Genomic mutations underlie uncontrolled cell growth but new research demonstrate how to turn these mutations into a tumor’s weakness

For some cancers it takes only three sequential mutations3 in key genes to spark tumorigenesis but as the tumor grows, this number can reach 1012 or more4. DNA mutations are processed into neo-antigens which can be recognized by T cells. These ‘abnormal flags’ allow T cells to specifically attack tumor cells while leaving neighboring healthy cells untouched. The more ‘visible’ tumors are to T cells (i.e. the greater number of neo-antigens they express), the greater their immunogenicity. Importantly for cancer therapy, the authors have devised a model which correlates tumor immunogenicity to clinical responses of immune checkpoint inhibitors.

All somatic cells digest proteins found in the cytosol into short peptides which are then loaded onto HLA class I and displayed on the cell surface. These peptides, or antigens, may be derived from pathogens, self-proteins or mutation-associated cancer neo-antigens. Cancer-associated DNA mutations are translated into proteins classed as neo-antigens when presented to a T cell via HLA. T cells that recognize neo-antigens usually unleash an immune attack by perforating the cancer cell forcing it into apoptotic cell death.

A Fitness Test for Tumors
 
T cells exhibit specificity for a vast array of antigens, which should allow them to eliminate transformed cells; however tumor cells hijack an immune dampening mechanism normally designed to protect the body against uncontrolled immune responses.These immune checkpoint pathways cloak the tumor cells, enabling them to evade the immune system. PD-L1 is one such checkpoint ligand which is often highly expressed by tumors and PD-L1 expression levels have been successfully used to predict therapeutic responses to immune checkpoint blockade.

Yet, not all tumors that regress in response to checkpoint therapies are PD-L1 positive. Balachandran, Łuksza and colleagues have devised a tumor fitness model that calculates the probability of a tumor being recognized by the immune system and whether checkpoint inhibitors will work. Crucially, not all neo-antigens are created equal. By calculating expected binding affinities between neo-antigens and HLA class I, a scoring system was devised to predict the chance of a successful neo-antigen/HLA class I pair being recognized by the TCR. The model was predictive of an immune attack against the tumor and is important as a similar analysis of neo-antigen load did not correlate with efficacy.

Predicting antigen recognition is still a challenge but the model was improved by the assumption that a neo-antigen sharing similarities with an infectious disease epitope known to stimulate T cells was more likely to evoke an immune response. Although this is a work in progress, the model has been designed to avoid ‘over fitting’ such that it applies to other data sets and is future-proofed to move with new discoveries.

  1. Balachandran, V. P. et al. Nature 551, 512–516 (2017).
  2. Łuksza, M. et al. Nature 551, 517–520 (2017).
  3. Tomasetti, C. et al. PNAS vol. 112, 118-123 (2015).
  4. Tomlinson, I. et al. Am J Pathol, 160, 755-758 (2012).