Abstract Introduction and Hypothesis: Prostate cancer (PC) bone metastases (BM) is a debilitating disease morbidity that primarily affects the axial skeleton. PCBM are associated to a variety of bone alterations from increased bone density (osteosclerotic), to reduced bone density (osteolytic). Structurally, PCBM present with irregular bone distribution, loss of collagen alignment and increased porosity that mimic features of woven bone that forms during development, but is not replaced by normal trabecular bone. The irregular woven bone structure persists and adversely affects bone strength and predisposing to fractures. We hypothesize that PCBMs induce dysregulated repair-like bone remodeling activity leading to distinct protein signatures that correlate with the extent of osteosclerotic and osteolytic characteristics in different PCBM subtypes. Methods: We performed tandem mass spectrometry to measure the protein composition of lumbar vertebrae specimens from 3 age-matched controls and 32 cadaveric PCBM specimens: 17 defined as osteosclerotic and 15 as osteolytic based on microcomputed tomography bone volume vs total volume measurements. We employed unsupervised hierarchical clustering and dimension reduction techniques to cluster specimens with distinct protein composition profiles. Differential expression analysis was correlated with PC histopathological in the respective vertebrae. A 20-protein signature was assessed for the ability to segregate PCBM specimens based on expression of the respective transcripts in publicly available RNA sequencing databases of primary and metastatic prostate cancer specimens. Results: Osteosclerotic PCBM specimens are enriched in bone sialoprotein 2, osteopontin, osteonectin, and the proteoglycans osteomodulin, lumican, biglycan and decorin. While osteolytic specimens exhibit elevated levels of cathepsin G, myeloblastin, and neutrophil elastase compared to the osteosclerotic specimens. Unsupervised clustering segregated the 3 control samples and 17 osteosclerotic samples into distinct groups. The 15 osteolytic samples separated into three subgroups: one clustering with controls (n=3), one exhibiting features that overlap with osteosclerotic lesions (n=6), and one unique osteolytic group (n=6). The differences in protein distribution are also associated to the histological characterisation of the specimens and surprisingly, with lesion transcript profiles from PCBMs in 3 annotated datasets. Conclusions: Proteomic profiling revealed differences between predominantly osteosclerotic and osteolytic PCBM that highlight altered mineralization processes, and suggesting potential biomarkers PCBM. The osteosclerotic samples are characterized by proteins that promote bone production and mineralization. The osteolytic lesions displayed signatures of immune cell activity. The proteomic segregation of three PCBM molecular subtypes demonstrates the sensitivity of MS-based analysis, highlighting its promise for further characterizing PCBM subtypes. Citation Format: Dennis Xie, Felipe Eltit, Bita Mojtahedzadeh, Raphaele Charest-Morin, Colm Morriessey, Eva Corey, Lawrence D. True, Michael C. Haffner, Michael E. Cox. Proteomics Classification of Prostate Cancer Bone Metastases abstract. In: Proceedings of the AACR Special Conference in Cancer Research: Innovations in Prostate Cancer Research and Treatment; 2026 Jan 20-22; Philadelphia PA. Philadelphia (PA): AACR; Cancer Res 2026;86 (2Suppl): Abstract nr A076.
Xie et al. (Tue,) studied this question.