Mortality and morbidity of hepatectomy, radiofrequency ablation, and embolization for hepatocellular carcinoma: a national survey of 54,145 patients
- First Online:
- Cite this article as:
- Sato, M., Tateishi, R., Yasunaga, H. et al. J Gastroenterol (2012) 47: 1125. doi:10.1007/s00535-012-0569-0
- 575 Views
Reported mortalities and morbidities of therapeutic procedures for liver tumors vary between studies, because of different designs and small sample sizes. We investigated the mortalities and complication rates for hepatectomy, radiofrequency ablation (RFA), and trans-catheter arterial embolization (TAE) for hepatocellular carcinoma (HCC) in a large sample, using a nationwide Japanese database (the Diagnosis Procedure Combination database).
Data from the Diagnosis Procedure Combination database were analyzed for July 1 to December 31, 2007 and the same period in 2008. We identified 54,145 patients with HCC who underwent hepatectomy (n = 5,270), RFA (n = 11,688), or TAE (n = 37,187). In-hospital mortality and morbidity were analyzed for each procedure. The relationships between mortality and factors including patient characteristics and procedural backgrounds were assessed.
In-hospital mortalities associated with hepatectomy, RFA, and TAE were 2.6 % [95 % confidence interval (CI) 2.2–3.1], 0.3 % (0.2–0.4), and 1.0 % (0.9–1.1), and post-procedural complication rates were 14.5 % (13.5–15.5), 4.5 % (4.2–4.9), and 4.5 % (4.3–4.7), respectively. Increased mortality following hepatectomy was significantly associated with older age, extended lobectomy (vs. partial hepatectomy; odds ratio [OR] 3.80, p < 0.001), lower hospital volume (OR 2.74, p < 0.001), and renal comorbidity (OR 3.01, p = 0.02). Older age and cardiac comorbidity (OR 5.14, p = 0.001) were significantly associated with RFA-related mortality, and lower hospital volume was significantly associated with TAE-related mortality (OR 1.60, p < 0.001).
Mortalities and morbidities associated with therapeutic procedures for liver tumors were acceptably low in Japan, but were affected by patient and institutional characteristics.
KeywordsLiver tumorHospital volumeNationwide database
Trans-catheter arterial embolization
Diagnosis Procedure Combination
International Classification of Diseases and Related Health Problems, Tenth Revision
The liver is one of the commonest sites of primary and metastatic tumors [1, 2]. Hepatectomy has been considered as a treatment of choice in patients with liver tumors, and can offer survival for 5 years in 50–70 % of patients with early hepatocellular carcinoma (HCC) [3–7], and in 50–60 % with liver metastases of colorectal carcinoma [8–10]. However, image-guided minimally invasive techniques have been widely used for the treatment of HCC in the past decade, and radiofrequency ablation (RFA) has yielded promising clinical results, with survival rates comparable with those of hepatectomy. During the past decade, there has been growing interest in the use of RFA in patients with liver tumors considered to be unresectable because of impaired hepatic function [11–14]. Despite the fact that surveillance programs have reduced the proportion of HCC detected at an advanced stage in certain populations [15, 16], the majority of patients with HCC are still not eligible for curative treatments, such as hepatectomy and RFA, because of advanced features, and trans-catheter arterial embolization (TAE) has been widely used as a palliative treatment in such patients. A recent meta-analysis showed improved overall survival in patients with well-preserved liver function who were treated with TAE .
Before the 1980s, hepatectomy-related mortality was reported to be as high as 10 %. In recent years, however, this has decreased to less than 5 % at some surgical centers, and several recent studies have reported large series of successful hepatectomies with no mortality [18, 19]. On the other hand, although the safety of image-guided therapies has been generally accepted, various complications have been reported. The reported complication rates vary substantially, primarily because of the small sample sizes used in most studies. Published studies may also be liable to publication bias, i.e., authors may be less enthusiastic about reporting studies with higher complication rates.
The Diagnosis Procedure Combination (DPC) database is a discharge abstract and administrative claims database of inpatient admissions to secondary and tertiary care hospitals in Japan [20–22], representing approximately 40 % of inpatient admissions to such hospitals. This database represents a large number of samples, and can thus be used to investigate the mortalities and morbidities associated with different treatment modalities, on an objective basis. The aim of this study was to investigate the mortalities and complication rates associated with hepatectomy, RFA, and TAE for HCC in a large patient sample, using the DPC database.
