ESTRO 2024 - Abstract Book

S938

Clinical - CNS

ESTRO 2024

of tumour progression. Several cell types contribute to producing an environment permissive to, and supportive of, tumour growth, invasion and immune escape and cell-cell communication also promotes tumour progression.

We utilised the Cardiff University Brain Research Imaging Centre’s (CUBRIC’s) capacity for advanced magnetic resonance imaging (MRI) by using the 3T CONNECTOM microstructure MRI scanner. This comprises the UK National Microstructure Imaging Facility and is one of only 4 such systems in the world. This advanced system allows tissue microstructure to be visualised in fine detail and, globally to our knowledge, these microstructure MRI scanners have not previously been used to image brain tumours.

Material/Methods:

We performed ultra-strong gradient diffusion MRI using the 3T CONNECTOM scanner in patients with brain metastases (BM) receiving linear accelerator-delivered stereotactic radiosurgery (SRS) at Velindre Cancer Centre in Cardiff. We performed microstructure MRI scans at baseline prior to treatment and at 1- and 3-months following SRS treatment. MRI scans focused on assessment of the BM and surrounding normal brain tissue. We utilised the vascular, extracellular, and restricted diffusion for cytometry in tumours (VERDICT) MRI sequence. This sequence is based on DTI sequences and obtains multi-shell DTI images in 60 directions to investigate brain microstructure. It has not been utilised in patients with BM before. We applied VERDICT to give estimates of the following parameters within the BM: Intracellular Space (fIC), Vascular Space (fVASC), Extracellular, Extravascular Space (fEES) and Cell Radius. We also applied our Soma And Neurite Density Imaging (SANDI) model(1) to the microstructure scans to develop quantitative maps to estimate cell size, density and anisotropy and extracellular volume fraction and we combined this with our SPAARC radiomic analysis(2,3) with the microstructural MRI maps. In total, 17,13 and 8 patients completed baseline,1-month and 3-month MRI assessments respectively. Using VERDICT, we demonstrated differences in cell radius, fIC and fEES compared to the normal brain tissue at baseline and following SRS, albeit without demonstrating statistical significance. In this cohort of patients, the BM showed higher fIC and lower fEES compared to oedema. In a patient with disease progression, we observed reducing intratumoural fIC with increasing fEES and cell radius measurements at 1 and 3-months compared with baseline. Conversely, in a patient who had a complete response to treatment, we observed increasing intratumoural fIC and decreasing fEES and cell radius measurements at 1-month with resolution of these changes to that of normal brain at 3-months. We also demonstrated a trend toward reduced median fIC at 1-month in patients who had progression at their first MRI scan compared with those with stable disease. Using the SANDI approach in a patient with a BM from colorectal cancer at baseline and 1- and 3-months post SRS our quantitative mapping highlights (figure 1A) microstructural heterogeneity in both oedema (white contour) and BM (black contour), which follows different temporal evolution after treatment. Results suggest that oedema is characterised by enlarged extracellular space that progressively reduces with time, whilst the BM is characterised by small cells of elongated shape that reduce in density 1-month and increase back to the baseline levels after 3 months. The quantitative maps were analysed with our radiomics SPAARC software to extract standardised radiomic features, and using hierarchical clustering algorithms we identified sets of features with similar temporal Results: