Humana Press M E T H O D S I N M O L E C U L A R M E D I C I N E TM Humana Press M E T H O D S I N M O L E C U L A R M E D I C I N E TM Hepatocellular Carcinoma Edited by Nagy A. Habib Methods and Protocols Hepatocellular Carcinoma Edited by Nagy A. Habib Methods and Protocols HCC: The Clinical Problem 3 Hepatocellular Carcinoma The Clinical Problem Valery Usatoff and Nagy A. Habib 1. Introduction Hepatocellular carcinoma (HCC) is one of the most common malignancies, responsible for over one million deaths annually world wide. The causal relationship between HCC and cirrhosis is clear, with the majority of cases of HCC occurring in patients with cirrhotic liver disease. Survival of untreated individuals is poor. At this stage, surgical resection provides the only chance of cure, but it is not suitable for the majority of patients in whom the tumor pathology or the underlying liver disease makes surgery hazardous. For most patients, nonsurgical treatment is the only option and this has led to the popularization of various regional and systemic modalities. Tumor stage is a significant predictor of survival and screening high-risk groups allows detection of tumors at an earlier stage, thereby increasing the likelihood of effective treatment. This chapter sets the clinical scene for the rest of this book by outlining the important issues in the management of HCC with particular focus on the limitations of the current treatment modalities. 2. Epidemiology The incidence of HCC in Western countries is low, but it remains a serious health problem globally, causing an estimated 1,250,000 deaths every year world- wide (1). Western countries have an incidence of about 4 in 100,000, whereas areas of Africa and Asia have an incidence of up to 150 in 100,000 (2). In Zimba- bwe and Ethiopia, it accounts for up to 50% of all malignancies, whereas in Europe 1 From: Methods in Molecular Medicine, vol. 45: Hepatocellular Carcinoma Methods and Protocols Edited by: N. A. Habib © Humana Press Inc., Totowa, NJ 3 4 Usatoff and Habib and North America, it accounts for less than 2%. These geographical variations closely reflect the influence of local risk factors. It is clear that cirrhosis is the prime risk factor for HCC with up to 90% of patients having liver cirrhosis (3), but hepatocarcinogenesis also depends on the underlying cause of cirrhosis with inter- play from secondary risk factors. Patients at high risk of developing HCC are those with cirrhosis caused by viral hepatitis and hemochromatosis. Those at moderate risk have alcohol, α-1-antitrypsin deficiency, and autoimmune hepatitis-induced cirrhosis, whereas cirrhosis from Wilson’s disease, primary biliary cirrhosis, and sclerosing cholangitits have a relatively low risk (4). Male sex and cigarette smok- ing provide secondary risk factors for HCC in patients with cirrhosis (5). The incidence of HCC among patients with cirrhosis was found to be 12.5% over a 3 yr compared to 3.8% among patients with chronic active hepatitis with- out cirrhosis (5). Recent interest has turned to the group of patients without cir- rhosis that develop HCC . Aflatoxin B1 and thorotrast (a contrast material) are two carcinogens that induce HCC in the noncirrhotic liver. The relationship between HCC and human steroids is disputable and the effect of these agents is probably very minor (6). It would appear that both hepatitis B (HBV) and hepa- titis C (HBC) virus can cause HCC in the noncirrhotic liver, but the HCV-in- duced HCC is more likely to be in older patients with cirrhosis (7). 3. Pathology The stepwise development of tumors has been well established and is usually a complex process involving at least three steps: initiation, promotion, and progression. Each step can only occur if there is a breakdown of a natural protective barrier and the oncogene is the tool by which the tumor breaks down these barriers (8). In colorectal cancer, Vogelstein defined the stepwise progres- sion from hyperplasia to adenoma to carcinoma and concluded that the process involves activation of an oncogene, with the loss of several tumor suppressor genes (9). No single oncogene has been defined as a constant occurrence for HCC but several tumor suppressor genes do appear more commonly. Arakawa et al. (10) first suggested that HCC emerges in an adenomatous hyperplastic nod- ule. Differentiation between regenerative nodules and low-grade dysplastic nod- ules and then between high-grade dyplastic nodules and overt HCC is difficult. An increase in size