Progressing Precision Healthcare

12 November 2024

What are the most significant trends shaping the field of precision medicine and how do you see these trends evolving over the next five to ten years?
I believe the most important developments have been in two important complementary areas. The first is our ability to deepen our understanding of the human molecular makeup. We started with genotyping and sequencing small numbers of individuals, but the range of molecular technologies available has accelerated at a pace that few could have anticipated. Costs have decreased, and the number of analyses that can be performed in short timeframes has increased, making it feasible to conduct very large-scale research. This enables us to explore the unique molecular characteristics of humans and patients in a truly revolutionary way.

Additionally, we are now in a much better position than just a few years ago to examine real world patient and population-level clinical data. The availability of data in electronic health records, which are already digitised, makes it possible to use this information for research much more rapidly. When we have opportunities to combine both molecular and clinical data, that is where precision medicine can make significant strides.

Furthermore, a crucial development is our improved computational ability and the advanced methods we now use to analyse this data. All of these advancements make it an exciting time to access, utilise, and examine data in ways that were previously impossible.

Is the UK in a leading position with precision medicine, and how does it compare to other countries?
I am originally from Germany and came to the UK a few years after completing my medical training, largely because of its leadership position and advantages in population health sciences. One crucial factor is the willingness of the UK population to engage with and contribute their data to research, which is not always matched in other countries. The altruism of individuals who contribute to research that may benefit others, even if not themselves, is truly remarkable.

Scientifically, the UK is leading the way in this field. The team I work with is truly inspiring, and my institution is perfectly positioned to drive advancements in precision medicine.

The UK’s capability to conduct exceptionally large studies while maintaining unparalleled resolution of molecular and other data is a testament to its leadership. The next step is to increase the diversity of these studies on a larger scale, which is where Queen Mary University of London plays a key role, given the extremely diverse local population we serve.

A significant advantage of the UK’s approach is its reliance on NHS real clinical data, generated as part of healthcare. Having a single provider and a personal identifier allows for a comprehensive analysis of individuals’ health trajectories over time. This is much more challenging in countries like Germany, where data is fragmented across numerous insurance providers and cannot be integrated as seamlessly.

Additionally, the digitisation and entrepreneurial spirit surrounding genomic data in the UK further contribute to its competitive edge. This combination of factors makes the UK an extremely competitive and attractive place for advancing precision medicine.

How central a role do you believe data-driven approaches and AI are playing in the advancement of precision medicine? What data sets do you believe will be the most important?
They are absolutely essential. It’s not just about using data to predict disease but about improving the accuracy of predictions. Machine learning algorithms and the ability to enhance our models are advancing almost daily. While we are not machine learning experts, we utilise established and emerging tools to distil data into something tangible and potentially implementable, thus having translational capacity.

The advantage of starting with a broad dataset—such as measuring an individual’s entire genome and all detectable proteins in their blood—is that data-driven methods and machine learning algorithms can help identify the most relevant molecules without prior bias or assumptions. This process of distilling data is crucial for moving towards practical clinical applications, rather than ending up with excessively complex models that might not lead to specific tests for particular diseases.

On both fronts, making use of and deriving insights from vast datasets, as well as distilling large-scale multimodal data into what is most informative, is key. This approach helps mitigate the biases inherent in human investigators, ultimately advancing the field of precision medicine.

I should also add that reducing investigative biases does not necessarily mean that the data will always lead to the truth. The quality and validity of results depend on the quality of the input data, which is something we need to consider carefully.

What are some of the challenges of deploying precision medicine approaches at scale?
One major challenge is integrating diverse types of data. Advances in precision medicine largely depend on the ability to harness various datasets within a large and diverse population. Different datasets may offer unique advantages, but unless you can combine them or have access to a large-scale dataset with high-quality clinical data, you face difficulties.

Another challenge is the tendency to remain focused on specific domains or clinical specialties. Many discoveries in precision medicine have been unanticipated, and in areas or clinical specialties where we were not necessarily experts. From a genomics perspective, our approach is to remain agnostic not only about what we investigate but also about the totality of the data we collect from individuals for a particular question. While the question may be specific, our data analysis does not take any particular view, and we strive to investigate across different diseases. This broad approach is crucial because it’s easy to draw false conclusions about the specificity of a finding for a given disease if you haven’t considered its potential relevance to others. It encourages a more comprehensive view of the discoveries we make.

