fax: +886 2 2873 9318
Article first published online: 7 JAN 2013
Copyright © 2012 American Association for the Study of Liver Diseases
Volume 57, Issue 1, pages 112–119, January 2013
How to Cite
Hsu, C.-Y., Lee, Y.-H., Hsia, C.-Y., Huang, Y.-H., Su, C.-W., Lin, H.-C., Lee, R.-C., Chiou, Y.-Y., Lee, F.-Y. and Huo, T.-I. (2013), Performance status in patients with hepatocellular carcinoma: Determinants, prognostic impact, and ability to improve the Barcelona Clinic Liver Cancer system. Hepatology, 57: 112–119. doi: 10.1002/hep.25950
Potential conflict of interest: Nothing to report.
Supported by grants from the Center of Excellence for Cancer Research at Taipei Veterans General Hospital (DOH101-TD-C-111-007), Taiwan, from Taipei Veterans General Hospital (V101C-170), Taipei, Taiwan, and from the Ministry of Education, Aim for the Top University Plan (101AC-D101), Taiwan.
- Issue published online: 7 JAN 2013
- Article first published online: 7 JAN 2013
- Accepted manuscript online: 13 JUL 2012 04:24PM EST
- Manuscript Accepted: 25 JUN 2012
- Manuscript Received: 17 APR 2012
Performance status is included in the Barcelona Clinic Liver Cancer (BCLC) system for hepatocellular carcinoma (HCC). Few studies specifically evaluated the role of performance status in patients with HCC. This study investigated its distribution, determinants, and prognostic impact, aiming to improve the performance of the BCLC system. A total of 2,381 HCC patients were enrolled. Performance status was determined according to the Eastern Cooperative Oncology Group scale. The prognostic ability of the original and three modified BCLC systems in HCC patients was compared by the Akaike information criterion (AIC). There were 60, 17, 11, 8, and 4% of patients who were classified as performance status 0, 1, 2, 3, and 4, respectively. A worse performance status significantly correlated with age, alcoholism, hypoalbuminemia, hyperbilirubinemia, renal insufficiency, hyponatremia, and prothrombin time prolongation (all P < 0.001). Larger tumor burden, poorer residual liver function, more frequent vascular invasion, and diabetes mellitus were also observed in patients with worse performance status (all P < 0.001). Patients with poorer performance status more often received best supportive care (P < 0.001). In the Cox proportional hazards model, performance status was an independent prognostic predictor and the long-term survival tended to be worse in patients with progressively poor performance status (all P < 0.05). Reassigning patients with performance status 0 or 1 to stage B provided the lowest AIC among the four BCLC-based staging systems.
Performance status is strongly associated with both tumoral and cirrhotic factors and accurately predicts long-term survival in HCC patients. Modification of the BCLC system based on performance status may further enhance its prognostic ability in patients with early to advanced cancer stage. (HEPATOLOGY 2013)
Hepatocellular carcinoma (HCC) is a major malignancy worldwide and increasing incidence of HCC has been observed.1, 2 The majority of symptomatic HCC patients are diagnosed at an intermediate or advanced stage,3 and curative treatments (resection, transplantation, and local ablation) are unlikely to be perform because of extensive tumor invasion, decompensated liver function, or poor general condition.
The performance status scale developed by the Eastern Cooperative Oncology Group (ECOG) is recorded from 0 (fully active, able to carry on all predisease performance without restriction) to 5 (dead).4 These criteria and scales are extensively used by physicians to evaluate the progression of diseases and how the daily living ability of patients is affected in patients with a variety of malignancies. In addition, performance status is used as an indicator of treatment and predictor of long-term survival. However, few studies specifically investigated the influence of performance status on HCC patients. Factors associated with performance status in HCC patients have not been determined. More important, the prognostic ability of the performance status has not been systematically evaluated.
