Clinical features, biochemical parameters, and virological profiles of patients with hepatocellular carcinoma in Hong Kong

Authors


Dr M. F. Yuen, Department of Medicine, University of Hong Kong, Queen Mary Hospital, Hong Kong.
E-mail: mfyuen@hkucc.hku.hk

Summary

Background  Clinical features of hepatocellular carcinoma patients are changing because of screening.

Aim  To examine the clinical features of hepatocellular carcinoma patients in Hong Kong and validity of different staging systems.

Methods  A total of 223 Chinese patients with hepatocellular carcinoma were studied.

Results  Seventy-eight percent of hepatocellular carcinoma patients had chronic hepatitis B (43% diagnosed by screening). Hepatitis B positivity, weight loss, jaundice, encephalopathy, alpha-fetoprotein level, portal vein thrombosis, extrahepatic metastasis, and treatment were shown to be independent factors affecting survival. Of chronic hepatitis B patients, hepatitis B virus DNA levels (P = 0.001) and portal vein thrombosis (P = 0.008) were independent factors affecting survival. Seventy-six percent of chronic hepatitis B patients with hepatocellular carcinoma were hepatitis B e antigen negative. Screening patients had hepatocellular carcinoma detected at an earlier stage and better survival (median survival: 21 vs. 4 months, P < 0.0001). All staging systems had good stratification of survival. Prognosis and median survival generated were different when compared with the US data.

Conclusions  Chronic hepatitis B was the most common cause of hepatocellular carcinoma in Hong Kong. High-risk chronic hepatitis B patients should be followed irrespective of the hepatitis B e antigen status. Hepatitis B virus DNA levels at the time of diagnosis are an important survival predictor. Screening detected hepatocellular carcinoma at an earlier stage and prolonged survival. Staging systems should be validated in different populations.

Introduction

Hepatocellular carcinoma (HCC) is one of the major health problems of the world. Its incidence is increasing in the developed countries.1 HCC currently represents the third cause of cancer-related death and fifth most common tumour worldwide.2 The overall survival of patients with HCC is still unsatisfactory despite the advances in diagnostic technology and treatment. The problem is even more overwhelming in Asia where the prevalence of chronic hepatitis B virus (HBV) infection is high.3 In Hong Kong, the annual incidence of HCC is 23/100 000 persons per year.4

The clinical presentation of patients with HCC has changed significantly in the last one to two decades with the adoption of screening tests, ultrasonography and alpha-fetoprotein (AFP). In the 1970s, most HCC presented with symptoms directly related to the tumours or to hepatic decompensation.5 The most common symptoms were abdominal pain, abdominal distension and weight loss. HCC could also lead to hepatic insufficiency by displacement of non-cancerous hepatocytes in a cirrhotic liver. Paraneoplastic symptoms, such as diarrhoea, polycythaemia, hypercalcaemia, hypoglycaemia had been described usually in late-stage disease.6 Tumour rupture with catastrophic bleeding was rare. With the HCC surveillance/screening in chronic HBV patients, increasing numbers of HCC nowadays are detected early.7

Clinical staging of cancers provides a guide to assess prognosis and to direct therapeutic interventions. Several prognostic staging systems have been proposed for HCC. There has been debate on which is the best staging system. The tumour-lymph node-metastasis (TNM) staging system is widely used for most solid tumours. This system does not incorporate non-tumour related factors.8 As patients with HCC often have poor liver function because of cirrhosis influencing the prognosis, newer staging systems have incorporated hepatic function status into the staging of patients with HCC. The Okuda staging system used ascites, albumin and bilirubin levels as indicators of liver function and an estimate of the percentage of primary tumour involvement in the liver.9 More recently, the Cancer of the Liver Italian Program (CLIP) investigators developed another prognostic scoring system (the CLIP score).10 They include Child-Turcotte-Pugh (CTP) grading, distribution of tumour(s), AFP level, and portal vein thrombosis into the calculation of their prognostic scores. All these systems have been validated to be useful in prognostic stratification in different cohort of patients.11–13

The primary aim of the present study was to identify and examine the clinical features, modes of presentation, biochemical parameters and virological data of a cohort of HCC patients in Hong Kong presented between 2001 and 2003. The secondary aim was to assess the validity of the different clinical staging in our cohort of patients.

