Hepatitis C and Hepatocellular Carcinoma: Grist for the Mill

November 28, 2008

See “Incidence of hepatocellular carcinoma and associated risk factors in hepatitis C-related advanced liver disease,” by Lok AS, Seeff LB, Morgan TR, et al, on page 138.

The Hepatitis C Antiviral Long-Term Treatment against Cirrhosis (HALT-C) study is a prospective, randomized, investigator-initiated study. The study was conducted to determine whether maintenance therapy with low-dose pegylated interferon alpha in patients with chronic hepatitis C who had failed to clear virus following a standard course of treatment would slow the progression of disease.1 Progression of disease was determined on biopsy and by the development of the end-stage complications of progressive hepatitis C, namely development of hepatocellular carcinoma (HCC), liver failure and the need for liver transplant.

Previous reports from the HALT-C study have indicated this treatment regimen did not halt progression of liver disease.1 In a report in this issue of GASTROENTEROLOGY, Lok et al2 provide evidence the incidence of hepatocellular carcinoma is also not different in those who were on maintenance therapy versus untreated controls. These results are in contradistinction to an earlier report on two-year maintenance therapy in nonresponders.3 However, the sample size in this study was small and may have led to type 1 error.3

HCC is a common complication of cirrhosis in patients with chronic hepatitis C. The reported incidence of HCC varies between less than one percent and greater than 10 percent in some series.4–6 The quality of the studies looking at this issue varies, with a paucity of prospective studies, and none that compare the incidence of HCC in patients with hepatitis C cirrhosis with an uninfected control group. Some factors that identify patients with hepatitis C at risk for HCC have been identified, including older age, cirrhosis and a low platelet count. Thus far, however, there has been no attempt to use these factors to predict HCC occurrence within a defined time period, such as three to five years.

Risk of HCC in hepatitis C cirrhosis is not fixed, but increases with time as disease progresses. Despite this, quoted incidence figures in cirrhosis assume all patients have equal risk. This is relevant when the issue of surveillance for HCC is considered. It would be very helpful to have some means of more accurately identifying those hepatitis C cirrhotics with a higher risk of developing cancer within a short period of time. For example, if the five-year risk of HCC were five percent, then patients might warrant surveillance. If the five-year risk was only 0.5 percent, surveillance might not be recommended.

There are several issues raised by this report from the HALT-C study. These are related to the incidence of HCC and the assessment of HCC risk, the efficacy of surveillance, and HCC staging.

The incidence of HCC in noncirrhotic patients in the HALT-C study was unexpectedly high. In the cirrhotic group, the incidence of HCC was 7.0 percent. In the patients who were noncirrhotic, the incidence of HCC was 4.1%. In North America, until this report, HCC developing in a noncirrhotic hepatitis C patient was considered uncommon. Although the patients were not cirrhotic, they did have advanced fibrosis as a precondition for inclusion in the study.

These results can be interpreted in two ways.

One explanation, as the authors note, is the severity of fibrosis taken during the initial biopsy was undercalled, because of sampling error, and these patients actually had cirrhosis. Some patients clearly progressed to cirrhosis during the course of the study, but it is unlikely 47 percent of initial biopsies were underinterpreted. Others developed thrombocytopenia, suggesting they had portal hypertension, and thus were indeed cirrhotic, even if not identified on biopsy.

Alternatively, these results indicate the cancer risk starts to increase before cirrhosis develops. There were some patients who developed HCC who did not have cirrhosis by any clinical, biochemical or hematologic criteria. Whatever the true explanation, it is irrelevant to the clinical situation. Clinicians are faced with biopsies in patients with hepatitis C that show advanced fibrosis but not cirrhosis. These HALT-C study results mean a patient with hepatitis C who has bridging fibrosis must now be considered to be at risk for HCC.

Whether these biopsies are underinterpreted because of sampling or not, the clinician has to decide whether the patient warrants HCC surveillance. This is a difficult question to answer and is discussed in more detail.

The objective of HCC surveillance is to decrease mortality from that disease. There are no randomized, controlled trials of HCC surveillance in hepatitis C that address this endpoint. Therefore, to identify patients who might benefit from surveillance, we have to rely on modeling studies to determine the efficacy and cost efficacy of HCC surveillance. There is a handful of such studies7–9; as expected, each has utilized different disease models and has included different assumptions. There does seem to be some agreement, however, that at an annual HCC incidence of about 1.5 percent surveillance becomes both effective in terms of life-years saved, and cost-effective (i.e., the cost of surveillance was less than $50,000 per life-year saved).

