

Discover more from Where Tech Meets Bio
Beyond Diagnosis: Digital Pathology's Emerging Role in Drug Development
ALSO: Weekly Highlights; Pick of the Week: the first broadly applicable method to study interactions with proteins in living cells; Digging Gems: biotech business models, science books for children
Hi! I am Andrii Buvailo, and this is my weekly newsletter, ‘Where Tech Meets Bio,’ where I talk about technologies, breakthroughs, and great companies moving the industry forward.
If you've received it, then you either subscribed or someone forwarded it to you. If the latter is the case, subscribe by pressing this button:
Now, let’s get to this week’s topics!

In a notable move towards advancing precision medicine, Alimentiv, AcelaBio, and PharmaNest have joined forces to advance digital pathology solutions in metabolic dysfunction-associated steatohepatitis (MASH) clinical trials.
This collaboration harnesses the power of artificial intelligence and spatial transcriptomics to deepen our understanding of MASH therapies. AcelaBio takes the lead in tissue sample analysis, generating crucial whole slide images and molecular data.
Alimentiv contributes its expertise in endpoint assessments and AI-powered digital pathology biomarker quantitation, while PharmaNest brings to the table its prowess in high-resolution, single-fiber, and single-cell quantitative image analysis.
Together, these organizations aim to elevate the quality of histological endpoints, expedite drug development, and hopefully enhance patient outcomes in the fight against MASH.
Why is the topic of digital pathology important for R&D folks?
The field of pathology has experienced a major shift over the past few decades. This evolution began as early as the 1960s with pioneering efforts in telepathology. However, the concept of virtual microscopy, also known as Whole-Slide Imaging (WSI), only started gaining momentum in the life science research community during the 1990s.
Moving to the present day, a 2023 survey commissioned by Proscia®, a software firm focused on the digital transformation of pathology, revealed that around 70% of significant pharmaceutical companies and contract research organizations (CROs) surveyed have embedded digital pathology within their research and development (R&D) operations.
However, even with this extensive adoption, over half of these organizations continue to depend on outdated software systems. Though adequate for basic image viewing, these older systems might not be equipped to fully capitalize on the opportunities presented by AI-driven R&D.
Around 55% of respondents of the Proscia survey identified the ability to generate new data assets as a primary motivator behind their investment in digital pathology.
Given the substantial amount of information enclosed within each whole slide image, it's evident that digital pathology has a crucial part to play in drug development and clinical research.
So, data from digital pathology is an increasingly used asset in modern landscape of therapeutic development, as it offers a nuanced and detailed analysis of disease mechanisms, essential for the advancement of drug development and translational research.
In the realm of clinical trials, for instance, digital pathology enables precise histopathological evaluations, crucial for assessing drug efficacy and safety.
For instance, companies like PathAI leverage AI in digital pathology to identify subtle histological features that correlate with patient responses to specific therapies, thereby refining the drug development process.
Similarly, Aiforia's AI algorithms assist in quantifying biomarkers from tissue samples, facilitating the discovery of potential therapeutic targets and the evaluation of their clinical relevance.
Furthermore, digital pathology plays a role in advancing personalized medicine by enabling the analysis of large datasets of pathology images, thereby uncovering patterns and biomarkers specific to individual patient profiles. This level of detail and precision not only provides value for therapeutic discovery but also may enhance the accuracy of clinical trial outcomes.
Delving into the topic of digital pathology, it became obvious to me that AI capabilities are a good fit here, as digital pathology heavily relies on visual data. Image analysis is among the most well-developed cases for modern deep learning systems, after all.
However, the progress of AI tools in this space is not without nuances, which is something to consider for both medtech founders and investors alike.
I just came across an article, Why Digital Pathology’s Ecosystem Won’t Support AI Startups, by Imogen Fitt, Senior Market Analyst at Signify Research, and I’d like to share key ideas about this space dynamic:
Market Growth and Challenges: The digital pathology market, including AI, is growing rapidly. However, the market faces challenges like cannibalization and consolidation, which could hinder the success of individual AI vendors. Investors need to be cautious about their investment choices (like in any other area, of course).
Varied Adoption Rates: The adoption of digital pathology varies significantly across regions. While places like the US and North America show lower adoption rates (below 10% for clinical use), Western Europe, especially the UK, is nearing saturation. Asia Pacific regions like Japan and South Korea have seen slower adoption, while China is expected to catch up soon.
