Revisiting neoantigen depletion signal in the untreated cancer genome

This study is arising from Van den Eynden J. et al. Nature Genetics. Lack of detectable neoantigen depletion signals in the untreated cancer genome. Van den Eynden J. et al. tried to address a very important scientific question: could the immune system eliminate cancer cells with immunogenic mutations in untreated situation? Van den Eynden J. et al. first annotated the human exome into "HLA-binding regions" and "non HLA-binding regions" based on the predicted binding affinity of nonapeptides translated from the un-mutated reference coding genome with type I HLA alleles. They hypothesized that if neoantigen depletion signal exist, the nonsynonymous mutations in "HLA-binding regions" will be negatively selected during cancer evolution, while nonsynonymous mutation in "non HLA-binding regions" will not be negatively selected. This will lead to decreased nonsynonymous vs synonymous mutation ratio (n/s) in "HLA-binding regions" compared with "non HLA-binding regions". They defined HLA-binding mutation ratio (HBMR) as the ratio of n/s in "HLA-binding regions" to "non HLA-binding regions", and reported that HBMRs are close to 1 in different types of cancer after background corrections, meaning neoantigen depletion signals are not detectable in different types of cancer. The fundamental problem of their hypothesis lies in that the actual neoantigens with immunogenicity do not overlap with their defined "HLA-binding regions". Actually, most neoantigens with immunogenicity are not located in "HLA-binding regions", when dissimilarity between mutant and wild type peptide are considered. It is the neoantigen with immunogenicity, but not nonsynonymous mutation in their defined "HLA-binding regions" undergo immunoediting based negative selection. Thus the results reported in that study are fundamentally flawed, and at this current stage we could not draw a solid conclusion as to whether the neoantigen depletion signal exists or not.