Texture analysis of non-contrast-enhanced computed tomography for assessing angiogenesis and survival of soft tissue sarcoma

Objective: To evaluate the role of computed tomographic (CT) texture analysis in assessing tumor angiogenesis and survival of soft tissue sarcoma (STS). Methods: In 20 patients with STSs, tumor texture parameters, which were measured on pretherapeutic CT using CT texture analysis software with the spatial scale filter extracting fine to coarse texture, were compared with microvessel density, plasma vascular endothelial growth factor (VEGF), soluble VEGF receptor-1, and overall survival (OS). Results: Mean of positive pixels (MPP) showed a positive correlation with microvessel density (P = 0.02). Entropy at medium texture scales (spatial scale filter = 3, 4, 5) showed positive correlations with VEGF (P = 0.03, P = 0.009, and P = 0.02, respectively), and entropy without filtration showed a positive correlation with soluble VEGF receptor-1 (P = 0.02). In the univariate analysis, kurtosis at a medium texture scale and MPP showed significant correlations with OS (P = 0.04 and P = 0.007), andmultivariate analysis demonstrated thatMPP was an independent prognostic factor (P = 0.01). Conclusion: Texture parameters are associated with tumor angiogenesis and OS in STS.