S127\u2005A comparison of CT and MRI volumetric assessment of malignant pleural mesothelioma

Abstract

Introduction Primary tumour (T-) staging in malignant pleural mesothelioma (MPM) is difficult due to complex tumour morphology. Volumetric assessment is a potential alternative to current T-staging. Computed Tomography (CT) volumetry has been limited by laborious manual segmentation methods or high inter-observer variability, frequently due to perception error or insufficient contrast between tumour and adjacent tissues. Magnetic Resonance Imaging (MRI) offers naturally high contrast between effusion and pleura and functional elements of MRI (e.g. perfusion) could offer additional advantages. Methods T1-weighted, isotropic, gadobutrol-enhanced 3-Tesla MRI and iodinated contrast-enhanced CT scans were acquired in patients with MPM. CT images were acquired as part of routine clinical activity. Non-contrast or pulmonary arterial-phase CT examinations were excluded. Images were acquired prior to biopsy in all cases. MRI analysis involved semi-automated generation of a contour mask, followed by perfusion-tuned tumour segmentation using Mryian® segmentation software (figure 1). This utilised signal intensity limits for region-growing derived from previous MRI perfusion studies in the same cohort. CT volume analysis involved manual segmentation using Myrian® software (figure 1). Inter-observer agreement was compared and the relationship between overall survival (OS) and MRI and CT T-volume was examined. Results 31/31 and 28/31 patients had MRI and CT volume analyses respectively. Using MRI, mean analysis time was 16 minutes, mean T-volume was 370 (SD 137) cm3 and inter-observer agreement was excellent (ICC 0.962). Patients with high MRI-derived T-volume (≥300cm3) had a poorer median OS (20 months versus 8.5 months, p=0.009). MRI T-volume was an independent predictor of OS at multi-variable analysis (HR 2.11 (95% CI 1.05 – 4.3). Using CT, mean analysis time was 151 minutes, mean T-volume was 302 (SD 102) cm3 and inter-observer agreement was moderate (ICC 0.72). There was no significant relationship between CT T-volume and OS (20 versus 12 months in patients with high (≥300cm3) and low T-volume (<300cm3) respectively, p=0.13). CT T-volume was not predictive of OS at univariable (HR 1.91 (95% CI 0.77 – 4.7), p=0.17 or multi-variable analysis. Conclusion MRI-derived T-volume appears to have superior reproducibility and shorter analysis time than segmentation using CT. MRI T-volume is an independent predictor of OS in patients with MPM.Abstract S127 Figure 1 Segmented primary tumour volume in a patient with Malignant Pleural Mesothelioma at contrast-enhanced MRI (figure la, segmented volume highlighted in blue) and at contrast-enhanced CT (figure lb, segmented volume highlighted in green)