Transarterial chemoembolization versus sorafenib in patients with hepatocellular carcinoma and extrahepatic disease

Abstract

Background

Sorafenib is the recommended treatment for advanced hepatocellular carcinoma (HCC), but transarterial chemoembolization (TACE) is performed in individual cases with limited extrahepatic spread. The aim of this study was to compare the outcome of patients with HCC and extrahepatic disease (EHD) treated with sorafenib and TACE.

Methods

A total of 172 patients with HCC and EHD treated with sorafenib (n = 98) or TACE (n = 74) at three German referral centers (Hannover, Mainz and Hamburg) were included in this study. In order to reduce selection bias, patients were matched for significant demographic differences using a propensity score analysis.

Results

Patients with liver cirrhosis, higher extrahepatic tumor burden and/or infiltration of adjacent organs/structures were significantly more often treated with sorafenib. Median overall survival (OS) was similar for sorafenib- and TACE-treated patients (7 versus 8 months, p = 0.312). In a propensity score analysis matched for demographic differences, median OS remained similar with 4 versus 8 months for sorafenib versus TACE (p = 0.613).

Conclusion

Treatment with TACE is not inferior to treatment with sorafenib in patients with limited EHD of HCC. TACE represents an effective therapeutic option in selected patients with EHD.

Introduction

Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide.^1,2 The prognosis of patients with HCC is still dismal and mortality and incidence rates are almost identical.³ In most patients, diagnosis is not made before intermediate or advanced stage disease.⁴ The latter comprises a heterogeneous population of patients with high hepatic tumor burden, vascular invasion, impaired hepatic function, reduced performance status and/or extrahepatic disease (EHD).⁵

EHD is detected in 9–13% of patients at the time of diagnosis of HCC.^4,6,7 EHD has consistently been associated with a poor prognosis.^8,9 Natural history studies and a recent meta-analysis of 30 randomized controlled trials, in which the untreated or placebo-controlled arms were evaluated, revealed a median overall survival (OS) of less than 7 months.^10–12

Since 2008, systemic treatment with the multi-tyrosine kinase inhibitor sorafenib has been established as the standard of care for patients with HCC and EHD.¹³ Within the SHARP trial, 53% of the patients treated with sorafenib had EHD. In addition, several studies revealed a consistent median OS of 10–11 months for sorafenib-treated patients with and without EHD.^14–16

In contrast to sorafenib, the role of hepatic locoregional therapy has not been established in this setting. Several studies suggest that hepatic tumor burden significantly contributes to hepatic decompensation and to death.^10,11,17,18 Prolongation of survival may be achieved by locoregional treatment.¹⁹ Based on these findings, local treatment with transarterial chemoembolization (TACE) may also be considered for patients with HCC and limited EHD according to the current German practice guidelines.²⁰

However, it remains unclear, whether sorafenib or TACE is the more effective treatment option for patients with HCC and EHD. The aim of this study therefore was (1) to characterize patients with HCC and EHD treated with sorafenib and TACE at three German referral centers, (2) to analyze treatment efficacy in terms of OS and to identify prognostic factors, and (3) to compare the OS of patients with EHD treated with sorafenib and TACE.

Patients and methods

Patient population and data selection

Patients with histologically or radiologically-diagnosed advanced HCC according to the current practice guidelines^20–22 with EHD, treated with TACE or sorafenib were included in this study. All sorafenib patients were treated at Hannover Medical School, Germany. As TACE was only performed in individual cases with EHD in contrast to treatment with sorafenib, patients treated with TACE were included from three German referral centers (Hannover Medical School, University Medical Center of the Johannes Gutenberg University Mainz, University Medical Center Hamburg-Eppendorf). Patient data were retrospectively evaluated for baseline characteristics including hepatic function, tumor burden and characteristics of transarterial procedures using clinical, imaging, laboratory and histological reports. Available laboratory values for bilirubin, albumin, prothrombin time, C-reactive protein (CRP) and alpha-feto-protein (AFP) were documented. OS was analyzed either from start of sorafenib or from first TACE until last follow up or death. Information about deaths was obtained from registration offices. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the appropriate institutional ethical boards of Hannover, Mainz and Hamburg (January 2011). Written informed consents were not obtained from the patients in agreement with the ethical boards due to the retrospective design of the study and anonymized data analysis.

