Understanding CAFs to improve radiotherapy outcomes

In patient tumours, we see a striking association between enrichment in cancer-associated fibroblasts (CAFs) pre-radiotherapy and poor outcomes.

This has been a consistent observation using both AI approaches to analyse routine H&E samples in rectal cancer and transcriptomic analysis in bladder cancer. To me, it’s a fascinating association because we’ve known for decades that radiation generates dose-limiting fibrosis in many normal tissues, but the biology of tumour fibrosis and radiation hasn’t been unravelled in detail.

To begin to gain this understanding, I think there are two important questions: Firstly, how do CAFs in the tumour microenvironment before radiation contribute to the poor outcomes we see after radiation; Secondly, how are CAFs reprogrammed during radiation – in the context of the wider tumour microenvironment – to impact radiotherapy responses?

deCAFinating bladder tumours

It’s an exciting time to be addressing these questions as CAF heterogeneity is a rapidly advancing field, in part because of recent technical advances in highly multiplex immunofluorescence and spatial transcriptomics. These powerful technologies mean we can, with single cell resolution, specifically address which CAF subtypes are most relevant to radiotherapy responses and how they are signalling in the wider tumour microenvironment.

We have clues from previous work that direct CAF-tumour cell crosstalk involving the MAP kinase pathway, as well as signalling via focal adhesion kinase/β1 integrin, may be important. There’s also a detailed literature pointing to a radiation-induced increase in transforming growth factor beta (TGFβ). This TGFβ can drive increased extracellular matrix production by CAFs and the trapping and suppression of effector immune cells away from tumour cells.

So, it looks as though CAFs contribute to both immune and non-immune mechanisms of tumour cell survival during radiation. Only a deeper mechanistic understanding of this will provide targets to combine with radiation in the clinic.

And that is what I am hoping to achieve – in my case in the context of bladder cancer, a tumour type that has historically been somewhat neglected in research. The UK has been a global leader in the development of curative bladder radiotherapy so there are mature clinical trials with samples we are able to interrogate. The particular benefit of radiotherapy in this context is that it enables patients to avoid radical cystectomy i.e. surgical removal of their bladder, a major operation with life-changing consequences.

Picture perfect

Spatial resolution is absolutely critical for our analysis of patient bladder cancer samples. We’ve been surprised by how few of the effector CD8+ T cells are getting to directly contact tumour cells; they are largely stuck in the surrounding stroma.

Taking time to look at tumour images continues to be illuminating. We were expecting that alpha-smooth muscle actin positive (aSMA+) CAFs, driven by TGFβ, would be the most common subtype; instead, in over 170 patient samples a different picture is emerging where other subtypes dominate. We are now working to discern how these subtypes are functioning and their key signalling networks with tumour cells and other cells in the TME.

Our experiments using genetically engineered bladder cancer murine models indicate that lymphocytes are an important part of the radiotherapy response. Targeting CAFs effectively is likely to mean augmenting this anti-tumour immune response because CAFs have many important immunomodulatory functions. However, lymphocytes are highly sensitive to radiotherapy, especially when in a proliferative state. As a result, the key here is the optimal timing of targeted agents, including agents that target CAFs and immunotherapy.

Colleagues at the ICR and Royal Marsden, have led early phase trials combining radiation and immunotherapy in bladder cancer. From these, we are mapping the longitudinal immune response in both blood and urine samples to best understand when the sweet spot is to introduce novel agents.

Using patient samples here is essential – intra-tumoural heterogeneity is more complicated in humans than in murine models.

A further exciting aspect of the project is utilising new proteomic approaches to discover novel targets. Initially we’ll centre proteomic analysis on CAF tumour cell crosstalk before and after radiation in 3D in vitro co-culture systems, but the plan is to expand this to more complex organotypic systems over time.

As for the future – well, my goal is to be using AI algorithms applied to diagnostic samples to identify patients with CAF-enriched bladder tumours in the clinic. The huge expansion in cutting-edge AI approaches in digital pathology means I think there is real potential to be doing this in the next few years.

Alongside this, I want to be able to offer these patients a novel CAF-targeting agent that combines synergistically with radiation and enables patients to keep their bladder and obviously be free of cancer. I hope an ongoing dialogue between CAF biology elucidated from patient samples and mechanistic interrogation in preclinical models will enable us to offer this to patients soon.