To date, at least nine staging systems have been proposed for HCC from different research groups.5-7 Among them, the Barcelona Clinic Liver Cancer (BCLC) system exhibited excellent discrimination of survival in patients with HCC,8, 9 and was suggested as the guideline for treatment allocation. Performance status is exclusively included in the BCLC system and considered a pivotal attribute of the superiority of the BCLC system over other prognostic models.3 Nevertheless, there were few analyses focusing on the rationality of this designation and the prognostic ability of performance status in the BCLC system has not been fully elucidated. In this study we investigated the distribution and associated factors of performance status in patients with HCC and analyzed its prognostic impact in a large cohort of patients. Three modified BCLC models based on performance status were proposed and compared with the original BCLC system, aiming to further enhance its prognostic accuracy.
Patients and Methods
Patients admitted to Taipei Veterans General Hospital during a 10-year period from 2002 to 2011 with newly diagnosed HCC were prospectively evaluated. A total of 2,381 treatment-naïve HCC patients were identified and retrospectively analyzed in this study. The duration of follow-up was defined as the time from the initial diagnosis until the end of the patient's last follow-up or death. The baseline information, including patient demographics, causes of chronic liver disease, severity of cirrhosis, serum biochemistry, performance status, cancer stage, and treatment modality was recorded at the time of diagnosis. Current ethical guidelines and the standards of the Declaration of Helsinki were strictly followed in this study.
Diagnosis and Definitions.
The diagnosis of HCC was according to the results of typical radiological features in at least two imaging modalities including ultrasound, magnetic resonance imaging (MRI), contrast-enhanced dynamic computed tomography (CT), and hepatic arterial angiography, or by a single positive imaging technique accompanied with serum α-fetoprotein (AFP) level >400 ng/mL or histologically proven.3, 10 Alcoholism was diagnosed in patients with daily consumption of at least 40 g of alcohol for 5 years or more.11 Performance status was recorded at enrollment according to the ECOG criteria.4 Total tumor volume (TTV) was calculated according to mathematical equations as described.6 The estimated glomerular filtration rate (eGFR) was calculated using the modification of diet in renal disease (MDRD) formula, which is considered better than other equations for estimating eGFR from creatinine in adults and in patients with cirrhosis.12, 13
Criteria of surgical resection for patients with HCC were (1) tumor invasion less than 4 Healey's segments; (2) Child-Turcotte-Pugh (CTP) class A or B and less than 25% retention of indocyanine green at 15 minutes after injection; and (3) no distant metastasis or main portal vein trunk involvement. Liver transplantation was considered in patients within the Milan criteria with coexisting CTP class B or C cirrhosis.14 For patients with inoperable tumor(s), locoregional therapy including percutaneous ethanol or acetic acid injection, radiofrequency ablation, or transarterial chemoembolization (TACE) was performed according to the number and size of tumor nodules as reported.15, 16 Targeted therapy was given to selected patients with distant tumor metastasis and preserved liver function. Best supportive care was provided for patients with extensive tumor invasion, severe liver decompensation, distant metastasis, main portal vein thrombosis, or when treatment efficacy was considered limited.
Categorical data were compared using the chi-squared test (two-tailed). Continuous data were compared among the five groups (performance stats 0, 1, 2, 3, and 4) of patients by the Kruskal-Wallis ranked sum test. In the analysis for prognostic predictors, continuous variables were split by the median values and treated as dichotomous covariates. The Kaplan-Meier method with a log-rank test was used to compare the survival distribution and prognostic factors. Factors that were significant in the univariate survival analysis were introduced into the multivariate Cox proportional hazards model with a forward stepwise method to determine the adjusted hazard ratio. P < 0.05 was considered statistically significant.
The prognostic accuracy of four BCLC-based staging systems was compared to decide which system possessed the most accurate prediction for long-term survival (monotonicity of the score). Homogeneity (small difference in survival among patients in the same classification within each staging system) was determined by the likelihood ratio (LR) χ2, which was generated by the Cox proportional hazard model.6, 7 The consequences of the Cox regression were presented with the Akaike information criterion (AIC), which exhibited how the staging systems affected the dependent variable (patient survival). The lower the AIC, the more explanatory and informative the model is.17, 18
The baseline demographics of 2,381 patients are shown in Table 1. The majority of the patients were men (77%) and the mean age was 64 years. The most frequent etiologies of underlying liver disease were hepatitis B (55%), hepatitis C (31%), and alcoholism (18%). According to the ECOG performance status criteria, 1,425 (60%), 402 (17%), 262 (11%), 188 (8%), and 104 (4%) patients were designated with status 0, 1, 2, 3, and 4, respectively. There were 7%, 20%, 14%, 45%, and 14% of patients belonging to BCLC stages 0, A, B, C, and D, respectively.
