Endocrine Abstracts

Computerized tomography (CT) texture analysis (TA) provides a valuable amount of quantitative data on single voxels and their relationship in a target lesions, giving additional value to the CT scan. CTTA has been applied to several oncologic diseases. We aimed to correlate the data derived from CTTA of pheochromocytoma to clinical, biochemical and histopathological data in 30 consecutive patients. Pheochromocytoma was confirmed at histology. The CTTA was performed on unenhanced, late arterial, venous, and delayed phases images. We analyzed first and second order features and performed correlation with blood pressure levels, urinary metanephrines and histopathological parameters: Pheochromocytoma of the Adrenal gland Scaled (PASS) and Grading of Adrenal Pheochromocytoma and Paraganglioma (GAPP) scores, Ki67 values, and SDHB expression at immunohistochemistry. We performed a multivariate regression model to assess the correlations and a cluster analysis to identify a “radiomic signature”. Urinary metanephrines correlated with skewness (b = 0.905, P <0.001), kurtosis (b = 0.337, P = 0.015), and mean value of the distribution of Hounsfield Unit (HU) in arterial phase (b = -0.007, P = 0.030). Urinary normetanephrines correlated with the maximum HU value (b = 31.649, P <0.001) and the energy in arterial phase (b = 97030.7, P = 0.007). Ki-67 showed a significant correlation with the standard deviation of HU values distribution in portal phase (b = 0.155, P = 0.001) and the skewness of distribution in unenhanced phase (b = -1.118, P = 0.047). PASS score showed weak correlation (R² = 0.182) with the tumor volume (b = 5.33x10^-5, P = 0.037). GAPP score was correlated with energy (b = 112.27, P = 0.006) and maximum HU value (b = 0.002, P = 0.003) in arterial phase, kurtosis in HU distribution of portal phase (b = -0.829, P = 0.023), and tumor volume (b = 2.25x10^-5, P = 0.008). Tissue expression of SDHB correlated with kurtosis of HU distribution in portal phase (b = 0.453, P = 0.031), and shape compacity (b = 3.856, P = 0.001). Two clusters were identified, based on radiomic features. When compared to cluster 2 (n = 7), cluster 1 (n = 18) showed higher compacity (15.4 vs 8.9, P = 0.001), lower volume (10.2 ml vs 42.7 ml, P <0.001), lower maximum HU value in unenhanced (86.1 vs 94.7, P = 0.040) and arterial (197.4 vs 241.9, P = 0.021) phase, and higher minimum HU value in portal phase (-24.2 vs -53.3, P = 0.046). Patients in cluster 2 had higher systolic and diastolic blood pressure at diagnosis, and a higher GAPP score (4 vs 6, P = 0.050). All patients with metastatic disease were included in cluster 2 (0% vs 28.6%), even though the difference was not significant (P = 0.070). CTTA provided useful parameter that may be helpful for a thorough preoperative characterization of the pheochromocytoma.