Enrichment-Free Rare Cell Profiling in Pediatric Osteosarcoma Reveals a Circulating Cell Population Associated with Relapse-Free Survival

Background: Outcomes for patients with osteosarcoma (OS) presenting with metastases at diagnosis or relapsed remain poor with five-year survival rate of only 20–30%. Risk stratification at diagnosis remains limited, and no validated prognostic liquid biopsy analyte currently exists for this disease. OS lacks the recurring mutational hotspots and epithelial surface markers that underpin liquid biopsy strategies developed for carcinomas, necessitating fundamentally different approaches to circulating biomarker detection. Methods: Peripheral blood samples from 25 pediatric and young adult OS patients at clinically apparent disease timepoints (n = 29 samples) and 76 adult healthy donors (ND) were processed without cell enrichment using fluorescent whole-slide imaging. An outlier detection pipeline employing deep contrastive learned features classified rare cells across eight immunophenotypic categories defined by DAPI, vimentin (V), cytokeratin (CK), and CD45/CD31 (CD) expression. From the same patient cohort, plasma cell-free DNA (cfDNA) was analyzed by ichorCNA-based tumor fraction estimation across 42 samples (10 clinically apparent disease timepoints shared with the rare cell cohort and 32 alternative timepoints) from ten patients, and single-cell copy number alteration (scCNA) profiling was performed on 80 rare cells from five patients. Results: Rare cells were detected at a median of approximately 127 cells/mL (range: 29–1114) in OS patients at clinically apparent disease timepoints versus 34 cells/mL (range: 4–225) in healthy donors. DAPI-only (D) cells, identified using a data-driven optimal cutpoint (38.32 cells/mL) that is subject to optimization bias and requires independent validation, were associated with relapse-free survival in this exploratory, single-center OS cohort (log-rank p = 0.0003), with high D cell burden associated with markedly shorter relapse-free survival (median 393 days, 95% CI: 106–572) compared to patients with low D cell counts (median 835 days, 95% CI: 494–1671). D cells also showed the broadest clinical associations, correlating with metastatic burden and disease subgroup. The Vimentin+/CD45− (D|V) population was the strongest discriminator between OS patients and healthy donors (p = 4.64 × 10−12) but did not distinguish relapse-free survival groups within the OS cohort (log-rank p = 0.8367). Genomic analyses at the plasma, single-cell, and tumor tissue levels were largely uninformative. Conclusions: Enrichment-free rare cell profiling identifies a DAPI-only circulating rare cell population whose abundance is associated with relapse-free survival in this exploratory, single-center OS cohort, accessible by nuclear morphology alone without biomarker enrichment. This is a novel, hypothesis-generating observation that, together with the mesenchymal-forward D|V population characterized here, motivates a larger, independent, prospective validation cohort and dedicated single-cell molecular (proteomic) characterization of the D-cell population, rather than establishing DAPI-only rare cell enumeration as a validated liquid biopsy stratification tool at this stage.