These broad and open-minded approaches present opportunities but also highlight the challenges. Often, data are not integrated or utilised in this way due to traditional thinking. Scientists may still adhere to the notion that a specific hypothesis is required to request data, which can limit the utility of the data and overlook the relevance of pathways across multiple diseases.

In today’s context, where ageing populations frequently present with multiple long-term conditions, it is particularly important to move away from a specialty-focused mindset and consider the relevance of causes that are shared between diseases, i.e. underly the development of related as well as seemingly unrelated diseases. This includes genetic, but also other important modifiable risk factors.

Tell me about the Precision Healthcare University Research Institute in North-East London, and your ambition for the future.
The Precision Healthcare University Research Institute (PHURI) is an institute that spans all faculties, and its ambition is to bring people together to work on precision healthcare-related questions. I was appointed in September 2022 to direct this new institute and build it from scratch. It’s an extremely exciting opportunity, and I am honoured to have been chosen for the role. I love QMUL’s international outlook and strategy, and I’m very grateful that they have given me the opportunity to keep a close to link to Berlin and the part of our team based there.

At PHURI, we are in the process of establishing four centres, each focusing on a specific but inter-related area. We’ve already made progress by recruiting outstanding full-time professors who will lead these centres in healthcare data analytics, genomics and multi-omics, MedTech and devices, and therapeutic innovation. We have also recruited international candidates on a part-time basis to build capacity in areas where we do not yet have a specific centre but wish to expand our expertise. Our aim is to drive innovation in precision healthcare and focus on areas that have not received the attention they deserve. While some diseases are relatively well-researched, that is not true for the vast majority of diagnoses that patients in our community have, this is important to address.

PHURI is now two years old and is currently temporarily based in the same building as the Digital Environment Research Institute, which focuses on AI. However, we are working on developing and planning our move to a new Life Sciences building, in close proximity to the Royal London hospital, which is an extremely exciting investment from Queen Mary University of London. Once developed, this new building will have state of the art facilities, including wet and dry labs and education space, providing us with an amazing capacity to grow.

You recently published a paper in Nature Medicine – tell me about the impact of that research.
The paper (https://www.nature.com/articles/s41591-024-03142-z) explored the potential of large-scale proteomics to measure as many proteins as possible in human blood. This study was conducted in collaboration with both GlaxoSmithKline and several other universities and aimed to use an agnostic approach to test these proteins across various diseases to see if we could improve or predict the onset of disease.

We focused on the first occurrence of these diseases in individuals who were disease-free at baseline. Rather than testing proteins in isolation, we compared our results with other data, including basic risk factors typically included in prediction models and many other blood tests already in clinical use. The results were astounding. We also tested our findings against polygenic scores, which estimate an individual’s genetic risk for specific diseases—a tool that has gained much attention and is used a lot in research, but with hardly any established clinical applications.

For over 60 diseases, the protein models—though derived from an initial pool of over 3,000 proteins—selected between 5 to 20 of the most significant proteins, which were then added to baseline or clinical predictor models. We examined the performance of these models to predict disease onset over a 10-year period. For many of the tested diseases, there are currently no prediction models at all; this includes common and severe conditions. While good prediction models exist for diseases like heart disease and diabetes, many other conditions lack such models, making it essential to improve our ability to predict risk for a broader range of diseases.

The performance of these protein models was impressive and, in most cases, outperformed genetic prediction models. For some diseases, existing clinical diagnostic markers, like blood sugar for diabetes or uric acid for gout, were as good as or better than our models, which would be expected and can almost act as a sanity check.
You asked about the impact, and currently, I must say the impact is limited – as the study was published very recently and the proteomics platform used is a research tool. Surprisingly, I received several emails from GPs whose patients had seen the research in the news and were requesting these tests. It is difficult to disappoint them, because we are not at that stage yet. However, we are excited to collaborate with Professor Mark Bradley, PHURI’s Chair for Therapeutic Innovation to develop tests and methods suitable for clinical use, moving beyond research frameworks with relative quantification to something more directly applicable in healthcare.
One thing I would like to add is that we have made all the results available for people who are interested, through a free online tool. We want to accelerate access to results globally, so if you’re interested, please look at Omic Science (https://omicscience.org/) where this is available.

