Abstract 4583: <i>In-situ c</i>haracterization of small blue round cell tumor proteome and N-glycome

Abstract Background: Small blue round cell tumors (SBRCT) is an umbrella term for different tumor types with similar histological presentation that can originate from entirely different tissues. Correct diagnosis is crucial, though challenging, as treatment strategies and prognosis vary significantly between tumor entities. Matrix-assisted laser desorption/ionization (MALDI) mass spectrometry imaging (MSI) is a label-free analytical technique to analyze and visualize the spatial distribution of different classes of molecules, like peptides, glycans, lipids, and metabolites. In this study, MALDI-MSI was used to profile tryptic peptides and N-glycans from archival SBRCTs samples. The data was used to create classification algorithms to reliably distinguish the different entities and to identify discriminatory features between the tumor types. Methods: SBRCT cases (n = 26) were assembled in a tissue microarray: Ewing sarcoma (n = 5), rhabdomyosarcoma (n = 5), neuroendocrine carcinoma (n = 5), acute lymphoblastic leukemia/lymphoma (n = 5), nephroblastoma (n = 3), and neuroblastoma (n = 3). Every case was represented by 3 to 4 tissue cores to account for tumor variability. One TMA section was subjected to on-tissue N-glycosidase F enzymatic digestions, and a second section with trypsin, followed by matrix application and measurement on a mass spectrometer (rapifleX, Bruker). Afterwards, the sections were H&amp;E stained, digitized and underwent histopathological annotation (QuPath, version 0.4.3)Data analysis was performed using SCiLS Lab (Bruker, v.2024a), Weave (v.1.0, Aspect Analytics), and Python (v.3.9). Weave was used to co-register all datasets, combining the peptide and glycan data into a single dataset for comparison of analytes and histology, and interactive assessment of data analysis. Results: MSI data was employed for the training and validation of different classification algorithms (gradient boosting, support vector machine, k-nearest neighbor, and linear discriminant analysis) using 10-fold cross-validation on pixel level. Multiclass classification was implemented as a one vs. rest (one entity vs. all other entities) and one vs. one (one entity vs. every other entity) approach. For the one vs. rest strategy, mean accuracy of all classifiers was 90.5 % with a mean accuracy of 88.7 % for peptide analysis, 92.4 % for glycan analysis and 95 % for the combined dataset. For the one vs. one approach, mean accuracy of all classifiers was 92.4 % with a mean accuracy of 88.8 % for peptide analysis, 96.1 % for glycan analysis and 97 % for the combined dataset. Conclusions: Peptide and N-glycan mass spectrometry imaging, particularly in combined form, has huge potential to aid in distinguishing between different SBRCT entities. Future work will aim to expand the patient cohort and explore alternative statistical methods to further improve classification performance. Citation Format: Christine Bollwein, Kristina Schwamborn, Thao Tran, Alice Ly, Nathan H. Patterson, Juliana P. Gonçalves. In-situ characterization of small blue round cell tumor proteome and N-glycome [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 4583.