Subjects, materials, and methods
The DPC database includes the following information: location of hospital; patient demographics; diagnosis, comorbidities at admission and complications after admission recorded with text in Japanese and the International Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes; therapeutic procedures coded by the Japanese original K-code; length of stay; and discharge status, including in-hospital death. The survey of the DPC hospitals is conducted by the DPC Research Group between July 1 and December 31 each year, and is funded by the Ministry of Health, Labour and Welfare, Japan. All 82 university teaching hospitals are obliged to adopt the DPC system, but adoption by community hospitals is voluntary. The survey started in 2003 with 82 teaching hospitals; 926 hospitals participated in 2007, and 855 in 2008. Data for 2.99 and 2.86 million patients were included in 2007 and 2008, respectively. The number in 2008 represented approximately 40 % of all the inpatient admissions to secondary and tertiary care hospitals in Japan.
The requirement for informed consent was waived in this study, because of the anonymous nature of the data. Study approval was obtained from the institutional review board of the University of Occupational and Environmental Health, Fukuoka.
The data used in this survey were derived from the DPC database for July 1 to December 31, 2007 and the same period in 2008. Patients with a diagnosis of HCC (ICD-10 code C220) were identified. We then selected patients who underwent hepatectomy (DPC procedure code K695), RFA (K697-3), or TAE (K615). Hepatectomy (K695) patients were divided into five sub-categories (K695-1, partial hepatectomy; K695-2, hepatic segmentectomy; K695-3, hepatic lobectomy; K695-4, extended hepatic lobectomy; K695-5, extended hepatic lobectomy with revascularization procedure). Finally, patients who underwent two or more types of the above procedures, and those who underwent TAE for controlling tumor bleeding during emergency hospitalization were excluded.
The primary endpoint was the in-hospital mortality after each procedure. The secondary endpoint was the occurrence of post-procedural complications during hospitalization, including hemorrhage, pneumothorax, liver abscess, gastrointestinal perforation, peritonitis, and hepatic infarction.
Patient characteristics were analyzed in terms of sex, age, and comorbidities. The characteristics of patients who underwent hepatectomy were analyzed in terms of each operative method. Hospital volume for each therapeutic procedure during the survey period was determined using the unique identifier for each hospital, and categorized into three (low-, intermediate-, and high-volume) groups, such that the numbers of patients in each group were almost equal. We assessed the relationships between mortality and various factors, including patient characteristics and hospital volume. The univariate association between each factor and in-hospital mortality was evaluated using the χ2 test or analysis of variance, as appropriate. Stepwise logistic regression analysis was used to model the concurrent effects of procedures and other factors on in-hospital mortality. Statistical analyses were performed using PASW version 18.0 (SPSS, Chicago, IL, USA). The threshold of reported p values for significance was accepted as <0.05.
HCC patient characteristics and details of procedures
Total (n = 54,145)
Hepatectomy (n = 5,270)
RFA (n = 11,688)
TAE (n = 37,187)
Sex, n (%)
Age (years), mean ± SD
70.7 ± 8.8
67.7 ± 9.5
70.7 ± 8.6
71.2 ± 8.6
Age (years), n (%)
Comorbidities, n (%)
Chronic renal diseases
The numbers of in-hospital deaths among patients who underwent hepatectomy, RFA, and TAE were 137 (2.60 %), 29 (0.25 %), and 383 (1.03 %), respectively.
In-hospital mortality of hepatectomy, RFA, and TAE
% (95 % CI)
% (95 % CI)
% (95 % CI)
Chronic renal diseases
The in-hospital mortality associated with RFA was significantly higher in older patients (p = 0.001) and those with cardiac diseases (p < 0.001) (Table 2). Both of these features were identified as significant factors associated with increased mortality (Fig. 1b).
The in-hospital mortality associated with TAE was 1.03 % (Table 2). Univariate comparison showed that in-hospital mortality was significantly higher in hospitals with lower procedure volumes (p < 0.001) and in younger patients (p = 0.04). Multivariate logistic regression analysis identified higher procedure volume as a significant factor associated with lower mortality (OR 0.62, p < 0.001) (Fig. 1c).
