We recently had the pleasure of speaking with Jan Lichtenberg, CEO of InSphero, to understand how changing market dynamics are enabling broader adoption of human in vitro assays.
InSphero develops organotypic biological 3D microtissues and microtissue-based assays for biochemical compound screening and predictive drug testing applications. It offers human microtissues reflecting organs like the liver, pancreas, brain, and others. The company provides product platforms for in-house use and research partnerships for the outsourced use of its technology. To this end, it provides advanced readouts like high-content imaging, transcriptomics, proteomics, and metabolomics. The company also provides custom development services, including microtissue development and their recently launched Akura flow organ-on-a-chip. InSphero AG was founded in 2009 and is based in Zurich, Switzerland.
• A combination of increasingly expensive and precise treatment modalities (e.g., cell & gene therapies, bi-specific antibodies), shift away from animal testing, and more sensitive omics methods/readouts has primed the market for the adoption of more biologically relevant, sophisticated models
• Spheroids and organoids offer an excellent balance between biological relevance and available readout methods on the one hand, and scalability, robustness, and seamless integration into existing workflows on the other hand
• While currently limited by cost, scalability, and lack of workflow integration, organ-on-a-chip devices hold promise as more complex and relevant models for studying challenging therapeutic areas (e.g., neurology)
• As vendors increasingly consolidate and begin offering E2E solutions to enable greater utility from these models, there is a shift towards using such models in clinical settings and therapeutic applications, especially oncology
Thanks for taking the time to speak with me today, Jan! To start, could you please provide some background on yourself and how you became a biotech founder?
My initial education was in microtechnology, but I always used microfluidics and microsensors in the context of genomics and cell assays. We had done our initial work on 3D cell culture more than 20 years ago at the Swiss Federal Institute of Technology in Zurich at ETH, where I also met my two co-founders. It took about eight years, until we had a technological breakthrough that allowed us to build 3D cell culture models at scale. Our first contacts with Novartis convinced us it was the right time to start a company even though it was 2010 and we were still in the middle of the financial crisis. There were a lot of indicators that showed us that the industry needed more physiologically relevant, human-centric models, in part due to skyrocketing drug development costs but also a clear need to move from animal testing to human testing. And so, we felt it was the right moment for us to take the leap.
That's a very interesting career path! How did your experiences translate into what InSphero is today?
Our mission is to make human and predictive 3D cell models available in all shapes or forms, both fast and scalable to advance science - that's why we started the company. What sets us apart from a lot of our competitors is our desire to bridge biological relevance with scalability and automation as we are moving towards more complex organ-on-a-chip devices. There's a lot of great technology out there, but I don't believe that it meets the industry needs in terms of scalability and compatibility with existing workflows. The engineer in me tries to make things simple and reliable and that's what we pursued with InSphero. We're about 70 people today between our Swiss headquarter and US subsidiary with a full range of solutions for advanced cell-based assays and assay-ready microtissues, which we produce and ship to the end user. Our second business is our collaboration and partnership with pharma where we run whole assay programs, both on the discovery and on the safety side, for our partners and provide the data that they need to make decisions in their workflow.
I've always thought about the major difference between organoids, organ-on-a-chip models, and microtissues as the amount of variability introduced vs the amount of biological relevance you can extract, but it sounds like scalability is a core factor you're thinking about. Would you agree with this or how would you differentiate between these terms?
Researchers in the field need a toolbox of different technologies so that they can pick what they believe is the most relevant and economic solution for any given point in their journey towards bringing the drug to the market. What's great about spheroids and organoids is that they create a substantial improvement in terms of biological value and insight over the classical 2D cell culture or suspension cultures. If done well, they will even retain that scalability and robustness that people have learned to appreciate in 2D cell culture. In terms of the ratio between biological relevance and scalability, I think spheroids are probably the most attractive value proposition on the market. We use the term microtissues for spheroids which are aggregated 3D constructs that are produced from primary cells and are typically co-cultures since most human tissues contain multiple cell types. In fact, it's not only about rebuilding that three-dimensional environment, but also rebuilding the cellular composition and the interaction between different cell types that makes our model so relevant.
We are also tapping into the organoid space, but many people use organoids and spheroids interchangeably. For me, however, organoids are developmental systems that grow and differentiate as they as they mature, while the spheroids are aggregates of already mature cells, which can be produced in a very uniform way. Organoids are often considered to be more biologically relevant because they differentiate into many cell types but they're a lot trickier to handle, require several weeks of production, and they differ in size. This introduces variability from well to well, which you don't have in the spheroid case.
Organ-on-a-chip add even more complexity with flow or mechanical strain along with other external factors. If these factors lead to more productivity, that's a great solution but if it’s just a technical add-on, that doesn't increase the model’s predictivity, then it is a difficult value proposition. Therefore, we stay as simple as possible and as complex as needed. For instance, we choose organ-on-a-chip devices for assessing the interactions between different microtissue types.
To summarize, it sounds like static, well-based methods are more scalable, cheaper, and thus more suitable to early-stage R&D vs organ-on-a-chip methods are more dynamic and better for modelling tissue-to-tissue interactions vs organoids are most expensive, arguably most representative, but with the highest variability. Based on this, what kind of interplay do you see between these offerings?