Patients and Methods

Patients

All patients who were diagnosed with HCC from January 2000 to December 2003 in the University Department of Medicine, University of Hong Kong, Queen Mary Hospital, Hong Kong, were recruited into the study.

Diagnosis and staging of HCC

The diagnosis of HCC was based on imaging techniques including ultrasound (USG), computerized tomography (CT), hepatic angiography (HAG), magnetic resonance imaging (MRI), AFP levels, and biopsy and/or histology from surgical specimen. The diagnostic criteria for HCC were either a confirmative histology or the presence of a space-occupying hepatic lesion with characteristic features on imaging studies (e.g. characteristic MRI imaging, neovascularization and arterio-venous shunting in HAG and/or characteristic lipiodal uptake in CT) together with an elevated AFP (>20 ng/mL)

For all patients, demographic information, mode of presentation, aetiology of liver disease, biochemical data, haematological data were recorded. Assessment of hepatic function was calculated based on CTP score and model for end-stage liver disease (MELD) score. Tumour characteristics on imaging (e.g. CT, MRI, HAG) were assessed by experienced radiologists. Their size, number, multifocal/diffuse pattern, portal vein thrombosis (PVT), and the presence of extrahepatic metastasis were recorded. All data, including staging of the tumour, were determined at the time of HCC diagnosis. The treatments received by patients and the dates of death or last follow-up were documented.

Three subgroup analyses are performed:

  • (i) Comparison of HBV related and non-HBV related HCC: patients were categorized into two groups according the hepatitis B surface antigen (HBsAg) status.
  • (ii) Comparison of hepatitis B e antigen (HBeAg) status in HBsAg-positive patients: patients who are HBsAg-positive were divided into two groups (HBeAg-positive vs. HBeAg-negative).
  • (iii) Comparison of their mode of presentation: patients were categorized according to their mode of presentation (screening vs. non-screening).

Statistical analysis

Statistical analysis was performed using the Statistical Program for Social Sciences (SPSS 12.0 for windows, SPSS Inc., Chicago, IL). Mann–Whitney U-test and Student's t-test were used for continuous ordinal variables with skew and normal distribution respectively. Chi-square test with Yates’ correction and the Fisher's exact test were used for discrete variables. Survival curves were calculated using the Kaplan–Meier methods and comparison was made with the log-rank test. Multivariate analysis was performed using the Cox regression analysis. In all cases, survival analysis was calculated from the date of diagnosis to the date of the most recent follow-up examination or to the date of death. Statistical significance is defined as P < 0.05.

Results

Patient characteristics

A total of 223 patients were diagnosed with HCC during the study period; all were recruited into the study. All patients were Chinese. Table 1 shows the demographics, clinical and tumour information for all patients. The clinical staging results are shown in Table 2. The overall median age at the time of HCC was 63, being 61 for males and 70 for females (P < 0.001). The male to female ratio was 4:1. The modes of presentation were as follows: 43.5% by screening; 39.5% with symptoms, and 17% present with unrelated condition and incidental HCC.

Table 1.   Demographics, clinical, biochemical information and tumour characteristics of the 223 patients with hepatocellular carcinoma (HCC)
 HCC patients (n = 223)
  1. HBV, Hepatitis B virus; HBeAg, hepatocellular carcinoma were e antigen; HCV, hepatitis C virus; ALB, albumin; AST, aspartate transaminase; ALT, alanine transaminase; BIL, bilirubin; PT, prothrombin time; AFP, alpha-fetoprotein; CTP, Child-Turcotte-Pugh; MELD, model for end-stage liver disease.

