fax: (81) 3-3814-0021
Liver Failure and Liver Disease
Prediction of recurrence of hepatocellular carcinoma after curative ablation using three tumor markers†
Article first published online: 28 NOV 2006
Copyright © 2006 American Association for the Study of Liver Diseases
Volume 44, Issue 6, pages 1518–1527, December 2006
How to Cite
Tateishi, R., Shiina, S., Yoshida, H., Teratani, T., Obi, S., Yamashiki, N., Yoshida, H., Akamatsu, M., Kawabe, T. and Omata, M. (2006), Prediction of recurrence of hepatocellular carcinoma after curative ablation using three tumor markers. Hepatology, 44: 1518–1527. doi: 10.1002/hep.21408
Potential conflict of interest: Nothing to report.
- Issue published online: 28 NOV 2006
- Article first published online: 28 NOV 2006
- Manuscript Accepted: 28 AUG 2006
- Manuscript Received: 30 MAR 2006
Three tumor markers for hepatocellular carcinoma (HCC) are available in daily practice in Japan: alpha-fetoprotein (AFP), des-gamma-carboxy prothrombin (DCP), and lens culinaris agglutinin-reactive fraction of alpha-fetoprotein (AFP-L3). To elucidate the predictability of these tumor markers on HCC recurrence after curative ablation, we enrolled 416 consecutive patients with naïve HCC who had been treated by percutaneous ablation at our department from July 1997 to December 2002. Tumor marker levels were determined immediately before and 2 months after the treatment. Complete ablation was defined on CT findings as nonenhancement in the entire lesion with a safety margin. Tumor recurrence was also defined as newly developed lesions on CT that showed hyperattenuation in the arterial phase with washout in the late phase. We assessed the predictability of recurrence via tumor markers in multivariate analysis, using proportional hazard regression after adjusting for other significant factors in univariate analysis. Until the end of follow-up, tumor recurrence was identified in 277 patients. Univariate analysis revealed the following factors to be significant for recurrence: platelet count; size and number of tumors; AFP, AFP-L3, and DCP preablation; and AFP and AFP-L3 postablation. Multivariate analysis indicated that AFP >100 ng/mL and AFP-L3 >15%, both pre- and postablation, were significant predictors. The positivity of AFP and AFP-L3 preablation that turned negative postablation was not significant. In conclusion, tumor markers pre- and post-ablation were significant predictors for HCC recurrence and can complement imaging modalities in the evaluation of treatment efficacy. (HEPATOLOGY 2006;44:1518–1527.)
Hepatocellular carcinoma (HCC) is a common malignancy worldwide, and its incidence is increasing in the United States and elsewhere.1, 2 Current options for the treatment of this cancer consist of surgical resection, orthotropic liver transplantation, transcatheter arterial embolization, and percutaneous ablation therapy. Although surgical resection is usually the first choice for treatment,3, 4 it is frequently contraindicated by underlying chronic liver disease based on hepatitis B or C virus infection.5, 6 Orthotopic liver transplantation is a strategy that can treat both cancer and liver dysfunction, and indeed has shown an excellent survival rate in patients at an early stage of the cancer (e.g., single nodule measuring ≤5 cm in diameter or fewer than three nodules measuring ≤3 cm in diameter).7, 8 However, in countries such as Japan, where cadaveric donor organs are scarce, application of liver transplantation is limited. Percutaneous ablative methods—including percutaneous ethanol injection therapy, percutaneous microwave coagulation therapy, and radiofrequency ablation—can achieve high local cure without deteriorating background liver function9–15 and have played an important role in the treatment of HCC.
