Article Text


Predicting survival in hepatitis B
  1. M Sherman
  1. Correspondence to:
    Dr M Sherman
    University of Toronto and University Health Network, Toronto General Hospital, 200 Elizabeth St, Toronto, Ontario M5G 2C4, Canada;

Statistics from

Alanine aminotransferase (ALT) level alone is not an appropriate indication for therapy in chronic hepatitis B infection, and other criteria in addition to ALT must be used to determine eligibility for therapy

Predictors of survival in chronic hepatitis B infection are surprisingly not well described. Various studies have identified different factors that were associated with adverse outcomes. For example, Niederau and colleagues,1 in a cohort of European patients, identified lack of clearance of hepatitis B e antigen (HBeAg) as a predictor of decreased survival. Others have identified older age, presence of cirrhosis, and the persistence of alanine aminotransferase (ALT) elevations as adverse prognostic signs in an antibody to hepatitis B e antigen (anti-HBe) positive cohort.2 In patients undergoing a flare of hepatitis B activity, whether spontaneous or chemotherapy induced, the presence of jaundice is an ominous sign.3 None of these adverse predictive factors are unexpected. Clearly, jaundice, cirrhosis, older age, and elevated ALT are obvious adverse predictive factors but until recently we have not had the tools to predict, years in advance, the outcome of chronic hepatitis B infection. This is important because we would prefer to offer treatment only to those who are likely to develop complications of the disease, and not to those whose disease will become inactive without long term sequelae. Recently, new predictors of outcome have been identified. At last year’s American Association for Study of Liver Disease (AASLD) meeting and the recent European Association for Study of the Liver (EASL) meeting, new data were presented that contribute to this debate. In addition, the article by Yuen and colleagues4 in this issue of Gut also forces us to re-examine some of our assumptions about hepatitis B prognosis and therefore treatment (see page 1610).

To some extent our current management algorithms have been directed by the pharmaceutical industry because management algorithms tended to follow the design of clinical trials used to license products. For example, the initial registration trials with lamivudine lasted one year.5,6 However, it has since become clear that for most patients one year of therapy with lamivudine (or any nucleoside analogue) provides inadequate seroconversion rates in HBeAg positive patients. It is also clear that most anti-HBe positive patients require extended, perhaps lifelong, therapy. Trials in HBeAg positive subjects chose HBeAg seroconversion as an end point because historically, seroconversion to anti-HBe positive was considered to be associated with inactivation of disease, with lower levels of viral replication and less inflammation.6 It now seems that e antigen seroconversion may be an inappropriate end point for many patients. At best interferon treatment only induces e antigen seroconversion in 30–35% of patients.7 The durability of this seroconversion is poor.8 The proportion of “sero-reverters” is even higher after seroconversion induced by one year of lamivudine therapy.8,9 Thus although randomised controlled studies have shown an increased seroconversion rate compared with controls in the short term, it is not clear that treatment induces more patients to undergo HBeAg seroconversion over the long term than would be expected without treatment.

Is it possible that treatment merely advances seroconversion by some uncertain period of time in those who are destined to seroconvert spontaneously at some future point? Perhaps therapy induced seroconversion merely advances spontaneous seroconversion by some years. An argument in favour of this hypothesis is that whether patients are treated with interferon or nucleoside analogues, those who have the highest likelihood of HBeAg seroconversion, namely those who have high ALT and low hepatitis B virus (HBV) DNA concentrations, are also those who have the highest likelihood of spontaneous seroconversion. Furthermore, Yuen and colleagues10 have observed matched cohorts of patients treated or not treated with interferon over an extended follow up period and have shown that beyond five years of follow up the proportion of treated patients who underwent seroconversion was the same as in the untreated group (that is, interferon therapy did not increase the overall number of patients undergoing seroconversion). It did however advance seroconversion by about five years. Nor did interferon therapy affect outcome. A study in children reached the same conclusion.11 Thus although in some populations treatment induced seroconversion appears to improve prognosis in the short term, it is not certain that treatment increases the overall number of patients undergoing e antigen seroconversions over the long term or improves the prognosis over the longer term.

