years.
GSK has just entered into an exclusive license agreement with China-based Hansoh for HS-20093, a B7-H3 targeted ADC utilising a clinically validated topoisomerase inhibitor (TOPOi) payload. Under the agreement, GSK will obtain exclusive worldwide rights (excluding China’s mainland, Hong Kong, Macau, and Taiwan) to progress the clinical development and commercialization of HS-20093.
One interesting case is C4 Therapeutics, with its degrader-antibody conjugates (DACs).
The field of ADCs is, however, facing challenges too. Sanofi has recently halted the development of tusamitamab ravtansine, an antibody-drug conjugate (ADC), for non-small cell lung cancer (NSCLC) after it failed to meet the primary endpoint in the Phase III CARMEN-LC03 trial.
The complexity of manufacturing these drugs and the uncertainty of their performance in combination with other treatments like immunotherapies are concerns. Additionally, geopolitical tensions could impact collaborations with Chinese partners.
The arsenal of chemical modalities is expanding
Here are five ‘hottest’ trends in medicinal chemistry, highlighted in a 2023 C&EN article authored by Laura Howes:
Covalent Inhibitors: These represent a shift from traditional drug designs, binding covalently to proteins, even in the absence of a preformed pocket. This approach has gained traction, with researchers targeting various reactive amino acid side chains, not just cysteine.
PROTACs (Proteolysis-Targeting Chimeras): Initially developed by Yale University, these molecules harness the cell’s own degradation machinery to eliminate unwanted proteins, marking them for destruction through ubiquitination.
Molecular Glues: These small entities induce protein degradation by altering a protein’s affinity for another, promoting targeted cellular responses. This approach is gaining interest, as evidenced by recent partnerships between molecular-glue firms and major pharmaceutical companies.
Macrocycles: Represented by cyclic peptides and other large molecules, these compounds exceed the traditional size limits of oral drugs but show promise in clinical settings due to their complex structures and potent activity.
RNA-Targeting Small Molecules: Once thought undruggable, RNA is now a target for small molecules that can induce degradation or modify its function, with companies like Arrakis Therapeutics leading the charge in this innovative space.
The evolution of these modalities often goes beyond the “rule of 5,” and it also illustrates the expansion of what is considered “druggable,” shifting from proteins to protein-protein interactions, RNAs, and other novel target classes.
So, while it seems like “biology is eating” the world, medicinal chemists can still sleep well too, knowing they are not losing jobs anytime soon.
Some more really nice reading to complement your end of year reflections:
The Most Promising Longevity Drugs To Date — for those who are into longevity field.
The Year in Biology — a really entertaining and joyful read about 2023 advances in basic biology research. Pure science!
- SEO Powered Content & PR Distribution. Get Amplified Today.
- PlatoData.Network Vertical Generative Ai. Empower Yourself. Access Here.
- PlatoAiStream. Web3 Intelligence. Knowledge Amplified. Access Here.
- PlatoESG. Carbon, CleanTech, Energy, Environment, Solar, Waste Management. Access Here.
- PlatoHealth. Biotech and Clinical Trials Intelligence. Access Here.
- Source: https://www.biopharmatrend.com/post/725-11-biopharma-trends-to-watch-in-2024/