Alpha-fetoprotein (AFP) has served as a diagnostic test for HCC since the 1970s, when most patients with HCC were diagnosed at an advanced stage with clinical symptoms.16 Presently, small (e.g., ≤3 cm) HCCs can often be detected due to advances that have been made in the various imaging techniques.17, 18 They do not usually secrete a diagnostic level of AFP,19 rendering the role of AFP as a diagnostic test less significant.20 On the other hand, AFP is known to be an important predictor of prognosis,21–23 recurrence in particular. To date, several tumor markers have been proposed as a complement or substitute for AFP in HCC diagnosis, such as des-gamma-carboxy prothrombin (DCP)24, 25 and lens culinaris agglutinin-reactive fraction of AFP (AFP-L3),26, 27 but no studies have compared the predictability of these tumor markers simultaneously on HCC recurrence after curative treatment. The aim of this study was to elucidate the predictability of these tumor markers on the recurrence after a curative ablation of HCC.
Patients and Methods
We determined the tumor markers AFP, DCP, and AFP-L3 in each patient with HCC admitted to the Department of Gastroenterology at the University of Tokyo Hospital since July 1997. A total of 449 patients received percutaneous ablation as the initial treatment for HCC from July 1997 to December 2002. Among them, 416 were enrolled in this study, excluding 29 in whom percutaneous ablation was intended to reduce tumor burden because of multinodularity with some nodules left unablated, and 4 who took oral warfarin, a DCP-inducing agent (Fig. 1). Informed written consent was obtained from each patient before treatment.
Diagnosis of HCC.
Treatment and Follow-up.
Inclusion criteria for percutaneous ablation were as follows: total bilirubin concentration <3 mg/dL; platelet count no less than 50 × 103/mm3; and prothrombin activity no less than 50%. Patients with portal vein tumor thrombosis, refractory ascites, or extrahepatic metastasis were excluded. In general, we performed percutaneous ablation on patients with three or fewer lesions, all of which were ≤3 cm in diameter. However, we also performed ablation on patients with more than three lesions or lesions >3 cm in diameter if the procedure could be assumed to be clinically effective. The procedure has been meticulously described elsewhere.30
After several sessions of percutaneous ablation, dynamic CT was performed 1 to 3 days after the last session with a slice thickness of 5 mm to evaluate treatment efficacy. The interval between the initiation of contrast material infusion and CT image recording was 30 and 120 s for single-detector row spiral CT (Highspeed Advantage; GE Medical Systems, Milwaukee, WI) and 25, 40, and 120 s for multidetector-row CT (LightSpeed QX/I; GE Medical Systems). The images were presented after axial reconstruction with a slice thickness of 5 mm. Complete ablation was defined on CT findings as nonenhancement in the entire lesion with a safety margin in the surrounding liver parenchyma. Patients received additional sessions of ablation until complete ablation was confirmed in each nodule. We also assessed the changes in three tumor markers before and after curative ablation. The levels obtained before therapy and 2 months after the confirmation of complete ablation were adopted for analysis.
The follow-up consisted of monthly blood tests and monitoring of tumor markers at the outpatient clinic, and ultrasonography and dynamic CT scan were performed every 4 months. Tumor recurrence was defined according to the same criteria applied to the initial HCC. Intrahepatic HCC recurrence was classified as either recurrence at a site distant from the primary tumor or adjacent to the treated site (local tumor progression).31 The tumor markers were assessed also at the diagnosis of recurrence.
Analysis of Recurrence.
Time to recurrence was defined as the interval between the first ablation and the detection of tumor recurrence, death without recurrence, or the last examination until December 31, 2004, whichever came first. Because tumor recurrence and death without recurrence are “competing risks,” we used cumulative incidence estimation with competing risks described by Gray.32
We first plotted the cumulative rate of HCC recurrence and that of death without recurrence stratified by the following variables obtained at the time of initial ablation therapy: age, sex, tumor size, number of nodules, pathological grade, Child-Pugh class, alanine aminotransferase level, platelet count, three tumor markers, positivity for viral markers (hepatitis B surface antigen and anti–hepatitis C antibody), and alcohol consumption. Polychotomous categorical data were represented by corresponding binary dummy variables. Continuous variables were transformed into ordinal categories. The predictability of each tumor marker before and after ablation was assessed with a multivariate proportional hazard regression model described by Fine and Gray and adjusted by factors shown to be significant in the univariate analysis.33 The differences with a P value of less than .05 were considered statistically significant. All statistical analyses were performed with S-plus 2000 software (MathSoft Inc., Seattle, WA).
