Fax: (011) 81-584-75-5715
Changes in the characteristics and survival rate of hepatocellular carcinoma from 1976 to 2000
Analysis of 1365 patients in a single institution in Japan
Article first published online: 4 MAY 2004
Copyright © 2004 American Cancer Society
Volume 100, Issue 11, pages 2415–2421, 1 June 2004
How to Cite
Toyoda, H., Kumada, T., Kiriyama, S., Sone, Y., Tanikawa, M., Hisanaga, Y., Hayashi, K., Honda, T., Kitabatake, S., Kuzuya, T., Nonogaki, K., Kasugai, T. and Shimizu, J. (2004), Changes in the characteristics and survival rate of hepatocellular carcinoma from 1976 to 2000. Cancer, 100: 2415–2421. doi: 10.1002/cncr.20289
- Issue published online: 18 MAY 2004
- Article first published online: 4 MAY 2004
- Manuscript Revised: 18 MAR 2004
- Manuscript Accepted: 18 MAR 2004
- Manuscript Received: 28 APR 2003
- hepatocellular carcinoma (HCC);
- tumor size;
- Child–Pugh class;
- tumor stage
The authors analyzed changes in the characteristics and survival rate of patients with hepatocellular carcinoma (HCC) in the past 25 years.
Trends in clinical characteristics and survival rate of patients with HCC were evaluated retrospectively based on data from 1365 patients who were diagnosed, treated, and followed between 1976 and 2000.
Between 1976–1995, the number of patients with smaller tumors, a less advanced tumor stage, and with a lower Child–Pugh class increased markedly. No differences were observed in the distributions of these three factors between the periods 1991–1995 and 1996–2000. The year of HCC diagnosis, tumor size, tumor stage, Child–Pugh class, and the kind of initial treatment received correlated significantly with patient survival rates by multivariate analysis. The year of HCC diagnosis was found to contribute independently to the improvement in patient survival rates. Using the Kaplan–Meier comparison, the time periods during which the highest patient survival rates occurred were found to be 1991–1995/1996–2000, 1986–1990, and 1976–1985, in that order. The authors did not observe a difference with regard to survival rates between patients in the 1991–1995 and 1996–2000 groups.
The characteristics of patients with HCC changed dramatically from 1976 to 1995 (but not in the past 10 years) toward the earlier detection of HCC. This contributed to the improvement noted in patient survival rates during this period. The year of HCC diagnosis was found to be an independent factor for the improved survival rates by multivariate analysis. This indicated that the progress of treatment and care for patients with HCC contributed to the annual improvement in patient survival rates. Cancer 2004. © 2004 American Cancer Society.
Hepatocellular carcinoma (HCC) is one of the most common malignancies, especially in southern and eastern Asia. In Japan, the incidence of HCC has been increasing for 30 years and has more than doubled in the last 10 years. HCC is the third leading cause of death from cancer in Japan. Tremendous effort has been invested in trials aimed at improving the survival rate of patients with HCC. Hepatectomy remains one of the primary treatments for HCC with high curability.1 Cadaveric or living-related liver transplantation is another treatment option.2, 3 Several nonsurgical treatments have been attempted, including transcatheter arterial chemoembolization (TACE), transcatheter arterial embolization (TAE),4, 5 percutaneous ethanol injection (PEIT),6, 7 percutaneous microwave coagulation therapy (PMCT),8 or most recently, radiofrequency ablation (RFA).9, 10 Systemic or hepatic arterial infusion chemotherapy11, 12 is also attempted for patients unable to undergo other surgical or nonsurgical treatment mainly because they have advanced stage disease.
Efforts also have been made to improve the early detection of HCC, including the development of various scanning techniques and imaging apparatuses13, 14 and identification of highly sensitive tumor markers.15, 16 However, to our knowledge, any improvement in patient survival rates during the past 20–30 years have not been confirmed to date and factors contributing to the possible improvement have not been verified.