correlates with malignancy, and benign nodules are rarely larger than 2 cm. As these nodules grow in size, there is a loss of normal histo- logical architecture and the portal supply is replaced by newly formed arterial vasculature. The most widely used staging system for HCC is the tumor, nodes, metastasis (TNM) staging system (Fig. 1) and is based on the size, number and distribution of the primary lesion and also on the presence of vascular invasion, lymph node involvement, and distant metastases. HCC: The Clinical Problem 5 1. The TNM Staging of HCC Tx Primary cannot be assessed T 0 No evidence of primary tumor. T 1 Solitary tumor, ≤ 2cm diameter, without vascular invasion T 2 Solitary tumor, ≤2 cm, with vascular invasion or Multiple tumors, limited to one lobe all ≤ 2cm without vascular invasion or Solitary tumor, >2 cm, without vascular invasion T 3 Solitary tumor, >2 cm, with vascular invasion or Multiple tumors, limited to one lobe, ≤2 cm, with vascular invasion or Multiple tumors, limited to one lobe, >2 cm, with or without vascular invasion T 4 Multiple tumors involving more than one lobe or Any invasion of major branch of portal or hepatic vein Nx Cannot assess nodes M x Cannot assess metastasis N 0 No regional node involved. M 0 No distant metastasis N 1 Regional nodes involved M 1 Distant metastasis Stage I T 1 N 0 M 0 Stage II T 2 N 0 M 0 Stage III T 1/2 N 1 M 0 T 3 N 0/1 M 0 Stage IV A T 4 N 0/1 M 0 Stage IV B T 1-4 N 0/1 M 1 Fig.1. TNM staging of HCC, data from ref. 17. 4. Natural History The mean survival of patients with HCC from diagnosis is approx 3 mo (11). There is however, a wide range, and the survival of patients is closely related to 6 Usatoff and Habib the stage of the tumor and to the extent of underlying liver disease. The simplest system of incorporating both tumor stage and liver impairment was put for- ward by Okuda et al. (12), which was based on the presence or absence of ascites, tumor volume >50% of liver, serum albumin < 30 g/L, and serum bilirubin >30 mg/L. Recently, several prognostic models have been put forward based on the complex multivariate analysis of untreated or ineffectively treated patients (13,14). Although these models do not easily lend themselves to clini- cal practice, they give a good estimate of the median survival from time of diagnosis to be in the range of 1 mo to 12 mo. Another, somewhat retrospective way of looking at prognosis is to categorize patients into resectable or nonresectable. Patients with resectable tumors who undergo adequate surgery have a survival of 20–30 mo. Those with nonresectable disease have a survival of 3–10 mo (15). 5. Diagnosis and Staging Once the suspicion of HCC is raised by clinical symptoms, ultrasound (US) scanning or elevated α-fetoprotein (AFP) levels, the aim of further investiga- tions is to confirm the diagnosis, stage the tumor, and assess the underlying liver disease. Percutaneous biopsy may lead to tumor dissemination along the needle tract and convert a favorable tumor to an inoperable one. Biopsy can be useful in certain situations, but only after liver resection or transplant have been excluded as treatment options. For HCCs <2 cm, the detection rate by computer tomography (CT) scanning is 72%, whereas it is 93% by magnetic resonance imaging (MRI) (16). Hepatic angiography is very useful in confirm- ing the diagnosis of HCC because of the very particular vascular features the tumor exhibits. CT angiography/portography and delayed CT scanning after intra-arterial injection of lipiodol are also very sensitive and specific imaging modalities to confirm the diagnosis and detect small lesions. Staging the tumor depends on documenting the number, site, and size of the tumor(s) and their relationship to the major vessels (17). Before curative surgery can be consid- ered, extrahepatic disease needs to be excluded and the common sites are local nodes, chest, and bone. Hepatic resection can only be considered in patients with adequate functional reserve and this can be estimated preoperatively by a variety of methods. These tests rely on measuring either the synthetic function of the liver, its excretion of certain metabolites, or on an estimate of the likely remaining liver volume after resection. However a common standard does not exist. 