Paediatric thalamic tumours: a predictive MR phenotype

To utilise MR phenotyping in accurately predicting paediatric thalamic tumour type via qualitative MR analysis and quantitative diffusion histogram analysis, with the aim of optimising pre-surgical planning and patient outcomes.

Patients with tumours epicentred in the thalamus were identified through an electronic patient record search (2000 to 2024). Radiological features evaluated included tumour margins, homogeneity, cystic components, T1 and T2 signal, contrast enhancement, and haemorrhage. Histopathological tumour type, WHO grade, and molecular subtype were recorded. Tumour masks incorporating both solid and cystic components were manually drawn on T2-weighted b0 images and ADC histograms were constructed and analysed.

34 patients (12 male, 22 female) were included, with a mean age of 10.4 years (range 2.4 months to 16 years). Diffuse midline gliomas (DMG) accounted for 24% and pilocytic astrocytomas (PA) for 28% of the cohort. Further tumour types included low-grade astrocytomas (9%), DNETs, PXAs, and embryonal tumours (8.8% each), with germinoma, glioblastoma, supratentorial ependymoma, ganglioglioma, lymphoma, and ATRT each comprising 3%. ADC p10 was significantly lower in DMG than PA (877 ± 152 vs. 1053 ± 152 × 10⁻⁶ mm²/s; t = 2.25, df = 13, p = 0.042, Cohen's d = 1.16). ADC mean showed a large but non-significant difference (p = 0.068). PAs were well-defined and microcystic with brush-like enhancement. DMGs were ill-defined with patchy enhancement, lower ADC values, and frequent susceptibility artefact.

Distinctive qualitative and quantitative MR features, particularly ADC p10, can guide the neuroradiologist in predicting tumour type and informing the decision between safe maximal resection and biopsy-first surgical pathways.