|Number of patients||2,381|
|Age (years, mean ± SD)||64 ± 13|
|Etiology of cirrhosis (%)|
|Serum biochemistry (mean ± SD)|
|Albumin (g/dL)||3.7 ± 0.7|
|Bilirubin (mg/dL)||1.6 ± 2.8|
|Creatinine (mg/dL)||1.2 ± 1.0|
|INR of prothrombin time||1.1 ± 0.2|
|Sodium (mmol/L)||138 ± 4|
|Estimated glomerular filtration rate (mL/min/1.73m2)||75 ± 30|
|α-fetoprotein (ng/mL, mean ± SD, [median])||25,504 ± 244,939 (47)|
|Performance status 0/1/2/3/4 (%)||60/17/11/8/4|
|Mean CTP score (median)||6.2 ± 1.7 (5)|
|MELD score (median)||9.8 ± 4.2 (8.3)|
|Total tumor volume (cm3, mean ± SD, [median])||373 ± 722 (49)|
|Number and size of tumor (%)|
|Vascular invasion (%)||926 (39)|
|Ascites (%)||586 (25)|
|Diabetes mellitus (%)||566 (24)|
|Treatment modality (%)|
|Local ablation||465 (18)|
|Targeted therapy||162 (7)|
|Supportive care||428 (18)|
|BCLC stage 0/A/B/C/D (%)||7/20/14/45/14|
Associated Factors of Performance Status.
Comparison of baseline demographics in HCC patients according to performance status is shown in Table 2. In comparison with patients who were fully active (performance status 0), patients with decreased performance status (1-4) were significantly older (P < 0.001) and more often had alcoholism (P < 0.001), lower serum albumin and sodium levels (both P < 0.001), higher serum bilirubin and creatinine levels (both P < 0.001), and longer prothrombin time (P < 0.001). Patients with performance status 0 had significantly lower serum AFP levels (P < 0.001). In addition, HCC patients with worse performance status had higher CTP and model for endstage liver disease (MELD) scores, larger total tumor volume, and more frequent vascular invasion and diabetes mellitus (all P < 0.001). Patients with performance status 0 more frequently received curative treatments (including surgical resection, transplantation, and local ablation; P < 0.001).
|0 (n = 1425)||1 (n = 402)||2 (n = 262)||3 (n = 188)||4 (n = 104)||P|
|Age (year)||63 ± 13||63 ± 13||67 ± 13||66 ± 16||69 ± 14||<0.001|
|Sex (male, %)||78||76||75||81||75||0.474|
|Albumin (g/dL)||3.9 ± 0.5||3.5 ± 0.6||3.4 ± 1.3||3.1 ± 0.6||3.0 ± 0.6||<0.001|
|Bilirubin (mg/dL)||1.1 ± 1.3||1.4 ± 1.7||1.9 ± 2.9||3.7 ± 5.4||4.5 ± 7.0||<0.001|
|Creatinine (mg/dL)||1.1 ± 0.8||1.1 ± 1.1||1.4 ± 1.3||1.4 ± 1.2||1.6 ± 1.4||<0.001|
|eGFR (mL/min/1.73m2)||75 ± 23||82 ± 42||70 ± 32||70 ± 34||60 ± 37||<0.001|
|Sodium (mmol/L)||139 ± 3||138 ± 3.6||137 ± 4.5||135 ± 4.9||135 ± 5.9||<0.001|
|INR of PT||1.0 ± 0.1||1.1 ± 0.2||1.1 ± 0.2||1.2 ± 0.2||1.2 ± 0.3||<0.001|
|α-fetoprotein (ng/mL)||11,200 ± 125,134||23,841 ± 99,644||66,431 ± 626,657||54,021 ± 174,077||74,183 ± 264,197||<0.001|
|CTP classification A/B/C (%)||89/11/0||64/31/5||48/41/11||26/43/31||10/57/34||<0.001|
|CTP score||5.5 ± 0.9||6.4 ± 1.6||7 ± 1.9||8.4 ± 2.2||8.7 ± 2||<0.001|
|MELD score||8.6 ± 2.7||9.7 ± 3.5||11 ± 4.8||14 ± 6||15 ± 7||<0.001|
|Total tumor volume (cm3; mean ± SD)||224 ± 523||510 ± 865||590 ± 869||655 ± 942||832 ± 975||<0.001|
|Tumor size <3/≥3 cm (%)||41/59||28/72||19/81||14/86||9/91||<0.001|
|Single/multiple tumor (%)||63/37||56/44||57/43||50/50||59/41||0.002|
|Vascular invasion (%)||30||44||52||63||73||<0.001|
|Diabetes mellitus (%)||20||30||31||28||30||<0.001|
|Treatment (curative treatments/ TACE/targeted therapy/SC, %)||58/33/3/6||42/29/11/18||24/31/13/32||12/14/15/59||3/6/12/79||<0.001|
