Surveillance for hepatocellular carcinoma in cirrhotic patients: From official recommendations to the real life

Jean-Claude Trinchet
2011 Journal of Hepatology  
COMMENTARY ON: Use of surveillance for hepatocellular carcinoma among patients with cirrhosis in the United States. Davila JA, Morgan RO, Richardson PA, Du XL, McGlynn KA, El-Serag HB. Hepatology 2010 Jul;52(1):132-41. Ó 2011 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. Hepatocellular carcinoma (HCC) occurs in more than 90% of cases in patients with cirrhosis. As only small tumours are eligible for
more » ... are eligible for curative treatment (radiofrequency, resection, transplantation), surveillance in cirrhotic patients is recommended by European and American societies [1, 2] . Surveillance is usually based on liver ultrasonographic examination (US) and serum alphafetoprotein (AFP) assay on a semi-annual basis [1, 2] . All patients with cirrhosis are candidates for surveillance regardless of the cause of the underlying liver disease [1]. Nevertheless, curative treatment of HCC being the main objective, surveillance is not recommended in patients with decompensated cirrhosis (usually not eligible for resection or radiofrequency) when transplantation is impossible [3] . Concerning the modalities of surveillance, there is a large consensus that US is the most prevalent and reliable tool and that serum AFP is not so useful due to the high rate of false-negative and -positive results [4] . Surveillance periodicity has been established on an empirical basis [1] . Recent studies suggest that a 6-month interval is superior to a 12-month interval [5] and not inferior to a 3-month interval at least with the relatively low sensitivity of current diagnostic procedures [6] . The clinical benefit of surveillance is controversial. The only randomized trial comparing surveillance with no surveillance has been performed in China in a large number of patients with HBV infection (cirrhotic or not) and a lower mortality in surveillance group (À37%) was demonstrated in spite of relatively low compliance [7] . To perform a new randomized trial including a non surveillance group seems unrealistic. In a recent attempt, 84% of the 192 informed patients chose surveillance and only one accepted randomization [8] . Despite the lack of evidence-based demonstration, several convergent clues suggest that surveillance of cirrhotic patients is an effective procedure. While the rate of small HCC is less than 20% without surveillance, it is close to 70% in surveillance programs in the more recent studies [9] . Consequently the percentage of patients eligible for curative treatments has markedly risen. A study from Hong Kong [9] reports better survival among patients who underwent surveillance in the period 1998-2004 than those in 1991-1997, likely due to higher rate of curative treatment. Another study from France [10] reports better survival in patients treated during 1996-2005 than during 1976-1985, when surveillance was rarely performed and modern curative treatments were not available. Moreover, recent studies taking into account modern curative treatments have concluded that surveillance is cost-effective [11, 12] . If surveillance is really effective, a large implementation at a population scale should result in a significant improvement in the survival of cirrhotic patients who develop HCC. However such an improvement has not been reported so far. The recent study from the United States by Davila et al. [13] suggests that implementation of surveillance in cirrhotic patients is inadequate in real life. The study included 1873 patients with HCC and previously diagnosed cirrhosis, recruited during 1994-2002 from Medicare databases. All patients were over 65 years of age. The main result was that regular surveillance (annual AFP and/or US test during at least 2 of the 3 years prior to HCC diagnosis) was performed in only 17% of patients before HCC diagnosis. Moreover, US was performed in only 54% of patients who underwent surveillance, serum AFP assay being the single test used in the remaining 46% of patients. Patients belonging to Child-Pugh class C had a higher rate of regular surveillance than patients belonging to classes A and B (22% versus 15%). The most important factors linked to surveillance were practice and affiliation of physicians and some demographic characteristics of patients such as income and education. Chances of benefitting from regular surveillance were higher when patients were followed-up by a gastroenterologist/hepatologist by comparison with an internal medicine/primary care physician, or by a physician with academic affiliation by comparison with solo practice (4.5-and 3.7-fold, respectively). As emphasized by authors, some bias could affect the results of the study. In particular it was not possible to identify precisely the tests (US and serum AFP) performed in surveillance intent. Using data extracted from medical records
doi:10.1016/j.jhep.2010.11.036 pmid:21296113 fatcat:ie4r3ovmlbhpvio2plu5nf54ge