6. Screening Certain characteristics of HCC make screening and early detection an attractive proposition: The at-risk population can be identified, resection of early stage tumors can be curative (18), HCC tends to grow slowly and stay confined to the liver (19), HCC: The Clinical Problem 7 and early detection is possible with US and AFP monitoring (20). However, the sensitivity and specificity of AFP is limited. Only 50–70% of patients with HCC have elevated levels of AFP (15). Only approximately one-third of patients with small HCCs (<5 cm) have a serum AFP above 200 ng/mL (21). At a cutoff point of 100 ng/mL, the sensitivity is 60% and the specificity is 95%. The assay becomes more useful if repeated tests show increasing levels (22). Ultrasound is widely used for screening because it is noninvasive and inexpen- sive, but the cirrhotic background of the liver makes detection of small tumors difficult. In a group of patients in which 50% were ultimately transplanted for cirrhosis, Dodd et al. (23) reported a detection rate of HCC nodules of 45%. Groups that have performed repeated ultrasound examinations are able to achieve sensitivity and specificity levels of 90% for tumors larger than 1 cm (24). Screening with AFP measurements every 2 mo and US every 3 mo has been shown to significantly increase the detection rate of HCCs less than 3 cm in diameter and with less portal vein invasion. As a result, more patients were deemed resectable and the clinical value of this approach was evidenced by a significant improvement in survival (25). Despite this apparent success of early detection, no screening program has so far succeeded in demonstrating a cost-effective way of detecting curable HCC (26). 7. Treatment Because only 20% of patients are considered suitable for hepatic resection at the time of diagnosis, based on tumor stage or underlying cirrhosis, the nonresectional modalities of treatment are an important part of the clinician’s armamentarium. In the following sections, we will review the different treat- ment modalities commonly employed for resectable and nonresectable HCC. 7.1. Liver Resection Hepatic resection has long been considered the only potentially curative treatment of HCC. However, as a result of advanced tumor stage and underly- ing cirrhosis, less than 20% of patients are eligible for resection at the time of presentation. A tumor diameter of less than 5 cm is often used as a cutoff for resection because of the increased risk of additional nodules and, consequently, incomplete resection. Newer imaging can more confidently exclude secondary nodules, and, hence, even large tumors that are truly solitary may still be suitable for successful resection despite their size (27). There is often a balance between the radicality of resection for cure and the conservation of liver paren- chyma to avoid postoperative hepatic decompensation. As summarized in Table 1, published series report a 1-yr survival rate between 56% and 88% and a 5-yr survival rate between 28% and 59%. Operative mortality in these series varies from 2% to 16% but needs to be considered in relation to the number and degree of cirrhotic patients in the series. The operative mortality rates are 8 Usatoff and Habib less than 3% for noncirrhotics and have been reported as high as 25% for cir- rhotic patients (38,39). It has been put forward that patients with Child’s A cirrhosis could safely undergo extensive resections, but this is proving to be an unreliable measure of preoperative liver function. More than half of these patients will develop hepatic decompensation after resection, and if persistant, it indicates a poor prognosis (40). Patient selection can be further refined using various preoperative assess- ments. Recent studies have redefined the cut-off level of indocyanine green, (ICG) retention at which a major hepatectomy can be safely performed (41) and others have shown that significant portal hypertension (hepatic venous pressure gradient ≥ 10mm Hg) is an accurate predictor of persistent postopera- tive hepatic decompensation (40). The main problem with liver resection for HCC is the high recurrence rate, which has a median survival rate of 1 yr (37). The incidence of recurrence is reported as between 20% and 64% within the first year, between 57% and 81% at 3 yr, and between 75% and 100% at 5 yr (33–35,42,43). The majority (80–90%) of patients that develop recurrence do so within the liver and only 10–20% have distant metastases. The pattern of recurrence is such that 12–26% occur at the resected margin, 40–50% away from the margin, and in 20–25% of cases there is widespread multinodular recurrence (18). This reflects the potential sources of recurrence, namely positive margins, undiagnosed multifocal disease, and ongoing malignant potential in the remaining cirrhotic liver. As a result, it is difficult to confidently classify a resection as curative. Even with the most favor- able tumors, tumor-free survival at 5 yr ranges from 20–30% in Asian patients to nearly 0% in Western patients (33,36,43,44). 