2026 Poster Presentations
P517: MANUAL ABC/2 ELLIPSOID APPROXIMATION VERSUS GOLD-STANDARD DELINEATION FOR VOLUMETRIC ASSESSMENT OF RESIDUAL VESTIBULAR SCHWANNOMA FOLLOWING SUBTOTAL RESECTION
Stephanie M Younan, MPH, BS; Rithvik Ramesh; Lourdes Kaufman, BA; Ruben Hernandez, BS; Nadeem Al-Adli, MD; Philip V Theodosopoulos, MD; Steven W Cheung, MD; Ramin Morshed, MD; Nicole T Jiam, MD; UCSF
Background: In the modern management of vestibular schwannoma (VS), function-preserving subtotal resection (STR) is a frequently employed strategy, necessitating accurate surveillance of residual tumor. The ABC/2 ellipsoid approximation is a time-efficient method for volumetric assessment, but its validity for quantifying irregular, post-surgical tumor remnants remains unestablished. This study aimed to validate the manual ABC/2 method against gold-standard manual delineation for volumetric assessment of residual VS after STR and to determine its impact on clinical classification.
Methods: Patients who underwent subtotal resection (STR) for vestibular schwannoma at a single tertiary center between 2006 and 2024 were retrospectively identified. For the ABC/2 method, maximal anteroposterior and mediolateral dimensions were measured on axial images and craniocaudal dimension on coronal images from contrast-enhanced T1-weighted MRIs. For manual delineation, volumes were independently quantified by 3 institutional raters (SY, RR, LK) using semi-automated software (BrainLab) on MRI scans. Agreement was assessed using Pearson correlation and Bland-Altman analysis.
Results: Among 285 patients with preoperative data, the ABC/2 method demonstrated a strong correlation with manual delineation (r = 0.898, p<0.001) but systematically overestimated tumor volume by a mean of 21.2%. A size-stratified analysis revealed this correlation was weakest for giant tumors (>20 cm3; r = 0.572). For 278 postoperative residual volumes, the correlation was substantially weaker (r = 0.679, p<0.001), and the bias inverted to a clinically significant underestimation of residual volume by a mean of 22.8%. This systematic underestimation led to a substantial misclassification of resection extent. For the extent of resection using categorical buckets (<50%, 51-75%, 76-95%, >95%), the ABC/2 method demonstrated 50.9% agreement with manual delineation.
Conclusions: The manual ABC/2 method systematically and significantly underestimates residual tumor volume following STR of vestibular schwannomas. This inherent bias leads to the over-diagnosis of near-total resection, which may confer a false sense of security and adversely affect decisions regarding surveillance intervals and adjuvant radiosurgery. While a practical tool, clinicians should be cognizant of this limitation. Manual delineation remains the gold standard for accurate postoperative assessment, and correction factors for the ABC/2 method may be warranted when it is used for longitudinal follow-up.
