Inverse relationship between cirrhosis and massive tumours in hepatocellular carcinoma

Abstract

Background

A subset of patients with hepatocellular carcinoma (HCC) present with massive tumours. It is unknown why certain patients develop these massive tumours, and whether this presentation is specific to the underlying viral aetiology or patient demographics such as gender, race and age.

Methods

All patients with HCC at Bellevue Hospital Center, New York from 1998 to 2012 were identified and relevant demographic and clinical information was collected. Computed tomography/magnetic resonance imaging (CT/MRI) images were reviewed and the maximal tumour diameter on axial sections was recorded. Cirrhosis was defined histologically or by radiographical criteria. The two cohorts of massive and non-massive HCC were compared.

Results

A total of 361 patients with HCC were identified, of which 58 were categorized as having a massive HCC using a 13 cm size cut-off. Univariate and multivariate analysis demonstrated a significant association of massive HCC with age <40 years; hepatitis B or Asian ethnicity; and a lack of cirrhosis or platelet count >100.

Discussion

Massive HCC represents a tumour subtype that is associated with young, chronic hepatitis B carriers with non-cirrhotic livers. The clinical implications of this finding are that patients with massive HCC are typically excellent resection candidates barring the presence of gross vascular invasion or distant metastases.

Background

In the United States and Europe, hepatocellular carcinoma (HCC) is almost always a consequence of cirrhosis, most frequently as a result of hepatitis C infection (HCV).¹ Carcinogenesis occurs as a result of chronic hepatocyte injury, leading to inflammation, cell-regeneration and eventual dysplasia. By the time HCC develops, patients frequently have portal hypertension and poor hepatic function, severely limiting treatment options.²

By contrast, in most of Asia and sub-Saharan Africa, HCC is generally as a result of infection with the hepatitis B virus (HBV).¹ As a DNA virus, hepatitis B is thought to induce carcinogenesis by viral integration into the hepatocyte genome leading to viral promoter-driven transcriptional up-regulation, and induction of genomic instability.³ As a result, patients with chronic HBV infection can develop HCC in the absence of cirrhosis.⁴ Furthermore, patients from these regions generally develop HCC much earlier, since the HBV infection occurs perinatally, and thus the hepatic insult begins at a younger age.⁴

On imaging, HCC and can vary greatly in its size at presentation and in its rate of growth. Tumour size is an important factor for treatment decisions. For example, a size greater than 5 cm disqualifies a patient from liver transplantation⁵ and makes ablative treatments less effective.² A large size also renders transarterial chemoembolization ineffective, since massive tumours usually parasitize their blood supply from several sites in addition to the hepatic artery. While the efficacy of sorafenib has not been specifically tested in patients with massive HCC, there is some hesitancy that its use in such large tumours can induce tumour necrosis and rupture. Therefore, hepatic resection remains the only treatment for HCC for which there is no strict size restriction, although large tumours can clearly make surgery more challenging owing to their mass effect.

A subset of patients with HCC present with markedly large tumours which can exceed 25 cm in diameter (Fig. 1). It is unknown why certain patients develop these massive tumours, and whether this presentation is specific to the underlying viral aetiology or patient demographics such as gender, race and age. The present study describes the demographic and clinical features associated with massive tumours in hepatocellular carcinoma. The inverse relationship between cirrhosis and these massive tumours is noted, and the treatment options and outcomes for these patients are reviewed.

Figure 1.

Massive hepatocellular carcinoma (HCC) in a non-cirrhotic liver

Methods

After obtaining Institutional Review Board approval, billing and diagnostic codes were used to identify all patients with HCC at Bellevue Hospital Center in New York City from January 1998 to January 2012. A retrospective chart review was then performed. A diagnosis of HCC was confirmed by either histology or established radiological criteria of arterial enhancement and venous washout.² Associated socio-demographic data were collected, including age, gender and race/ethnicity. Race/ethnicity was determined using either self-declared ethnicity or documentation by the healthcare provider, as indicated in the electronic medical records.

Baseline laboratory values recorded upon diagnosis included total bilirubin, platelets, international normalized ratio (INR), creatinine and albumin. The presence of underlying liver disease was noted, including HBV, HCV, alcohol abuse, HIV infection, diabetes and non-alcoholic fatty liver disease. Hepatitis B diagnosis was determined by positive serology results for HBV-surface antigen or core antibody. Regular alcohol use (i.e. more than one drink per day) was considered significant.

All computed tomography (CT) or magnetic resonance imaging (MRI) images were reviewed by one investigator (D.A.), and the results independently confirmed by a single hepatobiliary surgeon (U.S.). Cirrhosis was defined either histologically or using radiographical criteria (i.e. nodularity or other changes in liver morphology, splenomegaly, the presence of varices or ascites). The maximal tumour diameter on axial sections was recorded as a simple surrogate for tumour volume. Gross vascular invasion, tumour number, and/or distant metastases were noted.