Therefore, in the current retrospective study, we investigated changes in the survival rate of patients with HCC in Japan and analyzed factors contributing to these changes with data from 1365 patients treated and followed in a single institution between 1976–2000.
MATERIALS AND METHODS
Between 1976 and 2000, 1365 patients were diagnosed with initial HCC (not recurrence). Diagnosis was confirmed by histologic findings from the resected or ultrasonography-guided needle biopsy specimens. For patients who could not undergo biopsy because of the progression of HCC, the diagnosis of HCC was based on the radiologic findings of selective hepatic angiography and computed tomography (CT) scans. The study included 1036 men and 329 women with a mean age of 63.0 ± 9.3 years.
The survival rates of patients were analyzed on the basis of their clinical records. The maximum tumor diameter, Child–Pugh classification of remnant liver function,17 etiology of liver disease, tumor stage according to the American Joint Committee on Cancer, and initial treatment of HCC were also investigated. Patients were stratified by years of diagnosis into 4 groups: 1976–1985 (n = 247), 1986–1990 (n = 314), 1991–1995 (n = 366), and 1996–2000 (n = 408). The survival rate and other factors were compared among groups.
For the analysis of patient characteristics, data were expressed as the mean ± the standard deviation. Differences in proportions between the groups were analyzed by the chi-square test. Mean quantitative values were compared by the Student t test. The date of HCC diagnosis was defined as Time 0 for calculations of patient survival rates. Surviving patients and patients who died of causes other than liver disease were defined as censored cases. Patients who died of HCC-related causes or liver failure were defined as noncensored cases. The Kaplan–Meier method18 was used to calculate survival rates and the log-rank test19 was used to analyze differences.
The Cox proportional hazards model20 was used for multivariate analysis. The variables analyzed were the year of the diagnosis of HCC (1976–2000), tumor size (≤ 2 cm vs. > 2 cm and ≤ 5 cm, > 5 cm), Child–Pugh class (A vs. B, C), tumor stage (Stage I vs. II, III, and IV), and initial therapy (no treatment vs. surgery, PEIT/PMCT/RFA, TACE/TAE, repeated hepatic arterial infusion chemotherapy [RHAIC], and systemic chemotherapy [SC]. Data analyses were performed with the JMP statistical software package, Version 4.0 (Macintosh version; SAS Institute, Cary, NC). All P values were derived from two-tailed tests and P < 0.05 was accepted as statistically significant. The entire protocol was approved by the hospital ethics committee and was performed in compliance with the Helsinki Declaration.
Comparison of Patient Backgrounds among Groups
Table 1 shows the backgrounds of all patients and those according to the period. For factors such as age, gender, and treatment received for HCC, the distribution of the patients did not appear to vary significantly among periods, except that the use of PEIT/PMCT/RFA and RHAIC increased and the use of SC decreased from 1976 to 1995 and surgery increased from 1991 to 2000. With regard to the etiology of underlying liver disease, the number of patients with hepatitis B virus (HBV) infection gradually decreased between 1976 and 1995. In contrast, the number of patients with hepatitis C virus (HCV) infection gradually increased during the same period, considering the finding that most non-HBV patients before 1990 would have been patients with HCV, including some non-HBV and non-HCV patients.