The comparison of long-term survival according to performance status in HCC patients is demonstrated in Fig. 1. During a mean follow-up period of 19 ± 18 months, 571 (40%), 129 (32%), 108 (41%), 111 (59%), and 55 (53%) patients with performance status 0, 1, 2, 3, and 4 died, respectively. Significant survival differences were found between patients with performance status 0 versus 1 (P < 0.001), 1 versus 2 (P < 0.001), and 2 versus 3 (P < 0.001). Potential prognostic factors, including age, gender, etiologies of chronic liver disease, serum biochemistry and AFP level, number and size of HCC nodules, total tumor volume, performance status, presence of ascites, vascular invasion, diabetes mellitus, and treatment modalities were investigated in the univariate survival analysis. As shown in Table 3, prognostic predictors associated with increased risk of mortality included alcoholism, lower serum levels of albumin, eGFR and sodium, higher levels of serum bilirubin, prothrombin time prolongation and serum AFP, multiple tumors, tumor size of 3 cm or larger, larger tumor volume, ascites, vascular invasion, diabetes mellitus, and performance status (all P < 0.05). Additionally, patients undergoing curative treatments had the best long-term survival, followed by TACE, targeted therapy, and best supportive care group (P < 0.001). In the multivariate Cox model, patients with performance status 1, 2, 3, and 4 had an adjusted hazard ratio of 1.342 (95% confidence interval [CI]: 1.091-1.651, P = 0.005), 1.582 (95% CI: 1.252-1.999, P < 0.001), 2.162 (95% CI: 1.675-2.791, P < 0.001), and 2.295 (95% CI: 1.639-3.213, P < 0.001), respectively, in comparison to patients with performance status 0. Furthermore, 10 factors, including serum bilirubin and sodium level, eGFR, international normalized ratio (INR) of prothrombin time, AFP concentration, number of tumors, total tumor volume, ascites, vascular invasion, and treatment modalities were also identified as independent prognostic predictors (all P < 0.05).
|Univariate Analysis||Multivariate Analysis|
|N||1-Year Survival (%)||3-Year Survival (%)||P||Hazard Ratio||95% CI||P|
|Age (<65/≥65 years)||1162/1219||77/81||48/47||0.737|
|Albumin (<3.7/≥3.7 g/dL)||1080/1301||70/87||38/53||<0.001|
|Bilirubin (<0.9/≥0.9 mg/dL)||1164/1217||87/72||54/40||<0.001||1.23||1.072-1.411||0.003|
|eGFR (≥73/<73 mL/min/1.73m2)||1198/1183||81/78||52/43||0.004||1.238||1.086-1.41||0.001|
|INR of PT (<1/≥1)||714/1667||90/74||57/43||<0.001||1.377||1.179-1.609||<0.001|
|Sodium (≥139/<139 mmol/L)||1331/1050||87/69||55/34||<0.001||1.168||1.009-1.351||0.037|
|No. of tumor (single/multiple)||1428/953||83/75||50/42||<0.001||1.162||1.019-1.326||0.025|
|Tumor size (<3/≥3 cm)||776/1665||92/72||61/39||<0.001|
|α-fetoprotein (<47/≥47 ng/mL)||1189/1192||88/70||56/36||<0.001||1.457||1.274-1.666||<0.001|
|Total tumor volume (<49/≥49 cm3)||1188/1193||91/66||59/30||<0.001||1.508||1.297-1.753||<0.001|
|Vascular invasion (no/yes)||1455/926||89/61||54/33||<0.001||1.504||1.28-1.766||<0.001|
|Diabetes mellitus (no/yes)||1815/566||80/77||48/44||0.034|
|Best supportive care||428||35||7||3.169||2.505-4.01||<0.001|
Proposal of the Modified BCLC Systems Based on Performance Status and Comparison of the Prognostic Ability of Four BCLC-Based Systems.
