Background and objectives: In patients with colorectal cancer, it is important to diagnose peritoneal carcinomatosis as well as to detect location and size of peritoneal tumor dissemination in view of treatment planning. The aim of this study was to investigate the detection accuracy of computed tomography (CT).
Methods: Preoperative CT-scans from 25 consecutive patients with peritoneal carcinomatosis from colorectal or appendiceal origin were independently blindly reviewed by 2 radiologists. The presence and diameter of tumor deposits were noted in seven abdominopelvic areas. Intraoperative findings were regarded as the gold standard. Agreement was assessed using the Kappa index and the chi-square test.
Results: The presence of peritoneal carcinomatosis was detected in 60 and 76% of those patients by each of the radiologist. Detection of individual peritoneal implants was poor (kappa = 0.11/0.23) and varied from 9.1%/24.3% for tumor size <1 cm to 59.3%/66.7% for tumor size >5 cm. Overall sensitivity, specificity, accuracy, positive (PPV) and negative predictive value (NPV) for tumor involvement per area were 24.5%/37.3%, 94.5%/90.4%, 53.0%/60.0%, 86.2%/84.4%, and 47.3%/50.8%, respectively. Accuracy of tumor detection varied widely per anatomic site. Statistically significant interobserver differences were noted, specifically for tumor size of 1-5 cm (P = 0.007) and localization on mesentery and small bowel (kappa = 0.30, P = 0.04).
Conclusions: In colorectal cancer, CT detection of peritoneal carcinomatosis is moderate and of individual peritoneal tumor deposits poor. Interobserver differences are statistically significant. Therefore, preoperative CT seems not to be a reliable tool for detection of presence, size, and location of peritoneal tumor implants in view of treatment planning in patients with colorectal cancer.
Copyright 2004 Wiley-Liss, Inc.