By this criterion, even patients with cirrhosis in this study would not quite qualify for surveillance, because the overall annual incidence was only about 1.4 percent, and in the noncirrhotic group the incidence was only 0.8 percent. Some of the modeling studies are old; others rely heavily on retrospective data for transition probabilities. Some do not include modern approaches to the management of HCC (e.g., liver transplantation). These models are probably no longer trustworthy, given new data, and should be revisited.

The HALT-C study, among others, now provides prospective data that can be inserted into any disease model. In addition, there has been significant progress in the management of small HCC, which none of the earlier analyses capture. However, until we have new models we should continue to provide surveillance for our patients with cirrhosis. Whether patients with bridging fibrosis should also undergo surveillance cannot be answered at present, because we do not know whether the incidence is less or more than 1.5 percent.

The second important outcome of this analysis was the development of a predictive model for HCC using Cox proportional hazard techniques. The Cox proportional hazards technique assumes the effect of a risk factor remains proportional to the outcome during the time period analyzed. In other words, if age, platelet count or other factors influence outcome, the magnitude of this influence does not change with time. The analysis is a mathematical calculation that determines whether the variables examined in the model are associated with the outcome, and whether they are associated independently of other variables as well as the magnitude of that association.

By calculating a numerical value representing the strength of the association for each of the variables associated with development of HCC, a score can be calculated for each patient. In this analysis, Lok et al2 divided the range of possible scores into three risk strata; low, medium and high. The incidence of HCC predicted by the Cox proportional hazard model was calculated for patients in each stratum and compared with the actual incidence (Table 1).

For patients with the lowest risk scores, the expected incidence of HCC was less than one percent and the observed incidence between 0.0 and 0.4 percent at three and five years. For patients in the highest risk group, the calculated incidence should have been greater than five percent, and the observed three-and five-year incidences were 6.1 and 17.8 percent, respectively.

We are not given the annual incidence figures in the risk strata, so we cannot know whether any of these groups fall within the 1.5 percent incidence rate at which surveillance becomes cost-effective. However, it is likely the high-risk group does fall within that category and should undergo surveillance. The intermediate risk group has a three-year HCC incidence of 1.5 percent and a five-year incidence of 4.8 percent. This group probably does not get in under the 1.5 percent wire. That is not to say physicians looking after these patients should not offer surveillance, only that as public policy (or from a third-party payer point of view) it may not be worthwhile. The decision to offer surveillance or not for individual patients is always a decision between the patient and the physician.

The HALT-C data on HCC risk and the stratification into risk groups requires validation in another cohort. In addition, we need to know how accurate these predictions are, that is, the sensitivity, specificity and positive predictive values of the scores as well as the area under the receiver operating characteristic curve. This is a method to determine how good a diagnostic test really is at predicting the presence of a diagnosis. Sensitivity is plotted against 1/specificity, producing a curve above a diagonal line (Figure 1). The greater the area under the curve, the better the test. A useful test should have an AUC of greater than or equal to 0.8.

In addition to independent validation and accurate description of the performance characteristics of the model, it is important to know whether the model is applicable through time, as disease progresses and as cancer risk likely increases. Does someone who is initially in a low-risk category progress to higher risk categories with time, and does this actually reflect an increased risk? If this can be confirmed, we would have an excellent tool to determine who should and who should not undergo HCC surveillance.

Until recently, there has been no way of accurately predicting patients at risk for HCC except in broad strokes, cirrhotics, patients with hepatitis B and so on. However, the HALT-C data and data presented elsewhere looking at HCC risk in chronic hepatitis B will improve our predictive ability. Chen et al10 have developed a nomogram of HCC risk in chronic hepatitis B using data from the REVEAL study. Data such as HBV DNA concentration, HBV genotype and age were analyzed in a similar fashion as described herein to construct a nomogram resembling the cholesterol/heart attack risk nomogram developed from the Framingham study. This nomogram can be used as patients progress through the stages of their disease. However, the effect of hepatitis B therapy on the cancer risk has not been factored in.

One factor in common between the hepatitis B risk nomogram and that developed by Lok et al is the contribution of smoking to cancer risk. Perhaps it is time hepatologists and gastroenterologists jumped on the antismoking bandwagon with pulmonologists and cardiologists, and encouraged patients to stop smoking to reduce cancer risk.

