Fax: (011) 886 228739318
The model for end-stage liver disease-based Japan Integrated Scoring system may have a better predictive ability for patients with hepatocellular carcinoma undergoing locoregional therapy
Article first published online: 17 MAY 2006
Copyright © 2006 American Cancer Society
Volume 107, Issue 1, pages 141–148, 1 July 2006
How to Cite
Huo, T.-I., Lin, H.-C., Huang, Y.-H., Wu, J.-C., Chiang, J.-H., Lee, P.-C. and Lee, S.-D. (2006), The model for end-stage liver disease-based Japan Integrated Scoring system may have a better predictive ability for patients with hepatocellular carcinoma undergoing locoregional therapy. Cancer, 107: 141–148. doi: 10.1002/cncr.21972
- Issue published online: 16 JUN 2006
- Article first published online: 17 MAY 2006
- Manuscript Accepted: 27 MAR 2006
- Manuscript Revised: 18 JAN 2006
- Manuscript Received: 28 NOV 2005
- National Science Council, Executive Yuan, Taiwan. Grant Number: NSC 94-2314-B-010-049
- Cancer of Liver Italian Program;
- hepatocellular carcinoma;
- Japan Integrated Scoring;
- Model for End-Stage Liver Disease
The Japan Integrated Scoring (JIS) system was revealed as a better model for outcome prediction compared with the Cancer of Liver Italian Program system for hepatocellular carcinoma (HCC), and the Model for End-Stage Liver Disease (MELD) was better as a prognostic predictor for patients with cirrhosis compared with the Child–Turcotte–Pugh (CTP) system, which is a parameter used in the JIS system. The objective of the current study was to investigate the performance of the modified MELD-based JIS system.
In the modified JIS system, the CTP class in the original JIS was replaced with MELD cut-off scores of <10, 10 to 14, and >14. The modified JIS system was compared with the original system in 276 patients with HCC who underwent locoregional therapy (transarterial chemoembolization or percutaneous injection).
The mean ± standard error original JIS score was 1.8 ± 1.0 (range, 0-4), compared with 2.0 ± 1.1 (range, 0-5) for the modified JIS system (P<.001). Using mortality as the endpoint, the area under receiver operating characteristic curve (AUC) for the modified JIS system was 0.804 compared with 0.741 for the original JIS system (P = .008) at 12 months, and the AUC was 0.853 and 0.765, respectively (P<.001), at 24 months. Survival analysis showed that the modified JIS system had a better discriminatory ability for patients in different score groups and was more accurate for outcome prediction in the Cox multivariate model.
The current results indicated that the MELD-based, modified JIS system has improved predictive ability compared with the original system and is a more feasible model for clinical staging in patients with HCC who are undergoing locoregional therapy. Cancer 2006. © 2006 American Cancer Society.
Hepatocellular carcinoma (HCC) is among of the common malignant tumors in the world with a global increasing annual incidence, as reported recently.1, 2 To our knowledge to date, at least 7 clinical staging systems have been proposed from different countries to serve as a predictive model for patients with HCC.3 However, these staging systems may have variable predictive ability according to different published series, and the choice of the optimal staging system for HCC has remained a highly debatable issue in recent years.4, 5 Among these available staging systems, the Cancer of the Liver Italian Program (CLIP) system has been suggested as the primary staging system for HCC, because it is simple to implement in the evaluation of new patients and has been validated prospectively.6, 7 Conversely, Japanese investigators have proposed a new staging system, the Japan Integrated Scoring (JIS) system, which directly combines the tumor, lymph node, metastasis (TNM) system and Child–Turcotte–Pugh (CTP) system for scoring. The JIS system was identified as a better model than the CLIP system in a study of 4525 patients with HCC.8 Another independent study group consistently showed that the JIS system was a preferred model for outcome prediction, especially in areas in which the early detection and treatment of HCC are possible.9
The majority of patients with HCC have coexisting cirrhosis at the time of diagnosis, and the severity of cirrhosis most likely is the most important nontumor factor that affects survival. The Model for End-Stage Liver Disease (MELD), which is based on a calculation of 3 biochemical variables (serum bilirubin, prothrombin time, and creatinine), showed more accuracy for predicting survival than the CTP score for cirrhotic patients who were awaiting liver transplantation in a multicenter study in the U.S.10 The liver-allocation system subsequently has changed from a status-based algorithm to an algorithm that uses a continuous MELD severity scale to prioritize adult patients on the wait list.11, 12 The accuracy of MELD for outcome prediction in patients with decompensated cirrhosis also has been validated in other centers and in Europe.12–14
The MELD has been a major reference system for cirrhotic patients with or without HCC. Based on this information, it is not clear whether the CTP class, which is the sole marker in the JIS system to indicate the severity of cirrhosis, should be replaced with the MELD to be more accurate in terms of outcome prediction for patients with HCC. To evaluate the performance of the MELD score in the JIS model, we have proposed a modified MELD-based JIS system and compared its predictive ability with the original model in patients with HCC who underwent locoregional therapy.