The staging criteria were redefined based on the performance status in three modified BCLC models (Fig. 2). In modified BCLC model A, patients with performance status 2, who were originally assigned to BCLC stage C, were changed to BCLC stage D. In modified model B, all patients with performance status more than 0 were assigned to BCLC stage D. In modified model C, patients who were classified as performance status 0 or 1 without vascular invasion or extrahepatic spread were assigned as BCLC stage B. Other factors, including vascular invasion, distant metastasis, tumor burden, and CTP class in BCLC stage C and D remained the same in all four systems. The distributions according to the modified BCLC systems are shown in Table 4.
|N = 2,381||Distribution of Patients (0/A/B/C/D, %)||Homogeneity (Likelihood Ratio χ2)||Akaike Information Criterion|
|Modified BCLC model A||7/20/14/35/23||270.2||12631.4|
|Modified BCLC model B||7/20/14/19/40||239.7||12661.5|
|Modified BCLC model C||7/20/23/37/14||288.6||12612.6|
The comparison of survival distribution according to the four BCLC-based staging systems is shown in Fig. 3. Patients classified as BCLC stage 0 and A had comparable survival across all four staging systems (all P > 0.05). The original BCLC and modified BCLC model A and model C showed significant survival difference across patients from stage A to D (all P < 0.05). Among these four BCLC-based staging systems, the modified model C showed the lowest AIC value, followed by the modified model A and the original BCLC system, and lastly, the modified model B (Table 4).
The performance status scale proposed by the ECOG is widely used to quantify general well-being and activities of daily life in cancer patients.4 In our study, we found that 40% of HCC patients had performance status more than 0 at the time of diagnosis. Notably, 8% and 4% of patients were classified as status 3 and 4, respectively, which are defined by compromised capability of self-care. Importantly, deterioration of performance status was identified as an independent predictor of decreased survival, and patients with performance status 1-4 had 34%-130% increased risk of mortality. Considering its powerful influence on survival, we modified the criteria of BCLC system and found that the adaptation based on performance status may further enhance its prognostic ability in HCC patients.
Performance status is regarded as a competent and comprehensive parameter to determine the degree of health status in cancer patients.19, 20 In this study, HCC patients with poor performance status had significantly higher CTP and MELD scores. Only 7% of patients with performance status 0 had ascites; on the other hand, approximately half of patients with performance status 1-4 had ascites, which was closely associated with loss of daily activity and increased time restricted to bed.21 Additionally, we found that patients with deteriorated performance status had lower serum sodium level and eGFR.22 These two features, along with ascites, are likely the result of portal hypertension and peripheral vasodilatation in patients with advanced cirrhosis.23, 24 These findings highlight that performance status may comprehensively reflect a variety of complications of liver cirrhosis in patients with HCC.
As expected, HCC patients with worse performance status had larger or more HCC nodules, more frequent vascular invasion, and higher serum AFP level. Cachexia resulted from larger tumor burden, worsened portal hypertension, and aggressive phenotype of HCC could all or in part contribute to a poor performance status.25, 26 Furthermore, poorer performance status also correlated with older age, alcoholism, and diabetes mellitus, which are often associated reduced long-term survival in patients with HCC.27-30 Taken together, our findings suggest that performance status is an indispensable surrogate marker to describe the influence of most prognostic predictors in HCC.