Subjects in the HALT-C study underwent regular surveillance using ultrasonography. The surveillance interval was between six and 12 months. At the time the study was designed, the optimal surveillance interval was not known. More recently, data have emerged suggesting a six-month surveillance interval is associated with a better survival than a 12-month interval.11 No doubt as a result of surveillance, 75 percent of HCC patients in the HALT-C study were diagnosed as early-stage disease, Model for End-Stage Liver Disease (MELD) Tumor–Node–Metastasis (TNM) T1 or T2.

The MELD TNM staging is commonly used to stage HCC in the United States, but was developed primarily to help assess patients with HCC for liver transplantation.12 It is not a true staging system, because it was developed to compare survival in patients with HCC and those without HCC on the transplant waiting list. It should not be used in patients who are not undergoing liver transplantation. There are several other HCC staging systems available, but the most widely accepted is the Barcelona Cancer of the Liver Clinic (BCLC) staging system,13 which combines tumor characteristics with liver function and performance status.

The MELD TNM ignores liver function and performance status, appropriately; the liver would be removed at transplantation. Poor liver function is not a restriction on this modality of treatment. However, in the nontransplant setting, liver function is a major determinant of what treatment might be provided. MELD TNM should only be used for the specific purpose for what it was designed.

Regular surveillance identified 75 percent of HCCs at T1 or T2. We are not told what proportion were T1 or T2, which is an important distinction. The objective of HCC surveillance should be to find lesions less than 2 cm (T1) because the likelihood of a cure with treatment is greatest for these lesions. For example, studies have documented complete ablation using radiofrequency is possible in almost 100 percent of lesions less than 2 cm,14,15 but as the lesion approaches 3 cm in size, the complete ablation rate decreases. Because tumor size is a surrogate for vascular invasion, and because vascular invasion is a poor prognostic sign, the smaller the HCC at diagnosis the better.

Unfortunately, in the United States HCC surveillance is not applied as widely as it is in many European and Far Eastern countries. Most gastroenterologists offer some form of surveillance to their patients,16 but few other caregivers do.17 In one study, only 60 percent of patients presenting with HCC received any form of surveillance.18 Seventeen percent of patients with HCC were known to be cirrhotic and did not receive surveillance, and 18 percent were not even known to be cirrhotic. When surveillance is offered, it is most frequently with periodic testing of alpha-fetoprotein,17 whereas the recommendation is not to use alpha-fetoprotein, but to rely on ultrasound.19

Finally, despite universal surveillance in the HALT-C study, only 54 percent of patients with HCC underwent attempt at cure with resection, transplantation or radiofrequency ablation. In another study,18 only 35 percent of patients thought to be suitable for liver transplantation actually received a new liver. Since the rate of liver transplantation in the HALT-C cohort is so high is no doubt a reflection of the academic nature of the centers involved in the study. However, liver transplantation is not the answer for the majority of patients with HCC, not even in the United States. There are simply not enough donor livers to go round.

In the HALT-C study, an additional small group of patients underwent chemoembolization. This is not a curative therapy. We are not given sufficient detail to know why these patients presented with such advanced disease despite surveillance. Ideally, surveillance should identify the majority of HCCs at a stage when curative treatment is possible. The hepatology group in Toronto has been providing surveillance for a large cohort of patients at risk for HCC for several years now. In our experience, it is rare not to be able to offer an attempt at curative therapy (resection, local ablation or transplantation) for these HCC found on surveillance. In Europe and Japan 50 to 60 percent of patients undergoing surveillance present with single lesions, usually less than 2 cm.20

In summary, these results from the HALT-C study have shown clearly that maintenance hepatitis C treatment does not reduce the incidence of HCC. HCC can occur at an appreciable frequency in patients with advanced fibrosis but who do not have cirrhosis. The risk factors for HCC can be identified and an algorithm has been developed that identifies patients at risk for HCC more accurately. These patients can then be targeted for surveillance. However, the intensity of surveillance in the United States has to increase to a level that exists elsewhere in the world before HCC surveillance will result in mortality reduction.

MORRIS SHERMAN
University of Toronto
Toronto General Hospital
Toronto, Ontario, Canada

References

1. Di Bisceglie AM, Shiffman ML, Everson GT, et al. Prolonged therapy of advanced chronic hepatitis C with peginterferon: the Hepatitis C Antiviral Long-Term Treatment against Cirrhosis (HALT-C) Trial. N Engl J Med. In press.


