Accuracy of investigations for asymptomatic colorectal liver metastases

PURPOSE: The aim of this study was to prospectively assess the accuracy of the most promising imaging and tumor marker tests in liver metastasis diagnosis on follow-up of asymptomatic colorectal cancer patients during a median of 571 months after primary tumor resection. METHODS: One hundred patients, who were considered free of liver metastases after primary colorectal cancer resection and conventional follow-up, were screened for liver metastases by computerized tomography, magnetic resonance and ultrasound scans, ultrasound Doppler and isotope assessment of changes in hepatic arterial and portal venous flow, and serum estimation of carcinoembryonic antigen. Patients were followed up during a median of 41 months to identify those who developed liver metastases. RESULTS The most sensitive technique was computerized tomography (sensitivity 0.67, specificity 0.91). Computerized tomography and magnetic resonance but not ultrasound were 100 percent accurate in differentiating liver metastases from other hepatic lesions. Techniques based on changes in hepatic arterial and portal venous flow had lower diagnostic accuracies (Doppler perfusion index, sensitivity 0.58, specificity 0.57 hepatic perfusion index, sensitivity 0.50, specificity 0.55), whereas ultrasound scanning identified only 43 percent (sensitivity 0.43, specificity 0.96) and serum carcinoembryonic antigen 33 percent (sensitivity 0.33, specificity 0.81) of patients with asymptomatic liver metastasis. Sensitivity could be improved by using tests in combination but this reduced specificity. CONCLUSIONS: Computerized tomography was the most sensitive test for asymptomatic colorectal liver metastases, but only 67 percent of affected patients were identified.