Implementation Complexities: Implementing digital pathology can be a long process, often taking 6 months to 2 years. Early stages of adoption usually prioritize scanner purchases over other products, affecting the initial market for AI tools. This slow implementation process is causing a 'lag' for AI adoption in the field.
Funding vs. Market Size: Despite significant venture capital (VC) funding in recent years, the return on investment in the digital pathology AI market remains low. The total market size seems disproportionate compared to the level of funding, suggesting a potential sustainability issue.
Market Consolidation: There's an ongoing consolidation in the market, with notable acquisitions like Crosscope and KeenEye. This trend is expected to continue, indicating a shake-up in the vendor landscape.
Comparing with Adjacent Markets: When compared to the medical imaging market, which is more mature in digital technology usage, digital pathology still has a long way to go in terms of market size and vendor success. A positive side of this -- opportunities!
Vendor Strategies: Different vendors are adopting various strategies to survive in the market. Some are going public, like Aiforia, while others are diversifying their business models, including PathAI which acquired a laboratory business. The market is also crowded, making it hard for customers to distinguish between vendors.
Market Resilience and Opportunities: Despite challenges, the market is expected to continue growing. Recent developments in both research and clinical domains show promise. The research segment of digital pathology is more advanced than the clinical, and there is a shift towards platform-based approaches in AI, potentially opening up new opportunities for investment and growth.
Focus on Research-Based Applications: Many vendors are now focusing on research-based applications to sustain themselves, with the research market for digital pathology transitioning to more platform-based AI approaches. This shift could present shorter ROI timelines, making it an attractive area for investors.
In summary, while the digital pathology market is growing, it faces challenges of slow adoption rates, implementation complexities, and market consolidation. However, opportunities exist, particularly in research-based applications and platform-based AI approaches, which could be promising areas for future investment and growth.
Weekly Highlights
Recursion and Bayer to wind down fibrosis R&D pact after three years, refocusing on cancer
There is one more report from Andrew Dunn (Endpoints News) focuses on spotting the failures of “AI companies“ to deliver on their promises. Following his earlier coverage of Atomwise’s recent pivot to the ‘traditional‘ drug discovery model and some of Exscientia’s unrealized hopes regarding their AI-driven drug discovery candidates, he also pointed to Recursion Pharmaceuticals as another example of an AI company with overly optimistic forward-looking statements in the past.
In my opinion, the above coverage does not diminish the obvious disruptive potential of AI in drug discovery (I am writing an editorial explaining why), but it does show that doing marketing and pitching investors is easier than discovering drugs. Even with the help of AI.
In this context, I’d like to hint that BiopharmaTrend is soon launching a comprehensive industry report covering AI in pharma and biotech, with quite technical insights and a realistic picture of the sector.
AI in drug discovery is prone to hype, and it is important to have a realistic picture of where AI’s impact is transformative and where it is marginal, and who are the actual industry movers.
The report is in PDF format (100+ pages), delving into various modalities (small molecules, proximity inducers, antibodies, RNA/DNA therapies, and other biologics), and various areas of biopharmaceutical research, from early drug discovery to clinical research and pharmacovigilance. It also includes AI in synbio, biomanufacturing, and convergence of technologies.
PDF comes with yearly access to our SaaS platform, tracking more than 450+ AI companies. The package also includes consulting hours. If you are interested, request more details and an early-bird quote at info@biopharmatrend.com.
We have lofty discounts for those who express interest before the official annoucement (within 2 weeks).
AI predicts response rate in immunotherapies up to 70-80%
At the Society for Immunotherapy of Cancer (SITC) annual meeting, significant developments were shared by GE HealthCare, Vanderbilt University Medical Center (VUMC), and Universitätsmedizin Essen (UME) in Germany. They have developed Artificial Intelligence (AI) models capable of predicting how patients will respond to immunotherapies with an accuracy between 70-80%.
This breakthrough was achieved through an analysis involving over 3,000 patients who underwent immunotherapy at VUMC, with further validation using a separate cohort of 4,000 patients from UME.