2. Lok AS, Seeff LB, Morgan TR, et al. Incidence of hepatocellular carcinoma and associated risk factors in hepatitis C-related advanced liver disease. Gastroenterology 2009;136:138–148.


3. Shiffman ML, Hofmann CM, Contos MJ, et al. A randomized, controlled trial of maintenance interferon therapy for patients with chronic hepatitis C virus and persistent viremia. Gastroenterology 1999;117:1164–1172.


4. Simonetti RG, Camma C, Fiorello F, et al. Hepatitis C virus infection as a risk factor for hepatocellular carcinoma in patients with cirrhosis. A case-control study. Ann Intern Med 1992;116: 97–102.


5. Benvegnu L, Gios M, Boccato S, et al. Natural history of compensated viral cirrhosis: a prospective study on the incidence and hierarchy of major complications. Gut 2004;53:744–749.


6. Fattovich G, Giustina G, Degos F, et al. Morbidity and mortality in compensated cirrhosis type C: a retrospective follow-up study of 384 patients. Gastroenterology 1997;112:463–472.


7. Sarasin FP, Giostra E, Hadengue A. Cost-effectiveness of screening for detection of small hepatocellular carcinoma in western patients with Child-Pugh class A cirrhosis. Am J Med 1996;101: 422–434.


8. Arguedas MR, Chen VK, Eloubeidi MA, et al. Screening for hepatocellular carcinoma in patients with hepatitis C cirrhosis: a costutility analysis. Am J Gastroenterol 2003;98:679–690.


9. Thompson Coon J, Rogers G, Hewson P, et al. Surveillance of cirrhosis for hepatocellular carcinoma: a cost-utility analysis. Br J Cancer 2008;98:1166–1175.


10. Chen CJ, Yang HI, Iloeje UH, et al. A risk function nomogram for predicting HCC in patients with chronic hepatitis b: the Reveal- HBV study. J Hepatol 2007;46(Suppl 1):S180.


11. Kim DY, Han KH, Sang HA, et al. Semiannual surveillance for hepatocellular carcinoma improved patient survival compared to annual surveillance (Korean experience). Hepatology 2004; 6(Suppl 1):368.


12. Marsh JW, Dvorchik I, Bonham CA, et al. Is the pathologic TNM staging system for patients with hepatoma predictive of outcome? Cancer 2000;88:538–543.


13. Llovet JM, Brú C, Bruix J. Prognosis of hepatocellular carcinoma: the BCLC staging classification. Semin Liver Dis 1999;19:329– 338.


14. Sala M, Llovet JM, Vilana R, et al. Initial response to percutaneous ablation predicts survival in patients with hepatocellular carcinoma. Hepatology 2004;40:1352–1360.


15. Livraghi T, Meloni F, Di Stasi M, et al. Sustained complete response and complications rates after radiofrequency ablation of very early hepatocellular carcinoma in cirrhosis: is resection still the treatment of choice? Hepatology 2008;47:82–89.


16. Chalasani N, Said A, Ness R, et al. Screening for hepatocellular carcinoma in patients with cirrhosis in the United States: results of a national survey. Am J Gastroenterol 1999;94:2224–2229.


17. Davila JA, Weston A, Smalley W, et al. Utilization of screening for hepatocellular carcinoma in the United States. J Clin Gastroenterol 2007;41:777–782.


18. Stravitz RT, Heuman DM, Chand N, et al. Surveillance for hepatocellular carcinoma in patients with cirrhosis improves outcome. Am J Med 2008;121:119–126.


19. Bruix J, Sherman M. Management of hepatocellular carcinoma. Hepatology 2005;42:1208–1236.


20. Tanaka H, Nouso K, Kobashi H, et al. Surveillance of hepatocellular carcinoma in patients with hepatitis C virus infection may improve patient survival. Liver Int 2006;26:543–551.

Address requests for reprints to: Morris Sherman, MB, BCh, PhD, FRCP(C), Associate Professor of Medicine, University of Toronto, Toronto General Hospital, Rm NCSB 11C 1252, 585 University Avenue, Toronto, ON, M5G 2N2, Canada; e-mail: morris.sherman@uhn.on.ca; fax: 416-591-2107.

The author discloses the following: The author has received honoraria not exceeding $10,000 from Hoffmann la Roche for speaking and for acting as a consultant to an advisory board. The author has received research support from Hoffman LaRoche.

© 2009 by the AGA Institute
0016-5085/09/$36.00
doi:10.1053/j.gastro.2008.11.023