These proteomic platforms are still relatively expensive, so the fact that this pharma consortium generated these data enabled us to use them for this specific question. If we were designing this from scratch, I would say there are other questions or applications for which we would want to test this. Spending a huge amount of money on a test before someone is ill and has symptoms is quite different from dealing with patients. Such tests need to fulfil important criteria, as they can also cause harm, are not necessarily affordable for many healthcare settings. However, if you have people who are already ill, being able to predict who has the poorest prognosis or is likely to have a severe or fulminant course is much more tangible.

I am extremely optimistic that proteomics will help us predict which patients, i.e. people already in the care of the healthcare system, are likely to have a poor prognoses and more severe course of their disease. It is important to know what the complications and what the excess mortality is for a given disease. Currently, for almost no diseases, except cancer (where we have five-year risk models), we do not have good prognostic models and developing such models is a high priority for us, so we need to design studies that can support this.

There are a lot of universities engaged in Precision Healthcare, tell me what will set PHURI apart? Are there specific opportunities you would highlight in North-East London?

The reason I came to Queen Mary University of London and why I think this is the best place to do this is because of the very close and exciting collaboration between the university and the clinical world. That is extremely rare, and it is urgently needed, and Barts Life Sciences provides that opportunity. The hospital is huge (the Trust is the largest in London and 2nd largest nationally), the patient community is extremely diverse, not just in terms of their backgrounds and their needs, but also in terms of the diagnoses that they have. This means we can study diseases which don’t appear in other walks of life or in other geographies. It has been described as ‘a window into the world’ and that really is true. We have this ability and a diverse local population combined with hospital data which is amazing. The headway that Barts Health have made with digitisation and trying to enable use of this data, along with the investment Barts Charity has made to support this, makes North-East London one of the best places in the world to do this. The next step is to enable the integration of molecular data, given the promise we have seen, and this requires special consent and access to sample – in principle an extension of the Barts Bioresource across many specialties.

How important is the Barts Health Data Platform for improving patient outcomes?
It is absolutely crucial for our success. People use different terms such as “precision medicine,” “precision healthcare,” or “personalised medicine.” We refer to it as “precision healthcare” because our aim is to be more precise in our understanding of population differences with regard to disease causes, treatment effects, disease risk, or prognostic outcomes, with the ultimate goal of achieving better health and care for our patients.
In many ways, this isn’t really new because medicine and standard clinical practice often already implement these approaches; for example, in cancer treatment where we target specific groups of patients and sometimes even individual patients. This is of course complementary to population interventions that are universally applicable, like better population health due to everyone wearing seatbelts or a smoking ban.
For example, we don’t screen everyone for breast cancer; we target people who are at specific risk because that risk increases after a certain age. Precision healthcare is a refinement of these targeted strategies, aiming to enhance their effectiveness.

Are there any specific initiatives that you believe would significantly accelerate progress in precision medicine in the UK?
Locally, at Barts Life Sciences, the alliance between the university and the hospital prioritises bringing together clinical data with biological samples. This should be our utmost priority.

The second priority is more universal. Across the country, many studies link data from various sources, such as secondary care, hospital data, prescriptions, and cancer registries. However, accessing primary care data is currently more challenging. During COVID-19, primary care data was made available for research focused on COVID-19. I believe that the ability to catalyse research into diseases that predominantly present in primary care will depend on having access to these data for research purposes. This, for me, is one of the top national priorities and will enable the UK to make significant progress in diagnosing many common and severe conditions that are still predominantly managed in primary care.

About the author
Professor Claudia Langenberg is the Director of PHURI and Professor of Computational Medicine at the Berlin Institute of Health at Charité Universitätsmedizin, Germany. Claudia is a public health specialist combining her expertise with research focused on molecular epidemiology. Her work integrates large-scale genomic and metabolomic data to discover, prioritise and characterise metabolic pathways and test their causal relevance and specificity across a range of diseases, uncovering genetic influences on thousands of molecules in the blood, using this knowledge to better understand human disease such as type 2 diabetes. In 2018 Claudia received the Helmholtz International Fellow Award and was named in the top 20 female scientists in the UK in 2022, 2023 and 2024. In 2024, Claudia was awarded membership of the European Molecular Biology Organisation (EMBO).

Original Source
Progressing Precision Healthcare - Barts Life Sciences