Given the uncertainty about appropriate treatment end points, it is perhaps not surprising that the indications chosen for starting therapy have also been unsatisfactory. As the overall objective of treatment is to decrease inflammation, and hopefully thereby prevent all of the other complications of chronic hepatitis B, it seems logical to use an elevated ALT, a marker of hepatic inflammation, as an indication to start treatment. All of the registration trials have used this indication. All of the guidelines issued by various continental hepatology societies (AASLD, EASL, Asian Pacific Association for Study of the Liver)12–14 supported the use of this indication. However, using elevated ALT as an indication for therapy means that only a small minority of all hepatitis B carriers will ever be treated, far fewer than the 20–25% of hepatitis B carriers who will die of complications of their liver disease. Using ALT as an indication for therapy will leave many patients who are destined to develop cirrhosis or hepatocellular carcinoma (HCC) untreated. The study by Yuen and colleagues4 in this issue of Gut, together with conclusions from a large scale prospective study in Taiwan,15 supports the conclusion that ALT alone is not an appropriate indication for therapy. In the Yuen study, patients who had an ALT concentration in the upper range of normal had an increased mortality from HCC compared with those in whom ALT levels were less than <0.5× ULN. Paradoxically, the risk of HCC in patients with ALT levels that were higher than 2× ULN was lower than in patients with lower ALT levels. Yuen et al explain this anomaly by assuming that higher levels of ALT represent flares similar to acute hepatitis, which resolve without leaving significant permanent injury. However, there may also be some selection bias influencing these results, in that patients with ALT >2× ULN that persisted over time might have undergone therapy and been excluded from this study. Others might have been in the process of seroconverting and going into remission. Yuen and colleagues4 also showed that the HBeAg status and ALT on follow up did not correlate with the incidence of complications. When they analysed HBV DNA they found, as did Chu and colleagues,16 that HBV DNA below 105 copies/ml did not protect against the development of complications.

The REVEAL study is one of three similar large scale, long term, prospective, cohort studies of hepatitis B carriers, all of which provide new evidence about identifying patients at risk of a poor outcome. The studies were performed in a cohort in Philadelphia,17 a cohort in Haimen City China,18,19 and in a cohort from several townships in Taiwan.15,20–22 Each cohort included more than 3000–4000 subjects who were recruited and followed for more than 10 years. The end points in all three were similar: incidence of HCC, cirrhosis, and death from liver disease. All three studies came to similar conclusions. The best predictor of an adverse outcome in a hepatitis B carrier was the HBV DNA concentration. The higher the HBV DNA the higher the incidence of an adverse outcome. Neither ALT15,22 nor e antigen status20 at recruitment was correlated with outcome. These results, together with the data of Yuen and colleagues,4 support the concept that patients with normal ALT levels can no longer be excluded from therapy, and that criteria other than the ALT must be used to determine eligibility for therapy.

How then can we distinguish between those hepatitis B carriers who need therapy and those who do not? Are there other markers of a poor prognosis that might be more useful? Liver biopsy showing fibrosis or cirrhosis may be one such marker but we cannot biopsy every hepatitis B carrier. Perhaps non-invasive measurements of fibrosis will help answer this question. However, if treatment is restricted to patients with advanced fibrosis the incidence of cancer might not fall a great deal. The only other logical marker is HBV DNA concentration. The three large cohort studies strongly suggest (but do not prove) that suppression of viral replication will decrease the incidence of adverse outcomes. There is other evidence to support this notion. Long term suppression of woodchuck hepatitis virus replication by entecavir leads to a reduction in the incidence of HCC.23 Suppression of viral replication in cirrhotic patients decreases the incidence of liver related adverse events (HCC and cirrhosis).24 (Whether suppression of HBV DNA before the onset of cirrhosis would have a similar effect is not known, and is unlikely to be studied, given the difficulty of maintaining a patient with active disease off therapy for an extended period of follow up.) Patients in whom viral replication is spontaneously suppressed early in the course of their disease are at a much lower risk of cirrhosis and HCC compared with those in whom viral replication persists.25,26 Finally, in the Taiwan cohort, patients in whom HBV DNA spontaneously improved had a lower incidence of adverse outcomes than those in whom HBV DNA levels remained high.27