Treatment procedures

In all patients, TACE or sorafenib were regarded as the most appropriate therapy after discussion within a multidisciplinary local tumor board. The standard of care was conventional TACE (cTACE) or TACE using drug-eluting beads (DEB-TACE). cTACE was performed using doxorubicin, cisplatin and/or mitomycin mixed with lipiodol either alone or in combination with degradable starch microspheres (Spherex® or polyvinyl alcohol (PVA) embolization particles, Contour®). Sorafenib was administered at a dose of 2 × 200 mg twice daily.

Contraindications for re-treatment/continuation of treatment with sorafenib or TACE were progressive disease or intolerance.

Statistical analysis

Statistical analyses were performed using SPSS 24.0 (SPSS Inc., Chicago, IL, USA). A p-value <0.05 was considered significant. Data were expressed as number/percentages or the median with the interquartile range (IQR). Differences between categorical variables were calculated using Pearson's Chi-square test. OS was assessed using the Kaplan–Meier estimation. Comparison was made using the Log rank (Mantel–Cox) test. Parameters with a p-value <0.05 in the univariate analyses were entered in a multivariate Cox's regression analysis. In order to reduce selection bias and potential confounders of the therapeutic effect, the propensity score matching (PSM) method was applied. Significantly different baseline variables were included in the propensity score model. In cases of more than two values, a dummy transformation was performed. A propensity score for each patient was calculated by logistic regression. Patients that were treated with sorafenib or TACE were optimally matched at a one-to-many-ratio using the nearest neighbor matching method. Distributions of these covariates within the matched patient cohorts were analyzed using Pearson's Chi-square test and were examined for statistical significance.

Results

Characteristics of the study population

A total of 172 patients with HCC and EHD treated with sorafenib (n = 98, between 2007–2015) or TACE (n = 74; Hannover: n = 22, Mainz: n = 37, Hamburg: n = 15) between 2000 and 2015 were included in this study. Demographics of the study population are summarized in Table 1. Demographics and treatment details of the TACE population are summarized and differentiated by center in supplemental Table 1.

Table 1.

Demographics in total and by treatment.

	Total		TACE		Sorafenib		Chi2
Characteristics	n	%	n	%	n	%	p
Total	172	100	74	100	98	100
Sex
Male	143	83.1	65	87.8	78	79.6
Female	29	16.9	9	12.2	20	20.4	ns
Age ≥ 60 years
Yes	113	65.7	50	67.6	63	64.3
No	59	34.3	24	32.4	35	35.7	ns
Etiology
ALD	52	30.2	24	32.4	28	28.6
Hepatitis C	32	18.6	16	21.6	16	16.3
Hepatitis B	23	13.4	10	13.5	13	13.3
ALD + Hepatitis C	5	2.9	1	1.4	4	4.1
ALD + Hepatitis B	3	1.7	2	2.7	1	1.0
NAFLD	24	14.0	3	4.1	21	21.4
Hemochromatosis	2	1.2	1	1.4	1	1.0
Autoimmune hepatitis	1	0.6	1	1.4	0	0
No liver disease	4	2.3	4	5.4	0	0
Cryptogenic	12	7	8	10.8	4	4.1
Others	3	1.7	0	0	3	3.1
Unknown	17	9.9	4	5.4	7	7.1	<0.05
Cirrhosis
Yes	123	76.9	50	68.5	73	83.9
No	37	23.1	23	31.5	14	16.1	<0.05
CPS
A	73	42.4	37	50.0	36	36.7
B	40	23.3	24	32.4	16	16.3
C	0	0	0	0	0	0
Unknown/no cirrhosis	59	34.3	13	17.6	46	46.9	ns
ALBI
1	24	21.2	11	19.0	13	23.6
2	75	66.4	39	67.2	36	65.5
3	14	12.4	8	13.8	6	10.9	ns
AFP
<200 µg/l	65	52.8	25	53.2	40	52.6
≥200 µg/l	58	47.2	22	46.8	36	47.4	ns
CRP
≤8 mg/l	56	38.1	25	41.7	31	35.6
>8 mg/l	91	61.9	35	58.3	56	64.4	ns
Platelets
≤150 Thsd/µl	72	44.4	33	45.2	39	43.8
>150 Thsd/µl	90	55.6	40	54.8	50	56.2	ns
Number of lesions (hepatic)
0	4	2.3	0	0	4	4.4
1	36	23.7	18	29.0	18	20.0
2	17	11.2	8	12.9	9	10.0
3	11	7.2	8	8.1	6	6.7
>3	84	55.3	31	50.0	53	58.9	ns
Tumor size (hepatic)
>7cm	54	61.2	27	40.9	27	37.0
≤7cm	84	38.8	39	59.1	46	63.0	ns
Number of lesions (extrahepatic)
1	67	48.9	31	64.6	36	40.4
2	12	8.8	4	8.3	8	9.0
3	6	4.4	2	4.2	4	4.5
>3	52	38.0	11	22.9	41	46.1	<0.05
Tumor size (extrahepatic)
>2cm	53	51.0	22	44.0	36	66.7
≤2cm	51	49.0	28	56.0	18	33.3	<0.01
Localization (extrahepatic)
PUL	54	31.4	25	33.8	29	29.6
Non-PUL	18	10.5	8	10.8	10	10.2	ns
Portal vein invasion
Yes	42	25.1	18	25.0	23	24.2
No	125	74.9	54	75.0	72	75.8	ns