The baseline characteristics of the patients (276 males and 140 females) are shown in Table 1. The median age was 67 years, with the 25th and 75th percentiles being 62 and 71 years. Most (80.8%) were HCV-positive. The mean size and number of nodules were 2.7 cm and 1.7, respectively. The mean number of treatment sessions required for complete ablation was 7.7, 2.0, and 2.0 for percutaneous ethanol injection therapy, percutaneous microwave coagulation therapy, and radiofrequency ablation, respectively. Until the end of the follow-up, tumor recurrence was identified in 277 patients. Of these, 248 had intrahepatic recurrence distant from the primary site of ablation, 19 had local tumor progression, 7 had distant recurrence and local tumor progression simultaneously, and 3 had extrahepatic recurrence. Forty patients died with no sign of recurrence. Only 5 patients were lost to follow-up. The cumulative probability of overall recurrence was 18.6%, 44.7%, 59.0%, 67.2%, and 72.0% at 1, 2, 3, 4, and 5 years, respectively. The cumulative probability of local tumor progression was 2.2%, 5.1%, 6.1%, 6.5%, and 6.5% at 1, 2, 3, 4, and 5 years, respectively. The cumulative probability of death without recurrence was 2.9%, 5.8%, 8.2%, 9.6%, and 10.6% at 1, 2, 3, 4, and 5 years, respectively.
|Both positive||7 (1.7)|
|Both negative||38 (9.1)|
|Drinking >80 g/d||71 (17.1)|
|Severity of background liver disease|
|Class A||301 (72.3)|
|Class B||111 (26.7)|
|Class C||4 (0.96)|
|AST (IU/L)*||57 (41–81)|
|ALT (IU/L)*||54 (34–80)|
|Platelet count (× 103/mm3)*||103 (76–140)|
|≤2 cm||140 (33.7)|
|2.1–3 cm||161 (38.7)|
|3.1–5 cm||94 (22.6)|
|>5 cm||21 (5.0)|
|Number of nodules|
|Edmondson grade (N/A, n = 31)|
|Grade 1||90 (21.6)|
|Grade 2||260 (62.5)|
|Grades 3–4||35 (8.4)|
|≤100 ng/mL||319 (76.7)|
|101–400 ng/mL||66 (15.9)|
|>400 ng/mL||31 (7.5)|
Tumor Markers Before and After Ablation.
We plotted the value of each tumor maker immediately before ablation against the value 2 months after ablation (Fig. 2). With cutoff values of 100 ng/mLfor AFP, 100 mAU/mL for DCP, and 15% for AFP-L3, the sensitivities were 23.3%, 16.3%, and 13.7%, respectively. At least one marker was positive in 151 (36.3%) patients. After curative ablation, the value decreased below the cutoff level in 67 of 97 (69.1%), 61 of 68 (89.7%), and 46 of 57 (80.7%) patients for AFP, DCP, and AFP-L3, respectively. In Fig. 2A, dots were clustered near the diagonal line when AFP before ablation was <100 ng/mL, indicating that the changes through ablation were minimal. In those cases, AFP before ablation seems to have been produced mainly in the background liver. Although the cutoff value of 20 ng/mL is often adopted for AFP as a HCC biomarker, this value may be too low in the evaluation of efficacy of HCC treatment. In contrast, the ordinary cutoff values of 100 mAU/mL for DCP and 15% for AFP-L3 also appear to be suitable for the evaluation of treatment (Fig. 2B–C).
Predicting Factor for Recurrence.
