ESTRO 2024 - Abstract Book

S3825

Physics - Image acquisition and processing

ESTRO 2024

Purpose/Objective:

Angiogenesis is a hallmark of cancer. In solid tumours this process results in tortuous and leaky vessels which can impede oxygen delivery resulting in tissue hypoxia. Dynamic contrast-enhanced (DCE)-MRI is a widely used technique for assessing tumour microvasculature properties such as perfusion and vessel permeability and has the potential to monitor response to therapy and predict outcome [1]. MR Linac systems are being used increasingly for the treatment of head and neck squamous cell carcinoma (HNSCC). Quantitative imaging techniques such as DCE-MRI could be used to aid treatment adaptation in patients with HNSCC on the MR Linac. Additionally, MR Linac systems offer a unique opportunity for research application of DCE-MRI since many systems have equivalent hardware and software, making delivery of reproducible multi-centre studies viable. However, to date, only one study has been published on the use of DCE-MRI on the MR Linac system [2]. In this work we set out to assess the translational potential of DCE-MRI between a diagnostic and an MR Linac system. We compared parameter value repeatability on both platforms and then assessed radiotherapy response in patients with HNSCC with commonly reported DCE-MRI derived parameters [1]. 25 patients with HNSCC undergoing curative radiotherapy were recruited, following informed written consent, into a prospective clinical trial (ClinicalTrails.gov NCT03646747). Imaging was carried out on either a Philips Ingenia 1.5 T MR system (MR Sim) or an Elekta Unity 1.5 T MR Linac system. Patients underwent scanning at baseline (B1) and at week 2 (W2) of radiotherapy. All patients received between 55 Gy in 20 fractions (#) to 70 Gy in 35#. A subset of patients had two baseline scans (B0 and B1). DCE-MRI was acquired using a 3D T1-weighted fast field echo (FFE) Dixon sequence (3x3x5 mm3, flip angle (FA) = 5°), with IV contrast agent injection (Dotarem, 0.2 ml/kg (0.1 mmol/kg) at 3 ml/s), delivered by contrast power injector (MRXperion, Bayer) at the 8th of 45 dynamic measurement time points. TR, TE and temporal resolution (Tres) were matched as closely as possible; (Diagnostic MR, TR/TE1/TE2/Tres = 3.2/1.2/2.0 ms / 3.8 s), (MR Linac, TR/TE/Tres = 3.9/1.6/2.6 ms / 4.2 s). Native T1 mapping was achieved using the same sequence with FAs = 2, 5, 15°. The DIXON water-only images were used in subsequent analysis. Ktrans and ve estimation, was carried out using Madym open-source software [3], using the standard Tofts model and a population arterial input function. Lesion DCE-MRI parameter repeatability was assessed from double-baseline measurements by calculating the percentage within-subject coefficient of variation (wCV) and the repeatability coefficient (RC). Reproducibility was evaluated by assessing overlap of RC limits of agreement (LOA). Treatment response was assessed via Wilcoxon signed rank test at W2 compared to treatment baseline parameter estimates. An average baseline was taken if patients attended both B0 and B1 visits. P values of <0.05 were considered significant. Material/Methods:

Results:

DCE-MRI data acquired on the MR Linac in patients with HNSCC were comparable to those acquired on the MR Sim. Figure 1 shows parameter estimates as the standard error on the mean at each imaging timepoint across the cohort