Our goal is to provide a model that can be used along the whole discovery journey starting with our microtissues in a 384-well format for automated screening, with simple endpoints to create a set of relevant hits. You then take these hits and employ more complex readouts, high content imaging and omics to get a better understanding of your compounds. From there, you can select your candidate compounds and again use the same tissues from the same primary cell composition and put them into our organ-on-a-chip device for an even more complex and detail-rich setting. One of the big advantages of organoids is that you can build them patient specific which presents a huge opportunity to look at a more diverse patient population or even at a specific patient to understand whether a drug might be efficacious for that specific patient.
It’s very important for researchers in the field to have a good understanding of what system has which advantages and for us as technology providers to be clear and transparent about where these technologies can be used and where others might be better to move the field forward.
I can see how the application might dictate which model is used. Is there any trend in terms of where within the drug development pipeline each platform is used?
For drug discovery, target ID and validation, as well as pre-clinical safety studies, spheroids and organoids seem to take center stage more and more. Their excellent balance of biological relevance and breadth of readouts on the one hand, and acceptable cost, easy implementation, and scalability on the other, make them a fantastic solution. More complex organ-on-a-chip systems are useful for mode-of-action deconvolution and other mechanistic studies if spheroids are not sufficiently answering these questions.
Many of the more complex solutions are still in the exploratory phase: industry customers are experimenting with them to understand their capabilities and reliability. For spheroids, however, there is a growing number of customers running this technology on a daily or weekly basis and using the information to make critical decisions in their business. We are one of the few examples where this is happening, partially because our technology is so amenable to industrial use.
What do you see as the major inflection point for clinical applications, thinking about CDxs and IVDs?
We have used tumor biopsies to assess the efficacy of treatments, especially with combinations of different treatment modalities for a specific cancer type, with some very encouraging results. However, we also are not naive when it comes to the complexity of entering the IVD or CDx market. A lot of regulators, payers, hospitals, and stakeholders are involved in this so it’s not something that's moving very fast. One successful example is Hans Clevers’ group, one of the pioneers of organoid technology. They developed a CDx for cystic fibrosis therapy to identify whether a patient will respond to an expensive treatment.
Overall, there are a couple of drivers that make the use of human in vitro cell-based assays, in discovery, safety, and clinical phases, which are going to accelerate the inevitable transformation. The most important one is the switch to novel treatment modalities like cell & gene therapy, novel molecule types, ASOs, bi-specific antibodies, etc. Not only are these treatments more precise and personalized but they are very expensive. And payers will require the drug company to prove the treatment will work so the need for a predictive CDx becomes very relevant. If your treatment is a quarter million dollars and a CDx opens the door to reimbursement, then people will spend $25,000 on that test - it still makes sense on a on a micro and macroeconomic level.
Those are all great points – especially with BlueBird Bio recently announcing a recent $3M gene therapy. On the omics side, we’ve seen somewhat of a push from genomics and transcriptomics towards proteomics and functional assays recently. To your point, do you see these as competing technologies or purely enabling? I could envision a scenario where an organoid CDx is translated into a cheaper functional assay or even genomic signature similar to how the goal of NGS CDxs is often to de-plex into smaller signatures or even PCR.
Having a cutting-edge toolbox of readout technologies at our disposition is a key requirement for being successful in this business. The tremendous progress in the past 10-15 years on the omics side, not just in terms of the technologies available but increased sensitivity, has been pivotal for us we work with varied, miniaturised tissues. Typically, we're talking anywhere between 1,000 - 2,000 cells which puts a hefty sensitivity requirement on these technologies. Biochemical readouts are the preferred type as they are very simple, cheap, and can be automated. You can look at viability, apoptosis, or specific, functional biomarkers that that characterise the fatality or viability of a microtissue. They are very simplistic, though. If you run an ATP assay and lyse the whole tissue, you get one data point, but the tissue is heterogeneous. So, when you are looking at different cell types, you must understand that cells on the outside of the tissue have different levels of oxygenation, nutrients, and drugs / drug metabolites compared to those on the tissue interior. To retain that information, which can be very valuable, we must move to better readout technologies. High content imaging has become a real driver for the industry because we can look at that whole tissue and understand what's proliferating vs what's not proliferating, which cell types are affected by a drug, etc. I think we'll see a lot more adoption as this whole translation from function to RNA expression to protein expression and then to lipid secretion is better understood and documented.
That certainly resonates as we’ve seen a boom in digital pathology, multiplex tissue analysis, and single cell omics recently. Where do you see these technologies in particular fitting in and how enabling are they for human in vitro models?
To me, digital pathology and single cell sequencing are great examples for disruptive technologies. We adopted the former quickly and work with a couple of players in the field to look at the structure of sections of our microtissues. This is relevant in part because we're quite strong in the fibrosis field, where the pathology of the collagen deposition in the tissue and its morphology plays an important role in understanding disease progression and how drugs can change that.
