Cancer cell intrinsic mechanisms of immune escape can be thought of in several ways. Evolving mutations may not be presented at all by the HLA complex or recognised by a CD8/CD4 T cell through the TCR, presenting no immediate fitness disadvantage to an emerging clone.
However, a productive HLA-peptide mutation interaction and efficient presentation to T cells may result in cancer cell cytotoxicity, attrition and a cancer cell fitness cost. Such mutations are referred to as neoantigens.
In theory, selection for fitter subclones might manifest in several ways.
Firstly, an emerging clone expressing a neoantigen could be extinguished entirely by the immune system, resulting from purifying selection. However, a mutation predicted to be a potent neoantigen may occur in a gene that is rarely, if ever, expressed in cancer. Such putative neoantigen may never present any fitness cost to the emerging subclone.
Therefore, taking into account whether or not the neoantigen is expressed is clearly an important component of understanding immune editing.
Secondly, if the neoantigen is expressed, and the emerging clone escapes immediate T cell mediated cytotoxicity, there may be alternative routes to immune evasion. Tumour cells may be selected for loss of HLA (HLA LOH) or mutations in vital components of the antigen presentation machinery. Tumour cells may also lose neoantigens later in evolution through chromosomal instability. Alternatively, failing selection for any of the above mechanisms, tumour cells may suppress expression of the neoantigen altogether.
TRACERx, a longitudinal study of lung cancer is revealing hints for all these processes in early stage untreated disease.
Intriguingly, these processes seem to differ depending on the immune microenvironment and whether it is hot (heavily infiltrated with immune cells) or cold.
Early insight into the possibility of purifying selection against neoantigens in lung cancer expressed genes is seen across the cohort regardless of immune phenotype; hints of a fossil record of historical immune editing.
In immune hot tumours, we see signs of mutual exclusivity of HLA LOH events and mutations in the antigen presentation machinery, suggesting that selection occurs for one mechanism or the other, but rarely both. Alternatively, if selection for either of these events fails, transcriptional repression of the neoantigen can occur. Through RRBS analysis, we found that in approximately 25% of cases, this event could be attributed to promoter methylation of the neoantigen.
So, what about tumours with no immune infiltrate? In these tumours, we find evidence for copy number loss of previously clonal neoantigens or a declining evidence of immune editing later in the disease course, consistent with the absence of an immune predatory environment.
There is much to learn, including how dynamic transcriptional repression is, to what extent all these mechanisms may play a role in resistance to immunotherapy, and whether the dynamic nature of epigenetic silencing, could hold clues as to why some tumours respond a second time to checkpoint inhibitor blockade. Regardless, just as evolution held the key to understanding targeted therapy resistance, it is increasingly likely that viewing cancer through an evolutionary lens will provide further insight into how escape ensues from a predatory immune environment.