Validation of models in predicting residual disease in ovarian cancer: comparing CT urography with PET/CT.
Acta Radiol
; 64(6): 2190-2197, 2023 Jun.
Article
em En
| MEDLINE
| ID: mdl-37032426
ABSTRACT
BACKGROUND:
Many ovarian cancer (OC) residual-disease prediction models were not externally validated after being constructed, the clinical applicability needs to be evaluated.PURPOSE:
To compare computed tomography urography (CTU) with PET/CT in validating models for predicting residual disease in OC. MATERIAL ANDMETHODS:
A total of 250 patients were included during 2018-2021. The CTU and PET/CT scans were analyzed, generating CT-Suidan, PET-Suidan, CT-Peking Union Medical College Hospital (PUMC), and PET-PUMC models. All imagings were evaluated by two readers independently, then compared to pathology. According to surgical outcomes, all patients were divided into the R0 group, with no visible residual disease, and the R1 group, with any visible residual disease. Logistic regression was used to assess the discrimination and calibration abilities of each model.RESULTS:
CTU and PET/CT showed good diagnostic performance in predicting OC peritoneal metastases based on the Suidan and PUMC model (all the accuracies >0.8). As for model evaluation, the value of correct classification of the CT-Suidan, PET-Suidan, CT-PUMC, and PET-PUMC models was 0.89, 0.84, 0.88, and 0.83, respectively, representing stable calibration. The areas under the curve (AUC) of these models were 0.95, 0.90, 0.91, and 0.90, respectively. Furthermore, the accuracy of these models at the optimal threshold value (score 3) was 0.75, 0.78, 0.80, and 0.80, respectively. All two-paired comparisons of the AUCs and accuracies did not show a significant difference (all P > 0.05).CONCLUSION:
CT-Suidan, CT-PUMC, PET-Suidan, and PET-PUMC models had equal abilities in predicting the residual disease of OC. The CT-PUMC model was recommended for its economic and user-friendly characteristics.Palavras-chave
Texto completo:
1
Coleções:
01-internacional
Temas:
Geral
/
Tipos_de_cancer
/
Outros_tipos
Base de dados:
MEDLINE
Assunto principal:
Neoplasias Ovarianas
/
Tomografia por Emissão de Pósitrons combinada à Tomografia Computadorizada
Tipo de estudo:
Prognostic_studies
/
Risk_factors_studies
Limite:
Female
/
Humans
Idioma:
En
Revista:
Acta Radiol
Ano de publicação:
2023
Tipo de documento:
Article