AFP: alpha-feto-protein; ALBI: albumin-bilirubin; ALD: alcoholic liver disease; CPS: Child–Pugh score; CRP: C-reactive protein; NAFLD: non-alcoholic fatty liver disease; ns: not significant; PUL: pulmonary; TACE: transarterial chemoembolization.

The first diagnosis of HCC was made at a median age of 63 years (25–75 IQR, 56.25–69.75) in mainly male patients (83.1%). The underlying liver disease was alcoholic liver disease (ALD) in most cases (30.2%), followed by viral hepatitis C (18.6%) and B (13.4%) and non-alcoholic steatohepatitis/fatty liver disease (14%). Liver cirrhosis was diagnosed in 50 patients by imaging, biochemistry and/or histology (76.9%). Hepatic function was sufficiently preserved with most patients classified as Child–Pugh score (CPS) A/B (42.4%/ 23.3%) and albumin-bilirubin (ALBI) grade 1/2/3 (21.2%/66.4%/12.4%). There was no patient classified as CPS C.

Intrahepatic tumor burden was high with multiple lesions (median number of lesions of >3, 25–75 IQR, 1 to ≥4) and a median size of the largest lesion of 5.6 cm (3.3–9.2). Portal vein invasion was evident in 42 patients (25.1%). Median AFP value was 128 µg/l (8.75–4031.25). Overall, 61.9% (n = 91) of the patients had CRP levels >8 mg/l (median 12.6 mg/l, 5.2–34), which is associated with a poor tumor-related prognosis in HCC.^23–25 Regarding extrahepatic spread, most patients had only one metastasis (n = 67; 48.9%) with a median size of 2.1 cm (1.3–3.4). Pulmonary metastases were the most common site of EHD (n = 54, 31.4%), followed by lymph node (n = 48, 27.9%) and bone metastases (n = 18; 10.5%). Infiltration of adjacent organs or structures was found in 15 patients (8.7%).

Patients treated with sorafenib were significantly more often diagnosed with cirrhosis compared with TACE-treated patients (84.9% versus 68.5%, p < 0.05). Moreover, extrahepatic tumor burden was significantly higher in number (p < 0.05) and size (p < 0.01) of extrahepatic lesions in sorafenib- versus TACE-treated patients. The site of EHD was in general similar between sorafenib- and TACE-treated patients, but only 2 TACE-treated patients had infiltration of adjacent organs or structures in contrast to 13 patients treated with sorafenib (13.3 versus 2.7%, p < 0.05). In addition, non-alcoholic fatty liver disease (NAFLD) was significantly more often diagnosed in sorafenib-treated patients compared with TACE-treated patients (4.1% versus 21.4%, p < 0.05).

