Neuroimaging-Based Classification Algorithm for Predicting 1p/19q-Codeletion Status in IDH-Mutant Lower Grade Gliomas

Area of Science:

• Neuro-oncology
• Radiology
• Genetics

Background:

• Isocitrate dehydrogenase (IDH)-mutant lower grade gliomas are classified based on 1p/19q-codeletion status.
• Accurate classification is crucial for determining appropriate treatment strategies.

Purpose of the Study:

• To test and validate neuroradiologists' performance in predicting 1p/19q-codeletion status using simple neuroimaging metrics.
• To develop and validate a classification algorithm for predicting codeletion status in IDH-mutant lower grade gliomas.

Main Methods:

• A training dataset of 102 IDH-mutant lower grade gliomas from The Cancer Genome Atlas was analyzed for imaging features.
• A 2-step classification algorithm was developed using features like T2-FLAIR mismatch, texture, patient age, T2* susceptibility, primary lobe, and hydrocephalus.
• The algorithm was validated on an independent cohort of 106 institutional IDH-mutant lower grade gliomas by two independent neuroradiologists.

Main Results:

• The classification algorithm achieved 86.3% accuracy in predicting codeletion status on the training dataset.
• Independent validation demonstrated prediction accuracies of 81.1% and 79.2% for the two readers.
• The algorithm's metrics showed moderate to substantial interreader agreement (κ = 0.56–0.79).

Conclusions:

• A classification algorithm based on reproducible neuroimaging metrics and patient age was validated.
• The algorithm demonstrates moderate accuracy in predicting 1p/19q-codeletion status in IDH-mutant lower grade gliomas.
• This tool can aid in classifying gliomas and guiding treatment decisions.