Differentiation between non-hypervascular pancreatic neuroendocrine tumour and pancreatic ductal adenocarcinoma on dynamic computed tomography and non-enhanced magnetic resonance imaging.
Pol J Radiol
; 84: e153-e161, 2019.
Article
em En
| MEDLINE
| ID: mdl-31019610
PURPOSE: To determine the differentiating features between non-hypervascular pancreatic neuroendocrine tumour (PNET) and pancreatic ductal adenocarcinoma (PDAC) on dynamic computed tomography (CT) and non-enhanced magnetic resonance imaging (MRI). MATERIAL AND METHODS: We enrolled 102 patients with non-hypervascular PNET (n = 15) or PDAC (n = 87), who had undergone dynamic CT and non-enhanced MRI. One radiologist evaluated all images, and the results were subjected to univariate and multivariate analyses. To investigate reproducibility, a second radiologist re-evaluated features that were significantly different between PNET and PDAC on multivariate analysis. RESULTS: Tumour margin (well-defined or ill-defined) and enhancement ratio of tumour (ERT) showed significant differences in univariate and multivariate analyses. Multivariate analysis revealed a predominance of well-defined tumour margins in non-hypervascular PNET, with an odds ratio of 168.86 (95% confidence interval [CI]: 10.62-2685.29; p < 0.001). Furthermore, ERT was significantly lower in non-hypervascular PNET than in PDAC, with an odds ratio of 85.80 (95% CI: 2.57-2860.95; p = 0.01). Sensitivity, specificity, and accuracy were 86.7%, 96.6%, and 95.1%, respectively, when the tumour margin was used as the criteria. The values for ERT were 66.7%, 98.9%, and 94.1%, respectively. In reproducibility tests, both tumour margin and ERT showed substantial agreement (margin of tumour, κ = 0.6356; ERT, intraclass correlation coefficients (ICC) = 0.6155). CONCLUSIONS: Non-hypervascular PNET showed well-defined margins and lower ERT compared to PDAC, with significant differences. Our results showed that non-hypervascular PNET can be differentiated from PDAC via dynamic CT and non-enhanced MRI.
Texto completo:
1
Coleções:
01-internacional
Base de dados:
MEDLINE
Idioma:
En
Ano de publicação:
2019
Tipo de documento:
Article