Predicting outcome in childhood diffuse midline gliomas using magnetic resonance imaging based texture analysis

Area of Science:

• Pediatric Oncology
• Neuroradiology
• Medical Imaging Analysis

Background:

• Diffuse midline gliomas (DMG), formerly diffuse intrinsic pontine gliomas (DIPG), are aggressive pediatric brain tumors with poor overall survival (OS).
• Accurate prediction of OS is crucial for refining treatment strategies in pediatric DMG patients.
• MRI-based texture analysis (MRTA) offers a novel approach to quantify tumor heterogeneity and may serve as a valuable imaging biomarker.

Purpose of the Study:

• To investigate the utility of MRTA in predicting OS in pediatric patients diagnosed with DMG.
• To identify specific MRTA parameters that correlate with prognosis in childhood DMG.

Main Methods:

• Retrospective analysis of 32 pediatric DMG patients treated between 2007-2017.
• MRTA performed on T2W and Apparent Diffusion Coefficient (ADC) MRI sequences acquired at diagnosis and post-radiotherapy.
• Filtration-histogram technique using TexRAD software quantified texture parameters; Kaplan-Meier analysis assessed association with OS.

Main Results:

• T2W-MRTA parameters demonstrated superior prognostic capability compared to ADC-MRTA.
• Homogeneous tumor texture on diagnostic T2W MRI, particularly at a medium scale, was associated with worse prognosis.
• The Mean of Positive Pixels (MPP) texture parameter effectively stratified patients into poor (median survival 7.5 months) and good (median survival 17.5 months) prognostic groups.

Conclusions:

• A more homogeneous tumor texture on diagnostic MRI in pediatric DMG indicates a worse prognosis.
• The texture parameter MPP is identified as the most significant predictor of overall survival in childhood DMG.