Method of treatment was recorded, including resection, transplantation, chemoembolization, ablation, sorafenib and other modalities. In patients who underwent a resection, the histological appearance of the tumour was examined.

Statistical analysis was performed using SPSS version 17.0 and R package (R Development Core Team).⁶ Reports in the literature suggested that massive tumours tended to occur in patients without cirrhosis.^7,8 This was investigated by plotting the distribution of tumour size by the presence or absence of cirrhosis, and demonstrated that the majority of tumours in cirrhotic patients were smaller than 10 cm (Fig. 2). A preliminary chi-square test was performed using 10 cm as a cut-off, suggested by previous publications on massive tumours.⁸ This confirmed that the incidence of cirrhosis was significantly less in patients with tumours smaller than 10 cm in size (P < 0.05). To more accurately define the tumour size at which point the association with hepatic cirrhosis became significant, serial tumour size cut-offs were used to create dichotomized groups. Chi-square tests were used to compare the incidence of cirrhosis for each cut-off. The value corresponding to the statistically significant divergence in rates of cirrhosis resulted in a definition of ‘massive HCC’ as ≥13 cm. This size corresponded well to the size cut-off seen with the other clinical and demographic features known to be associated with a lack of cirrhosis, e.g. young age, platelets >100 000, hepatitis B and Asian ethnicity⁴ (Fig. 3). Using this resulting definition of 13 cm, the two cohorts of massive and non-massive HCC were compared. Univariate analysis was conducted using chi-square tests for categorical variables, and Mann–Whitney U-tests for non-parametric continuous variables. Normality was examined using quantile–quantile plots and the Kruskal-Wallis test. Logistic regression multivariate analysis was conducted in the standard fashion using variables that were clinically important base on a priori knowledge. Multivariate models were constructed with independent variables that were not associated with each other.

Figure 2.

Weighted plot of tumour size distribution by the presence or absence of cirrhosis

Figure 3.

A plot of P-values derived from chi-square tests comparing various cut-offs for tumour size and the presence of cirrhosis, age, hepatitis B, Asian race/ethnicity and thrombocytopenia, demonstrating a divergence in significance above 13 cm

Survival was measured in months from the date of HCC diagnosis. Death was confirmed by the Social Security Death Index or our institutional Tumor Registry. As a result of the small number of patients with massive HCC undergoing resection, survival was dichotomized as survival >1 year, and survival >2 years. Fisher's exact test was used to compare rates between the massive and non-massive cohorts.

Results

A total of 361 patients with HCC were identified (Table 1). The mean age at HCC diagnosis was 55 years [standard deviation (SD) ± 13]. Eighty-three per cent (n = 300) were male. In 355 patients, information on ethnicity was available. Patients defined themselves as Asian (48%) Hispanic (25%), African American (18%), Caucasian (3%) or other (4%).

Table 1.

Features of massive vs. non-massive HCC

	Massive (n = 58)	Non-massive (n = 303)	P-value
Demographical features
Age	47 years	57 years	<0.001
Male gender	51 (88%)	249 (82%)	NS
Asian ethnicity	45 (78%)	129 (42%)	<0.001
Clinical features
Hepatitis B	48 (83%)	139 (46%)	<0.001
Radiologic cirrhosis	16 (29%)	216 (72%)	<0.001
Platelets <100	4 (7%)	92 (30%)	<0.001
Tumour features
Gross vascular invasion	31 (53%)	58 (19%)	<0.001
Extrahepatic metastases	9 (16%)	16 (5%)	<0.008
Treatment
No treatment	29 (53%)	105 (35%)	<0.001
Sorafenib	13 (24%)	19 (6%)
Chemotherapy	3 (5%)	24 (8%)
Embolization	3 (5%)	47 (16%)
Ablation	0	34 (11%)
Radiation	1 (2%)	0
Transplantation	0	4 (1%)
Resection	6 (11%)	66 (22%)

Aetiology of the underlying liver disease was defined as HBV in 52% (n = 187/361), HCV in 35% (n = 125/361) alcohol without HCV in 3%, diabetes or non-alcoholic fatty liver disease in 3%, multiple aetiologies in 4%, and in 4% (n = 13) of patients no identifiable aetiology was documented. In the 307 patients where information on symptoms was available, 45% of patients (n = 163/307) were asymptomatic at diagnosis, 35% (n = 126/307) complained of abdominal pain, 2% (n = 8/307) presented with weight loss, 1% (n = 2/307) presented with jaundice, and 2% (n = 8/307) had other complaints which led to abdominal imaging.

