Heterogeneity assessment and protein pathway prediction via spatial Lipidomic and proteomic correlation: advancing dry proteomics concept for human Glioblastoma prognosis

Prediction of protein pathways from lipid analyses via MALDI MSI is a pressing challenge. We introduced "dry proteomics," using MALDI MSI to validate spatial localization of identified optimal clusters in lipid or protein imaging. Consistent cluster appearance across omics images suggests association with specific lipid and protein pathways, forming the basis of dry proteomics.

The methodology was refined using rat brain cerebellum tissue as a model, then applied to human glioblastoma, a highly heterogeneous cancer. Sequential tissue sections underwent omics MALDI MSI and unsupervised clustering. A MatLab script was developed to process omics image datasets, with SVD data compression, providing unsupervised clustering, independently for each data set. The integration of the Silhouette criterion prediction allowed to determine the most optimized segmentation number of cluster. Differentiated lipid and protein clusters, with distinct spatial locations, were identified. Spatial omics analysis facilitated lipid and protein characterization, leading to a predictive model identifying clusters in any tissue based on unique lipid signatures and predicting associated protein pathways. Application to rat brain horizontal slices revealed diverse tissue subpopulations, including successfully predicted cerebellum areas with distinct lipid and protein pathways.

Dry proteomic workflow was applied on a prospective and retrospective cohort of 50 glioblastoma patients tissues, re-using MALDI imaging and proteomic data from previous study. Resulting analysis confirmed lipid-protein associations on intricate and heterogeneous pathology samples. By correlating data obtained from lipid MALDI MSI cluster identification with patient stratification and prognosis data, we aimed to establish a robust link between the lipid and protein signatures identified through the optimized method and the clinical outcomes of glioblastoma patients. Thus, this study has culminated in the development of a sophisticated prognostic prediction model for glioblastoma patients, amalgamating lipid imaging clusters with corresponding proteomic data to stratify distinct prognostic categories. This crucial validation step not only enhances confidence in the reliability of this approach but also holds significant promise for advancing personalized medicine strategies in the management of this challenging disease.