Univariate analysis identified the following variables as significant predictors for recurrence: platelet count, size, and number of tumor nodules, AFP pre- and postablation (P < .001 and P < .001), and AFP-L3 pre- and postablation (P = .044 and P < .001) as significant predictors for recurrence (Table 2, Fig. 3). DCP preablation was also a significant predictor (P = .020) when a cutoff value of 100 mAU/mL was adopted. Child-Pugh classification was not significant for recurrence, whereas it strongly affected the probability of death without recurrence (Fig. 3A). Multivariate analysis with proportional hazard model revealed that AFP >100 ng/mL preablation and AFP-L3 >15% preablation were significant predictors of recurrence after adjusting for other significant predictors—namely, platelet count and size and number of tumor nodules (Table 3). DCP >100 mAU/mL preablation showed marginal significance (P = .069).
|Variable (n)||Cumulative Probability of Recurrence (%)||P||Cumulative Probability of Death Without Recurrence (%)||P|
|1 Year||2 Years||1 Year||2 Years|
|Age ≤67 (214)||16.8||43.5||.13||2.8||4.7||.32|
|Age >67 (202)||20.4||45.9||3.0||7.0|
|Drinking ≤80 g/d (345)||18.3||43.7||.62||3.2||6.1||.66|
|Drinking >80 g/d (71)||19.7||49.3||1.4||4.2|
|Child-Pugh A (301)||17.3||42.2||.45||1.0||2.3||<.001|
|Child-Pugh B/C (115)||21.9||51.1||7.9||15.0|
|≤100 × 103/mm3 (198)||20.2||50.2||.030||3.0||8.1||.061|
|101-150 × 103/mm3 (127)||18.9||43.5||3.9||4.7|
|>150 × 103/mm3 (91)||14.4||34.4||1.1||2.2|
|ALT ≤80 IU/L (312)||18.0||43.8||.29||2.9||6.1||.81|
|ALT >80 IU/L (104)||20.2||47.4||2.9||4.8|
|Size of tumor|
|≤2.0 cm (140)||10.0||35.1||.009||0.7||4.3||.55|
|2.1–3.0 cm (161)||16.8||46.0||4.3||6.2|
|3.1–5.0 cm (94)||27.7||53.5||4.3||8.6|
|>5.0 cm (21)||50.0||60.0||0||0|
|Single nodular (249)||16.5||38.0||<.001||2.8||6.1||.44|
|Edmondson grade 1 (90)||13.3||36.8||.18||3.3||3.3||.11|
|Edmondson grade ≥ 2 (295)||19.3||47.2||3.1||7.1|
|≤100 ng/mL (319)||15.7||41.2||<.001||3.4||5.7||.82|
|101–400 ng/mL (66)||19.7||48.5||1.5||7.6|
|>400 ng/mL (31)||46.7||73.3||0||3.3|
|≤100 mAU/mL (348)||15.5||41.8||.099||2.6||5.5||.43|
|101–400 mAU/mL (45)||33.3||60.0||6.7||6.7|
|>400 mAU/mL (23)||36.4||59.1||0||9.0|
|≤100 ng/mL (385)||17.1||42.7||<.001||3.1||6.0||.80|
|101–400 ng/mL (25)||29.2||66.7||0||4.2|
|>400 ng/mL (6)||66.7||83.3||0||0|
|≤100 mAU/mL (405)||17.8||44.4||.35||2.5||5.2||.12|
|>100 mAU/mL (9)||44.4||44.4||22.2||33.3|
|>400 mAU/mL (2)||50||NA||0||NA|
|Hazard Ratio (95% CI)||P Value|
|≤100 ng/mL preablation||1|
|>100 ng/mL preablation||1.45 (1.09–1.94)||.01|
|& ≤100 ng/mL postablation||1.23 (0.88–1.72)||.22|
|& >100 ng/mL postablation||2.31 (1.45–3.70)||<.001|
|≤100 mAU/mL preablation||1|
|>100 mAU/mL preablation||1.37 (0.98–1.93)||.069|
|& ≤100 mAU/mL postablation||1.44 (1.02–2.04)||.041|
|& >100 mAU/mL postablation||0.91 (0.27–3.13)||.89|
|>15% preablation||1.52 (1.06–2.18)||.023|
|& ≤15% postablation||1.24 (0.85–1.81)||.26|
|& >15% postablation||4.25 (1.42–12.74)||.0096|
In subgroup analysis, AFP >100 ng/mL preablation was not a significant predictor among patients in whom the value had decreased below 100 ng/mL after ablation. On the other hand AFP >100 ng/mL both pre- and postablation was a significant risk for recurrence. Similarly, AFP-L3 >15% preablation was not significant in patients whose AFP-L3 became ≤15% after ablation, whereas AFP-L3 >15% pre- and postablation was the strongest predictor of recurrence (hazard ratio, 4.25; P = .0096). Those whose DCP value was > 100 mAU/mL before ablation but below 100 mAU/mL after ablation had a higher risk of recurrence than those whose DCP had been < 100 mAU/mL preablation.
