DDW Editor Reece Armstrong speaks to Benedikt Nilges, Head of Technology and Data Analytics at OMAPiX about spatial biology’s use in drug discovery and bettering our understanding of disease.
RA: How does spatial biology enhance our understanding of the human body and associated diseases?
BN: Our bodies consist of an incredibly complex network of about 35 trillion cells divided into as many as 300 cell types that fulfil distinct roles in development, homeostasis, and disease. To truly understand the function of these cells and ultimately how the human body works, we need to know how cells interact with each other, with tissue structures, and with pathogens in their local environments. This can only be achieved by identifying cells within their native tissue context at single-cell resolution while simultaneously characterizing them at the molecular level.
Today, spatial omics technologies enable imaging-based identification of cells in their native tissue context while simultaneously detecting hundreds of RNAs or tens of protein markers in the same cells. Spatial omics technologies enable us to reveal how cells are organised in tissues and how their molecular makeup changes in the multitude of tissue niches. Analysing spatial changes during disease progression or treatment response is key to fully understand pathologies.
RA: With spatial biology being cross-disciplinary – in that it uses techniques in both gene expression analysis and immunofluorescence – is expertise in the field limited?
BN: Expression analysis and imaging-based analyses have indeed been performed by completely separate groups of researchers so far, with little to no interaction between them. Fortunately, we have learned a lot about omics and imaging analysis over the last few decades and all the expertise is there. For an efficient spatial experiment, we have to make sure that we start with an interdisciplinary group of researchers from the beginning, with experts like pathologists and image analysis professionals for sample selection and cell segmentation and bioinformaticians for expression analysis. In the medium term, we will also see the emergence of full-on spatial biologists who combine expertise in both fields, but till then working together is more than just a stopgap solution. The expertise in the field is growing and a talented pool of scientists who have the experience and knowledge to deliver cross-disciplinary data is a niche but growing group.
RA: Can spatial biology studies help drive advances in precision medicine?
BN: Yes, spatial analysis will help develop precision medicine, initially through advancing the speed of research and progression of therapeutic approaches and later as part of the treatment regime.
Spatial omics technologies are ideally suited for characterising cells and their interactions within a tissue. This is an invaluable approach for precision medicine because it promises a detailed view of desired and undesired effects of treatments with cell and gene therapy agents or antibodies. Let’s take CAR T cells as an example. Here we are still facing unresolved challenges for their application in solid tumours, such as poor infiltration into the tissue, immunosuppressive environments, CAR T exhaustion, and the potential for off-target effects. Spatial omics can gather information on all of these factors in a single experiment: identify which CAR T clones successfully infiltrate a tissue, how their activity is impacted by the local environments in the tissue, what their cell-state is, and how the surrounding cells are impacted by their presence. This can have a dramatic effect on CAR T R&D and accelerate the transition from bench to bedside.
Precision medicine is all about custom treatment options for patience, which requires deep knowledge of a patient’s individual pathology. Spatial omics can provide just that – for example by identifying the different subtypes of tumour cells and where in the tumour they are located to make sure that the right therapy and the correct target are chosen.
RA: What advantages does spatial biology have over techniques such as cell culturing in relation to studying diseases?
BN: Cell culture will remain important for drug screening and for understanding the basic mechanisms of action for therapeutics, but it lacks the complexity and spatial organisation of true tissues. This limits its efficacy in later stages of drug development or in the discovery and characterisation of biomarkers. This is where animal models and patient samples are absolutely required but have often not been used to their full potential due to analytical challenges. Spatial omics offers a new way of understanding these systems by providing a holistic view of healthy and diseased cells, pathogens, and the immune response.
RA: What have been some recent insights discovered using spatial biology?
BN: Most published spatial biology efforts have focused on a deeper understanding of homeostatic tissues to form the basis for more targeted endeavors into pathologies of these tissues. A good example of this is recent work from the labs of Martin Guilliams and Charlotte Scott, who used spatial protein and RNA readouts to generate an unprecedented atlas of human and mouse livers and were able to characterise the hepatic macrophages induced in fatty liver diseases1. Similarly, a combined effort from researchers at MD Anderson and UC Irvine was able to generate an atlas of human breast, including many of the resident immune cells that form the first line of defence against emerging breast cancers2.
