High-sensitivity HLA-I immunopeptidome profiling from limited clinical PBMC samples

Background Human leukocyte antigen (HLA) molecules are central to immune surveillance, presenting antigenic peptides from self and non-self proteins to T cells, and directing adaptive immune responses. Assessment of HLA-presented peptides is increasingly critical in oncology, both for understanding tumor immunogenicity and for guiding immunotherapy strategies. Peripheral blood mononuclear cells (PBMCs) offer a minimally invasive and clinically accessible matrix for such analyses, enabling serial sampling and longitudinal monitoring in clinical trials. Application of the immunopeptidomics profiling (IMPX) to PBMC samples has demonstrated clinical relevance, as exemplified by its use in the GRWD5769 trial where pharmacodynamic modulation of the immunopeptidome was observed in patients treated with an ERAP1 inhibitor (ESMO 2024, ASCO 2025). Despite these advances, isolating HLA-associated peptides from limited PBMC material remains technically challenging, underscoring the need for more sensitive and robust methods suitable for routine use in clinical settings. Methods We developed and optimized an unbiased immunopeptidomics (IMPX) workflow utilizing native lysis and magnetic bead-based immunoprecipitation to isolate HLA class I associated peptides. Lysis and pulldown parameters were systematically refined to maximize recovery and specificity of HLA complexes from limited input material across diverse sample types, including cell lines, tumor tissues, and PBMCs. Downstream liquid chromatography-mass spectrometry (LC-MS) conditions and data processing pipelines were further optimized. Results To evaluate the performance of the workflow under clinically relevant conditions, a ramping experiment was conducted using 1 to 10 million PBMCs, reflecting cell yields commonly obtained from trial samples. Incorporation of mild detergent (n-dodecyl beta-D-maltoside, DDM) during immunopeptide elution from HLA complexes significantly improved overall recovery, particularly at inputs of ≤5 million PBMCs. Following optimization of both the biochemical and computational pipelines, up to 2,300 HLA class I-associated peptides were confidently quantified from as few as 1 million PBMCs. When 5 million PBMCs were used as input, more than 9,500 unique HLA-bound peptides were identified. In summary, this high-sensitivity, optimized immunopeptidomics platform enables comprehensive class I HLA profiling from minimal PBMC input, supporting its translational application as a pharmacodynamic readout in clinical studies.