Tumor Markers at Recurrence.
Tumor markers were examined also at the diagnosis of recurrence in all 277 patients (Fig. 4). When cutoff values of 100 ng/mL, 100 mAU/mL, and 15% were applied for AFP, DCP, and AFP-L3, the sensitivities were 24.1%, 14.8%, and 22.4%, respectively. Among 71, 49, and 42 patients who were positive for AFP, DCP, and AFP-L3 before ablation, 43 (60.6%), 16 (32.7%), and 27 (64.3%) were positive for the corresponding marker also at the diagnosis of recurrence.
In this study, a recurrence of HCC was extremely frequent, although each initial treatment had been judged complete. In fact, various studies have shown a similar or higher incidence of HCC recurrence after successful medical and surgical treatments.34, 35 It has been assumed that there are two distinct types of intrahepatic recurrence of HCC: de novo carcinogenesis and intrahepatic metastasis.36 Those factors responsible for HCC development, such as fibrosis stage, age, sex, and presence of viral hepatitis, may also affect de novo carcinogenesis.34, 35 On the other hand, factors related to the primary tumor, such as the size and number of tumors, pathological grade, the presence of vascular invasion, and positivity of tumor markers, may affect the possibility of intrahepatic occult metastasis at the time of initial treatment.
We confirmed that AFP was independent from tumor size and number as a risk factor for recurrence. More than a few authors reported AFP as an independent predictor of poor prognosis, even in patients who had received curative resection.21–23, 37 High AFP values may indicate intra- or extrahepatic cancer spread and be associated with poor prognosis. On the other hand, it is well known that AFP may increase in some patients with acute or chronic hepatitis without HCC.38, 39 It is also reported that AFP levels correlate with inflammation of background disease.40 A high value of AFP postablation may also indicate active hepatic necrosis and subsequent regeneration, which may be associated with a high potential of de novo carcinogenesis in the background liver. The high level of AFP postablation is associated with a high risk of recurrence in at least two mechanisms, which are not easily distinguishable.
In the multivariate analysis on predictability of recurrence by tumor markers, we adopted a cutoff value of 100 ng/mL for AFP based on the distribution of AFP values pre- and postablation (Fig. 2A), where a large proportion of patients showed no changes in AFP levels through complete ablation when the preablation value was below 100 ng/mL. This may indicate that AFP was secreted mainly from the background liver rather than tumor itself when the level was below 100 ng/mL.41, 42
This study revealed that AFP-L3 positivity after ablation was the strongest predictor of recurrence. Those whose AFP-L3 was more than 15% postablation all encountered recurrence within 1.5 years, except for one who died of liver failure without recurrence 1.1 years after ablation. AFP-L3 is a fucosylated variant of AFP that reacts with Lens culinaris agglutinin A, and elevation in the AFP level accompanied by an increase in AFP-L3 fraction is highly specific to HCC.26, 27, 43 AFP-L3 positivity postablation strongly indicates residual cancer that cannot be detected with imaging techniques. More frequent checkups with dynamic CT may be beneficial for these patients. For overall prognosis, it is essential to detect recurrent HCC as early as possible via careful checkup and treat it effectively whenever possible, especially in countries such as Japan, where the application of liver transplantation is quite limited.
