Background
The discovery and characterization of tumor-specific antigens are critical for advancing cancer immunotherapies, including next-generation cancer vaccines. A mass spectrometry (MS)-based approach, such as immunopeptidomics, is widely recognized as the optimal method to identify expressed and presented antigens. However, despite its high selectivity, immunopeptidomics could benefit from improved sensitivity, as evidenced by the low (<1%) success rate of single amino acid variant (i.e., neoantigen) detection when augmented with next-generation sequencing. Infinitopes has developed a platform that integrates high-performance laboratory techniques and bespoke AI-driven computation to identify hundreds to thousands of novel cancer-specific antigens from primary tumors.
Methods
Infinitopes presents a cutting-edge immunopeptidomics workflow that combines artificial intelligence (AI), advanced mass spectrometry, and next-generation sequencing for antigen profiling. In Precision ImmunomicsTM, we directly analyze peptides presented by HLA molecules, providing a more relevant and immediate understanding of the tumor immunopeptidome with extremely small sample inputs. Our workflow ensures exceptional depth and sensitivity in antigen detection, performing robustly under various false discovery rate (FDR) controls for the confident identification of different MHC-associated peptide categories.
Results
Performance was first evaluated on the A375M cell line, identifying 15,469 unique peptides from 10 million cells at a 1% FDR in a single shot. Compared to a previously published dataset, we increased total acquired spectra by 25% and doubled the number of novel, non-canonical peptides from A375 cells. DenoVax, our in-house developed computational engine, identified 1,671 to 5,396 more spectra and 1,795 to 2,930 more unique peptides compared to known academic and commercial methods. Applying our pipeline to primary upper gastrointestinal (GI) tumor samples, we identified 1,684 peptides outside the canonical proteome, with 85.5% absent from prior HLA-I peptide lists. A shortlist of these peptides showed significantly higher cancer specificity than existing tumor-associated antigens and clinically vaccinated and published peptides (median 45.8x vs 2.7x cancer over-expression). These novel peptides were prevalent across tumors and could be validated as immunogenic, forming an effective off-the-shelf tumor antigen map. Overall, our pipeline discovers hundreds of novel non-canonical MHC-I presented cancer antigens, providing a fully integrated discovery workflow.
Conclusions
By combining sensitive immunopeptidomics with a bespoke computational engine, DenoVax, we can identify a best-in-class pool of therapeutic antigens with unprecedented precision and present large tumor target maps on a per-indication basis. We have combined our Precision ImmunomicsTM antigens with our proprietary vector platform in our lead asset, ITOP1, and are initiating a Phase 1/2a clinical trial in 2024 in the UK.
Acknowledgements
This work was supported by the Innovate UK [grant number 10086924].