Treatment procedures

In most patients treated with TACE, TACE was the first tumor-specific treatment (n = 57; 77.0%). Treatment with TACE was repeated in the absence of contraindications as described above at intervals of at least 8 weeks. The median number of TACE sessions was 2 (1–3). Most patients were treated with cTACE (n = 49; 73.1%), followed by DEB-TACE (n = 16; 23.9%). Each patient was treated with transarterial embolization and transarterial chemoperfusion (TAC). TACE was performed selectively by means of segmental application of the chemoembolizing agents whenever possible. Contraindications for selective TACE were multiple nodules or diffuse HCC. Overall, 29 patients (42.6%) were selectively treated, whereas TACE was not performed in a selective fashion in 39 patients (57.4%).

A total of 33 patients (44.6%) received a subsequent therapy. Among these, 31 patients (41.9%) were subsequently treated with sorafenib. Overall, four patients subsequently underwent resection of hepatic tumors. Another four patients had symptom-oriented, local treatment of metastases. Of these four patients with bone metastases, three were treated with radiation and one was treated with radiofrequency ablation. No patient underwent liver transplantation.

Within the sorafenib group, nine patients were treated at a reduced dose (seven patients at a dose of 1 × 200 mg twice daily, two patients at a daily dose of 200 mg) and in eight patients, treatment had to be interrupted due to intolerance/side effects including fatigue, diarrhea and hand-foot-syndrome. Re-escalation to the recommended dose of 2 × 200 mg twice daily was possible only in one patient after improvement of supportive therapy (skin moisturizer containing urea).

A total of 25 patients with sorafenib treatment were previously treated with TACE. Only seven patients (7.2%) had any subsequent treatment including TACE in five patients and a second-line systemic treatment within clinical trials in two patients. Overall, five patients underwent resection of extrahepatic spread including adrenal and pulmonary metastases and three patients with bone metastases were treated with radiation.

Overall survival and prognostic factors

Median OS was 7 months in the total study population. Within the univariate analysis including clinical and laboratory data, CPS A, AFP ≥ 200 µg/l and CRP > 8 mg/l were identified to be significantly associated with OS (Table 2). CRP > 8mg/l remained significantly prognostic in the multivariate analysis (hazard ratio (HR) 1.86, p < 0.05) (Figure 1(a)). There was a prognostic trend for CPS A (HR 0.44, p = 0.084) and AFP ≥ 200 µg/l (HR 1.48, p = 0.127) without reaching statistical significance (Figure 1(a)). Median OS was similar for patients treated with sorafenib or TACE with a median OS of 7 months and 8 months, respectively (p = 0.312) (Figure 1(b)).

Table 2.

Univariate analysis of prognostic factors.

Characteristics	n	mOS (months)	p
Total	171	7
Sex
Male	142	7
Female	29	8	ns
Age ≥ 60 years
Yes	112	6
No	59	10	ns
Etiology
ALD	51	7
Non-ALD	109	7	ns
NAFLD	24	9
Non-NAFLD	136	7	ns
Hepatitis C	32	7
Non-hepatitis C	128	7	ns
Cirrhosis
Yes	123	7
No	36	13	ns
CPS
A	73	8
Non-A	40	5	<0.001
ALBI
1	24	10
Non-1	89	5	ns
3	14	5
Non-3	99	6	ns
AFP
<200 µg/l	65	13
≥200 µg/l	57	4	<0.001
CRP
≤8 mg/l	55	14
>8 mg/l	91	5	<0.001
Platelets
≤150 Thsd/µl	72	7
>150 Thsd/µl	89	7	ns
Number of lesions (hepatic)
1–3	67	8
>3	84	6	ns
Tumor size (hepatic)
>7cm	54	5
≤7cm	84	5	ns
Number of lesions (extra- hepatic)
1	66	10
≥2	58	5	ns
Tumor size (extrahepatic)
>2cm	53	7
≤2cm	50	10	ns
Localization (extra- hepatic)
PUL	27	7
Non-PUL	46	7	ns
Portal vein invasion
Yes	42	5
No	124	8	ns

AFP: alpha-feto-protein; ALBI: albumin-bilirubin; ALD: alcoholic liver disease; CPS: Child–Pugh score; CRP: C-reactive protein; NAFLD: non-alcoholic fatty liver disease; ns: not significant; PUL: pulmonary; mOS: median overall survival.

Figure 1.