MATERIALS AND METHODS
From November 1997 to October 2002, 276 patients with HCC who received locoregional therapy in our hospital were evaluated prospectively, and their medical profiles were reviewed retrospectively. The methods of locoregional therapy included transarterial chemoembolization (TACE) (n = 168 patients), percutaneous ethanol injection (PEI) (n = 39 patients), and percutaneous acetic acid injection (PAI) (n = 69 patients). The objective of these therapies was to achieve a palliative or curative, tumor-ablative effect. All patients had a definite JIS score before treatment and known 2-year and long-term survival status to undergo the c-statistic and survival analysis. The baseline demographics of the study patients are shown in Table 1. For this study, we complied with the standards of the Declaration of Helsinki and current ethical guidelines.
|Characteristic||No. of Patients (%)|
|No. of patients||276|
|Age, y (mean ± SD)||67 ± 10|
|Male/female ratio (%)||80/20|
|Etiology of cirrhosis (%)|
|HBV and HCV||8 (2.9)|
|JIS score (%)|
|Serum biochemistry (mean ± SD)||1.5 ± 0.7|
|Bilirubin (mg/dL)||1.1 ± 0.1|
|INR of PT||1.2 ± 0.3|
|Mean ± SD||10.2 ± 3.1|
|CTP class (%)|
|Treatment modality (%)|
The diagnosis of HCC was verified histologically by needle biopsy or was based on the findings of typical radiologic features in at least 2 image examinations (including ultrasound, contrast-enhanced dynamic computed tomography, magnetic resonance imaging, and hepatic angiography) or by a single positive imaging technique associated with serum α-fetoprotein (AFP) level >400 ng/mL.15 The underlying etiology of liver disease was attributed either to hepatitis B virus (HBV) infection if patients were seropositive for hepatitis B surface antigen (radioimmunoassay kits; Abbott Laboratories, North Chicago, IL) or to hepatitis C virus (HCV) infection if patients were seropositive for antibody against HCV according to a second-generation enzyme immunoassay (Abbott Laboratories) on at least 2 occasions.
MELD, JIS, and the Modified JIS System
The MELD equation was used to calculate the severity score: 9.57 × loge (creatinine mg/dL) + 3.78 × loge (bilirubin mg/dL) + 11.2 × loge (INR) + 6.43.10 Minimal values were set to 1.0 for the purposes of calculation. The maximal serum creatinine level considered within the MELD score equation is 4.0 mg/dL. The severity of cirrhosis was assessed by using the CTP scoring system, which is part of the predictive parameters in the JIS system. We proposed a modified JIS system based on the MELD to reflect the extent of hepatic dysfunction. A comparison of the original and modified JIS system is shown in Table 2. In the modified JIS system, we used the MELD score instead of CTP class for scoring, with 0 points for scores <10, 1 point for scores of 10 to 14, and 2 points for scores >14. These cutoff scores were selected, because MELD scores >10 have been associated with poor survival in patients with HCC,16 and scores >14 are considered equivalent to CTP Class C.17
|Scoring System||Original Score||Modified Score|
|T1||Fulfilling 3 factors|
|T2||Fulfilling 2 factors|
|T3||Fulfilling 1 factor|
|T4||Fulfilling 0 factors|
|Stage IVA||T4N0M0, or any TN1M0|
|Stage IVB||Any TN0-N1M1|
Treatment and Follow-Up
The treatment procedures have been published previously in detail.18, 19 TACE was performed according to the Seldinger technique of arterial embolization. The equipment for selective common or proper hepatic artery angiography was LCN (GE Medical Systems) or Diagnost 3 (Philips, the Netherlands). Hepatic arteriography and superior mesenteric arterial portovenography were performed to define the size and location of tumor nodules. During sequential scanning of the liver, 100 mL to 150 mL of radiocontrast medium (Ultravist, Germany) were injected by using a power injector (CT9000 ADV; Liebel-Flarsheim, U.K.) to evaluate the vascularity of the tumor. The arteries supplying the tumor were catheterized superselectively, and this was followed by the infusion of a mixture of 20 mg to 30 mg doxorubicin (Carlo Erba, Milan, Italy) and 5 mL to 10 mL lipiodol (Laboratoire Guerbet, Aulnay-Sous-Bois, France). The objective was to deliver a sufficient amount of emulsion to the tumor areas without retrograde flow in each treatment cycle. Under fluoroscopic control, the feeding arteries subsequently were embolized with 2-mm to 3-mm strips of Gelfoam (Upjohn, Kalamazoo, MI) until complete flow stagnation was achieved.
