NIMG-70. COMPARISON OF VOLUMETRIC AND CROSS-SECTIONAL MEASUREMENTS IN DIFFUSE MIDLINE GLIOMAS: RETHINKING RESPONSE ASSESSMENT

Abstract

Abstract BACKGROUND The Response Assessment in Pediatric Neuro-Oncology (RAPNO) criteria, which are based on bidimensional or cross-sectional product (CP) measurements of tumor size, are the standard for assessing treatment response in diffuse midline glioma (DMG). This study aims to investigate the correlation between CP and volumetric measurements in progressive disease (PD) and compare their prognostic impact in pediatric patients with DMG. METHODS We retrospectively reviewed clinical and imaging data from 27 subjects with newly diagnosed DMG, encompassing 16 subjects from the PNOC003 and PNOC007 clinical trials and 11 subjects from the Children’s Hospital of Philadelphia. Tumors were segmented using a pretrained pediatric-specific autosegmentation model followed by manual revisions. Tumor volume and CP measurements were automatically derived from the outlined tumor. Cox regression analysis was used to compare the performance of CP and volumetric measurements in predicting overall survival (OS). The correlation between CP and volumetric thresholds for PD was assessed by linear regression. Comparison of survival of patients classified as PD versus non-PD by CP and volumetric measurements at 7 months post-baseline scan was performed using log-rank analysis. RESULTS Cox regression analysis demonstrated that volumetric measurements were significantly associated with OS (p=0.04), whereas CP measurements were not a significant predictor (p=0.07). An approximate 55% increase in segmented volume corresponded to a 25% increase in CP, which is the definition of PD according to RAPNO criteria (R2 = 0.71). At 7 months, volumetric classification of PD more accurately predicted survival than the CP method (p=0.06 for volumetric vs 0.14 for CP). CONCLUSION Our study showed that volumetric measurements are better predictors of survival compared to bidimensional measurements. We are extending this analysis to a larger sample size, examining survival at additional time points, and incorporating other factors influencing OS, such as treatment approaches.