  2. Continuous variables are quoted as median (range).

Demographics
 Age, years (range)63 (25–93)
 Age, years (M:F)61:70
 Sex, M:F179 (80%):44 (20%)
 Mode of diagnosis (screening:others)97 (43%):126 (57%)
 HBV: others175 (78%):48 (22%)
 HBeAg positive:negative:missing data35 (16%):113 (51%):27 (12%)
 HCV15 (6.7%) (one HBV co-infection)
 Alcohol ≥50 gm/day73 (33%)
 Smoking (>10 pack years)114 (51%)
 Diabetes mellitus52 (23%)
Laboratory values (median, range))
 AFP, ng/mL250 (2–1 100 000)
 ≤20, n43 (19%)
 21–200, n62 (28%)
 >200, n116 (52%)
 ALB, g/L34 (16–47)
 BIL, μmol/L23 (2–788)
 AST, U/L77 (15–1540)
 ALT, U/L58 (5–1415)
 ALP, U/L130 (11–819)
 PT, (s)14 (10–139)
 Creatinine, μmol/L95 (57–875)
 Platelet count, ×109/L146 (30–730)
 CTP score6 (5–13)
 MELD score11 (2–40)
Tumour characteristics
 Diameter of largest tumour nodule cm (median), (range)5 (1–17)
 Unifocal/multifocal/diffuse94 (43%):103 (47%):22 (10%)
 Bilobar101 (45%)
 Portal vein thrombosis62 (28%)
 Extrahepatic metastasis23 (10%)
 Median survival (months)11
Table 2.   Clinical staging information and the median survival of our cohort of 223 patients with hepatocellular carcinoma (HCC) and comparisons of the median survivals with a recent study from the United States17
  HCC patients (n = 223)Median survival of our study cohort (months)Median survival of a recent study from the United States17 (months)
  1. Values are expressed (%).

  2. n/a - data not available.

Staging
Child Pugh (A:B:C)
 A119 (53.4)18n/a
 B52 (23.3)7.5n/a
 C52 (23.3)3n/a
Okuda (I:II:III)
 I80 (36)2918
 II106 (48)6.516
 III36 (16)2.54
CLIP (0:1:2:3:4:)
 025 (11.4)3053
 145 (20.5)2116.5
 249 (22.4)2015
 335 (16)610.5
 424 (11)3.52
 525 (11.4)2n/a
 616 (7.3)1.5n/a
TNM (I,II,III,IV)
 I16 (7.2)34n/a
 II57 (25.6)2822
 III11 (4.9)15.518
 IV139 (62.3)58

Clinical features

For patients presenting with symptoms, the most common symptom was abdominal pain (30.2%) followed by abdominal distensions or ascites (20.2%), weight loss (10.9%), dyspnoea (5.4%) and jaundice (3.1%). One patient presented with tumour rupture and severe abdominal pain. One patient presented with symptoms of polycythaemia.

Seventy-three (32%) patients had greater than 50 g alcohol intake per day and 114 (51%) patients had more than 20 cigarettes per day for over 10 years.

Biochemical parameters

The biochemical parameters of all patients are summarized in Table 1. The median AFP was 250 ng/mL. Forty-three patients (19%) had an AFP level 20 ng/mL or less. Both the median aspartate transaminase (AST) and alanine transaminase (ALT) were only mildly elevated with AST/ALT ratio of 1.3:1. The liver function, as reflected by CTP and MELD scores, was relatively well preserved with the median CTP score of 6 with MELD score of 11.