The mean intervals between the date of procedure and death in fatal cases were 43.1 days (range 0–167 days) for hepatectomy, 42.1 days (range 0–178 days) for RFA, and 40.5 days (range 0–204 days) for TAE. The mortalities within the 30 days following the procedure were 1.08 % for hepatectomy, 0.14 % for RFA, and 0.45 % for TAE.
Complications related to each procedure
Hepatectomy (n = 5270)
RFA (n = 11688)
TAE (n = 37187)
Overall, n (%)a
Hemorrhage, n (%)
Bile duct stenosis, n (%)
Liver abscess, n (%)
Pneumothorax, n (%)
Perforation of gastrointestinal tract, n (%)
Peritonitis, n (%)
Heat burn, n (%)
Hepatic infarction, n (%)
Liver failure, n (%)
Cardiac complication, n (%)
Ruptured suture, n (%)
Renal failure, n (%)
Pulmonary embolism, n (%)
Wound infection, n (%)
Pneumonia, n (%)
Allergy to anesthetic agents, n (%)
Hepatectomy, RFA, and TAE are widely applied therapeutic procedures which account for more than 90 % of all procedures performed in patients with primary HCC in Japan . When deciding on a treatment strategy, both therapeutic efficacy and the risks associated with each treatment modality should be considered. Whereas the effectiveness of a procedure can be evaluated based on established criteria such as survival, local recurrence, and tumor necrosis, the complication rate as an index of safety varies greatly among reports, mainly because of differences in the definitions used. In-hospital mortality is a more reliable indicator of safety, but a large number of samples are required to obtain an accurate estimate of mortality, because death is a relatively rare event. In the present study, we investigated the mortalities and morbidities associated with various therapeutic procedures for HCC, using information from the nationally representative Japanese DPC database. It is particularly worth noting that the data for in-hospital mortality were expected to be 100 % reliable and free from recall bias because outcome was a required item on discharge.
In this study, the mortality following hepatectomy was 2.60 %, out of a total of 54,145 patients with HCC. Previously reported in-hospital mortalities following hepatectomy for primary and metastatic liver tumors at major high-volume centers were 3.8–8 and 0–7.0 % [4, 5, 24–30], respectively. For example, one report from the United States, using data from a nationwide inpatient sample over a 9-year period, showed a mortality of 6 % . However, the background of that study may be different from that of the present study; for example, lower invasive procedures such as partial hepatectomy and segmentectomy accounted for the major part of the present study. This kind of factor may have influenced the lower mortality rate in the present study.
The mortality associated with RFA in the present study was 0.25 %, which was similar to that noted in previous reports (0.2–0.6 %) [32–39]. Multivariate analysis identified older age and cardiac comorbidity as factors significantly related to high mortality in patients undergoing RFA.
A recent systematic review of the safety of TAE, based on 37 trials with 2,858 patients, reported a median periprocedural mortality (≤30 days) of 2.4 % (range 0–9.5 %) , which was higher than the in-hospital mortality for TAE in the present study (1.03 %). Some previous reports defined mortality as death within the 30 days following the procedure. In the present study, the 30-day mortalities for hepatectomy, RFA, and TAE were 1.08, 0.14, and 0.45 %, respectively.
A number of studies identified hospital procedure volume as an important determinant of postoperative mortality following advanced surgical procedures [31, 41–46]. In the present study, hospital procedure volume was significantly associated with in-hospital mortality for hepatectomy and TAE, but although the RFA-associated in-hospital mortality tended to be lower in high-volume hospitals, the difference was not significant. Despite the large sample size, it is still possible that this study was too underpowered to show any significant association between hospital volume and RFA mortality, because of the exceptionally low mortality rate of RFA. These results suggest that concentrating patients indicated for hepatectomy in high-volume centers should be considered on safety grounds.