At diagnosis, the median tumour size was 4.7 cm with a positive skew, and ranged from 1.0 to 24.9 cm. Unifocal tumours were present in 193 patients (53%), 61 patients (17%) had two tumours, 14 patients had three tumours (4%) and 93 patients (26%) had greater than three tumours present at diagnosis. Gross vascular invasion into the portal or hepatic veins was present in 89 patients (24.7%) at the time of diagnosis. Using the Child–Pugh scoring system to assess liver function, 254 patients (72%) were class A, 87 patients (25%) were class B and 13 patients (4%) were class C.

Using the definition of ≥13 cm obtained as described in the methods, 58 of the 361 patients (16.1%) were categorized as having massive HCC. Univariate analysis demonstrated significant differences between these two cohorts, as shown in the Table 1.

Multivariate analysis with logistic regression was performed. To avoid multicollinearity, Asian race/ethnicity and HBV infection were not used in the same model as these factors were found to be associated. Similarly, cirrhosis and thrombocytopenia were not used in the same model as these factors were also associated. Logistic regression demonstrated a significant association of massive HCC with age <40 years [odds ratio (OR) = 0.22, P < 0.001], hepatitis B (OR = 5.8, P < 0.001) or Asian ethnicity (4.5, P < 0.001), and lack of cirrhosis (OR = 0.22, P < 0.001) or platelet count >100 (OR = 4.7, P = 0.004).

Treatment modality also differed significantly between the cohorts, as shown in the Table 1. In looking specifically at the outcome of all patients who underwent a resection, of the six patients with massive HCC who underwent a resection, five were alive at 1 year, and two were alive at 2 years from the date of diagnosis. This was compared with the 66 patients in the non-massive group who underwent a resection, of whom 49 (74%) were alive at 1 year and 39 (59%) who were alive at 2 years. Fisher's exact test did not demonstrate any difference in the likelihood of survival to 1 or 2 years from the date of diagnosis. Resection specimens were available from three patients with massive HCC. These specimens were notable for the presence of a pseudocapsule encompassing the tumour (Fig. 4).

Figure 4.

An example of the tumour pseudocapsule seen with massive HCC

Discussion

The term ‘massive HCC’ was first used by Eggel in 1901 to describe large tumours which replaced much of the entire right or left hepatic lobe.⁹ This term was later used by Yuki and colleagues in 1990 to describe tumours which were poorly demarcated from the surrounding parenchyma.⁷ While this definition was irrespective of tumour size, the majority of patients which they categorized as massive HCC were larger than 10 cm.⁷ A greater proportion of patients with massive HCC were younger than 40 years of age and more often had extrahepatic metastases when compared with other tumour types – consistent with the current findings.⁷ In 1993, Trevisani and colleagues reported on patients with massive HCC, defined by a tumour greater than 10 cm with poorly demarcated borders, and found a significant association of massive tumours with non-cirrhotic livers, as confirmed by the present findings.⁸ More recently, Barazani and colleagues reported that HBV patients presented with larger maximum tumour diameters, and more HBV patients had tumours greater than 5 cm compared with HCV patients.¹⁰

The aim of this study was to confirm the inverse relationship between cirrhosis and massive HCC and to more accurately describe massive HCC. The present results reveal that patients with tumours larger than 13 cm can be defined as having massive HCC and that this tumour subtype is associated with young, chronic hepatitis B carriers with non-cirrhotic livers. This number should not be thought of as a strict size cut-off, but instead reflects the point at which the clinician can be confident that tumour characteristics may be different than expected.

The clinical implications of this finding are that patients with massive HCC are typically excellent resection candidates barring the presence of gross vascular invasion or distant metastases. In the current series, only a small number of patients with massive HCC underwent resection, but the results in these select patients were reasonable. By comparison, in a report on the natural history of Chinese patients with large HCC tumours (>5 cm), the median survival was only 2.7 months.¹¹

The present study was not designed to address the underlying mechanism of the development of massive HCC. However, the authors hypothesize that the compliant tissue of the non-cirrhotic liver may offer less resistance to expansive tumour growth.

In conclusion, patients with HCC tumours greater than 13 cm in size should be considered as having massive HCC, and should be evaluated in this context, understanding that their aetiology and treatment options differ from the majority of patients with HCC. Massive HCC represents a tumour subtype that is associated with young, chronic hepatitis B carriers with non-cirrhotic livers. Treatment options are limited in these individuals; however, the absence of cirrhosis makes them excellent resection candidates when metastases or prohibitive vascular invasion are not present. It is critical to aggressively pursue resection in candidate patients, as no other treatment options exist for tumours of such a large size.

Conflicts of interest

None declared.