Aetiology factors

One hundred and seventy-five patients (78%) were positive for HBsAg (24% HBeAg-positive and 76% hepatitis B e antibody (anti-HBe) positive). Eight-four HBsAg-positive patients had the measurements of serum HBV DNA levels using COBAS TaqMan HBV test (Roche Diagnostics, Branchburg, NJ, US) with a lower detection limit of 35 copies/mL. The median viral DNA was 4.5 × 105 copies/mL. Twenty-one patients (25%) had HBV DNA levels <104 copies/mL. Of these, four patients had undetectable HBV DNA; eight patients with levels between 102 and 103 copies/mL; and the remaining nine patients had levels between 103 and 104 copies/mL. Two patients had documented HBsAg seroclearance before the development of HCC.

Fourteen patients (6.3%) were positive for anti-HCV. One patient was co-infected with hepatitis B and hepatitis C. Twelve patients (5.8%) had alcohol intake of more than 50 g per day. Twenty-one (9.4%) had unknown aetiology.

Tumour characteristics

The tumour characteristics on imaging are summarized in Table 1. A considerable proportion of patients had advance disease with 28% (n = 62) of the patients having PVT and 10% (n = 23) having extrahepatic metastasis.

Treatments

The treatment modalities received by the patients included liver transplantation (n = 3), tumour resection (n = 21), radiofrequency ablation (RFA) (n = 21), transcatheter arterial chemoembolization (TACE) (n = 83), and systemic chemotherapy (n = 5). One hundred and sixteen patients (52%) received more than one modalities of treatment.

Clinical staging and survival analysis

Table 2 lists the number of patients in each of the clinical staging group with their median survival in their respective stages. The median survival of our cohort was 11 months. Figures 1–3 show that all of the systems (Okuda, CLIP, and TNM) had good stratification of survival across all stages.

Figure 1.

 Kaplan–Meier estimated survival curves by Okuda Staging system (P < 0.0001).

Figure 2.

 Kaplan–Meier estimated survival curves by Cancer of the Liver Italian Program Score (P < 0.0001).

Figure 3.

 Kaplan–Meier estimated survival curves by UNOS-TNM staging system (P < 0.0001).

Univariate analysis showed that mode of diagnosis (screening vs. no-screening), HBsAg positivity, smoking, alcohol intake, presentation with ascites, encephalopathy, abdominal pain, jaundice, weight loss, platelet count, prothrombin time (PT), alkaline phosphatase (ALP), ALT, AST, bilirubin, albumin, AFP, the presence of extrahepatic metastasis, PVT, diameters of largest nodule, unifocal, bilobar, diffuse and multifocal HCC and treatment were significant baseline predictors of patient survival (Table 3). Cox regression analysis identified HBsAg positivity (P = 0.049), weight loss (P = 0.005), jaundice (P = 0.009), encephalopathy (P = 0.014), elevated AFP level (P = 0.036), the presence of PVT (P = 0.028) and extrahepatic metastasis (P = 0.001) were independent predictors of poor survival. Treatment is independently associated with better survival (P < 0.001).

Table 3.   Univariate analysis of baseline predictors of survival in the cohort of patients with hepatocellular carcinoma (HCC)
VariablesMedian survival (months) P-value
  1. HBeAg, hepatitis B e antigen; HCV, hepatitis C virus; ALB, albumin; AST, aspartate transaminase; ALT, alanine transaminase; BIL, bilirubin; PT, prothrombin time; AFP, alpha-fetoprotein; CTP, Child-Turcotte-Pugh.