The equipment used for PEI or PAI was a commercially available ultrasound scanner with a puncture probe that had a guide device (Aloka, Tokyo, Japan). Sterile pure ethanol or 50% acetic acid (Merck, Germany) was injected with a 22-gauge spinal needle. Treatment was administered twice each week in an inpatient setting. One or 2 injections at a dose of 2 mL to 8 mL for ethanol and 2 mL to 4 mL for acetic acid were given during each treatment session. After the injection was complete, the needle was left in place for 1 or 2 minutes to prevent reflux of acetic acid into the peritoneal cavity. Two to 7 sessions were given in each treatment cycle, depending on the tumor size and the choice of ethanol or acetic acid. For patients who had multiple tumors, percutaneous injection therapy was performed first on the main tumor (the largest tumor), followed by treatment of the smaller tumors, until all tumors were treated successfully.
Posttreatment follow-up included ultrasound scans and measurements of serum AFP levels every 2 months and contrast-enhanced computed tomography scans every 3 to 4 months. Magnetic resonance imaging and/or hepatic angiography were done as supplemental examinations. Residual viable tumor or new tumor that occurred elsewhere in the liver was detected on dynamic computed tomography scans with contrast enhancement during the arterial phase or on hepatic angiography that showed tumor staining, and an additional treatment cycle was administered once tumor recurrence or viable tumor was found.
Chi-square tests or Fisher exact tests (2-tailed) were used for categorical data, and the Mann–Whitney rank sum test was used for continuous data. Spearman correlation analysis was used to estimate the correlation between baseline MELD and CTP scores of the study patients. To assess the ability of the original and modified JIS system in predicting the risk of mortality, our analysis was performed by measuring the concordance (c-statistic) equivalent to the area under the receiver operating characteristic curve (AUC).20 Comparisons of the AUCs were calculated by using the method of Hanley and McNeil.21 Outcomes were assessed as 6-month, 12-month, 18-month, and 24-month mortality.
Survival distributions were estimated by using the Kaplan–Meier method and were compared by using the log-rank test. The Cox proportional-hazards model was used to determine the adjusted risk of mortality per 1 unit increase of the original JIS and modified JIS system. We also compared the overall predictive power of survival for each scoring system to determine which system provided the most accurate prediction of survival (monotonicity of the score). The Cox model was used to calculate the likelihood ratio (LR) chi-square statistic to determine homogeneity (small differences in survival among patients in the same stage within each system).22). Then, the linear trend chi-square test was used to measure the discriminatory ability of each staging system (greater differences in survival among patients in different stages within each system).23 In addition, the results of the Cox regression analysis were expressed by using the Akaike information criterion, which shows how the staging systems affect the dependent variable (patient survival): the lower the Akaike information criterion, the more explanatory and informative the model.24 All statistical analyses were conducted by using SPSS for Windows (version 12; SPSS Inc., Chicago, IL), SAS (version 8.2; SAS Inc., Cary, NC), and MedCalc for Windows (version 4.2; MedCalc Software, Mariakerke, Belgium). For all tests, P values <.05 were considered statistically significant.
Patient Demographics and Change in the JIS Score after Modification
Table 1 shows that patients were predominantly male (80%), 62% of patients had chronic HBV infection, and 3.2% of patients were in CTP Class C, because they were considered tolerant for active treatment. The mean (±standard deviation) original JIS score for all patients was 1.8 ± 1.0 (range, 0-4). According to the proposed method, the mean score for modified JIS system was 2.0 ± 1.1 (range, 0-5; P<.001). The score distribution was categorized further to examine the significance of trend. There was a significant shift toward an increase in the JIS score from the original system to the modified system (P = .047) (Fig. 1).
Correlation of the MELD and CTP Scores in Patients with HCC
The mean baseline MELD and CTP scores were 10.1 ± 3.1 and 6.2 ± 1.4, respectively, in the study patients. There was a significant correlation between the MELD and the CTP score (P = .815., P<.001) (Fig. 2) in the Spearman correlation analysis.
