Researchers at the Hungarian Centre of Excellence for Molecular Medicine (HCEMM) have identified five dominant patterns of protein-altering mutations in cancer that help determine how tumors interact with the immune system. The research, published in Molecular Systems Biology, showed that rather than just simple tumor mutation burden, five distinct patterns of protein-altering mutations, called amino acid substitution signatures help determine whether a tumor is visible to the immune system or remains immunologically “cold.” The findings add a method to help determine which patients will respond to immunotherapy despite a high mutational burden.
“Despite the diversity of mutational processes, their protein-level consequences converge into just five recurring fingerprints, which can strongly influence immune recognition,” said lead author of the study Szilvia Juhász, PhD, head of the Cancer Microbiome Research Group at HCEMM.
Cancer genomes carry the imprint of environmental exposures such as tobacco smoke and ultraviolet radiation, as well as internal processes such DNA repair errors. These processes leave characteristic mutational signatures in DNA. While such signatures have been widely used to trace cancer origins and exposures, how these characteristics might affect how an individual’s tumor responds to treatment have not been clear. The HCEMM-led team sought to fill this gap by focusing not on nucleotide changes, but on how mutations alter amino acids in proteins—amino acid substitution signatures (AAS).
“We hypothesized that certain mutation sources preferentially generate amino acid substitutions that evade immune recognition, producing immune-cold tumors regardless of tissue or mutation load,” the researchers wrote.
To investigate this possibility, the team analyzed existing exome sequencing data from more than 9,000 tumors across diverse cancer types and then added to this their own data from mutagenesis experiments. Taking this approach allowed the researchers to map links between environmental mutagens, defects in DNA repair pathways, and recurring patterns of amino acid substitutions. This analysis showed that rather than a broad array of outcomes, tumors typically clustered into five dominant AAS categories (AAS1 through AAS5), each with its own distinct function reflecting different underlying mutational processes.
According to the research, AAS1 is associated with tobacco smoke mutagens and oxidative damage, AAS2 with ultraviolet light and base-excision repair, AAS3 and AAS4 with different types of mismatch repair defects, and AAS5 with APOBEC-induced mutagenesis.
In particular, the AAS4 signature was strongly associated with immune evasion. Caused by alkylating agents and mismatch repair deficiency, AAS4 was enriched in kidney and liver cancers and showed a reduced tendency to accumulate hydrophobic amino acids. This biochemical bias led to the production of neopeptides that are less likely to be recognized by immune cells. Tumors dominated by AAS4 displayed immune-desert microenvironments, with low lymphocyte infiltration and other markers of reduced immune activity, and they responded poorly to ICI therapies, regardless of mutational burden.
The discovery helps clarify why some mismatch repair–deficient cancers, which are often assumed to be good candidates for immunotherapy because of their high mutational burden, fail to show durable responses.
By comparison, AAS5-classified tumors accumulated more radical amino acid changes and were linked to immune-rich tumor microenvironments. “These results show that neoantigen quality, not merely quantity, dictates anti-tumor immunity,” the researchers wrote.
The study also noted that a subset of AAS4-dominated tumors still show evidence of immune activity. This is because certain HLA class I variants, particularly HLA-B07:02, were associated with increased immune recognition in these otherwise cold tumors. This allele, which is common in European populations, is capable of presenting proline-enriched neopeptides derived from AAS4 mutations to T cells.
“Tumor visibility to the immune system is not determined by mutation numbers alone, but also by the protein-level patterns those mutations create,” noted senior author Máté Manczinger, MD, PhD, group leader at HCEMM. “These findings support a new framework for truly personalized immunotherapy, integrating tumor genomics with the patient’s immunogenetic background.”
The team’s work was informed by earlier research that has shown a link between specific mutational processes to immune outcomes, including studies of APOBEC activity and UV-driven mutations in melanoma.
These new findings have could have potential clinical benefits. The researchers noted that using AAS signatures in conjunction with biomarker assays could help refine predictions of immunotherapy response and help avoid unnecessary treatments. The researchers noted that their analyses focused on HLA binding and T-cell recognition and that future work will expand to include additional mutagens, tumor models, and HLA alleles, all of which will be essential for translating the uses of AAS into prognostic tools to guide patient-specific treatments.
