Giorgio Maria Masci1, Fabio Ciccarelli1, Fabrizio Ivo Mattei1, Damiano Grasso1, Fabio Accarpio2, Carlo Catalano1, Andrea Laghi3, Paolo Sammartino2, Franco Iafrate4. 1. Department of Radiological, Oncological and Pathological Sciences, Policlinico Umberto I, Sapienza University of Rome, Viale Regina Elena 324, 00161, Rome, Italy. 2. Department of Surgery "Pietro Valdoni", Policlinico Umberto I, Sapienza University of Rome, Viale del Policlinico 155, 00161, Rome, Italy. 3. Department of Medical-Surgical and Translational Medicine-Radiology Unit, "Sapienza" University of Rome, Sant'Andrea Hospital, Rome, Italy. 4. Department of Radiological, Oncological and Pathological Sciences, Policlinico Umberto I, Sapienza University of Rome, Viale Regina Elena 324, 00161, Rome, Italy. francoiafrate@gmail.com.
Abstract
PURPOSE: Aim of the study was to perform CT texture analysis in patients with gastric cancer (GC) to investigate potential role of radiomics for predicting the occurrence of peritoneal metastases (PM). MATERIALS AND METHODS: In this single-centre retrospective study, patients with gastric adenocarcinoma and surgically confirmed presence or absence of PM were, respectively, enrolled in group PM and group non-PM. Patients with T1-staging, previous treatment or presence of imaging artifacts were excluded from the study. Pre-operative CT examinations were evaluated. Acquisition protocol consisted of gastric distension with water, pre-contrast and arterial phases on upper abdomen and portal phase on thorax and whole abdomen. Texture analysis was performed on portal phase images: the region of interest was manually drawn along the margins of the primitive lesion on each slice and the volume of interest of the whole tumour was obtained. A total of 38 texture parameters were extracted and analysed. ROC curves were performed on significant texture features (p < 0.05). Multiple logistic regression was conducted on features with the best AUC to identify differentiating variables for both groups. RESULTS: A total of 90 patients were evaluated (group PM, n = 45; group non-PM, n = 45). T2/T3 tumours were prevalent in group non-PM, T4 was significantly associated with group PM. Significant differences between the two groups were observed for 22/38 texture parameters. Volume and GLRLM_LRHGE showed the greatest AUC in ROC curve analysis (0.737 and 0.734, respectively) and were found to be independent differentiating variables of group PM in the multiple regression analysis (OR 8.44, [95% CI, 1.52-46.8] and OR 18.99 [95% CI, 84-195.31], respectively). CONCLUSIONS: Our preliminary results suggest the potential value of CT texture analysis for predicting the risk of PM from GC, which may be helpful to stratify patients and address them to the most appropriate treatment.
PURPOSE: Aim of the study was to perform CT texture analysis in patients with gastric cancer (GC) to investigate potential role of radiomics for predicting the occurrence of peritoneal metastases (PM). MATERIALS AND METHODS: In this single-centre retrospective study, patients with gastric adenocarcinoma and surgically confirmed presence or absence of PM were, respectively, enrolled in group PM and group non-PM. Patients with T1-staging, previous treatment or presence of imaging artifacts were excluded from the study. Pre-operative CT examinations were evaluated. Acquisition protocol consisted of gastric distension with water, pre-contrast and arterial phases on upper abdomen and portal phase on thorax and whole abdomen. Texture analysis was performed on portal phase images: the region of interest was manually drawn along the margins of the primitive lesion on each slice and the volume of interest of the whole tumour was obtained. A total of 38 texture parameters were extracted and analysed. ROC curves were performed on significant texture features (p < 0.05). Multiple logistic regression was conducted on features with the best AUC to identify differentiating variables for both groups. RESULTS: A total of 90 patients were evaluated (group PM, n = 45; group non-PM, n = 45). T2/T3 tumours were prevalent in group non-PM, T4 was significantly associated with group PM. Significant differences between the two groups were observed for 22/38 texture parameters. Volume and GLRLM_LRHGE showed the greatest AUC in ROC curve analysis (0.737 and 0.734, respectively) and were found to be independent differentiating variables of group PM in the multiple regression analysis (OR 8.44, [95% CI, 1.52-46.8] and OR 18.99 [95% CI, 84-195.31], respectively). CONCLUSIONS: Our preliminary results suggest the potential value of CT texture analysis for predicting the risk of PM from GC, which may be helpful to stratify patients and address them to the most appropriate treatment.
Authors: N Bernards; G J Creemers; G A P Nieuwenhuijzen; K Bosscha; J F M Pruijt; V E P P Lemmens Journal: Ann Oncol Date: 2013-10-11 Impact factor: 32.976
Authors: Irene Thomassen; Yvette R van Gestel; Bert van Ramshorst; Misha D Luyer; Koop Bosscha; Simon W Nienhuijs; Valery E Lemmens; Ignace H de Hingh Journal: Int J Cancer Date: 2013-08-05 Impact factor: 7.396
Authors: Freddie Bray; Jacques Ferlay; Isabelle Soerjomataram; Rebecca L Siegel; Lindsey A Torre; Ahmedin Jemal Journal: CA Cancer J Clin Date: 2018-09-12 Impact factor: 508.702
Authors: Willem J Koemans; Robin J Lurvink; Cecile Grootscholten; Rob H A Verhoeven; Ignace H de Hingh; Johanna W van Sandick Journal: Gastric Cancer Date: 2021-01-25 Impact factor: 7.370