Recommended management of patients with cirrhosis includes regular surveillance for hepatocellular carcinoma using noninvasive imaging methods. Currently, ultrasound with real-time and serum alpha-fetoprotein measurements are applied worldwide. Although the prevalence of hepatocellular carcinoma is not as high in the United States and Europe as it is in Asia and other endemic areas, a significant proportion of patients will be expected to develop this disease (with the incidence rising 10-15 years after diagnosis of cirrhosis, and an expected annual incidence of 0.5% to 1%, depending on the etiology and severity of the cirrhosis). In high-prevalence areas, contributing factors include exposure to aflatoxins and hepatitis B infection, with hepatitis C emerging as a major etiologic agent, particularly in association with ethanol abuse. Autopsy records show that 20% of patients with cirrhosis have hepatocellular carcinoma at the time of death.^[1] It has been suggested that increased frequency of evaluation (at 3- or 6-month intervals, as opposed to 1-year intervals) leads to detection of tumors at an earlier stage.^[2]

A recent survey of over 450 physicians specializing in management of liver disease in the United States indicates that ultrasound and alpha-fetoprotein measurements are the most commonly obtained screening tests (used with a frequency of 93% and 99.7%, respectively), with a majority (48%) obtaining alpha-fetoprotein levels at 6-month intervals and ultrasound at 12-month intervals (62%). However, considerable variability was found in screening practices among practitioners. The surveyors point out, however, that there is no official position on screening in North America, and the cost-effectiveness of screening has not been established in this population.^[10]

With a diagnostic sensitivity as low as 30%, measurement of serum alpha-fetoprotein is not always helpful.^[3] Specificity of this study is also not 100%, although persistently elevated alpha-fetoprotein is a risk factor for hepatocellular carcinoma.^[1]

Ultrasound, although widely applied and sensitive in skilled hands, may demonstrate nonmalignant masses such as focal fat or hemangioma, which on real-time appearances alone can be indistinguishable from neoplastic lesions. The prevalence of these lesions (with characteristic appearances on body computed tomography) was found to be on the order of 65% in a subgroup of patients with focal hepatic lesions detected by ultrasound but with normal alpha-fetoprotein levels.^[4] The presence of arterio-venous shunting, however, which is characteristic of malignant tumors, particularly hepatocellular carcinoma, is very helpful, and specificity increases as the degree of Doppler shift and shunt velocities increase. In a series of 115 patients screened for chronic liver disease in the Northeast United States, specificity for malignancy increased from 76% for a mass alone to 95% if associated with an arterio-venous shunt.^[5] In this population, all patients with shunt above 2.4 kHz had malignancy. It should be remembered that vascularity can also be demonstrated in other intrahepatic lesions such as metastatic disease, focal nodular hyperplasia, and liver cell adenoma. Neovascularity is not commonly demonstrated in lesions of less than 2 cm.

The dilemma for the clinician is the threshold for biopsy of focal masses in the liver. A new focal hypoechoic, avascular lesion (or with vascularity below the threshold of detection by currently available ultrasound systems) may represent a focus of dysplasia, which is a premalignant condition with a spectrum advancing to frank neoplasm, or a nodule of hepatocellular carcinoma, and would justify a biopsy. There is always the potential that it may represent a benign lesion such as a region of focal fatty sparing. It has been recommended that all ultrasonically detected lesions be biopsied, regardless of vascular characteristics.^[3] The diagnostic yield, however, is likely to become progressively higher if associated with vascularity, particularly if there are accompanying alpha-fetoprotein abnormalities. In general, fine-needle aspiration biopsy of focal liver masses has a very low complication rate but hemorrhage (rarely fatal), tract seeding, and arterio-venous fistulas are definite risks.^[6]

If portal vein thrombosis is detected, it is important to determine whether this is bland or related to neoplasm. In the case where ultrasound examination detects a portal venous thrombosis but no lesion, a dual-phase CT would be indicated, because occasionally, multifocal hepatoma may not be detectable on ultrasound due to marked heterogeneity of liver texture and lack of tumor vascularity.

Other diagnostic modalities -- such as MRI and CT -- may be used in evaluation of focal masses, and early comparative studies indicate a role for gadolinium-enhanced MRI in detection of hepatocellular carcinoma.^[7] Despite efforts to screen transplant recipients for underlying malignancy, it is still common to find foci of previously undetected hepatocellular carcinoma at transplant. The prevalence of hepatocellular carcinoma was 17.5% in one recent series, with a mean size of 11.6 mm.^[8] Unfortunately, no diagnostic tests have 100% accuracy. However, imaging screening of patients with chronic liver disease does provide the clinician with an opportunity to detect those with otherwise unsuspected neoplasia at a potentially treatable stage. This may lead to more effective management strategies because the management choices for the affected individual to undergo transplantation, surgery, or local or systemic therapy continue to broaden. Even though some patients with advanced cirrhosis may not be candidates for transplantation or surgery, they may still benefit from local control of tumor through percutaneous therapy.^[9]