Mobile Menu

Highly Multiplexed Tissue Imaging to combat Tumour Heterogeneity

Tumours are complex entities and current pathological techniques do not capture that complexity. However, the new suite of spatial tools, in this case spatial proteomics, present a light in the dark for properly analysing tumour samples.

Prof. Bernd Bodenmiller, the Director at the Department of Quantitative Biomedicine, University of Zurich is the primary developer of Imaging Mass Cytometry (IMC), alongside a number of computational tools, and recently spoke at a Front Line Genomics webinar.

In this article, we will give an overview of the challenge of tumour heterogeneity, discuss the IMC approach that the Bodenmillar lab pioneered and look at some application case studies.

To hear the full set of webinar talks for the Single-cell and Spatial ONLINE webinar series, please follow this webinar link.

Interested in Spatial Proteomics? Check out our free Spatial Playbook resource

Tumour heterogeneity – the main obstacle to cancer treatment

In the clinic, tumours are analysed based on morphological and molecular characteristics (a few proteins in-situ). This information can result in patients being stratified into different phenotypic groups and receiving different treatment plans. However, tumours are much more heterogeneous than a few markers can delineate. Each patient’s tumour is a unique and complex entity.

The aim of tumour research is to quantify a tumour with enough depth to distinguish the different cell types and the differences in their cell co-ordination networks. The ultimate goal of such work is to produce personalised treatment plans to reflect true tumour heterogeneity, rather than the approach we currently use. The goal is to always get the right drug to the right person (see Figure 1).

Figure 1. Getting the right drug to the right person through understanding the tumour. Image taken directly from FLG webinar (February 6th, 2024), full credit – Prof. Bernd Bodenmiller

This leads to some fundamental questions we’d love to ask for every tumour in the clinic (and feed the information to drug developers and oncologists) but currently are not equipped to:

  • Which cells are present in the tumour, and can they be targeted directly?
  • How can we deal with tumour heterogeneity?
  • Which proteins and pathways are deregulated?
  • How can we mobilise the immune system to attack the tumour?
  • How can we disrupt the tumour cell-cell communication?

How to get a comprehensive view of each tumour ecosystem

Spatial and single-cell proteomics presents the bridge between pathology and genomics to actually answer some of the above questions and pull apart tumour heterogeneity to inform treatment (see Figure 2). Pathology approaches typically allow some distinction of tumours through the few proteomic markers that are used. Genomic approaches allow the classification of hundreds and thousands of markers, but the DNA/RNA information is only an indirect measure of the functional state of the tumour.

With 40+ protein markers, spatial proteomics is the method that can allow much more to be deduced. This includes protein levels, activity of pathways, cell-communication networks and the nature of higher-level tissue structures. The method that the Bodenmiller lab created to meet this need was Imaging Mass Cytometry (IMC), the technology currently captured by the Hyperion imaging system from Standard Biotools.

Figure 2. Spatial and single-cell proteomics presents the bridge between genomics and classical pathology. Image taken directly from FLG webinar (February 6th, 2024), full credit – Prof. Bernd Bodenmiller

Imaging Mass Cytometry works by staining tissue with metal isotopes bound to antibodies. These can then be imaged in a multi-channel system. No special slides or tissue sections are needed, standard pathology tissue works fine. This creates rich data for research, drug development and diagnostics.

Impressive applications of Imaging Mass Cytometry

One example of IMC being applied came from a study analysing tumour samples from 281 breast cancer patient using IMC1. These patients were able to be grouped into a number of unique clusters based on the IMC data. These groups could also be grouped by survival and the IMC data meant that they could predict which patients would survive and those with worse outcomes. This is associated to the proteomic nature of the tumour identified by IMC.

More recently, and on a much larger scale, the Bodenmiller group analysed 1,000 samples from non-small cell lung cancer (NSCLC) patients2. When these patients were grouped based on proteomic IMC profile, 4 distinct clusters emerged based on the fibroblast composition of the tumour. Furthermore, the group were able to find cell types that, when present, predicted a poorer prognosis. Again, the different patient groups also had distinct survival curve patterns (see Figure 3).

Figure 3. Results from the non-small cell lung cancer study. Left – the four distinct patient groups based on fibroblast type composition. Centre – how the presence of a fibroblast cell type predicted prognosis. Right – the survival curves of the four patient subgroups. Image taken directly from FLG webinar (February 6th, 2024), full credit – Prof. Bernd Bodenmiller. Published research – Cords, et al. 2

Webinar Q&A’s (summarised – not direct quotes)

Q: How long does this diagnostic approach take for each patient?

A: From the moment we receive the sample, we have a turnaround time of 48 hours to provide the clinical report to the oncologist or pathologist. From the moment we receive the sample to data analysis and report generation, everything is fully automated using robotic systems.

Q: Is this applicable to other pathologies and tissues like COPD, and how can this be validated?

A: IMC can definitely be applied to other pathologies and tissues. The starting point is to identify the targets that you want to detect with the antibodies, build the panel and validate the panel accordingly. If the goal is to create a pathway to the clinic, then you should start with patient cohorts that are biobanked and further optimize. In the end, if you have antibodies that are relevant, you can apply this technology to any kind of pathology.

Bernd’s talk continued on to discuss his Tumour Profiler framework and the benefits of coupling his approach to AI . To hear his full talk and talks from the other speakers on this webinar series, please use this link.


1.            Jackson, H.W. et al. The single-cell pathology landscape of breast cancer. Nature 578, 615-620 (2020).

2.            Cords, L. et al. Cancer-associated fibroblast phenotypes are associated with patient outcome in non-small cell lung cancer. Cancer Cell (2024).