Differentiation of Clear Cell Renal Cell Carcinoma From Other Renal Cortical Tumors by Use of a Quantitative Multiparametric MRI Approach.
AJR Am J Roentgenol
; 208(3): W85-W91, 2017 Mar.
Article
em En
| MEDLINE
| ID: mdl-28095036
ABSTRACT
OBJECTIVE:
The purpose of this study was to develop a quantitative multiparametric MRI approach to differentiating clear cell renal cell carcinoma (RCC) from other renal cortical tumors. MATERIALS ANDMETHODS:
This retrospective study included 119 patients with 124 histopathologically confirmed renal cortical tumors who underwent preoperative MRI including DWI, contrast-enhanced, and chemical-shift sequences before nephrectomy. Two radiologists independently assessed each tumor volumetrically, and apparent diffusion coefficient values, parameters from multiphasic contrast-enhanced MRI (peak enhancement, upslope, downslope, AUC), and chemical-shift indexes were calculated. Univariate and multivariable logistic regression analyses were performed to identify parameters associated with clear cell RCC.RESULTS:
Interreader agreement was excellent (intraclass correlation coefficient, 0.815-0.994). The parameters apparent diffusion coefficient (reader 1 AUC, 0.804; reader 2, 0.807), peak enhancement (reader 1 AUC, 0.629; reader 2, 0.606), and downslope (reader 1 AUC, 0.575; reader 2, 0.561) were significantly associated with discriminating clear cell RCC from other renal cortical tumors. The combination of all three parameters further increased diagnostic accuracy (reader 1 AUC, 0.889; reader 2, 0.907; both p ≤ 0.001), yielding sensitivities of 0.897 for reader 1 and 0.897 for reader 2, and specificities of 0.762 for reader 1 and 0.738 for reader 2 in the identification of clear cell RCC. With maximized sensitivity, specificities of 0.429 and 0.262 were reached for readers 1 and 2, respectively.CONCLUSION:
A quantitative multiparametric approach statistically significantly improves diagnostic performance in differentiating clear cell RCC from other renal cortical tumors.Palavras-chave
Texto completo:
1
Base de dados:
MEDLINE
Assunto principal:
Algoritmos
/
Reconhecimento Automatizado de Padrão
/
Imageamento por Ressonância Magnética
/
Interpretação de Imagem Assistida por Computador
/
Carcinoma de Células Renais
/
Neoplasias Renais
Tipo de estudo:
Diagnostic_studies
/
Observational_studies
/
Prognostic_studies
/
Risk_factors_studies
Limite:
Adult
/
Aged
/
Aged80
/
Female
/
Humans
/
Male
/
Middle aged
Idioma:
En
Ano de publicação:
2017
Tipo de documento:
Article