First things first, the proposed classification are immune suppressing and immune escaping. Now let's go back for some specifics.
The whole immuno-oncology (IO) process can be briefly described as the following cycle.
Tumor cells release neoantigen during apoptosis. These 'foreign' epitopes get picked up by professional antigen presenting cells (APC), mainly dendritic cells. APCs migrate to lymph node and activate T cells. Mature T cells travel back to tumor site and try to lysis tumor cells.
Throughout the IO cycle, four steps in general have been known for their role to compromise IO process:
Factors affecting IO cycle
First, the tumor mutation burden (TMB) reflects the 'foreignness' of the tumor. Some studies use neoantigen instead of TMB by selecting those mutations harboring peptide with high binding affinity to patient's HLA. One thing worth mentioning is that addition of HLA can't guarantee the immunogenicity. The binding affinity between MHC-neoantigen complex and TCR should be measured as well. Alone the line of this recognition process, tumor can disrupt multiple factors to reduce the immunogenicity of the tumor:
Reduce its mutation burden through Darwin selection
Down-regulate HLA expression
Compromise TCR diversity / abundance
APC recognition and T cell activation / recruitment
Tumor can up-regulate a number of immunomodulators that compromise the T cells activation. OX40 and CD28 are two examples of such co-stimulators on the surface of T cells. The absence of these co-stimulators switch off T cells even MHC - epitope and CD4 / 8 perfectly bind.
During T cell differentiation, naive T cells can transform to functionally opposed mature T cell (Th1 or Th2) depending on the type of APC-epitope complex. Figure below shows the T cell differentiation rout and corresponding regulators.
Once T cell matured, they ought to migrate back to the tumor site. During this stage, tumor can disrupt the expression of certain chemokine to diminish the recruitment of T cells to tumor site.
Tumor can induce tumor associated fibroblast (TAF) which forms a 'physical wall' around the tumor lesion and blocks immune cells infiltration.
Tumor may create an either immune supporting environment or immune suppressive environment by expressing different set of genes. Ghost's previous blog have more detailed discussion on THIS topic.
Tumor cells recognition
Effector T cells finally make this far to bind to tumor cell for destruction. However, tumor can switch off effector T cells (mainly CD8) by expressing a number of inhibitory receptors. PD-L1 and CTLA-4 are the most well-known example of such molecules and corresponding blockers have shown stunning therapeutic effect.
Immune suppressing and immune escaping
Now let's take a closer look at all these factors mentioned above. In general, Tumor can either suppress immune activity by modulating TIL composition in tumor environment or escape from immune recognition by modulating all the rest of the process: immunogenicity, APC recognition, TIL localization and tumor cell recognition.
Current model for IO evaluation
When come down to predict tumor progress, we can assume that, as long as immune cells are observed in TME, the IO cycle at early step remains intact (tumor poses proper immunegenicity to initiate immune response and APC presentation / T cell activation remains intact). By doing this, we can simplify our model by only focusing on what is going on within tumor TME.
What's left is the battle between tumor cells and immune cells in TME. Tumor progression now largely depends on the balance between the strength of immune capacity and tumor's ability to escape from it. Luckily, much of these features can be reflected from gene expression. We can:
Dissect RNA-Seq data to infer the fraction of each immune cells (mainly for separating cells that act on opposite functional direction).
Measure HLA expression for tumor immunegenicity check
Measure expression of immunomodulators for chemotaxis capacity
Measure expression of immune checkpoint protein for T cells - tumor cells recognition.
Note that current models only estimate the relative abundance of immune cells. We imagine, for instance, minimum immune cells may not be able to make much difference no matter what they are. Some kind of quantification methods may yield better resolution.
TIL composition as a key determinant of IO status can be estimated by a number of tools using different algorithms. However, the results shows poor consistency.