This all suggests that a high HBV DNA concentration should be the prime indication for therapy. Several immediate questions arise. Firstly, at what point in the natural history is it appropriate to start treatment? This question cannot be answered at present and will have to await full publication of the cohort studies so that the relationship between HBV DNA, age at recruitment, and outcome can be evaluated. The next question is how low should the HBV DNA level be to have the highest likelihood of preventing complications of chronic HBV infection. In the three cohort studies, even HBV DNA concentrations <100 000 copies/ml were associated with a significant incidence of HCC, cirrhosis, and death from liver disease. Others have shown, in a small cohort, that all those with HCC had an HBV DNA concentration above 2×104.28 Until this level can be confirmed as “safe”, it is probably wise to aim to suppress HBV DNA maximally, to undetectable levels, or if that is not possible, to below 104 copies/ml.

Until recently we have not had adequate tools to achieve optimal suppression of viral replication indefinitely. Interferon based therapy only induces permanent suppression in a small proportion of patients.5,29,30 There are as yet no long term studies on the durability of pegylated interferon response, but at best, only 30–35% of HBeAg positive patients respond, leaving 65–70% requiring additional treatment.31 Lamivudine cannot be used long term because of the development of resistance.32 The prospect of using adefovir long term is also fading, as up to 18% resistance rates at four years have been reported.33 Entecavir is the most potent nucleoside analogue to complete phase III testing and, so far, no resistance has been reported in the treatment of naïve subjects.34 Tenofovir is more potent than adefovir35 but its HBV resistance profile and frequency of emergence of resistance remains unknown. However, it is inevitable that resistance will develop over time for both of these agents. There are as yet no data on the combined use of more than one nucleoside analogue, although this would seem to be a logical approach and perhaps the only approach that might provide long term viral suppression.

Given the above considerations, the time has come to abandon some of our old concepts about the management of hepatitis B. We should no longer rely on an elevated ALT to select patients for treatment. Patients with high viral loads and normal ALT levels are not good candidates for interferon and should probably only be treated with nucleos(t)ide analogues. Treatment should aim to suppress viral replication as much as possible. The target should be undetectable HBV DNA and if that is not possible, HBV DNA should not exceed 104 copies/ml. The availability of more potent drugs such as entecavir, telbivudine, and clevudine will make these targets feasible. To ensure long term suppression we have to prevent the emergence of resistance. It may be that if viral replication is suppressed deeply enough, even with monotherapy, resistance will be rare, but addition of a second agent that does not have cross reactive resistance might be necessary. Cost will be a huge issue once we contemplate long term therapy, which will have to be confronted head-on. I believe that we will have to bite the bullet and immediately undertake studies of combination therapy, or risk creating strains of virus that are resistant to everything.


  • ALT is a poor predictor of outcome and therefore cannot be used as the sole indication for therapy.

  • HBeAg positivity is associated with a higher risk of adverse outcomes than anti-HBe positivity.

  • HBV DNA concentration at recruitment and during follow up is the best predictor of an adverse outcome.

  • The higher the HBV DNA concentration (above 104 copies/ml), the greater the mortality.

  • Treatment should be aimed at suppressing HBV DNA maximally, to below 104, and preferably to below 103.

Alanine aminotransferase (ALT) level alone is not an appropriate indication for therapy in chronic hepatitis B infection, and other criteria in addition to ALT must be used to determine eligibility for therapy


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles