Purpose: To retrospectively measure the diagnostic performance of prospective computed tomographic (CT) results obtained by using 16- or 64-detector row scanners in the detection of peritoneal metastases (PMs) in patients with advanced gastric cancer.
Materials and methods: The institutional review board approved this retrospective study and waived the need to obtain patient consent. In 498 patients with gastric cancer (stage >or= T2) who were undergoing surgery, the presence of PM was prospectively rated as grade 0 (absent), grade 1 (equivocal), or grade 2 (present) on CT scans (0.67 or 2 mm thick) that were interpreted by interactively adjusting the viewing thickness and viewing plane. The CT readings were retrospectively compared with surgical and pathologic findings. In patients in whom the presence of PM was rated as grade 1 or less, factors predictive of PM were identified by testing variables, including patient and tumor characteristics, with univariate tests and multivariate logistic regression analysis.
Results: Fifty-three patients (10.6%) had confirmed PM. When only grade 2 was considered to indicate a positive CT reading, sensitivity and specificity were 28.3% (15 of 53) and 98.9% (440 of 445), respectively. With the threshold of grade 1 or greater indicating a positive reading, sensitivity and specificity were 50.9% (27 of 53) and 96.2% (428 of 445), respectively. In the patients with grades of 1 or less, the significant factors predictive of PM were greater tumor size and T stage.
Conclusion: The sensitivity of PM detection is limited, even with modern CT techniques. In patients whose CT results are not definitely positive for PM, staging laparoscopy is still recommended if the aforementioned two predictive factors (greater tumor size and T stage) are suspected.
(c) RSNA, 2009.