CT features of colorectal schwannomas: Differentiation from gastrointestinal stromal tumors

Ji Hee Kang, Se Hyung Kim, Young Hoon Kim, Sung Eun Rha, Bo Yun Hur, Joon Koo Han

Research output: Contribution to journalArticle

1 Scopus citations

Abstract

Purpose: To find differential CT features of colorectal schwannomas from gastrointestinal stromal tumors (GISTs). Materials and Methods: CT features of 13 pathologically proven colorectal schwannomas and 21 GISTs were retrospectively reviewed. The following CT items were analyzed: size, longitudinal and transverse location, shape, margin, homogeneity, necrosis, surface ulceration, calcification, degree of attenuation, the presence of enlarged lymph node (LN), and metastasis. Among the features, significant variables were evaluated using univariate statistical tests. The optimal cut-off point of tumor size was obtained by ROC analysis. Binary logistic regression analysis was used to find the most independent CT variables. Results: Small size, non-rectum location, smooth margin, homogeneous high attenuation without necrosis, and the presence of enlarged LNs were found to be significant variables to differentiate schwannomas from GISTs (P<0.05). The optimized cut-off point for tumor size in distinguishing GISTs from schwannomas was 3.9 cm (AUC = 0.808, sensitivity = 66.7%, specificity = 92.3%, P<0.0001). Binary regression analysis revealed that only non-rectum location remained independent predictor for schwannomas differentiated from GISTs (odds ratio = 31.667, P = 0.001). Conclusion: Colorectal schwannomas usually located in non-rectum and appear as small subepithelial nodules showing homogeneous high attenuation and smooth margin. Schwannomas exclusively accompany with enlarged LNs.

Original languageEnglish
Article numbere0166377
JournalPLoS ONE
Volume11
Issue number12
DOIs
StatePublished - Dec 2016

Fingerprint Dive into the research topics of 'CT features of colorectal schwannomas: Differentiation from gastrointestinal stromal tumors'. Together they form a unique fingerprint.

  • Cite this