7.2. Liver Transplantation In patients with HCC and cirrhosis, liver transplantation offers the prospect of treating the tumor and the underlying liver disease. This has the attraction of avoid- ing the morbidity of postoperative liver failure, eliminating the chance of further tumor occurrence in the remaining diseased liver, and preventing progression of portal hypertension. Early results were disappointing because the procedure was performed on patients with advanced disease, and the resulting high recurrence rate, consequent on immunosupression, led to poor long-term survival. This was in contrast to the much better results that were observed in patients who underwent liver transplantation for cirrhosis and were incidentally found to have HCC in the resected liver. It became clear that survival after liver transplantation for HCC was linked to tumor stage. Selby (45) details the 5-yr survival rates by TNM stage as follows: stage I—75%, stage II—68%; stage III—11%. If selec- tion criteria are applied, then consistently reasonable results can be obtained. HCC: The Clinical Problem 9 The ranges of 1-, 3-, and 5-yr survivals achieved in recent series are 45–71%, 21–45%, and 20–45%, respectively (Table 2). Most centers restrict transplantation to patients with less than three nodules, with tumors less than 3 cm in diameter and with no vascular invasion, although the most useful prognostic criteria have not yet been identified. These sorts of selection criteria have led to a reduction in the recurrence rate and a long-term survival rate comparable to non-HCC patients. Earlier recurrence rates were as high as 65% (51), but in a recent study of 48 patients, the actuarial survival rate was 75% at 4 yr with a recurrence rate of only 17% (52). Despite its apparent attractiveness, liver transplantation for HCC has several limitations. Organ shortage is probably the main factor leading to long waiting periods and inevitable progression of the disease to a less favorable tumor stage. The high risk of recurrent viral hepatitis and the potential for increased growth of residual/recurrent tumor in the setting of immunosupression also pose sig- nificant problems. 7.3. Transcatheter Arterial Chemoembolization Hepatocellular carcinoma derives the majority of its blood supply from the hepatic artery, compared to the surrounding normal liver, which is mainly supplied by the portal vein. This difference has been used to advantage in selectively treating tumor nodules with various forms of embolization. Arterial blood flow can be interrupted by selective catheterization of the hepatic artery branch feeding the tumor and instillation of embolizing agents such as metallic coils, gelfoam, or starch. To avoid extensive hepatic necrosis, patency of the portal vein needs to be confirmed before this can be undertaken. Simple embo- Table 1 Survival After Surgical Resection of HCC Authors No. of Cases Operative 1-yr 3-yr 5-yr Mortality (%) Survival Survival Survival Tsuzuki et al. (28) 119 9 80 47 39 Franco et al. (29) 72 7 68 51 — Gozzetti et al. (30) 168 8 — 57 36 Gennari et al. (31) 84 16 85 45 28 Nagasue et al. (32) 229 11 — 53 29 Izumi et al. (33) 104 7 88 65 59 Chen et al. (34) 205 4 56 36 28 Lai et al. (35) 194 12 n/a 44 35 Takenaka et al. (36) 280 2 88 70 50 Farges et al. (37) 226 8 82 59 39 10 Usatoff and Habib lization of the hepatic blood supply will result in ischemic necrosis of more than 80% of the tumor, in most patients (53). Where gelfoam is used rather than metallic coils, the segmental branch will recannalize and the procedure can be repeated at intervals of 6–12 wk. Embolization has been combined with a variety of chemotherapeutic agents and with lipiodol to potentially prolong the local concentration of these agents. Minor morbidity is common after this procedure, occurring in nearly 90% of patients (54) and has been termed the postembolization syndrome. The symptoms consist of abdominal pain, nausea, and fever and usually resolve within a week. Other complications include cholecystits. Mortality rates of less than 2% can be expected in patients with normal hepatic function, but this increases markedly in patients with poor liver