Age
 <65120.213
 ≥6511
Gender
 Male110.401
 Female19
Screening
 Yes21<0.00001
 No4
HBsAg
 Positive120.0481
 Negative6
HBeAg
 Positive90.42
 Negative15
HCV
 Positive180.61
 Negative7
Alcohol
 Yes80.0317
 No13
Smoking
 Yes60.0002
 No21
Diabetes mellitus
 Yes140.107
 No11
Ascites
 Yes3.4<0.00001
 No15.6
Varices
 Yes120.545
 No10.7
Bleeding varices
 Yes130.998
 No11
Encephalopathy
 Yes10.0039
 No12
Splenomegaly
 Yes100.0832
 No13
Abdominal pain
 Yes3<0.00001
 No14
Jaundice
 Yes2.7<0.00001
 No16
Weight loss
 Yes4<0.00001
 No13.5
ALB, (g/L)
 <3560.0006
 ≥3518
BIL, μmol/L
 <2519<0.00001
 ≥254
AST, (U/L)
 <5622<0.00001
 ≥566.3
ALT, (U/L)
 <56150.0218
 ≥569
ALP, (U/L)
 <12025<0.00001
 ≥1205.1
PT, (sec)
 <14150.0056
 ≥146.3
Creatinine,
 <95120.2442
 ≥958
Platelet count (×109/L)
 <150180.0004
 ≥1506
Size of largest tumour nodule
 <5 cm22.5<0.00001
 ≥5 cm6.6
Unifocal
 Yes170.0135
 No9
Multifocal
 Yes80.002
 No18
Diffuse
 Yes1<0.00001
 No13.5
Bilobar
 Yes70.0018
 No17
PVT
 Yes3.5<0.00001
 No18.7
Metastasis
 Yes2.5<0.00001
 No13.5
AFP, (ng/mL)
 (≤20: 21–200: >200)28:19:6<0.00001
Treatment received
 Yes116<0.00001
 No107

In the subgroup of HBsAg-positive patients with HBV DNA performed at the diagnosis of HCC (n = 84), Cox regression survival analysis was performed with the addition of HBV DNA levels to the independent survival predictors identified above. In this sub-analysis, high HBV DNA levels (P = 0.001) and the presence of portal vein thrombosis (P = 0.008) were the only independent predictors of poor survival.

Subgroup analysis

HBV-positive vs. non-HBV patients

The median age of the patients in the HBV group (62 years) was significantly younger than the non-HBV group (69 years) (P = 0.006). A higher proportion of HBV patients had tumours detected by screening (P = 0.002). In addition, the tumour nodules were smaller than those that were not detected by screening (median, 4.5 vs. 6.5 cm respectively) (P = 0.025). No significant differences were found in the biochemical parameters, clinical stagings or treatment received between the two groups. The median survival was longer in HBsAg-positive patients compared with non-HBV patients (12 vs. 6 months, respectively, P = 0.048).

HBeAg-positive vs. HBeAg-negative patients

A total of 148 patients were included in the analysis. Seventy-six percent were anti-HBe-positive by the time of presentation. The result is summarized in Table 4. HBeAg-positive patients were significantly younger than the HBeAg-negative group (57 vs. 62-years old, P = 0.037). They also had significantly higher HBV DNA levels (P = 0.03). In terms of clinical features, patients who were HBeAg-positive had significant more ascites and jaundice compared with the HBeAg-negative group. They also had much higher median AFP, AST, ALT, ALP and PT, but lower albumin. The hepatic function of HBeAg-positive subjects as indicated by CTP score were significantly worse than that of HBeAg-negative subjects but their MELD scores were not statistically different.

Table 4.   Demographics, clinical and biochemical, tumour characteristics, median survival and clinical staging information of the patient groups (HBeAg-positive vs. HBeAg-negative group)
 HBeAg+ve (n = 35)HBeAg−ve (n = 113) P-value
  1. HBV, Hepatitis B virus; ALB, albumin; AST, aspartate transaminase; ALT, alanine transaminase; BIL, bilirubin; PT, prothrombin time; CTP, Child-Turcotte-Pugh; MELD, model for end-stage liver disease; CLIP, Cancer of the Liver Italian Program; TNM, tumour-lymph node-metastasis.

