NEURORADIOLOGY OF ADULT-TYPE DIFFUSE GLIOMAS IN THE MOLECULAR ERA: A NARRATIVE REVIEW OF IMAGING BIOMARKERS, RESPONSE ASSESSMENT AND ARTIFICIAL INTELLIGENCE
Abstract
Background. The fifth edition of the World Health Organization (WHO) Classification of Tumours of the Central Nervous System, published in 2021, reorganised adult diffuse gliomas around molecular rather than purely histological criteria. Neuroradiology has consequently shifted from a descriptive discipline concerned with grade and mass effect to one expected to contribute genotype-relevant information, quantify residual and recurrent disease, and adjudicate ambiguous post-treatment findings. Objective. To synthesise, in narrative form, the current role of neuroradiology in adult-type diffuse gliomas, covering acquisition standardisation, structural and advanced imaging biomarkers, radiogenomics, response assessment criteria and artificial intelligence (AI). Methods. We searched PubMed for English-language articles indexed up to September 2026, combining terms for diffuse glioma, the WHO 2021 classification, individual imaging techniques, response assessment and machine learning. Priority was given to consensus statements of the Response Assessment in Neuro-Oncology (RANO) group and allied societies, diagnostic test-accuracy meta-analyses, multicentre validation studies and randomised trials with imaging endpoints. Selection was purposive rather than systematic, as befits a narrative review. Findings. Standardised acquisition remains the rate-limiting step for every quantitative biomarker. Among structural signs, the T2-FLAIR mismatch sign is near-specific but insensitive for astrocytoma, IDH-mutant (pooled sensitivity 42%, specificity 100% across 12 studies and 1,053 patients). Advanced techniques are complementary rather than interchangeable: relative cerebral blood volume distinguishes grades imperfectly when assessed against a rigorous reference standard, 2-hydroxyglutarate spectroscopy detects IDH mutation with wide confidence intervals, amide proton transfer-weighted imaging performs competitively for IDH status and grading, and amino acid positron emission tomography (PET) outperforms MRI for post-treatment recurrence and identifies measurable disease in far more patients than contrast enhancement does. RANO 2.0 unified response assessment across grades, and PET RANO 1.0 provides the first structured framework for amino acid PET. AI has reached credible multicentre performance for segmentation, volumetric response and molecular prediction, but external validation, protocol heterogeneity and reporting quality remain limiting. Conclusion. No single imaging biomarker substitutes for molecular pathology. The realistic contribution of neuroradiology is a standardised, multiparametric and increasingly automated workflow that stratifies patients, quantifies disease burden and resolves post-treatment ambiguity.