|Characteristics||Total (%)||1976–1985 (%)||1986–1990 (%)||1991–1995 (%)||1996–2000 (%)|
|No. of patients||1365||259||321||377||408|
|Age (yrs)||63.0 ± 9.3||60.5 ± 9.3||62.5 ± 9.7||63.1 ± 9.0||65.2 ± 8.6|
|Surgery||199 (14.6)||30 (11.6)||41 (12.8)||45 (11.9)||83 (20.3)|
|PEIT/PMCT/RFA||207 (15.2)||0||27 (8.4)||109 (28.9)||71 (17.4)|
|TACE/TAE||489 (35.8)||59 (22.8)||146 (45.5)||141 (37.4)||143 (35.1)|
|RHAIC||55 (4.0)||0||9 (2.8)||27 (7.2)||19 (4.7)|
|SC||120 (8.8)||82 (31.6)||24 (7.5)||5 (1.3)||9 (2.2)|
|None||295 (21.6)||88 (34.0)||74 (23.0)||50 (13.3)||83 (20.3)|
|Etiology of underlying liver disease|
|HBV||290 (21.2)||78 (30.1)||76 (23.7)||67 (17.8)||69 (16.9)|
|HCV||662 (48.5)||13 (5.0)||92 (28.7)||268 (71.1)||289 (70.8)|
|HBV, HCV||10 (0.7)||0||2 (0.6)||2 (0.5)||6 (1.5)|
|Non-HBV||259 (19.0)||126 (48.7)||133 (41.4)||0||0|
|Non-HBV, non-HCV||87 (6.4)||0||5 (1.5)||38 (10.1)||44 (10.8)|
|Not tested||57 (4.2)||42 (16.2)||13 (4.1)||2 (0.5)||0|
|Patients without cirrhosis||134 (9.8)||6 (2.3)||22 (6.9)||46 (12.2)||60 (14.7)|
Figure 1 (left panel) shows the change in the distribution of tumor size (the size of the largest tumor if a patient had multiple tumors). There was a significant shift of the distribution toward smaller tumors. The rate of tumors measuring ≤ 2 cm increased significantly from 1976 to 1995 (4.6%, 15.6%, and 36.6%, respectively; P < 0.0001) and the rate of tumors measuring > 5 cm decreased significantly during the same period (75.7%, 52.0%, and 31.8%, respectively; P < 0.0001). We found no change in the rate of tumors measuring ≤ 2 cm (P = 0.3014) or > 5 cm (P = 0.7204) between 1991 and 2000. Figure 1 (middle panel) shows the change in the distribution of tumor stage according to AJCC cancer staging. Tumor stage was not evaluated in 28 patients. The rate of patients with Stage I/II HCC significantly increased between period 1991–1995 compared with 1986–1990 (46.5% vs. 29.7%; P < 0.0001). In contrast, no difference in the rate of patients with Stage I/II HCC was observed between the periods 1976–1985 and 1986–1990 and between the periods 1991–1995 and 1996–2000 (P = 0.8715 and P = 0.6418, respectively). Figure 1 (right panel) shows the change in distribution of liver status (Child–Pugh class) of patients at the time of diagnosis of HCC. Child–Pugh class was not evaluated in 37 patients. The rate of patients with Child–Pugh Class A increased from 1976 to 1995 (1976–1985 [27.9%] vs. 1986–1990 [37.3%], P = 0.0252; 1986–1990 vs. 1991–1995 [46.3%], P = 0.0334). The rate of patients without cirrhosis accordingly increased with the increase of patients with Child–Pugh Class A (Table 1). The rate of patients with Child–Pugh Class C decreased significantly between the periods 1976–1985 (27.1%) and 1986–1990 (17.5%, P = 0.0084), but not between the periods 1986–1990 and 1991–1995 (14.4%; P = 0.2822). As well as the distribution of tumor size, we found no significant change between the periods 1991–1995 and 1996–2000 in either the rate of Child–Pugh Class A patients (P = 0.9983) or that of Child-Pugh Class C patients (P = 0.9265).
Factors Affecting Patient Survival
There were 1059 noncensored patients and 306 censored patients by the end of 2002. The median survival period was 18.6 months. Of the 1059 noncensored patients, 712 patients (67.2%) died of the progression of HCC and 317 (29.9%) patients died of liver-related diseases (189 of liver failure and 128 died of complications of ruptured varices). In the other 30 patients (2.8%), it was not known whether the cause of death was HCC related.
