Detection of nodules in liver cirrhosis: spiral computed tomography or magnetic resonance imaging? A prospective study of 88 nodules in 34 patients.

Detection and characterization of all focal lesions in the liver are critical for screening patients with chronic liver disease. The aim of this prospective study was to investigate the accuracy of magnetic resonance imaging (MRI) and spiral computed tomography for the diagnosis of hepatic nodules in cirrhotic patients when compared with pathological findings of the explanted liver. From February 1997 to July 1999, 34 cirrhotic patients waiting for orthotopic liver transplantation (OLT) (mean age, 53.5 +/- 9.3 years; 24 males) were included. All patients had MRI and spiral computed tomography examinations, and findings were matched with the histological findings. Data analyses were made using the McNemar chi-square test. Mean time between radiological examination (MRI or spiral computed tomography) and OLT was 43.8 +/- 39 days. A total of 88 nodules were found in the 34 patients: 54 hepatocellular carcinoma (HCC) (mean size, 18 +/- 10 mm) in 21 patients, 22 dysplastic nodules (mean size, 10.7 +/- 4.3 mm) in 11 patients, and 12 macroregenerative nodules in 13 patients. Lesion-by-lesion analyses showed that sensitivity of MRI and spiral computed tomography for nodule, HCC or dysplastic nodule diagnosis was 44.3 and 31.8% (P = 0.02), 61.1 and 51.9% (P = 0.2), and 27.3 and 0% (P = 0.04), respectively. Patient-by-patient analyses showed no statistical difference between spiral computed tomography and MRI for nodule diagnosis. In conclusion, in patients with liver cirrhosis, MRI is more accurate than spiral computed tomography for the detection of liver nodules and dysplastic nodules. However, tumour size is always a restricting factor for these two techniques, which are unable to detect small HCC in more than 60% of cases.