  2. Continuous variables are quoted as median (range).

Demographics
 Age, years57620.037
 Sex, M:F27:894:190.455
 Mode of diagnosis (screening:others)16:1964:490.332
 Alcohol ≥50 gm/day15 (43%)31 (27%)0.106
 Smoking (>10 pack years)17 (49%)53 (47%)0.84
 Diabetes mellitus9 (26%)24 (21%)0.64
 HBV DNA (copies/mL)6.5 × 1062.1 × 1050.03
Clinical features
 Ascites16 (46%)24 (21%)0.008
 Varices15 (43%)34 (30%)0.217
 Bleeding varices2 (6%)12 (11%)0.52
 Encephalopathy0 (0%)4 (4%)0.573
 Splenomegaly19 (54%)51 (45%)0.439
 Abdominal pain8 (23%)22 (19%)0.639
 Jaundice14 (40%)21 (19%)0.013
 Weight loss5 (14%)12 (11%)0.551
Laboratory values, median (range)
 AFP ng/mL911 (4–464 000)165 (2–812 900)0.017
 ALB, g/L32 (19–43)35 (16–46)0.005
 BIL, μmol/L33 (6–418)21 (2–788)0.129
 AST, U/L96 (27–1540)61 (15–794)0.004
 ALT, U/L64 (21–746)53 (4–286)0.022
 ALP, U/L151 (39–734)119 (16–550)0.006
 PT, (s)15 (11–139)13 (11–114)0.031
 Creatinine, μmol/L94 (61–169)97 (57–875)0.573
 Platelet count, ×109/L134 (38–730)117 (30–549)0.392
CTP score8 (5–12)6 (5–13)0.003
MELD12 (3–40)11 (2–36)0.248
Tumour characteristics
 Diameter of largest tumour nodule, cm (range)5 (2–16)4 (1–17)0.3
 Unifocal11 (31%)57 (50%)0.054
 Multifocal22 (63%)44 (39%)0.019
 Diffuse7 (20%)6 (5%)0.014
 Bilobar20 (57%)44 (39%)0.078
 Portal vein thrombosis9 (26%)24 (21%)0.64
 Extrahepatic metastasis4 (11%)10 (9%)0.741
Staging
 Child Pugh (A:B:C)13:8:1473:21:190.006
 Okuda (I:II:III)8:17:1051:50:100.007
 CLIP (0:1:2:3:4:≥5)1:6:9:4:5:1018:26:31:20:5:120.009
 TNM (I, II, III, IV)1:7:3:2413:33:6:610.221
 Treatment received20 (57%)71 (63%)0.557
 Median survival (months)9150.42

In terms of their tumour characteristics, the patients who were HBeAg-positive had a higher percentage of multifocal and diffuse HCC compared with those who were HBeAg-negative. A significantly higher proportion of subjects who were HBeAg-positive had higher stage/scores (CTP, Okuda and CLIP) compared with those who were HBeAg-negative. However, there was no significant difference in their TNM staging. The proportion of patients that could be treated was the same between the two groups. The median survival of the HBeAg-positive patients was 9 months compared with 15 months in those who were HBeAg-negative (P = 0.42).

Screening vs. non-screening patients

Of the 223 subjects, 97 (43%) patients were detected having HCC via screening/surveillance. The remaining 126 (57%) patients either presented with tumour-related symptoms/signs or were incidentally discovered to have HCC. The clinical features are summarized in Table 5. The screening group had significantly higher proportion of patients with chronic hepatitis B but fewer smokers. They had significantly less symptoms or signs (ascites, encephalopathy, abdominal pain, jaundice and weight loss) at presentation. Their laboratory values were also better (closer to normal) compared with the non-screening group (AFP, albumin, AST, and ALP). In terms of liver function as indicated by the CTP score, the patients detected via screening were significantly better but when using MELD score, there was no significant difference. The screening group had smaller tumour nodules, less diffuse HCC, less portal vein thrombosis and lower number of extrahepatic metastasis (all P < 0.01). The patients in the screening group were at earlier stages of HCC in all clinical staging systems (CTP, Okuda, CLIP and TNM) compared with the non-screening group. A significant higher proportion of the patients in the screening group were treatable (P < 0.0001). Their median survival was also statistically longer (21 months vs. 4 months) than the patient of the non-screening group (P < 0.0001).

