Endoscopic Classification of Severe Spontaneous Epistaxis: A Case-Series of 445 Patients

To identify the anatomical location and characterize the histopathological findings of severe spontaneous epistaxis through endoscopic evaluation and tissue biopsy.

A retrospective analysis of prospectively recruited 445 consecutive patients with severe spontaneous epistaxis was conducted at a tertiary medical center between January 2022 and December 2024. All patients underwent systematic endoscopic assessment to localize the bleeding site. Based on endoscopic morphological characteristics, cases were classified as papillary epithelioid elevation-related epistaxis (PEE) or mucosal erosive elevation-related epistaxis (MEE). Lesional tissue was excised when feasible for histopathological analysis. Clinical characteristics, hemorrhagic patterns, therapeutic responses, and clinical outcomes were compared between groups.

Among 445 patients, 358 (80.4%) presented with PEE, while 87 (19.6%) demonstrated MEE. Histopathological examination revealed lobular capillary hemangioma in all 177 biopsied PEE lesions, representing 39.8% of the total cohort. PEE exhibited higher proportions of arterial/pulsatile bleeding (52.2% vs. 24.1%, p < 0.001) and severe blood loss (26.5% vs. 11.5%, p = 0.009) compared to MEE. Anatomical distribution patterns differed significantly (p < 0.001): PEE demonstrated widespread involvement throughout the nasal cavity, including olfactory cleft (20.4%) and inferior meatus (11.7%), whereas MEE predominantly localized to the anterior nasal septum (80.5%). Surgical intervention achieved superior cure rates compared to conservative management in both PEE (95.0% vs. 13.8%, p < 0.001) and MEE (89.7% vs. 58.3%, p < 0.001) groups.

Endoscopic morphological classification into PEE and MEE phenotypes is associated with distinct bleeding patterns and treatment responses. Lobular capillary hemangioma was the most frequently identified histopathological finding among biopsied PEE lesions; however, given incomplete pathological sampling, these findings should be interpreted as hypothesis-generating rather than definitive evidence of causality.

4.