Comparison of the AUC at Different Times between the Original and Modified JIS System
Using the c-statistic and mortality as the endpoint, the estimated AUCs for the original and modified JIS systems were compared, and the results are shown in Table 3 and in Figure 3. At 6 months, there were no significant differences of the predictive ability between the 2 models. At 12 months, the modified JIS system had a significantly higher AUC compared with the original JIS system (0.804 vs. 0.741, respectively; P = .008). The modified JIS system consistently had a higher AUC at 18 months (0.849) and at 24 months (0.853) compared with the original JIS system (0.774 and 0.765, respectively; P<.001 for both), suggesting that the modified JIS system had better predictive ability for intermediate-term outcomes.
|Period||Original JIS||Modified JIS|
Staging Systems and Survival
The survival analysis was performed for both the original and modified JIS systems and was compared between patients with JIS scores of 0, 1, 2, 3, and >4. Patients were followed for an average of 28 months (median, 26 months; range, 1-71 months). The survival data were censored at their last visit or at the time of death. The original JIS system had fairly good ability to differentiate survival between various score categories except for comparisons between groups with scores of 0 versus 1 and between groups with scores of 3 versus 4 (Fig. 4A). For the modified JIS system, there was a significant survival difference between various score categories except between groups with scores of 0 versus 1 (Fig. 4B).
The long-term mortality risks for the original and modified JIS systems were calculated and adjusted for age, gender, etiology of cirrhosis, and treatment modality by using the Cox proportional-hazards model (Table 4). The estimated risk ratio of mortality per unit increase of the original JIS system was 2.11, compared with 2.60 for the modified JIS system (P<.001 for both). Table 5 shows that the modified JIS system had greater homogeneity (LR chi-square statistic, 125.01), indicating small differences in survival among patients in the same stages. The modified JIS also had a higher discriminatory scores (chi-square test for linear trend, 66.16) compared with the original system. The Akaike information criterion consistently was lower for the modified JIS system, indicating that this model was more informative regarding the survival of patients with HCC.
|Variables||Regression coefficient||SE||RR||95% CI||P|
|Scoring System||Discriminatory Ability Linear Trend Chi-Square Test||Homogeneity LR Chi-Square Test||Akaike Information Criterion|
The MELD scoring system, which has shown the ability to predict mortality across a broad spectrum of liver diseases,25, 26 has become the prevailing criteria for organ allocation in liver transplantation.10, 11 It is useful for characterizing which patients have the greatest need and will achieve the best survival benefit for a liver transplantation.27, 28 More important, the MELD has demonstrated equal or better ability compared with the CTP system for short-term or intermediate-term outcome prediction.10, 29 In addition, the MELD system has been useful as a model for predicting the outcomes of cirrhotic patients who are undergoing major surgical procedures.17, 30, 31
In the current study, we evaluated and compared the prognostic ability of the original and modified JIS systems by comparing the AUC and survival distributions among different score categories to individually assess short-term, intermediate-term, and long-term outcomes. It noteworthy that the modified JIS system consistently had a higher AUCs at or before 2 years compared with the original JIS system, suggesting that the incorporation of the MELD score into the cancer staging model may enhance its predictive ability. Because there was a close association between baseline CTP and MELD scores in patients with HCC, there still was a significant increase or shift in the JIS score from the original to the modified JIS system that helped identify a subgroup of patients who had suboptimal outcomes. For example, patients with original JIS scores of 1 would be grouped to score 2 in the modified JIS model provided they had a higher MELD score. These results suggest that the MELD is a better prognostic indicator of cirrhosis compared with the CTP class in patients with HCC.