Survival and prognostic factors. (a) Multivariate Cox's regression analysis. (b) Cumulative survival estimate to death from first TACE or start of sorafenib by treatment. (c) Cumulative survival estimate to death from first TACE or start of sorafenib by treatment in the propensity-score-matched populations.

AFP: alpha-feto-protein; CI: confidence interval; CPS: Child–Pugh score; Cum.: cumulative; CRP: C-reactive protein; HR: hazard ratio; TACE: transarterial chemoembolization.

Propensity score match (PSM) analysis

One-to-many PSM was applied to reduce selection bias including 91 patients. Etiology of the liver disease, cirrhosis and number and size of extrahepatic tumor were included in the analysis as these were significantly different baseline variables. After PSM, these factors were balanced and differed no longer (all p > 0.05, Table 3). In the matched cohorts, median OS was similar with 4 versus 8 months for sorafenib- versus TACE-treated patients (p = 0.613, Figure 1(c)). There was a trend for a prolonged median OS in the TACE group for CRP (≤8 mg/l) and AFP (<200 µg/l) low patients: 6 versus 17 months (CRP ≤ 8 mg/l, p = 0.505) and 6 versus 20 months (AFP < 200 µg/l, p = 0.266). There was no difference in median OS for CPS A patients (9 versus 11 months, p = 0.884).

Table 3.

Demographics in total and by treatment following propensity score matching.

	Total		TACE		Sorafenib		Chi2
Characteristics	n	%	n	%	n	%	p
Total	91	100	73	100	18	100
Etiology
NAFLD	5	5.5	3	4.1	2	11.1
Non-NAFLD	82	90.1	67	91.8	15	83.3
Unknown	4	4.4	3	4.1	1	5.6	ns
Cirrhosis
Yes	64	70.3	50	68.5	14	77.8
No	27	29.7	23	31.5	4	22.2	ns
Number of lesions (extra- hepatic)
≤3	48	72.7	37	77.1	11	61.1
>3	18	27.3	11	22.9	7	38.9	ns
Tumor size (extrahepatic)
>2cm	27	39.7	17	34.0	10	55.6
≤2cm	41	60.3	33	66.0	8	44.4	ns

NAFLD: non-alcoholic fatty liver disease; ns: not significant; TACE: transarterial chemoembolization.

Discussion

In this multi-center study, we compared the characteristics and outcomes of sorafenib- and TACE-treated patients with HCC and EHD. Median OS was similar for sorafenib and TACE treatment (7 versus 8 months) and remained similar after PSM for demographic differences. In selected patients with a prognostically good tumor biology as defined by low AFP and CRP levels, treatment with TACE resulted in a prolonged median OS of up to 20 months despite presence of EHD.

Prognosis of patients with HCC and EHD has consistently been reported to be very poor. Natural history studies and one recent meta-analysis of 30 randomized controlled trials of HCC, which evaluated the untreated/placebo arms, revealed a median OS of <7 months and emphasized the need for optimized therapeutic approaches.^10–12

Since 2008, sorafenib has been available for patients with advanced stage disease and has improved median OS to 10 months. Its efficacy in patients with EHD has been investigated in several pro- and retrospective studies.^13–16 In our study investigating patients with EHD, median OS was 7 months in the sorafenib group. In contrast with the above-referred studies, at least 16.3% of the sorafenib-treated patients from our cohort were classified as CPS B. There were mostly CPS A patients in the SHARP trial (95%)¹³ and in both the retrospective trials by Kawaoka et al. (100% of the patients were CPS A)¹⁵ or by Sohn et al. (92% of the patients were CPS A).¹⁶ In addition, our patients were characterized by high extrahepatic tumor burden. Compared with the retrospective study performed by Sohn et al., which revealed a median OS of 9.6 months, 78% of the patients treated with sorafenib had only one extrahepatic metastasis, whereas most of our patients (58.9%) had more than three lesions. In line with our findings, lung metastases were also the most common extrahepatic metastatic site within comparable studies.^14–16

