Differentiation between gastrointestinal schwannomas and gastrointestinal stromal tumors by computed tomography.

The aim of the present study was to identify computed tomography (CT) features to assist in differentiating gastrointestinal schwannomas from gastrointestinal stromal tumors (GISTs). CT images of gastrointestinal schwannomas (n=15) and GISTs (n=50) were analyzed. The absolute CT values of tumor/aorta during plain scan/arterial phase/venous phase were recorded as tumor plain scan (Tp)/aorta plain scan (Ap), tumor arterial phase (Ta)/aorta arterial phase (Aa) and tumor venous phase (Tv)/aorta venous phase (Av), respectively, and normalized CT values of the three phases were calculated as Sp=Tp/Ap, Sa=Ta/Aa and Sv=Tv/Av, respectively. The difference in tumor CT value between arterial and venous phases was calculated and recorded as Tv-a. CT data including tumor size, contour, margin, growth pattern, presence of calcification, cystic change, hemorrhage, ulceration, perilesional lymph nodes (PLNs), local invasion to surrounding structures, metastasis, ascites, vasculatures, enhancement pattern/degree, Tp/Ta/Tv and Sp/Sa/Sv were evaluated for each patient. Receiver operating characteristic (ROC) curve analysis was used to assess the ability of the CT data to differentiate gastrointestinal schwannomas from GISTs. Compared with GISTs, gastrointestinal schwannomas more frequently demonstrated round contouring, relatively smaller tumor size, a homogeneous enhancement pattern, with the presence of PLNs and a higher level of vasculature (P<0.05), whilst the presence of cystic changes were more common in GISTs compared with gastrointestinal schwannomas (P<0.05). The Sa, Ta and Tv-a of gastrointestinal schwannomas were less compared with those of GISTs (P<0.05). The difference in margin, growth pattern, intra-tumoral calcifications and hemorrhage were insignificant (P>0.05). ROC analysis indicated that tumor size, cystic change, the presence of PLNs, tumor enhancement pattern and Sa demonstrated improved diagnostic potential compared with others [area under the curve (AUC) >0.7], amongst which cystic change demonstrated the best diagnostic ability (AUC=0.82). Size exhibited the highest sensitivity, 90%, and cystic change, Sa exhibited the best specificity, 87%. Quantitative analysis indicated that certain features aided the differentiation between gastrointestinal schwannomas and GISTs using CT imaging.