Table 5.   Demographics, clinical and biochemical, tumour characteristics, median survival and clinical staging information of the patient groups (Screening vs. non-screening group)
 Screening (n = 97)Non-screening (n = 126) P-value
  1. HBV, Hepatitis B virus; ALB, albumin; AST, aspartate transaminase; ALT, alanine transaminase; BIL, bilirubin; PT, prothrombin time; CTP, Child-Turcotte-Pugh; MELD, model for end-stage liver disease; CLIP, Cancer of the Liver Italian Program; TNM, tumour-lymph node-metastasis.

  2. Continuous variables are quoted as median (range).

Demographics
 Age, years62.565.50.25
 Sex, M:F73:24106:200.125
 Alcohol ≥50 gm/day27 (28%)460.226
 Smoking (>10 pack years)38 (39%)760.007
 Diabetes mellitus24 (25%)280.75
HBsAg-positive86 (89%)890.002
HCV antibody positive8 (8%)70.134
Clinical features
 Ascites16 (16%)50 (37%)<0.0001
 Varices36 (37%)32 (25%)0.078
 Bleeding varices10 (10%)8 (6%)0.328
 Encephalopathy0 (0%)7 (6%)0.019
 Splenomegaly49 (51%)53 (42%)0.34
 Abdominal pain3 (3%)47 (37%)<0.0001
 Jaundice14 (14%)40 (32%)0.003
 Weight loss2 (2%)24 (19%)<0.0001
Laboratory values, median (range)
 AFP ng/mL133 (3–40940)622.5 (2–1 100 000)0.002
 ALB, g/L37 (16–46)33 (19–47).<0.0001
 BIL, μmol/L21 (2–196)28.(5–788)0.04
 AST, U/L63 (15–545)98 (15–1540)0.001
 ALT, U/L59 (10–233)57 (4–1415)0.903
 ALP, U/L105 (16–485)173 (11–819)<0.0001
 PT, (s)13.5 (11–114)14 (10–139)0.113
 Creatinine, μmol/L95.5 (57–875)93.5 (59–518)0.905
 Platelet count, ×109/L109 (30–283)177 (40–730)<0.0001
CTP score6 (5–12)7 (5–13)<0.0001
MELD11 (2–36)12 (2–40)0.064
Tumour characteristics
 Diameter of largest tumour nodule, cm (range)3.1 (1–17)6.7 (1–17)<0.0001
 Unifocal46 (47%)48 (38%)0.22
 Multifocal38 (39%)65 (52%)0.057
 Diffuse0 (0%)22 (17%)<0.0001
 Bilobar38 (39%)63 (50%)0.079
 Portal vein thrombosis9 (9%)53 (42%)<0.0001
 Extrahepatic metastasis3 (3%)20 (16%)0.002
Staging
 Child Pugh (A:B:C)65:20:1254:32:40<0.0001
 Okuda (I:II:III)50:41:530:65:31<0.0001
 CLIP (0:1:2:3:4:≥5)17:26:33:15:4:28:19:16:20:20:39<0.0001
 TNM (I,II,III,IV)14:32:7:442:25:4:95<0.0001
Treatment received71 (73%)45 (36%)<0.0001
Median survival (months)214<0.0001

Discussion

This is a study of patients with HCC in a major referral center in Hong Kong. The demographics of the patients were consistent with previous epidemiological studies.14 Males not only had a higher chance of HCC, they also had HCC at an earlier age than females. Similar male to female ratio in Qidong, China had also been reported in a study.15 The reason for the male predominance is not completely understood although it has been postulated that males are more likely to be infected with HBV and HCV. In addition, they tend to consume more alcohol and cigarettes, and have increased iron stores. Androgenic hormones and increased genetic susceptibility has also been proposed as the reason.

