Purpose: To determine retrospectively the sensitivity and specificity of computed tomography (CT) for the differentiation of perihepatic metastases with and those without liver paren-chymal invasion (LPI) in patients with ovarian cancer by using interpretations of radiologists with different experi-ence levels and staging laparotomy and pathologic exami-nation findings as the reference standards.
Materials and Methods: Institutional review board approval and waiver of in-formed consent were obtained for this HIPAA- compliant study; 121 patients with ovarian cancer (age range, 29-94 years; mean age, 57.8 years) formed the study group. Two radiologists blinded to patient clinical data (radiologist 1, 6 months of experience; radiologist 2, 2 years 6 months of experience) retrospectively and independently recorded presence of perihepatic metastases, liver regions involved, and presence of LPI by perihepatic metastases visible on CT images. Sensitivities and specificities for detecting the presence of perihepatic metastases and liver regions in-volved and sensitivities for detecting LPI were calculated, Statistics were used to analyze interradiologist agreement.
Results: Pathologic examination results showed 66 perihepatic me-tastases in 43 (36%) of 121 patients. Sixty (91%) of 66 perihepatic metastases did not show signs of LPI and six (9%) did. Sensitivity and specificity combinations for radi-ologists 1 and 2 were 56% and 87% and 86% and 99%, respectively, for detecting the presence of perihepatic me-tastases and 46% and 97% and 82% and 100%, respec-tively, for determining liver regions involved. Radiologists 1 and 2 had sensitivities of 35% and 80%, respectively, for detecting regions with perihepatic metastases without LPI and sensitivities of 50% and 100%, respectively, for de-tecting regions with perihepatic metastases with LPI.
Conclusion: CT can be used to detect perihepatic metastases in patients with ovarian cancer and allows for distinction be-tween metastases that invade the liver and those that do not.