P03.06.A PERFUSION CHARACTERISATION OF GLIOMAS USING DYNAMIC [18F]FET PET/CT AND TIKHONOV MODEL-FREE DECONVOLUTION

Abstract BACKGROUND Accurate assessment of blood flow (BF) and blood-brain barrier (BBB) transport remains clinically relevant in gliomas, given their marked heterogeneity. Conventional PET imaging lacks the spatiotemporal resolution and kinetic scope needed for reliable quantification. However, the long axial field-of-view (LAFOV) PET/CT allows simultaneous high temporal resolution imaging of brain and aorta. Utilising these improvements we evaluated [¹⁸F]FET-derived estimates of BF and BBB extraction fraction (E) using model-free kinetic deconvolution. MATERIAL AND METHODS Seven glioma patients (median age 46, range 33-68, 3M/4F, 2 glioblastoma and 5 oligodendroglioma grade 3) underwent 40-minute dynamic [¹⁸F]FET LAFOV PET/CT. Images were reconstructed using 2-second frames across the bolus injection and subsequently modelled with an image-derived input function and Tikhonov model-free deconvolution to derive BF and E from the residue impulse response function. Tumour regions were delineated using a tumour-to-background ratio (TBR) of 1.6, with background defined in healthy-appearing contralateral cortex. Maximum TBR (TBRmax) was mean 3.7 (range 2.0-6.2), metabolic tumour volume (MTV) mean was 46 ml (range 14-123 ml). Tumour BF was compared to contralateral healthy-appearing grey matter (GM). RESULTS Tumours differed from contralateral GM in several key perfusion parameters. E was markedly elevated in tumours (mean 8.3 ± 3.3%, range 4.5-13.2) versus GM (2.4 ± 1.2%, range 1.1-4.9; p<0.001). Tumour BF was variable (46.4 ± 14.7 ml/100ml/min, range 25.2-65.4) but did not differ significantly from GM (61.5 ± 14.7 ml/100ml/min, range 35.0-79.3; p = 0.087). The unidirectional influx constant K1 was higher in tumours (3.6 ± 1.3 ml/100ml/min, range 2.4-6.0) than in GM (1.3 ± 0.3 ml/100ml/min, range 0.9-1.7; p = 0.0015). Mean transit time (MTT) was prolonged in tumours (93.3 ± 35.5 s, range 52.7-154.4) relative to GM (31.9 ± 11.9 s, range 16.5-53.4; p = 0.002). BF values varied substantially across tumours, reflecting underlying heterogeneity. CONCLUSION Quantitative perfusion imaging using dynamic [¹⁸F]FET PET/CT and Tikhonov model-free deconvolution enables estimation of BF and BBB transport in gliomas. Elevated E and K1, and prolonged MTT distinguished tumour tissue from healthy appearing GM. While MTT is likely underestimated due to tracer retention, these findings demonstrate the feasibility of non-invasive characterisation of glioma perfusion and underline the potential of kinetic modelling to capture intratumoural heterogeneity beyond conventional uptake metrics.