MRI features predict survival and molecular markers in diffuse lower-grade gliomas

Area of Science:

• Neuro-oncology
• Radiology
• Medical Imaging Analysis

Background:

• Previous studies demonstrated MR imaging's utility in predicting glioblastoma survival and molecular profiles.
• Such investigations have not been conducted for lower-grade gliomas (LGGs).

Purpose of the Study:

• To investigate the predictive value of MR imaging features and radiomic models for molecular subtypes and progression in LGGs.
• To assess the association of specific MRI features with progression-free survival (PFS) and overall survival (OS) in LGG patients.

Main Methods:

• Presurgical MRIs from 165 LGG patients (grades II and III) were analyzed using Visually Accessible Rembrandt Images (VASARI) annotations.
• Radiomic models incorporating automated texture analysis and VASARI features were developed to predict IDH1 mutation, 1p/19q codeletion, histological grade, and tumor progression.

Main Results:

• Interrater reliability for imaging features was high (k = 0.703-1.000).
• No enhancement and smooth non-enhancing margins on MRI predicted longer PFS and OS.
• Texture models demonstrated superior prediction accuracy for IDH1 mutation (AUC=0.86), 1p/19q codeletion (AUC=0.96), histological grade (AUC=0.86), and progression (AUC=0.80) compared to VASARI features.

Conclusions:

• Specific MRI characteristics, such as lack of enhancement and smooth margins, are significant predictors of improved survival in LGGs.
• Advanced texture analysis of MR imaging provides highly accurate predictions of molecular status and tumor progression in lower-grade gliomas.