reserve, being 37% in Child’s C patients (55). Nonrandomized controlled studies have shown significant improvement in survival (55–57); however, this has not been supported by randomized controlled studies (58). Other variations on the intra-arterial approach to treatment have been used, but no randomized controlled trials have yet shown advantage over other methods. Neocarzinostatin is a proteinaceous antibiotic with antitumor effect. The styrene–maleic acid form has enhanced cytotoxicity and other pharmaco- logical advantages. Styrene–maleic acid neocarzinostatin (SMANCS) has been used effectively in patients with unresectable HCC (59), but no comparative data are yet available. 7.4. Percutaneous Ethanol Injection Under ultrasound guidance, a fine needle can be introduced into the tumor within the liver and ethanol injected to cause coagulative necrosis. HCC is more sensitive to the effects of the percutaneous ethanol injection (PEI) because of the difference in density between the soft tumor and the cirrhotic liver. This Table 2 Results of Liver Transplantation for HCC Group No. of Cases Operative 1-yr 3-yr 5-yr Mortality(%) Survival Survival Survival Yokoyama et al. (46) 80 13 64 45 45 Ismail et al. (47) 21 38 45 21 21 Pichlmayr et al. (48) 87 20 55 30 20 Farmer et al. (49) 44 16 63 30 30 Haug et al. (50) 24 17 71 42 — HCC: The Clinical Problem 11 causes nearly complete destruction of the tumor with minimal injury to the surrounding liver. Although other agents (acetic acid, hot saline, and chemo- therapeutic agents) have been tried, ethanol is the most widely used because it is readily available, inexpensive, well tolerated by patients, has low systemic toxicity, is effective, and can be used repeatedly. The extent of necrosis is closely related to the size of the tumor, with small nodules (<3 cm) usually completely destroyed while larger ones only partially destroyed (60). It is usually performed as an outpatient procedure twice a week using 3–5 mL of ethanol on each occasion. Small lesions can be treated with 3–6 sessions, whereas larger tumors have been treated with up to 15 sessions (61). The assessment of tumor destruction is difficult and usually relies on repeated imaging and measurement of AFP levels (62). The patients most suited for this treatment are those with low-stage HCC (single tumor <5 cm or no more than three nodules, each <3 cm) and a liver function that limits resection (6). Portal vein invasion is not a contraindication (63), but patients with Child’s C cirrhosis should be treated with caution, especially because PEI seems not to modify the otherwise rapidly fatal outcome in these patients (64). The most common complications with this technique are transient abdominal pain and fever. Severe complications such as hemorrhage or hepatic abscess are infrequent, occurring in 1.7% of cases (64). The 1-yr sur- vival rate is more than 90% in many series, and for small tumors in cirrhotic patients, the 3-yr and 5-yr survival rates are 63% and 39%, respectively (65). Recurrence rates at 1 and 2 yr are 28.3% and 54%, respectively (61), although it has been suggested that the majority of the so-called recurrences actually represent new tumors (66). Larger tumors can be treated with PEI but with less effect, prognosis being related to the presence of cirrhosis, tumor size, and the number and the level of AFP (67). There are no prospective randomized trials comparing PEI to surgery, but retrospective matched control studies suggest that both modalities have similar outcomes for single HCCs less than 3 cm. The operative mortality of resection is offset by the greater recurrence rates after PEI. Surgery probably provides a better chance of cure for solitary lesions greater than 3 cm by adequately treat- ing the surrounding foci of microscopic tumor, but possesses a greater proce- dure-related risk than PEI in patients with Child’s B cirrhosis (37). 7.5. Radiotherapy The conventional approach of whole-liver irradiation is not effective. At the dose required to destroy the tumor, the surrounding liver also undergoes hepa- titis and even failure in a cirrhotic liver. Modern three-dimensional beam-fo- cusing methods can minimize beam scatter and deliver the required dose more specifically to the tumor only. This method of targeted radiotherapy has shown [...]