RA: Will AI and computing advances help researchers better analyse data from spatial biology studies?
BN: The analysis of spatial biology data is complex and computationally intensive. A single patient sample can yield terabytes of data. This makes the first phase of analysis – the transition from raw data to a map of cell types and interactions – very computationally demanding. Any advances on the hardware and software side can significantly boost the usability of these technologies. Machine learning is already used for this part of data analysis — for example, to enhance cell segmentation — but the main advantages of artificial intelligence lie in the interpretation of spatial maps across large sample cohorts with different backgrounds. This will be essential for spatial technologies to transition closer to the clinic, because we ultimately need to transfer complex data into simple conclusions and actionable treatment decisions.
Spatial data can also boost the performance of AI. Training of AI algorithms designed to interpret large imaging datasets in clinical pathology, for example, could be improved by annotating training data with the help of spatial omics instead of relying solely on human annotation of the imaging data.
RA: What are some of the technical limitations of spatial biology?
BN: Current spatial technologies strike a balance between the number of RNAs or proteins they can detect and the sensitivity or specificity of detection, making it difficult to navigate the technology landscape and select the right platform for a specific project. Here are some the major challenges:
- Integrating spatial multi-modal data can be challenging if it was generated with vastly different technologies and data formats.
- Computational power needed for analysis and storage of large datasets is a limitation today and needs to improve.
- There is a lack of data and case studies beyond pilot projects related to patient recruitment and clinical trials. This kind of information will be necessary for generating confidence in the technology.
- Education and awareness across the different stakeholders to drive adoption of spatial omics into clinical programmes.
RA: What does the future look like for spatial biology?
BN: The future of spatial biology will be shaped by the expansion of current capabilities, like the possibility to not only detect more RNAs or proteins in the same experiments, but to also detect both on the same sample with relative ease, lower cost, and higher throughput. At the same time, we will also see a maturation of spatial analysis with the rise of standard formats for images, localisation files for cells and molecules, and quality metrics. Together, these developments will enable a transition of spatial technologies towards the mainstream of research, later stages of clinical development, and ultimately the clinic. So, the future is bright indeed, but will take several years to emerge.
RA: Will spatial biology need to incorporate multi-omics approaches such as metabolomics and lipidomics to help better define tissues and cells?
BN: Yes, an integrated multi-omics approach will be critical for a complete and comprehensive analysis, depending on the use case. Spatial transcriptomics, spatial proteomics, spatial metabolomics, or spatial lipidomics utilised independently cannot be sufficient to understand the biology of disease.
The first step to reach this goal will be to generate protein and RNA data from the same sample. This enables precise identification of cells and tissue structures and combines a quantitative transcriptomics readout with detection of proteins that are key effectors in signaling pathways. The integration of lipids and small molecules on top of that will enable new use cases, like the spatial characterisation of the effect of small molecules, but is unlikely to be as widely adopted as transcriptomics and proteomics. Another layer of information that is missing so far and has implications for the use of spatial technologies in oncology is the ability to probe for mutations at the DNA level and connect them to changes in spatial gene expression and cellular activity.
RA: As a research tool, how accessible is spatial biology to teams throughout academia and pharma?
BN: Groups in academia and pharma are starting to bring spatial technologies into their labs, some with a specific use case in mind and others with the intention to benchmark and understand the possibilities these tools offer. However, the barrier to entry is still relatively high. Due to the interdisciplinary nature of the technology, even the decision about whether to place instruments in a sequencing or imaging core often causes some confusion. Samples need to be available, sectioned, processed, and analysed; each of these steps can be a significant hurdle for smaller labs. This is where we believe service providers can make an impact as they take over a lot of the complexities of technology selection, project execution, and data interpretation to make spatial omics accessible for everyone, from first-time users to experts.
DDW Volume 24 – Issue 4, Fall 2023
- Guilliams et al. 2022, https://doi.org/10.1016/j.cell.2021.12.018
- Kumar et al. 2023, https://www.nature.com/articles/s41586-023-06252-9
Since 2022, Benedikt Nilges has been leading the analysis and technology team at OMAPiX, where he strives to make the latest spatial omics technologies accessible to a wider audience. Prior to that, he worked in QIAGEN’s R&D division and was one the first scientists to join Resolve Biosciences.