Immunotherapy drugs continue to gain momentum across a growing range of disease areas with blockbusters such as MSD’s Keytruda (pembrolizumab) and BMS’s Opdivo (nivolumab) generating billions in annual revenue since their US Food and Drug Administration (FDA) approval in 2014.
Amid this sustained growth, British biotech IMU Biosciences is working to make immunotherapy drugs safer and more effective. Spun out from King’s College, London and the Francis Crick Institute in 2021, the UK company has designed a platform to assess how an individual’s immune system influences their responsiveness to immunotherapy treatments.
Go deeper with GlobalData
Discover B2B Marketing That Performs
Combine business intelligence and editorial excellence to reach engaged professionals across 36 leading media platforms.
To identify the immune signatures that predict disease progression, treatment response, and patient outcomes for immunotherapies, two things needed to happen. According to IMU CEO John Baker, instrumentation needed to get better, so more parameters could be analysed, while tools for the analysis of vast volumes of data needed to arise. In effect, the rise of artificial intelligence and machine learning (AI/ML) have had a major impact on IMU’s ability to develop and run its platform.
Medical Device Network sat down with Baker to learn more about the company’s mission to reshape the understanding of the immune system, and why.
Ross Law (RL): What factors motivated IMU’s establishment?
John Baker (JB): Our founding thesis was that the immune system is everywhere. If you look at biopharma trials, regardless of therapeutic area, a large proportion are targeting the immune system or aiming to fundamentally alter it. Despite this, going to the doctors, individuals tend to get five components of their immune system measured in terms of white blood cell counts, and that is nowhere near the level of resolution needed to understand an individual’s immune system and use those data to inform treatments. Whether that’s informing trial design, target discovery or disease status, you need a lot more information.
At IMU, we’re routinely looking at more than 2,500 cell compartments, and the platform the team has built starts with a routine blood draw and returns that level of information about one’s immune system within several hours. It does this by looking at around about 70 different proteins on the cell surface of every single immune cell in that sample. From that single blood sample, around 200-250 million datapoints are generated.
US Tariffs are shifting - will you react or anticipate?
Don’t let policy changes catch you off guard. Stay proactive with real-time data and expert analysis.
By GlobalDataRL: Discuss the role of AI and ML and how they factor into your platform.
JB: Classical immunology is a particularly manual process. Our team has built algorithms that effectively determine what is in a sample and classify its value in a more robust and high throughput way than a human could achieve.
The reality is that the immune system is hugely variable between individuals, and dynamic within any one individual over time. Therefore, to be able to put any of those data in context, you need a large dataset to understand, for example, what’s the expanse and shape of immune normal versus what are the specific cell types that are driving a disease state, an outcome, or a likelihood to respond to a drug. To understand these factors, the team have now profiled the immunomes of almost 20,000 individuals, and this is a diverse population in terms of age, ethnicity, and biological sex. Around half of this total are healthy individuals, with the other half from disease specific cohorts across indications including oncology, rheumatology, and neurology.
Having that baseline of ‘normal’ is essential. As will be obvious based on the volume of data, to be able to model which of the cell types matter, is another ML and AI challenge, and we’re now at a point where we can identify the variables within the immune system that are the most valuable, in terms of understanding how a disease is going to progress, across a range of indications.
RL: Can you elaborate on the purpose and vision of the Manifest consortium and explain your role within it.
JB: We joined the Manifest consortium in October 2024 and it’s a UK-based programme investigating patient response to cancer immunotherapies, with the aim to make them safer and more effective.
We’re leading the immune side of that. The dance that’s going on in these patients is a battle between the tumour and the patient’s immune system. However, everyone’s looked forensically at the tumour for the last 20 years or more. No one has characterised the other party in that dance, which is the immune system. Given you’re expecting the immune system to ‘do the business’ here, we’ve long thought it would be a good idea to measure it.
Therefore, we’re excited to be a part of Manifest, and the data are hugely exciting. We’re not far into that study, but we’re already able to show that from a standard blood sample from those patients, before they get their first dose of these drugs, we can predict which patients are going to respond with about 95% accuracy, whereas the best accuracy anyone could get to before was about 60%.
RL: What do you view as the key advantages of IMU’s platform and what are your current clinical programmes?
JB: We’re using the same assay stack and ML models for every indication we look at. For all we’re enabling personalised medicine approaches, we don’t have to tailor the test and the assay and the outcome to every patient that we run. This makes our offering more readily scalable versus next-generation sequencing (NGS) approaches.
For instance, with Manifest, we’re looking at the same markers for solid tumours. Our second programme is in blood cancers, and there we’ve got some extraordinary data on our ability to improve outcomes in the three nastiest leukaemia’s. This is likely our most advanced programme and is currently with the FDA for development track sign-off. We’re hoping to have that up and running in clinical trials in 2026.
Our third programme is in looking at patients who’ve received solid organ transplants and monitoring their immune systems to be able to better predict the likelihood of them rejecting or starting to reject those organs. A lot of the morbidity and mortality post-transplant is rejection, but another big chunk of it is a consequence of immunosuppression and infections and cancers that emerge in the medium term in those patients as a result of immunosuppression. Again, there’s a huge need to be able to routinely, stably and rationally monitor the immune status of those patients to maintain that fine balance between the health of the transplanted organ and the broader health of the patient.
RL: What are your future plans around integrating your platform into clinical care pathways?
JB: How we deploy is somewhat contingent on the particular disease state being looked at. One of our first applications is in haematology oncology, where we have some exciting immune signatures that we believe can help to fundamentally change patient outcomes. There we’re working on a central lab type model to deploy in the US and are currently in conversation with the FDA about how to get that moving at pace.
Other applications we’re developing are likely to be a good fit for more typical hospital type labs, and there we’re currently working with instrumentation partners in cytometry, to demonstrate that we can take signatures that we discover on our platform through into their clinical install bases.