In contrast to sorafenib, the role of locoregional therapy is not established in the setting of EHD. Following the recommendation of the Barcelona Clinic Liver Cancer (BCLC), which is the most frequently applied classification system, locoregional therapies are not recommended for patients with EHD.²⁶ However, natural history studies of HCC suggest that intrahepatic tumor burden significantly contributes to hepatic decompensation and to death and not extrahepatic spread.^10,11 Among others, a retrospective study including 240 patients with HCC and extrahepatic metastases, who were treated with locoregional and/or systemic treatment between 2004 and 2009, revealed progressive intrahepatic tumor as the leading cause of death¹⁸ providing a rationale to consider locoregional therapy even in the presence of EHD. By now, several studies have been performed that have investigated TACE in patients with HCC and EHD and have reported efficacy in this specific subgroup of patients. Accordingly, following the recommendation of the German practice guideline, TACE may be considered for patients with limited EHD.²⁰ Median OS times ranging from 8–13 months have been reported for patients with EHD treated with TACE: Median OS was 11.9 months within one large, retrospective study including 508 patients with advanced stage disease treated with TACE, among which 84 patients were diagnosed with EHD.⁹ A prospective Chinese trial, which compared TACE with TACE with ginsenoside Rg3 in patients with advanced HCC, revealed a median OS of 10.1 months for TACE in 76 treated patients, among which 32 patients had EHD.²⁷ Another small retrospective study compared TACE in patients with and without extrahepatic spread of HCC and revealed a considerable median OS of 13 months for patients with EHD (n = 39).⁸ The patients included in this study however had an excellent hepatic function with no liver cirrhosis in 22.7% of the patients and with only 2.2% of patients classified as CPS B, whereas the remaining patients were classified as CPS A. Similar to our results, another group showed a median OS of 8 months in 177 patients with EHD treated with TACE.²⁸ It is important to note that prognostically good patients (low AFP) treated with TACE had a median OS of 20 months in our study. Finally, the broad range of OS found in the literature may also be explained by the lack of standardization in procedures, repetitions and intervals of TACE treatments.²⁹

So far, there are only few studies comparing outcome of patients with advanced HCC treated with TACE or sorafenib.^30,31 Within a retrospective study including patients with any advanced disease – either with or without EHD – 55 patients were treated with TACE and 56 were treated with sorafenib. Median OS was 6.6 months in the TACE group and 9.2 months in the sorafenib group.³¹ Another retrospective analysis included 97 patients with advanced disease stage with or without EHD. The median OS was 9.2 months for patients treated with TACE and 7.4 months for those treated with sorafenib. This difference was non-significant.³⁰ However, in contrast to our analysis, these studies have not particularly included patients with EHD.

We have found no survival difference between sorafenib and TACE treatment in patients with HCC and EHD. Patients with more and larger extrahepatic tumor and cirrhosis were more often selected for treatment with sorafenib than treatment with TACE. In order to reduce this selection bias, we have performed a PSM analysis including these factors. In the matched cohorts with low levels of AFP and CRP, treatment with TACE resulted in a prolonged survival of up to 20 months.

One limitation of our study is its retrospective nature and the limited number of patients. By performing a PSM analysis, possible selection bias in the sorafenib and TACE group has been reduced. An additional selection bias caused by TACE treatments within three centers cannot completely be ruled out.

Conclusions

In conclusion, we provide evidence for efficacy of both sorafenib and TACE in patients with HCC and EHD. TACE was not inferior to sorafenib. Patients with a good prognosis defined by low AFP and CRP levels may benefit from treatment with TACE.

• Sorafenib is the standard of care for patients with HCC and EHD.

• Hepatic tumor burden significantly contributes to hepatic decompensation and to death providing rationale to consider locoregional therapy even in the presence of EHD.

• Treatment with TACE is not inferior to treatment with sorafenib in patients with HCC and limited EHD.

• TACE represents an effective therapeutic option in selected patients with EHD.

Funding

Martha M. Kirstein was supported by the Ellen Schmidt program from Hannover Medical School, Germany.

Declaration of conflicting interests

The authors declare the following conflicts of interest: AV and HW have received honoraria for Advisory Boards and speakers activities from Bayer. AW has received research funding and honoraria for speakers activities.

Ethics approval

The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the appropriate institutional ethical boards of Hannover, Mainz and Hamburg (January 2011).

Informed consent

Written informed consents were not obtained from the patients in agreement with the ethical boards due to the retrospective design of the study and anonymized data analysis.