In the survival analysis, the original JIS score showed fairly good ability to discriminate prognoses among different score categories except between groups with scores of 0 versus 1 and with scores of 3 versus 4. This defect was overcome in part with the modified JIS system, which showed improved differentiation for late cancer stages (scores of 3 vs. 4 or 5) (Fig. 4). A possible explanation for a better performance in the modified JIS system is that, because the CTP class is the sole marker to indicate the severity of cirrhosis, replacement with the MELD score may mean that the whole model will predict outcomes more accurately. In the Cox model, the modified JIS system was better for discriminating the survival of patients in different stages and had greater homogeneity of survival among patients within the same stage. However, the JIS system (either original or after modification) was not able to differentiate the long-term survival between patients with JIS scores of 0 and 1. There are a few possible explanations. Only a small proportion of patients (9.8%) in our study had a JIS score of 0, and the survival difference may not be identified readily. Second, the initial and follow-up treatment modalities may vary and were used interchangeably, even though these patients were diagnosed at a relatively early cancer stage. Third, because the JIS system uses a cut-off size of 2 cm to categorize tumors, this may be over-strict in clinical practice. For example, the treatment efficacy may not be distinguishable clearly for patients with small differences in tumor size, such as 1.8 cm and 2.2 cm, and the measurement of tumor size reportedly has been associated frequently with profound deviation using the contemporary imaging modalities.32 Alternative cut-off sizes, such as 3 cm or 5 cm, which are used in other staging systems, may be more feasible in the clinical setting.33, 34
The choice of the optimal staging system for HCC continues to remain highly debatable,4, 5, 35–37 and it is likely that different staging systems should be employed for patients undergoing different treatment strategy. The TNM system may be a preferred model for patients undergoing surgical resection, and the Milan criteria constitute an accepted system for liver transplantation.38 For the current study, we evaluated the performance of the original and modified JIS systems in patients with HCC who were receiving locoregional therapy: The JIS system has been used widely and has become the mainstay of treatment for patients with unresectable lesions in areas where donor livers are scare.15, 18, 39 Our results showed that there is still much room for the JIS system to improve its prognostic ability.
A potential limitation of this study is that the majority of our patients had chronic HBV infection, which is common in many Asian countries, and our results may not be applicable readily to patients in the U.S. or other countries. Because disease progression and tumor behavior may not be the same between HBV-related and HCV-related HCC, more studies are needed to address the potential impact of viral factors on patient outcomes.
A major shortcoming of the conventional CTP system is that it contains subjective variables that may make the interpretation less consistent and that vary from center to center. However, whereas MELD is gaining popularity and is more objective compared with CTP class at the population level, it can be anticipated that the CTP system, which does not require logarithmic transformation, may be more practical as an intuitive score for the initial bedside assessment of cirrhotic patients with or without HCC.40
In conclusion, the current results indicate the JIS system has a fairly good ability to predict the outcome of patients with HCC who are undergoing locoregional therapy. The modified MELD-based JIS system appears to have better accuracy for predicting short-term and long-term outcomes in these patients. Based on these data, we propose the modified MELD-based JIS system as a feasible tool for staging cancer in patients with HCC. Further studies will be required to determine whether the modified JIS system is an appropriate model for unselected patient populations with early-stage to advanced-stage cancer who are undergoing different treatment modalities. In addition, it remains to be clarified whether other, currently used, CTP-based staging systems also need to be modified by using the MELD to increase their prognostic ability.6, 9
We thank Ms. Wen-Yung Sheng (Institute of Neurology, Taipei Veterans General Hospital) for her help with the statistical analyses.
- 6The Cancer of the Liver Italian Program (CLIP) Investigators. Prospective validation of the CLIP score: a new prognostic system for patients with cirrhosis and hepatocellular carcinoma. Hepatology. 2000; 31: 840–845.
- 9Comparison of the usefulness of three staging systems for hepatocellular carcinoma (CLIP, BCLC, and JIS) in Japan. Am J Gastroenterol. 2005; 100: 1764–1771., , , et al.Direct Link:
- 16Trans-catheter arterial chemoembolisation for hepatocellular carcinoma in patients with viral cirrhosis: role of combined staging systems, Cancer Liver Italian Program (CLIP) and Model for End-Stage Liver Disease (MELD), in predicting outcome after treatment. Aliment Pharmacol Ther. 2003; 17: 1563–1569., , , et al.
- 19Comparison of transarterial chemoembolization and percutaneous acetic acid injection as the primary loco-regional therapy for unresectable hepatocellular carcinoma: a prospective survey. Aliment Pharmacol Ther. 2004; 19: 1391–1398., , , et al.
- 34GreeneFL, PageDL, FlemingID, et al., editors. The AJCC Cancer Staging Manual. 6th ed. New York: Springer-Verlag; 2002.
- 36Comparison of the usefulness of three staging systems for hepatocellular carcinoma (CLIP, BCLC, and JIS) in Japan. Am J Gastroenterol. 2005; 100: 1764–1771., , , et al.Direct Link:
- 37Construction of the Chinese University Prognostic Index for hepatocellular carcinoma and comparison with the TNM staging system, the Okuda staging system, and the Cancer of the Liver Italian Program staging system: a study based on 926 patients. Cancer. 2002; 94: 1760–1769., , , et al.