(872) Predicting Vestibular Schwannoma Growth Using Gadolinium-Enhanced MRI: Diagnostic Accuracy and Inter-rater Reliability

Introduction: Predicting vestibular schwannoma (VS) growth is critical for treatment planning. We previously reported that growing tumors exhibit significantly higher signal intensity on gadolinium (Gd)-enhanced MRI compared to non-growing tumors in a cross-sectional study (Yamada et al., Laryngoscope, 2021). Based on this finding, the current study aims to validate whether signal intensity ratios can serve as prospective predictors of future tumor growth and to evaluate the inter-rater reliability of this method to ensure its clinical reproducibility.

Methods: A retrospective review was conducted on 37 patients with sporadic VS managed between 2010 and 2020. "Growing" tumors were defined as those with ≥20% volume increase, while "non-growing" tumors showed < 20% change over an observation period of ≥5 years. Three examiners (one expert, two inexperienced with 7 and >20 years of practice) independently measured signal intensities. Regions of interest (5 mm²) were placed on the tumor, pons, and temporal muscle. Diagnostic cut-off values established by the expert (T/Np: 1.47, T/Nm: 1.41) were applied to predict growth. Inter-rater reliability was assessed using Fleiss’ kappa coefficient.

Results: The diagnostic accuracy for predicting tumor growth across the three examiners ranged from 75.7% to 86.5% for T/Np and 83.8% to 89.2% for T/Nm. The T/Nm ratio demonstrated consistently higher accuracy than T/Np. The inter-rater agreement for growth prediction was substantial for both metrics. Fleiss’ kappa coefficients were 0.687 (95% CI: 0.501-0.873, p< 0.001) for T/Np and 0.713 (95% CI: 0.527-0.899, p< 0.001) for T/Nm. Diagnostic agreement among all three examiners was achieved in 83.8% of cases using the T/Nm ratio.

Conclusions: Quantitative measurement of signal intensity on Gd-enhanced MRI, particularly the T/Nm ratio, demonstrates substantial reproducibility for predicting VS growth, regardless of examiner experience. This method may serve as a robust objective biomarker to assist in treatment decision-making.