Transient hepatitis B virus (HBV) infections, which generally last <6 months, do not cause cirrhosis and cause only minor increases in the risk of hepatocellular carcinoma (HCC) (3
). Chronic infections, typically lifelong, can cause cirrhosis and HCC (3
). Of the ~350 million HBV carriers now alive, ca. 60 million will die prematurely of cirrhosis and/or HCC. Cirrhosis, which usually develops late in infection, is a significant risk factor for HCC. Early reports stated that most HCCs occur on a background of cirrhosis. However, later studies suggested that as many as 50% of HCCs may occur in noncirrhotic liver (4
), that is, in patients in whom the progression of liver disease still appears rather mild. Thus, liver damage that appears severe by histologic examination is not a prerequisite for HCC.
Interestingly, during chronic HBV infections there is, in the face of persistent viremia, a decline over time in the fraction of infected hepatocytes, from 100% to as little as a few percent (5
). Along with HCC, this is perhaps the most surprising and unexplained outcome of chronic infection. The timing of this decline has not been systematically studied, but it is presumably gradual, occurring over years or decades, and dependent on persistent, albeit low-level, killing of infected hepatocytes by antiviral cytotoxic T lymphocytes (CTLs) (20
). It is believed that the liver is largely a closed, self-renewing population. Such a population might be expected to evolve under any strong or persistent selective pressure. In HBV-infected patients, the earliest and most persistent selective pressure is immune killing of infected hepatocytes, which should initially constitute the entire hepatocyte population. Persistent killing of HBV-infected hepatocytes could lead to clonal expansion of mutant or epigenetically altered hepatocytes that had lost the ability to support infection and that were not, therefore, targeted by antiviral CTLs.
Such a selective pressure may explain why foci of altered hepatocytes (FAH) and HCC are typically virus negative (1
). Normal or preneoplastic hepatocytes (e.g., in FAH) that have evaded the host immune response should undergo clonal expansion, because their death rate is lower than that of surrounding hepatocytes, even if they do not have a higher growth rate. Indeed, clonal expansion of hepatocytes has been detected, in the absence of cirrhosis, in woodchucks chronically infected with woodchuck hepatitis virus (WHV) (19
) and in chimpanzees chronically infected with HBV (21
). The presence of discrete foci of normal-appearing but virus-negative hepatocytes in chronically infected woodchuck livers (39
) suggested, but did not prove, that normal-appearing hepatocytes that had lost the ability to support virus replication might clonally expand.
The purpose of the present study was, therefore, to determine if normal-appearing hepatocytes undergo clonal expansion. To address this issue, we focused on noncirrhotic livers, because hepatocyte appearance and organization in many cirrhotic nodules are often considered to indicate premalignancy (7
), and this, together with the cellular environment in the cirrhotic liver, may explain why as many as 50% of cirrhotic nodules have been found to be made up of clonally expanded hepatocytes (2
). In older HBV patients, cirrhosis, the result of cumulative scarring due to ongoing tissue injury, presumably produces an evolutionary pressure on the hepatocyte population due to restricted blood flow and altered hepatic architecture.
Clonal expansion was detected by assaying for integrated HBV DNA by inverse PCR (19
). Because integration occurs at random sites in host DNA, each integration event provides a unique genetic marker for the cell in which it occurred, and for any daughter cells. Thus, the clonal expansion of these tagged hepatocytes can be measured by determining how many times a given virus-cell DNA junction is repeated in a liver fragment. Analysis of fragments of nontumorous liver from noncirrhotic HCC patients revealed that at least 1% of hepatocytes are present as clones of >1,000 cells. Examination of 5-μm-thick sections of paraffin-embedded livers from the same patients revealed that clonally expanded hepatocytes were present in liver sections lacking preneoplastic lesions or changes. Therefore, normal-appearing hepatocytes must have undergone clonal expansion. Although clonal expansion was detected by analysis of integrated HBV DNA, the expansion did not appear to be due to the site of integration of the viral DNA into host DNA.
These results are consistent with the hypothesis that immune selection and the later emergence of liver cirrhosis, with altered lobular organization and restricted blood flow, may constitute the two major selective pressures on the hepatocyte population that culminate in hepatocellular carcinoma. More-direct proof of the role, if any, of immune selection in hepatocyte evolution and HCC will require, first of all, an assay with a greater ability to detect clonally expanded hepatocytes. The present approach is limited by a number of factors, including a need for integration near a particular restriction endonuclease cleavage site in host DNA and for conservation of particular viral sequences so that the integrated DNA can be amplified using the PCR primers chosen. These issues may explain why the fraction of clonally expanded hepatocytes reported here is much less than that suggested by histologic data showing that more than 50% of hepatocytes appear negative for virus replication in long-term carriers. Further dissection of this issue will also require localization and determination of the virologic status of hepatocyte clones present in tissue sections.