A Simple Automated Method for Detecting Recurrence in High-Grade Glioma

Area of Science:

• Neuroimaging
• Oncology
• Radiology

Background:

• High-grade gliomas often recur, necessitating accurate identification of recurrent disease for effective management.
• Current methods for detecting recurrence can be subjective and time-consuming.
• Focal re-irradiation requires precise localization of recurrent tumor areas.

Purpose of the Study:

• To develop and validate an automated multiparametric MRI analysis for identifying focally recurrent high-grade glioma.
• To assess the utility of this automated method in guiding treatment decisions for recurrent gliomas.
• To establish a tool for improved treatment planning and post-treatment surveillance.

Main Methods:

• Retrospective review of MRI data (T1WI, FLAIR, DWI) from 141 patients with high-grade glioma treated with radiation therapy.
• Development of automated multiparametric subtraction maps to identify patterns of change indicative of recurrence.
• Validation of the method in a cohort of 12 patients with nodular recurrence and assessment in a separate cohort of 4 patients treated with radiosurgery.

Main Results:

• Automated subtraction maps accurately predicted radiologist-identified recurrence in cohort 1 (median AUC = 0.91).
• The model correctly identified recurrent lesions in cohort 2, guiding treatment decisions.
• In cohort 2, treated lesions showed control, while untreated progressing lesions aligned with model predictions.

Conclusions:

• Automated multiparametric MRI subtraction maps can reliably predict nodular progression in previously treated high-grade gliomas.
• This automated approach, utilizing routine imaging sequences, offers a valuable tool for clinical decision-making.
• Prospective validation is recommended for treatment planning and surveillance of high-grade glioma recurrence.