For patients with cirrhosis, adding serum alpha fetoprotein testing to ultrasound significantly boosted its ability to detect early-stage hepatocellular carcinoma, according to the results of a systematic review and meta-analysis reported in the May issue of Gastroenterology.
Used alone, ultrasound detected only 45% of early-stage hepatocellular carcinomas (95% confidence interval, 30%-62%), reported Kristina Tzartzeva, MD, of the University of Texas, Dallas, with her associates. Adding alpha fetoprotein (AFP) increased this sensitivity to 63% (95% CI, 48%-75%; P = .002). Few studies evaluated alternative surveillance tools, such as CT or MRI.
Diagnosing liver cancer early is key to survival and thus is a central issue in cirrhosis management. However, the best surveillance strategy remains uncertain, hinging as it does on sensitivity, specificity, and cost. The American Association for the Study of Liver Diseases and the European Association for the Study of the Liver recommend that cirrhotic patients undergo twice-yearly ultrasound to screen for hepatocellular carcinoma (HCC), but they disagree about the value of adding serum biomarker AFP testing. Meanwhile, more and more clinics are using CT and MRI because of concerns about the unreliability of ultrasound. “Given few direct comparative studies, we are forced to primarily rely on indirect comparisons across studies,” the reviewers wrote.
To do so, they searched MEDLINE and Scopus and identified 32 studies of HCC surveillance that comprised 13,367 patients, nearly all with baseline cirrhosis. The studies were published from 1990 to August 2016.
Ultrasound detected HCC of any stage with a sensitivity of 84% (95% CI, 76%-92%), but its sensitivity for detecting early-stage disease was less than 50%. In studies that performed direct comparisons, ultrasound alone was significantly less sensitive than ultrasound plus AFP for detecting all stages of HCC (relative risk, 0.80; 95% CI, 0.72-0.88) and early-stage disease (0.78; 0.66-0.92). However, ultrasound alone was more specific than ultrasound plus AFP (RR, 1.08; 95% CI, 1.05-1.09).
Four studies of about 900 patients evaluated cross-sectional imaging with CT or MRI. In one single-center, randomized trial, CT had a sensitivity of 63% for detecting early-stage disease, but the 95% CI for this estimate was very wide (30%-87%) and CT did not significantly outperform ultrasound (Aliment Pharmacol Ther. 2013;38:303-12). In another study, MRI and ultrasound had significantly different sensitivities of 84% and 26% for detecting (usually) early-stage disease (JAMA Oncol. 2017;3[4]:456-63).
