Joerg Bewersdorf, Yale University
The pandemic has revealed how much work there is to do in advancing and protecting human health. More than 2.5M people died after infection with SARS-CoV-2 and millions more suffer from its long-term effects. The costs to individuals, families and society are immeasurable.
But the pandemic has also revealed what a difference a scientific breakthrough can make. The scientific and pharmaceutical communities developed revolutionary mRNA vaccines on timescales 10 times faster than was previously thought possible. This advance is saving millions of lives and preventing millions of lost person-years of disability and distress.
We’ve witnessed a remarkable feat of science, collaboration, and global response that shows what is possible. It’s a proud moment for science, but there’s so much more to do.
Beyond the millions lost to COVID-19, 2020 and the years before it were tragically normal. In recent years, we have lost:
- 10M people/year to all forms of cancer
- 1.8M people/year to lung cancer, 685K people/year to breast cancer
- 1.4M people/year to tuberculosis (TB)
- 435K people/year to malaria
Without new advances, 2021 and the years that follow will look much the same.
These diseases have something in common. They are all ultimately caused by changes in the molecules and cells that define how a tissue functions and interacts with the other tissues in our bodies. If we could understand the physiological state of a tissue, we could explain the status of a disease in each person and better predict how that disease would progress.
In the case of TB, we know that immune system responses in more than 90% of the 1.7 billion people exposed to Mycobacterium tuberculosis (or “Mtb”), the active agent of tuberculosis, will successfully manage or even clear the disease. But, the remaining 10% of the world’s infected population propagate a disease that kills more people annually than any other. The antibiotics we have to fight TB require at least 6 months of regular treatment. Many people don’t complete the therapy and this has led to the development of antibiotic-resistant TB, which is cited by the WHO as one of the most significant threats to global health. It is a rapidly growing epidemic in India, China, the Russian Federation and much of Asia. We cannot tell who will clear TB on their own and who needs help, so we over-prescribe antibiotics, increase the risk of antibiotic resistance, and make the TB epidemic even worse.
In the case of cancer, we know that ~30% of the population in industrialized countries will be diagnosed with cancer in their lifetimes and 10% of all deaths will be due to the disease. Diagnosing many forms of cancer earlier in disease progression has improved survival rates, and in some cases, we can intervene successfully. But for a frustratingly large number of patients, our therapies extend life by mere months, instead of providing cures.
We must do better.
We know that infectious and noncommunicable diseases act at the levels of molecules, cells and tissues, so at least part of the solution resides in measuring and understanding their states. Whereas we used to think of cell transitions as one-way and irreversible, we now see that in many cases cells transit between states, sometimes reversing themselves as they respond to changing conditions. These cell and tissue dynamics are new territory that we must detect, quantify, and ultimately model. Measuring the properties of cells and tissues requires technologies that are expensive to buy and run, and require expert, well-trained staff who are in short supply — tools should be much more widely accessible to academic centers, start-ups, SMEs, the whole biopharma industry, and ultimately to clinicians and patients for diagnostic use.
We need a new platform – a ‘tissue time machine’ – that can profile tissue states and predict transitions between states (‘Delta Tissue’ or ‘ΔT’). The platform would provide quantitative, multi-scale, multi-modal information sufficient to build integrated prediction models of key cell and tissue states and transitions. If we are successful, we’ll be able to intervene in diseases earlier and with approaches that are targeted to the individual. We’ll also have an improved understanding of the mechanisms that drive disease, which, in turn, will provide more opportunities for intervention. If we succeed, we’ll begin to eradicate the stubbornly challenging diseases that cause so much suffering around the world.
Such a platform is now possible if we combine the latest cell and tissue profiling technologies with recent advances in machine learning and other computational methods. With this foundation, we can now imagine the tissue time machine, which assembles a rational set of profiling modalities, integrates their outputs and builds predictive models of tissue states and transitions.
To build this new platform, we will need to overcome key limitations and achieve three main goals:
1. Develop and optimize method(s) to select modalities that accurately profile tissue in a given state.
1a. The method(s) should quantitatively assess the value of a set of integrated modalities for predicting different tissue and disease states.
1b. The method should be demonstrably better than expert human judgement with respect to time, cost, resource requirements, and predictive value.
2. Develop new or improve existing individual molecular and structural profiling capabilities, with respect to spatial and/or temporal resolution, number of markers, volume of tissue and/or other assay properties, so as to reveal the states and transitions for the exemplar diseases described in the Platform Demonstration Areas (see full program description).
2a. New or enhanced profiling methods should improve one or all of the following in the context of a sample volume of at least 1 mm3:
i. Number of molecular markers or features routinely detected by 10-100x;
ii. Spatial resolution by 5-10x over what is achievable by conventional light microscopy;
iii. Sample processing time by 5-10x.
2b. A key goal is the expansion and linkage of markers and structures, e.g., establishing the relative value and linkage of molecular markers and features derived from an organelle or cell/tissue structure.
3. Develop a platform that integrates multi-scale, multi-modal data from different states and builds models that predict states and transitions. Inclusion of explainable models in the platform is of interest. Performers working in this goal will:
3a. Identify and implement methods to integrate models or knowledge of state gained in Goal 1, ultimately improving the prediction of profiling methods.
3b. Test the platform against the Platform Demonstration Areas at least annually.
3c. Construct an open data resource to share models and datasets, providing a route to integrate contributions from others and/or commercialize advances, as appropriate.
Platform Demonstration Areas. To demonstrate and validate the ‘tissue time machine’, we have chosen to develop, test, and validate our platform in biomedical contexts that are as broad as possible: an infectious disease, tuberculosis (TB), and two different cancers, triple-negative breast cancer (TNBC) and glioblastoma multiforme (GBM). Each represents a current, unmet biomedical challenge and features a complex, dynamic set of cell and tissue states and transitions. See the full program description for more information. Advances across models and measures should inform each other to improve and validate predictive markers, environmental influences and optimize the key ingredients necessary for promoting healthy network development. It is not necessary to form a large consortium or team to do this. Synergies and integrated system demonstrations will be facilitated by Wellcome Leap on an annual basis as we make progress together towards the program goals.
Jason Swedlow, PhD has expertise in mechanisms and regulation of chromosome segregation during mitotic cell division and the development of software tools for accessing, processing, sharing and publishing large scientific image datasets. He is co-founder of the Open Microscopy Environment (OME), a community-led open source software project that develops specifications and tools for biological imaging. He earned his PhD in Biophysics from the University of California San Francisco. In 2012, he was named Fellow of the Royal Society of Edinburgh.
Who are eligible Wellcome Leap program performers?
Performers are from universities and research institutions: small, medium and large companies (including venture-backed); and government or non-profit research organizations. We encourage individuals, research labs, companies, or small teams to apply in program areas best aligned with their expertise and capabilities. It is not necessary to form a large consortium or a single team to address all thrusts or an entire program goal in an abstract or proposal. Indeed, one of the benefits of our programs is that we actively facilitate collaboration and synergies dynamically among performers as we make progress together toward the program’s goals.