... scanning, From: Methods in Molecular Medicine, vol 45: Hepatocellular Carcinoma Methods and Protocols Edited by: N A Habib © Humana Press Inc., Totowa, NJ 21 22 Riordan and Williams the indocyanine green and bromosulfthalein retention rates, uptake of technetium-99m–diethylenetriaminepentaacetic acid–galactosyl human serum albumin, the serum lecithin aminotransferase level, the Child’s class and its individual... cirrhosis and may possibly be responsible for development of HCC On the other hand, a particular HCV protein(s), in concert with the other HCV proteins, may directly be involved in hepatocarcinogenesis by the virus For example, an amino-terFrom: Methods in Molecular Medicine, vol 45: Hepatocellular Carcinoma Methods and Protocols Edited by: N A Habib © Humana Press Inc., Totowa, NJ 37 38 Ishido, Fujita, and. .. Y., and Kawasaki, H (1992) Hepatic vascular and bile duct injury after ethanol injection therapy for hepatocellular carcinoma Gastrointest Radiol 17, 167–169 30 Taavitsainen, M., Vehmas, T., and Kaupilla, R (1993) Fatal liver necrosis following ethanol injection for hepatocellular carcinoma Abdom Imag 18, 357–359 31 Vehmas, T., Taari, K., Kivisaari, L., Laurila, P., and Taavitsainen, M (1993) Intra- and. .. treatment of hepatocellular carcinoma: a randomised controlled trial Lancet 352, 17–20 54 Manesis, E K., Giannoulis, G., Zoumboulis, P., Vafiadou, I., and Hadziyannis, S J (1995) Treatment of hepatocellular carcinoma with combined suppresion and inhibition of sex hormones, a randomized, controlled trial Hepatology 21, 1535–1542 55 Guechot, J., Peigney, N., Ballet, F., Vaubourdolle, M., Giboudeau, J., and Poupon,... 836–874 2 Rustgi, V K (1989) Epidemiology of hepatocellular carcinoma Gastroenterol Clin North Am 16, 545–551 3 Kew, M C and Popper, H (1984) Relationship between hepatocellular carcinoma and cirrhosis Semin Liver Dis 4, 136–146 4 De Bac, C., Stroffolini, T., Gaeta, G B., Taliani, G., and Giusti, G (1995) Pathogenic factors in cirrhosis with and without hepatocellular carcinoma: A multi–centre Italian... lipiodol chemoembolization and conservative treatment for unresectable hepatocellular carcinoma N Engl J Med 332, 1256–1261 23 Shah, S R., Riordan, S M., Karani, J., and Williams, R (1998) Tumor ablation and hepatic decompensation rates in multi–agent chemoembolization of hepatocellular carcinoma Q J Med 91, 821-828 24 Yu, Y Q., Xu, D B., Zhou, X D., Lu, J Z., Tang, Z Y., and Mack, P (1993) Experience... chemotherapy with implanted reservoir for hepatocellular carcinoma Cancer Chemother Pharmacol 31(Suppl 1), s93–s98 73 Venook, A P (1994) Treatment of hepatocellular carcinoma: too many options? J Clin Oncol 12(6), 1323–1334 74 Trinchet, J C and Beaugrand, M (1997) Treatment of hepatocellular carcinoma in patients with cirrhosis J Hepatol 27, 756–765 75 Zhou, X D., Yu, Y.Q., and Tang, Z Y (1992) An 18–year study... the pig liver Scand J Gastroenterol 28, 791–794 32 Shimada, M., Maeda, T., Saitoh, A., Morotomi, I., and Kano, T (1995) Needle track seeding after percutaneous ethanol injection therapy for small hepatocellular carcinoma J Surg Oncol 58, 278–281 33 Ebara, M., Kita, K., Yoshikawa, M., Sugiura, N., and Ohto, M (1992) Percutaneous ethanol injection for patients with small hepatocellular carcinoma, in Primary... of hepatic resection and transcatheter arterial embolization for the treatment of primary hepatocellular carcinoma Cancer 71(7), 2181–2186 88 Simonetti, R G., Liberati, A., Angiolini, C., and Pagliaro, L (1997) Treatment of hepatocellular carcinoma: a systematic review of randomized controlled trials Ann Oncol 8(2), 117–136 Medical Management of HCC 21 2 Medical Management of Hepatocellular Carcinoma... Pichlmayr, R., Weimann, A., Steinhoff, G., and Ringe, B (1992) Liver transplanta- 18 49 50 51 52 53 54 55 56 57 58 59 60 61 62 Usatoff and Habib tion for hepatocellular carcinoma: clinical results and future aspects Cancer Chemother Pharmacol 31(Suppl 1), s157–s161 Farmer, D G., Rosove, M H., Shaked, A., and Busuttil, R W (1994) Current treatment modalities for hepatocellular carcinoma Ann Surg 219(3), . C I N E TM Hepatocellular Carcinoma Edited by Nagy A. Habib Methods and Protocols Hepatocellular Carcinoma Edited by Nagy A. Habib Methods and Protocols HCC: The Clinical Problem 3 Hepatocellular. Methods in Molecular Medicine, vol. 45: Hepatocellular Carcinoma Methods and Protocols Edited by: N. A. Habib © Humana Press Inc., Totowa, NJ 3 4 Usatoff and Habib and North America, it accounts for. 20% and 64% within the first year, between 57% and 81% at 3 yr, and between 75% and 100% at 5 yr (33–35,42,43). The majority (80–90%) of patients that develop recurrence do so within the liver and