In a major study published in Nature Communications, scientists from French-American techbio Owkin and pathology labs in France validated MSIntuit™ CRC, an AI-driven digital pathology diagnostic tool for colorectal cancer. MSIntuit™ CRC, the only CE-marked AI diagnostic for Microsatellite Instability (MSI) testing from digital H&E slides, demonstrates a 96% sensitivity in identifying MSI in colorectal cancer, which is essential for guiding immunotherapy treatments.
This tool can rule out nearly half of the patients with microsatellite stable tumors, who are unlikely to respond to checkpoint inhibitor therapy, thus optimizing the screening process. The study's strength lies in its broad validation across multiple pathology labs and different slide scanners, showcasing the tool's potential to ease lab workload, reduce costs, and advance precision medicine by identifying actionable biomarkers from a single slide.
In another news from Owkin, it collaborated with 10x Genomics, to integrate 10x Genomics' spatial omics and single-cell technologies into the MOSAIC project, a landmark $50 million initiative focused on cancer research.
MOSAIC, unveiled in June 2023, aims to map tumor structures at single-cell resolution across 7,000 tumor samples using 10x Genomics' Visium and Chromium platforms for spatial and single-cell transcriptomics.
This integration, powered by OWKIN's AI, will enhance the understanding of cancer biology, patient responses, and resistance to treatments, and assist in the development of new therapies.
Recursion has advanced its collaboration with NVIDIA and Bayer and launched a new partnership with Tempus Labs, to bolster its precision oncology efforts.
The deal with Tempus grants Recursion access to over 20 petabytes of oncology data, aiming to enhance therapeutic development through AI.
Recursion will also expand its NVIDIA-powered supercomputer to rank among the world’s top 50, aiding its drug discovery pipeline.
The renewed focus with Bayer will drive up to seven oncology programs, with Recursion positioned to potentially receive over $1.5 billion in success-based payments.
Medidata, a Dassault Systèmes company, announced at NEXT NEW York two new experiences for enhancing and streamlining clinical data workflows: Clinical Data Studio and Health Record Connect.
Key highlights from these new products include:
Clinical Data Studio is an advanced, artificial intelligence (AI)-enhanced data experience to streamline data integration, standardization, and management across many use cases by pulling data from multiple sources like never before. This solution leverages AI to automate tasks for faster insights and better decision-making so that sites and sponsors can get to cleaner, submission-ready data even sooner.
Health Record Connect is the first scalable solution for connecting clinical research to healthcare data. This offering seeks to gain access to as many U.S.-wide patient health records as possible to help clinical trial sites to complete Medidata Rave EDC forms significantly faster using existing patient health record data. With this solution, users will have 93% connectivity to the top 100 clinical trial sites supported by Medidata Rave and access to data from over 400 other academic sites.
Evidence generation needed to prove AI’s value in pharma
Werngard Czechtizky, who leads the medical chemistry department for respiratory and immunology at AstraZeneca, emphasized the importance of educating and updating the company's workforce on AI technologies.
She noted that AstraZeneca employs AI throughout various divisions, including chemistry and biology. According to her, AI and machine learning are set to become integral components of all work areas.
mRNA’s next trick? Reprogramming off-the-shelf cell therapies for cancer and autoimmune diseases [Ryan Cross, Lei Lei Wu, Endpoints, 2023]
Carisma Tx, Capstan Tx, Orna Tx, and Moderna are innovating CAR-therapies by developing in vivo methods that transform immune cells using injectable mRNA therapies, eliminating the need for complex external cell engineering.
This approach, while potentially more efficient, requires multiple mRNA doses and is set to be first tested in humans by Orna in 2024, using its circular RNA technology that offers extended expression. Myeloid Therapeutics has already begun human trials with its therapy. Carisma and Moderna's collaboration focuses on engineering macrophages in vivo, targeting solid tumors in the tumor microenvironment (TME).
However, as Dylan from Zetta Ventures highlights, the major challenge lies in achieving targeted delivery to specific tissues and cell types, a hurdle Capstan is addressing by integrating antibodies into its lipid nanoparticle (LNP) delivery system to direct them to T-cells in the spleen.
One of the major setbacks for cell therapies has been the complexity in manufacturing operations, so in vivo CAR-based therapy could be the holy grail (with the exception that it requires multiple mRNA doses).