Single cell sequencing is one of the most disruptive technologies in the past 10-15 years. Applications are still very rare in the 3D space but I believe that it is of great importance to understand the heterogeneity of the response that we have in these tissues, looking at rare cell types, rare cell mutations, or short-lived progenitor cells.
Given your position commercially, it’s interesting that your perspective is framed from a patient-centric point-of-view, in which you view these technologies as purely enriching the field rather than competitive if we think about how pharma may spend a fixed budget on various technologies. What would you consider the major competitive technology to these models?
I see your point, but still, if you rely primarily on functional assays, at some point you want to validate your results with an orthogonal readout, so I see room for both technologies co-existing.
In the future, I expect our biggest competitor to be pure in-silico methods. They are not going to replace cell-based assays, but might reduce the number compounds to be screened in the wet lab after filtering in silico. A lot of the information that we generate with and for our customers, but also in our own R&D and our own discovery activities, are fed into bioinformatics systems which we can use different methods of regression analysis and AI to classify this information. In the future, we will not only ship the biological microtissue, but we will also provide the analysis algorithms and databases that will help our end users to interpret the generated raw data to make the right choice.
Increasingly providing E2E solutions and merging them with more software capabilities seems to be the trend across most of precision medicine and we’ve seen this alter market dynamics in various ways, with specialized vendors emerging balanced with increasing market consolidation. How do you see the market dynamics evolving in response to the drivers and changes you’ve mentioned?
The market pull is increasing substantially. For instance, our ASO collaborations have increased 800% over the past two years. The FDA Modernization Act has passed the US and will level the playing field between alternative testing methods and animal testing. This is long overdue and should be considered in Europe and Asia as well.
The big change that hat I foresee in the next few years is consolidation in the organ-on-a-chip and 3D cell culture provider side. I'm convinced that once this consolidation process starts, the willingness of the industry to go exclusively with one of the technology providers is going to increase. Some technologies, like organoids and spheroids, are quite compatible so you might be able to get those from the same source, but organ-on-a-chip companies might be separate.
Do you think this might negate that seamless, connected system you referred to earlier where a customer takes a compound through increasingly complex screening models from the same provider?
Consolidation will bring a seamless experience from discovery to later stage, mechanistic and investigative studies. The next two years are going to be very active in terms of supporting thisdecision making, closing deals and partnerships, and bringing the technologies productively into the industry labs.
What do you see as the major barriers moving forwards?
Generally, a big barrier, psychologically, is defining when is a new model better. What is the benchmark that you work against? Is it the animal model? Is it the 2D cell culture? Then, scalability is a big problem in my opinion. A lot of the organ-on-a-chip technologies are interesting but it's very hard to see somebody testing more than a few compounds with them. The cost, manual nature of the assay, and need for high amounts of the compound for flow-through systems, since you're perfusing for days, are all limiting.
Building a better understanding of the advantages and disadvantages of the various solutions and finding overlap for a specific application has been a bit of a hurdle. For instance, a lot of the organ-on-a-chip devices are made of polydimethylsiloxane (PDMS), which is a great polymer for prototyping. However, it's also highly absorbent and quite porous so the interaction with your new molecules is unknown. A few large companies have even decided to not work with PDMS-based systems anymore because of these concerns.
Finally, the biggest hurdle is the incompatibility with the needs and workflows in the industry. If you have a solution that isn’t scalable enough, too expensive, not robust enough, doesn't work with existing liquid handling or imaging systems, or doesn't produce endpoints of interest, this limits adoption. I think the industry is starting to realize this and is now placing its bets on specific technologies that tick the right boxes.
The workflow incompatibility is an issue I’ve heard of before. Are there any other technical barriers such as inherent heterogeneity, ex-vivo lifespan, etc.?
Heterogeneity depends on your production process. Batch to batch or even plate to plate, we have less than 3% variability in terms of cellular composition and size. These are highly uniform systems - more so than typical 2D cell cultures. Ex vivo life span also depends very much on the technology chosen. We've fully characterized our liver tissues for 28 days including gene expression and functionality, but we have customers that use them for 8+ weeks. However, there are few use cases outside maybe neurodegenerative diseases, where you would go for longer than that simply because of the cost of the assay.
You’ve brought up how the technology varies by application a few times. How do you envisage the future disease areas? To my knowledge, much of what’s done currently is liver toxicity screening, but oncology, rare diseases, and neurology seem like growing applications.
We have a lot of applications in the safety space for two reasons. When we started InSphero, liver safety was a major concern in the industry, because there was no good in vitro model for the liver that would retain liver functionality for more than 3-4 days. We've changed that with our technology. Second, the safety business by nature is recurring as pharma needs to assess compound safety every week while the discovery business is more project-based.
In terms of the disease areas, oncology is going to be a big market. Specific diseases like fibrosis will play an important role across a range of different tissue types. On the safety side, we'll see kidney, cardio, and neuro mainly. There will be more need for diversity as patients have different mutations, ethnic backgrounds, age, and sex. If we can generate this information in vitro, it could be helpful to stratify clinical trials and help bring these drugs successfully into the clinic.
Thank you so much for your insights today, Jan!