Multivariate analysis was performed for the potential factors that were associated with patient survival (Table 2). All factors analyzed were found to be associated significantly with patient survival (P = 0.0002 for the year of HCC diagnosis and P < 0.0001 for tumor size, Child–Pugh classification, tumor stage, and initial treatment, respectively). With regard to tumor size, tumor stage, and Child–Pugh class, a significant increase in the relative risk (RR) was observed according to the increase in tumor size, tumor stage, or Child–Pugh class, respectively, except that no difference was observed for patient survival between patients with tumors ≤ 2 cm and patients with tumors > 2 cm and ≤ 5 cm. In analysis for the treatments, surgery, PEIT/PMCT/RFA, TACE/TAE and RHAZC were found to reduce the RR. Patients who received SC did not demonstrate a higher survival rate compared with patients who received no treatment. The year of HCC diagnosis was found to be an independent factor affecting patient survival. The RR decreased year by year and was noted to be independent of tumor size, tumor stage, or Child–Pugh class.
|Factors||Parameter estimate||Standard error||Mean||Risk ratio (95% CI)||P value|
|Tumor size (diameter [cm])|
|> 2 and ≤ 5||0.0346||0.0585||0.35||1.0352 (0.9230–1.1611)||0.5527|
|> 5||0.3501||0.0603||36.56||1.4193 (1.2610–1.5974)||< 0.0001|
|III||0.4606||0.0691||45.44||1.5851 (1.3844–1.8148)||< 0.0001|
|IV||0.6572||0.1062||33.33||1.9294 (1.5603–2.3675)||< 0.0001|
|Class B||0.2229||0.0392||32.79||1.2497 (1.1573–1.3494)||< 0.0001|
|Class C||0.5652||0.0517||111.41||1.7599 (1.5903–1.9475)||< 0.0001|
|Surgery||−0.6163||0.0711||78.99||0.5399 (0.4697–0.6207)||< 0.0001|
|PEIT/PMCT/RFA||−0.6020||0.0653||88.32||0.5477 (0.4819–0.6225)||< 0.0001|
|TAE/TACE||−0.4540||0.0475||86.00||0.6351 (0.5786–0.6971)||< 0.0001|
Comparison of Survival Rate by Years of Diagnosis
The survival rate was compared according to the period of the year of HCC diagnosis (Fig. 2). Patients diagnosed during 1991–1995 as well as those diagnosed during 1996–2000 showed the highest survival rate, followed by patients diagnosed during 1986–1990 and during 1976–1985, in that order (P < 0.0001, respectively). In contrast, we found no difference in survival rates between patients diagnosed during the period 1991–1995 and patients diagnosed during the period 1996–2000 (P = 0.9715).
In a previous study of patients diagnosed with HCC between 1977–1996 in the U.S., the survival rate was lower than the rate we observed for the same period.21 This may be partly because all patients in the current study were followed at our institution and received at least palliative treatment for liver failure. In addition, the majority of patients in the current study (1070 of 1365 patients [78.4%]) received treatment for HCC. There also are many differences between patients in the U.S. and Japan that could affect survival rates, including varied medical and insurance systems, social circumstances, ethnicity, and lifestyle.
A potential increase in the incidence of HCC in the U.S. was suggested recently.22 A recent report23 also predicted that in a few decades, the incidence of HCV-related HCC in the U.S. will increase to equal the current incidence in Japan. The prediction was derived from a comparison by the molecular clock analysis of HCV in the U.S. and in Japan.23
The current study is based on data from patients treated and followed in a single institution. Therefore, the data could be influenced by the institution's treatment modality. In addition, liver transplantation was not available as a treatment option, which could cause the difference in survival rates noted between our patients and patients from Western countries. However, in the current study, follow-up on nearly all patients lasted throughout their lifetime, and detailed information regarding each patient's background, tumor status, and liver function was available to us.
