High‐Grade Atypia in Circulating Tumour‐Related Endothelial‐Like Cells Serves as a Specific Marker for Malignant Thoracic Tumours
Cytopathology · 2026-07 · 37 (4) · 344-352
DOI: 10.1111/cyt.70075 ↗Abstract
ABSTRACT Objective Conventional liquid biopsy primarily focuses on epithelial circulating tumour cells (CTCs). We previously demonstrated that a population of circulating cells in thoracic tumours express the vascular endothelial marker ERG. In this study, we defined these ERG‐positive cells with endothelial morphology in liquid‐based cytology (LBC) as “circulating tumor‐related endothelial‐like cells” (CELCs). Our aim was to establish a morphological grading system to distinguish malignancy‐specific High‐Grade CELC from benign or reactive CELC populations. Methods Preoperative arterial blood samples from 28 patients (25 malignant and 3 benign cases) were processed using a soft‐microporous filtration system (S‐MPF). Captured cells were evaluated by Papanicolaou staining. CELCs were categorized based on LBC morphology into “Low‐Grade” (single cells or clusters < 15 cells) and “High‐Grade” (three‐dimensional morula‐like clusters ≥ 15 cells with marked nuclear pleomorphism). Results CELCs were identified in 13 cases (46.4%). While Low‐Grade CELCs were observed in both cohorts (12 malignant, 1 benign), High‐Grade clusters—representing definitive CTEC signatures—were exclusively identified in 8 malignant cases (5 primary lung cancers, 2 metastatic tumours, and 1 thymic cancer). No High‐Grade features were detected in benign lesions, yielding a specificity of 100% (95% CI: 29.2%–100%) for malignancy. Conclusion Implementing an LBC‐based morphological threshold of ≥ 15 cells allows malignancy‐specific CELCs to be accurately distinguished from the broader population of CELCs. This grading system provides a practical and highly specific liquid biopsy framework, ensuring precise identification of tumour‐associated endothelial clusters even amidst reactive non‐malignant cells. Trial Registration UMIN Clinical Trials Registry (UMIN‐CTR, identifier: UMIN000058385)
EUVIMED