Tony Ng from GSK is spearheading the use of digital biological twins in oncology, leveraging advanced modeling to tailor cancer treatments to individual patients. This innovative approach involves creating detailed virtual reconstructions of tumors or organs using patient-derived data, encompassing a range of information from omics and medical imaging. These digital twins allow for the prediction of individual responses to cancer therapies and the likelihood of remission or relapse. The process involves building 3D models from patient tissue samples, enabling the testing of various drugs and dosages over time, thus generating a comprehensive set of data. This data is then analyzed using machine learning and Bayesian statistical models to identify key biomarkers that can predict treatment outcomes, aiding in personalized treatment planning and patient stratification in clinical trials.
New Zealand-based SRW Laboratories has partnered with Insilico Medicine, a clinical-stage AI drug discovery company, to develop advanced nutraceutical products focused on longevity, utilizing AI to analyze and process natural ingredients. The AI technology will be specifically trained to screen tens of thousands of natural molecules, assessing their potential in promoting healthy aging and extending healthspan, a process that significantly accelerates development in the nutraceutical space.
Pick of the Week
Founded in 2013, Pelago Bioscience introduced CEllular Thermal Shift Assay (CETSA) technique that enables researchers to bypass potential setbacks by gaining early insights into the mode of action and validating their drug discovery projects at an earlier stage.

CETSA, developed by Pär Nordlund and further refined by Daniel Martinez Molina at the Karolinska Institutet, is a novel method for studying protein interactions in living cells, based on the principle that protein melting temperatures vary with ligand binding.
This technology, which does not require labeling or tagging, making it suitable for studying challenging 'undruggable' targets like protein-protein interactions and transcription factors, has four formats:
CETSA Navigate and CETSA Navigate MS for probing a few proteins with multiple compounds,
CETSA Navigate HT for high-throughput screening,
CETSA Explore, a mass spectrometry-based method analyzing systemic effects on 6,000 to 8,000 proteins.
CETSA Explore's utility in pathway mapping allows for a comprehensive view of a drug's impact on target and downstream proteins, proving effective in identifying new therapeutic nodes and developing assays, as demonstrated in studies on inflamed PBMCs and their responses to anti-inflammatory drugs.
Digging Gems
I’ve just discovered The Century of Biology newsletter here on Substack, and I must say, it is certainly worth having a look at. The post that led me to uncover this gem is the one exploring various biotech business models, mainly ‘platform companies’:
Traditional Biopharma Companies: These companies often focus on developing a single drug or a small set of drug assets based on academic research. They face a strategic choice between raising capital through equity (dilutive funding) and financing development through partnerships (non-dilutive co-development). The goal is to increase the company's value based on clinical milestones before likely acquisition by a larger pharma company.
New Modality Platform Companies: These platforms, exemplified by Genentech with its recombinant DNA technology, aim to commercialize biotechnologies capable of producing multiple products. Their financing strategy differs from product companies, as early partnerships are more appealing given the distributed future value across various potential assets. The strategy includes creating a cycle of financing further platform development and advancing assets before entering new partnerships, aiming for a mix of internal pipeline development and external partnerships.
Insights Platform Companies: Companies like Millennium Pharma, which focus on generating therapeutic insights through technologies like high-throughput genetics, follow a similar partnering logic to modality platforms. They aim to partner early to avoid dilutive funding, relying on the significant value of technological breakthroughs. However, they face the challenge of their data potentially becoming commoditized and the pressure to quickly transition towards vertical integration.
Services Platforms Transitioning to Therapeutics Platforms: Many service platforms, which focus on external partnerships without an internal pipeline, tend to evolve into therapeutic platforms. This transition is driven by the limited applicability of a single platform across different therapeutic areas and the need to maximize value through internal drug development programs.
Nuances and Strategic Considerations: Each business model has its nuances and strategic considerations. For instance, the nature of the platform (modality or insights) influences its financing and partnering strategies. The post underscores the importance of understanding these nuances for effective value creation in biotech, as per Steven H. Holtzman's analyses and frameworks.
This summary does not catch even 1/10 of the original post ideas and great examples, so make sure to read it:
Another gem for today are these children’s books about synthetic biology! Unfortunately, Substack won’t allow me to embed the Twitter URL with an image, so make sure to check it on Twitter.
This reminded me of another book, Neural Networks for Babies, from the Baby University series. Great stuff, and I am planning to gift the whole series, Baby University, to my toddler daughter for her next birthday!