We found a significant change in the characteristics of patients with HCC in the periods 1976–1985, 1986–1990, and 1991–1995/1996–2000. In contrast, we observed no difference in the tumor size at detection, in tumor stage at detection, or in liver status at diagnosis (Child–Pugh class) between patients diagnosed during 1991–1995 and those diagnosed during 1996–2000. The development of various scanning techniques enabling detection of small tumors.13, 14 The identification of sensitive and specific tumor markers such as Lens culinaris agglutinin-reactive α-fetoprotein (AFP), or high-sensitivity des-γ-carboxy prothrombin (DCP) also has contributed to the detection of small tumors.15, 16 In addition, the program for the screening of HCC in the study hospital also has changed greatly during the period 1976–1985. Liver scintigraphy, ultrasonography (US), and CT scans were performed only for patients with symptoms such as abdominal pain in the right upper quadrant, hepatomegaly, splenomegaly, or ascites, and these patients were followed every 6 months with US or CT. All patients with cirrhosis were followed every 3–6 months with US or CT during the period 1986–1990. During the periods 1991–1995 and 1996–2000, all patients with cirrhosis and all patients without cirrhosis but with high fibrosis were followed with US, CT scans, and dynamic magnetic resonance imaging scans every 3 months. In addition, regular monitoring of tumor markers (AFP and DCP) was performed during these last two periods. The changes in the screening system also contributed strongly to the early detection of HCC.
In the multivariate analysis of potential factors affecting patient survival, factors that are associated with the early detection of HCC, such as tumor size, tumor stage, and Child–Pugh classification, were found to affect patient survival. The kind of treatments received also were found to independently affect patient survival; surgery, PEIT/PMCT/RFA, TACE/TAE, or RHAIC contributed to the improvement in patient survival rates. RHAIC with a subcutaneously implanted injection port was initiated at the study institution (Ogaki Municipal Hospital, Ogaki, Japan) in 1990.11 In Japan, it currently replaces systemic chemotherapy as a treatment option for patients who are unable to undergo surgery, PEIT/PMCT/RFA, or TACE/TAE.11, 12, 24–26
Using the Kaplan–Meier comparison, patient survival rates significantly improved according to the period of the year of HCC diagnosis among the first three periods. In contrast, we found no difference in survival rates between patients diagnosed between 1991–1995 and those diagnosed between 1996–2000. The lack of a trend toward the early detection of HCC during the periods 1991–1995 and 1996–2000 corresponded to the plateau of the improvement in patient survival noted during the same periods. Therefore, the improvement in the overall survival rate observed between the periods 1976–1985 and 1991–1995/1996–2000 is, in some part, based on the early detection of HCC. From this aspect, the improvement was made by lead-time bias and may not be a real improvement in patient survival rates.
However, the year in which HCC was diagnosed was found to be one of the independent factors associated with patient survival rates. Patients in whom HCC was detected, diagnosed, and treated more recently demonstrated a better survival rate regardless of the size or stage of the tumor, Child–Pugh class, or the kind of treatment received. This meant that the annual improvement in patient survival rates was independent of other factors and indicated that the progress of treatment and care for patients with HCC over time contributed to the improvement in patient survival rates.
In conclusion, the characteristics of patients with HCC changed greatly between 1976 to 1995, but not in the past 10 years, toward the earlier detection of HCC. In addition, the survival rate of patients with HCC has improved in Japan since 1976, except for during the past 10 years. This improvement in patient survival rates is partly due to the increase in the early detection of HCC. However, patient survival rates have improved year by year, independent of tumor size, tumor stage, or Child–Pugh class by multivariate analysis. This improvement may be attributable to the annual progression of treatment and care for patients with HCC. Further studies will be required to verify improvement in the survival rates of patients with HCC over time. In addition, the impact of liver transplantation on patient survival, which could not be evaluated in the current study because of the lack of patients, should be evaluated in further studies. For further improvement in the survival rates of patients with HCC, another screening program should be established to increase early detection among patients with HCC. In addition, a strategy for preventing HCC should be developed in the future.
- 1The Liver Cancer Study Group of Japan. Primary liver cancer in Japan. Clinicopathologic features and results of surgical treatment. Ann Surg. 1990; 211: 277–287.
- 20Regression models and life tables. J R Stat Soc. 1972; 34: 187–220..