Using ex vivo avatars to understand immune reactivation in cancer

Human ex vivo avatars can provide the missing link between classical animal models of cancer and descriptive translational work that uses tumour biopsies and histological techniques. The ex vivo tumour fragment platform (Nat Med 2021;27:1250–61) that we developed allows one to perturb pathways in the human tumour microenvironment, thereby making it possible to directly test how cancer (immuno)therapies influence intratumoural immune activity. 

The model involves growing small fragments of patient-derived tumour tissue in the laboratory and monitoring the effects following exposure of these samples to, for instance, anti-PD-1 agents. Using this approach, we have obtained some very exciting findings that shed new light on the mechanism of action of the PD-1 inhibitors that are so widely used in the clinic. For instance, prior work by others that involved comparison of pre- and post-therapy biopsies had demonstrated that PD-1 blockade was followed by an influx of novel T-cell clones into tumour tissue, and this had led to the suggestion that anti-PD-1 agents mediate their effects not so much at the tumour site, but rather in lymphoid tissue. However, using the avatar model, we were able to demonstrate convincingly that tumour tissue is a – if not the –primary site of activity. Firstly, PD-1 blockade led to immune reactivation in some tissue samples but not others, and such immune reactivation was strongly associated with clinical response. Secondly, we could demonstrate that anti-PD-1-induced immune reactivation consists of two phases in which an initial boosting of local T-cell activity results in an interferon-γ-dependent release of inflammatory chemokines. Furthermore, we speculated that this wave of secondary chemokine production drives the recruitment of novel T-cell clones to the tumour site, as previously reported by others. This study also revealed some findings that could have particular relevance for selecting which patients will benefit most from PD-1 inhibitors. Specifically, we found that reactivation of immune cells was predominantly observed in tumours harbouring tertiary lymphoid structures – miniature, organised immune infiltrates that resemble lymph nodes – and that this was independent of the tumour type. These findings underscore that a radical change is required in the way we describe tumours. We must try to move away from a categorisation based on cell of origin. A more rational and meaningful approach, particularly in immuno-oncology, is to define tumours based on an understanding of their tumour microenvironment composition and the factors that facilitate immune reactivation.

What does the future hold for ex vivo avatars? One avenue is to further explore the function of tertiary lymphoid structures to determine if these really are the engines that reignite a tumour-specific response. We therefore want to look at the spatial organisation of tumours, particularly with regard to tertiary lymphoid structures, to try to map where reactivation is localised. This will help us to further understand the mode of action of immunotherapies. And we are aiming to extend the use of ex vivo avatars beyond investigation of PD-1 inhibitors to other novel molecules and combination therapies, for example a combination of an immune blocking agent and a targeted cytokine. Looking at the platform itself, as with any experimental system, ex vivo avatars are not a perfect representation of the physiological system they are modelling. For example, while they enable the effective, detailed capture of parts of the immune response, this is, as yet, limited to short-term activity of just a few days. One important goal for the future is to investigate if the model can be developed to look at processes taking place over a longer time period.

Since the platform was created, increasing numbers of researchers have expressed interest in using ex vivo avatars to investigate cancer treatments. Personally, I do not feel that our ambition should be to develop this technologically demanding approach into a platform that can move into mainstream diagnostics. Rather, we should consider this technology to be an instrument we can use to understand how therapies influence immune activity at the cancer site and then apply that information to identify biomarkers and thereby develop simpler assay systems that are suitable for use in clinical diagnostics.