In the present study, the mortality rate for hepatectomy was lower in patients over 80 years old. The indication for hepatectomy is determined on the basis of several factors. Although there is no specific age limitation for hepatectomy in Japan, older patients have shorter long-term survival after hepatectomy compared to younger patients, because of their expected life span. Thus, the indication for hepatectomy in older patients, especially those over 80, is stricter in clinical practice. Taking these factors into consideration, it is possible that the patients over 80 years old from the DPC database who did undergo hepatectomy were in generally better than average health. That could explain the lower mortality associated with hepatectomy in patients of 80 years and over in the present study. Similarly, the in-hospital mortality rate for TAE was significantly lower in older patients according to the univariate analysis. This result also may be related to the indications for TAE in Japan. Moreover, the intensity and area of embolization for TAE can be regulated, and embolization is likely to be less intensive in older patients, possibly accounting for the lower mortality in older patients who underwent TAE. Cardiac comorbidity was significantly associated with in-hospital mortality for RFA. According to the database, three out of the five deceased patients with cardiac comorbidities were speculated to have died as a result of cardiac complications (e.g., myocardial infarction, angina pectoris, and heart failure). RFA is thought to be less invasive than hepatectomy, and is sometimes considered as an alternative therapy to hepatectomy in patients with relatively severe cardiac comorbidities. The cardiac comorbidities were thus likely to have been severe in the RFA group, which could account for the higher mortality after RFA in the present study.
The complication rates for hepatectomy, RFA, and TAE in the present study were 14.48, 4.54, and 4.49 %, respectively. Previously reported complication rates have varied among studies, ranging from 28.4 to 47.7 % [47–52], from 0 to 12.7 % [32–38], and from 4.3 to 10.8 % [53–55] for hepatectomy, RFA, and TAE, respectively. Our multivariate logistic regression analysis demonstrated that the complication rate was significantly higher after more invasive procedures in patients treated in hospitals with lower procedure volumes, in patients with diabetes mellitus, and in patients with cardiac diseases in the case of hepatectomy; and in patients treated in hospitals with higher procedure volumes, patients with diabetes mellitus, and patients with cardiac diseases for RFA and TAE (data not shown). However, complications are usually reported in the DPC database in relation to the reimbursement of medical fees, and the reported complications were therefore less objective than the reported mortality, and could have been underestimated. The complication rate was relatively low for hepatectomy, and the rates for the other procedures were similar to those in previous reports.
The present study had several limitations. First, although the DPC database represents approximately 40 % of all admissions to secondary and tertiary care hospitals in Japan, participating hospitals tend to be medium-to-large-sized institutions. The mortality could therefore have been underestimated by potentially excluding low-procedure-volume hospitals. Second, some important clinical data that may affect the risk of death related to treatments, such as the size and location of the tumor, and severity indexes of liver disease [e.g., the Child–Pugh and model for end-stage liver disease (MELD) scores] were unavailable in this database. Third, data on late-onset complications that appeared after discharge (i.e., biloma, biliary injury, or hepatic abscess) were also unavailable, because the database covers only inpatient data. This may have led to an underestimation of the complication rate in this study. However, according to previous reports, late-onset complications appear to have minimal effects on the mortality rates. Fourth, as noted above, as the DPC system was basically designed for assessing reimbursement, co-existing diseases are usually reported when a specific treatment is needed,; e.g., the proportion of patients with diabetes was higher in the hepatectomy group than that with the other procedures. This may be because patients who underwent hepatectomy were more likely to have been treated with intensive insulin therapy before surgery. Fifth, the immediate cause of death is not a required item in the DPC database. Accordingly, the in-hospital deaths recorded in this database have room for treatment unrelated deaths, the procedure-related mortality could therefore have been overestimated. However in-hospital mortality is more robust in terms of objectivity compared to treatment-related mortality assessed by operators. Finally, some complications, such as tumor seeding, were not covered by the ICD codes and could therefore not be evaluated.
In conclusion, this study confirmed that the therapeutic procedures used to treat liver tumors in Japan were associated with low mortalities and low complication rates. However, procedure-related mortality can be affected by patient and therapeutic backgrounds.
This work was supported by a Grant-in-Aid for Research on Policy Planning and Evaluation from the Ministry of Health, Labour and Welfare, Japan; and Health Sciences Research Grants of The Ministry of Health, Labour and Welfare (Research on Hepatitis).
Conflict of interest
The authors declare that they have no conflict of interest.