Survival analysis showed that HCC patients with poorer performance status tended to have worse long-term outcome, with the only exception of the comparison between patients with status 3 and 4. According to the criteria proposed by the ECOG, the difference of status 3 and 4 was defined by the degree of self-care patients could provide and how often they are stationary. The insignificant impact of performance status in this subset of patients is possibly because the interpretation might be subjective in borderline cases and the numbers of patients with performance status 3 and 4 were relatively small (8% and 4%, respectively).
In addition to the cirrhosis and tumoral factors (severity of cirrhosis, tumor burden, serum sodium and AFP level, renal function, and vascular invasion), consistent with previous studies,31-33 performance status was identified as an independent prognostic predictor in the multivariate Cox model. Notably, in comparison to patients with performance status 0, patients with status 1 to 4 had significantly increased of risk of mortality. Our data indicate that patients with performance status 2 had 58% increased risk of death, which outweighed most other variables in the Cox model. In addition, performance status 3 and 4 had an even more devastating influence on long-term survival. These results not only disclose the strong prognostic impact of performance status on survival, but also imply that the allocation method of the currently used BCLC system, partly based on performance status, might not be perfect. In this study, we proposed three modified BCLC-based staging systems with different strategies of patient allocation according to performance status to assess the prognosis in a large cohort of patients. By using the AIC analysis, the modified model C had the lowest AIC value, followed by the modified model A. These findings indicate that the current allocation strategy in the BCLC system might restrain the prognostic ability of performance status. Reallocating patients who are classified as performance status 0 or 1 without vascular invasion or extrahepatic spread into BCLC stage B could provide an enhanced prognostic accuracy.
Treatment modality is highly associated with long-term survival of HCC patients.7, 31, 32 Performance status may greatly influence the choice of treatment, and both performance status and treatment modalities play an important role in prognostic evaluation. To clarify their interaction, treatment modalities were entered into the univariate and multivariate analyses, which showed that the prognostic impact of performance status is independent of the selection of treatment.
There are a few limitations of this study. First, although the performance status was determined at the time of diagnosis, interobserver bias could still exist. Second, in this single-center study, more than half of our patients had HBV infection. This feature is distinctly different from most Western countries and Japan, where HCV infection is the predominant etiology of liver dysfunction.34 Lastly, only a minority of patients (0.3%) received liver transplantation in this cohort; therefore, the generalizability of this study might be limited, especially in centers with a high volume of liver transplantation.
In conclusion, our results indicate that performance status is associated with larger tumor burden and more severe cirrhosis, and is an indispensable surrogate to reflect the general condition of HCC patients. A poor performance status is an independent and strong prognostic predictor of decreased survival in HCC patients. Modification of the BCLC system by allocation rearrangement based on performance status could further enhance its prognostic ability.
- 9Comparison of seven staging systems in cirrhotic patients with hepatocellular carcinoma in a cohort of patients who underwent radiofrequency ablation with complete response. Am J Gastroenterol 2008; 103: 597-604., , , , , , et al.Direct Link:
- 15The 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 2006; 107: 141-148., , , , , , et al.
- 16Proposal of a modified Cancer of the Liver Italian Program staging system based on the model for end-stage liver disease for patients with hepatocellular carcinoma undergoing loco-regional therapy. Am J Gastroenterol 2006; 101: 975-982., , , , , , et al.Direct Link:
- 20Performance status is the most powerful risk factor for early death among patients with advanced soft tissue sarcoma: the European Organisation for Research and Treatment of Cancer-Soft Tissue and Bone Sarcoma Group (STBSG) and French Sarcoma Group (FSG) study. Br J Cancer 2011; 104: 1544-1550., , , , , , et al.
- 21Ascites in patients with hepatocellular carcinoma: prevalence, associated factors, prognostic impact and staging strategy. Hepatol Int 2012 [Epub ahead of print]., , , , , , et al.
- 27Hepatocellular carcinoma in the elderly: results of surgical and nonsurgical management. Am J Gastroenterol 1999; 94: 2460-2466., , , , , , et al.Direct Link:
- 30Alcoholism worsens the survival of patients with hepatitis B virus and C virus-related hepatocellular carcinoma. Hepatol Int 2012 [Epub ahead of print]., , , , , , et al.