HBeAg-positive patients, when compared with HBeAg-negative subjects, had HCC at an earlier age, worse clinical symptoms and signs, more deranged liver tests, and a higher percentage of multi-focal and diffuse HCC. The median survival of the HBeAg-positive patients was only 9 months compared with 15 months for HBeAg-negative patients. The clinical picture of HCC in HBeAg-positive patients was more aggressive than that in HBeAg-negative patients. These findings are consistent with a previous study16, 17, possibly because of the higher viral load. However, it must be emphasized that the majority of patients (76%) were anti-HBe positive when they developed HCC. Thus, achieving HBeAg seroconversion may not be enough to prevent disease progression and HCC development.

In this study, we observed that HCC can still develop in patients with HBV DNA less than 105 copies/mL. In fact, 25% of the HBsAg-positive patients had HBV DNA level below 104 copies/mL. Therefore, it is important to follow all patients with chronic HBV irrespective of their HBeAg status.

HBsAg positivity, weight loss, jaundice, encephalopathy, AFP level, the presence of portal vein thrombosis and extrahepatic metastasis, and treatment were shown to be independent factors affecting survival in this study. In the subgroup of HBV patients who had HBV DNA performed at the time of diagnosis, HBV DNA levels together with portal vein thrombosis were shown to be the only independent factors affecting survival. Viral load had also been shown to be an important prognostic factor in HBV-associated HCC in another study.17

For patients who presented with symptoms, their clinical features at presentation were similar to a previous study of Hong Kong patients in the 1970–1980s.5 Since then surveillance with ultrasound and AFP in patients with chronic hepatitis had altered the presenting features. In this study, 43% of the HCC were detected by screening similar to a study on HCC performed in our center in mid-1990.7 The screening group had a higher proportion of patients with HBV, probably because HBV carriers represent a high-risk group and are more likely to undergo screening/surveillance. It is not surprising that the tumour was larger, more advanced with more PVT and extrahepatic metastasis in the non-screening group. The non-screening group had more symptoms and signs and with more deranged laboratory profiles and liver function as a result of the larger size of the tumours exerting a space-occupying effect. The better survival in the screening group was also likely to be related to better liver function, smaller tumour, less metastasis and hence better chance of receiving HCC treatment.

Patients in this cohort with chronic HBV were younger, had smaller tumours and better survival, probably related to the fact that chronic HBV patients are more likely to undergo screening. Most of the independent factors mentioned above are included in the current staging systems. There has been much debate regarding which of the existing tumour staging systems has the best prognostic value. In our cohort of HCC patients, Okuda stage, CLIP scores, and UNOS-TNM staging all had good stratification of survival across all stages of the disease. When the prognostic staging systems were analysed separately using Kaplan–Meier survival analysis, each staging system showed a significant difference in the probability of survival across the different stages. However, the prognosis and the median survival in our cohort of patients were quite different from those observed in a recent study conducted in the United States18 (Table 2). Therefore, survival data generated are not necessarily applicable to different countries or area and each staging system should be validated locally.

In conclusion, chronic HBV infection was still the major aetiological factors for the development of HCC in Hong Kong. HBeAg-positive patients had HCC at an earlier age and more deranged liver function tests, but the majority of HCC patients were negative for HBeAg when they present with HCC. Therefore, it is important to follow all high-risk patients (males >40 years, females >50 years, those with active hepatitis or cirrhosis, family history of HCC) with chronic HBV irrespective of their HBeAg status. HBV DNA level at the time of diagnosis, as shown in the sub-analysis, is an important predictor of survival.19 Nearly half of the HCC presented from screening/surveillance program. Screening appeared to be able to detect HCC at an earlier stage and therefore prolong patient survival. However, despite that the survival rate with HCC was still poor. Although all staging systems had good stratification of survival across all stages of the disease, prognosis and survival data generated form them are not necessarily applicable to different populations and each should be validated locally.

Acknowledgements

No external funding was received for this study.

Ancillary