AFP-L3 value pretreatment has also been reported to be correlated with poorly differentiated cancer and poor prognosis.44, 45 In this study, AFP-L3 preablation was a significant predictor of recurrence in the multivariate analysis, but retained no significance in those whose AFP-L3 had decreased below 15% after ablation. This may indicate that ablation therapy is highly effective even for poorly differentiated HCC, and the poor prognosis of AFP-L3–positive HCC may be reversible if the marker can be turned negative by the complete ablation of tumor.
DCP, also known as prothrombin induced by vitamin K absence-II, is an abnormal prothrombin protein that is increased in the serum of HCC patients. Since the report by Liebman et al.,24 DCP has been recognized not only as a highly specific marker for HCC, but also as a predictor of the prognosis of HCC patients.46, 47 In this study, DCP >100 mAU/mL preablation showed marginal significance in multivariate analysis for the predictability of recurrence, though it is significant in univariate analysis. In contrast to AFP and AFP-L3, DCP >100 mAU/mL preablation was significant in patients whose DCP value decreased below 100 mAU/mL after curative ablation. DCP did not turn negative in 10% of patients whose DCP value was >100 mAU/mL before ablation. In those patients, DCP positivity postablation may indicate something other than residual cancer. It is known that patients with severely impaired liver function show a high level of DCP even without HCC.48 Indeed, patients with positive DCP postablation showed a high incidence of death without recurrence.
As diagnostic markers for HCC, the sensitivity of each of the three tumor markers was rather low in the present study because the patients had less-advanced disease suitable for curative ablation. However, the sensitivity of AFP-L3 had increased to 23.3% at the diagnosis of recurrence, from 13.7% before ablation, whereas sensitivity did not increase at recurrence in the other two markers. Fifteen percent of patients negative for AFP-L3 preablation turned positive at the time of recurrence. This may suggest that the diagnostic value of tumor markers increases during the clinical course of HCC.45
Recurrent tumor was sometimes detected adjacent to the primary site (local tumor progression), even if the treatment had been judged as complete on CT examination. We analyzed the risk of local tumor progression separately. The results showed that AFP >100 ng/mL pre- and postablation and AFP-L3 >15% postablation were significant risk factors for local tumor progression as well as for remote recurrence (data not shown). Local tumor progression and the majority of remote recurrence are similar in that they involve residual tumors undetectable via imaging evaluation of the treatment.
In conclusion, treatment efficacy in solid tumors has generally been based on findings obtained with imaging modalities.49 Because AFP-L3 value after curative ablation assessed via imaging could predict residual cancer, tumor markers may be a complement end point for the treatment of HCC.
- 6Interferon therapy reduces the risk for hepatocellular carcinoma: national surveillance program of cirrhotic and noncirrhotic patients with chronic hepatitis C in Japan. IHIT Study Group. Inhibition of Hepatocarcinogenesis by Interferon Therapy. Ann Intern Med 1999; 131: 174–181., , , , , , et al.
- 16Alpha-fetoprotein. In: ReadAE, ed. Modern Trends in Gastroenterology. Volume 5. London: Butterworths, 1975: 91..
- 21The Cancer of the Liver Italian Program (CLIP) Investigators. A new prognostic system for hepatocellular carcinoma: a retrospective study of 435 patients: the Cancer of the Liver Italian Program (CLIP) investigators. HEPATOLOGY 1998; 28: 751–755.
- 46Des-gamma-carboxy prothrombin as a useful predisposing factor for the development of portal venous invasion in patients with hepatocellular carcinoma: a prospective analysis of 227 patients. Cancer 2001; 91: 561–569., , , , , , et al.
- 49New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 2000; 92: 205–216., , , , , , et al.