Stereotactic body radiation therapy for inoperable hepatocellular carcinoma as a local salvage treatment after incomplete transarterial chemoembolization




The objective of this study was to evaluate the efficacy and safety of stereotactic body radiation therapy (SBRT) as a local salvage treatment after incomplete transarterial chemoembolization (TACE) for inoperable hepatocellular carcinoma (HCC).


The main eligibility criteria were a greatest tumor dimension (LD sum) <10 cm, inoperable HCC, and incomplete response after TACE. Prescribed SBRT doses were up to 60 gray (Gy) in 3 fractions, but doses were reduced until normal tissue constraints were allowed.


Between May 2008 and February 2011, 50 patients were enrolled in this phase 2 trial, of which 47 patients were evaluable. Forty-one patients had Child-Pugh class A disease (A5/A6 were 32/9), 6 patients had class B7 disease, and 5 patients had portal vein tumor thrombosis. All patients underwent TACE 1 to 5 times before SBRT. SBRT doses ranged from 42 to 60 Gy in 3 fractions (median dose, 57 Gy), and the median LD sum was 29 mm (range, 13-78 mm). Eighteen patients (38.3%) achieved complete remission within 6 months of completing of SBRT, and 18 patients (38.3%) had a partial response. The 2-year local control rate was 94.6%, the overall survival rate was 68.7%, and the progression-free survival rate was 33.8%. Three patients (6.4%) experienced grade 3 gastrointestinal toxicity, and 2 patients (4.3%) experienced grade 4 gastric ulcer perforation.


This trial demonstrated that SBRT after incomplete TACE for inoperable HCC achieves promising rates of response and local control. On the basis of these study results, a modified, multi-institutional, phase 2 trial to reduce gastrointestinal toxicity is recommended. Cancer 2012. © 2012 American Cancer Society.


The standard treatment for hepatocellular carcinoma (HCC) is surgery, including hepatic resection or liver transplantation, which result in 5-year survival rates of 30% to 70%.1-3 However, <20% of patients with HCC are suitable for surgery.4 For patients with HCC who are unsuitable for surgery, other local ablative therapies offer alternative treatment options, and of these, radiofrequency ablation (RFA) and percutaneous ethanol injection (PEI) reportedly have high response rates.5, 6 Unfortunately, not all patients are amenable to these local treatments because of tumor size or location or because they have a lesion that is not visible on abdominal ultrasonography. In these situations, transarterial chemoembolization (TACE) has been widely used as a local treatment option, but TACE alone rarely produces a complete response, which largely is attributed to the presence of remaining, viable tumor cells. However, although it has been suggested that additional treatment is required after TACE, no clear guidelines have been issued concerning suitable modalities for patients with HCC.

Stereotactic body radiation therapy (SBRT) can now be used to accurately deliver radiation to tumors at high doses in a small number of fractions and spare surrounding tissues. Furthermore, the liver is regarded as a parallel organ; thus, the delivery of high radiation doses to selected liver regions is considered safe. Blomgren et al pioneered the use of SBRT for the treatment of liver malignancies7; and subsequently, several authors reported high tumor response and local control rates after SBRT for intrahepatic tumors, including metastases.8-10 However, the safety and feasibility of SBRT for primary HCC have not been determined, and clinical data are limited.

In a previous study, we confirmed the feasibility of SBRT for the treatment of primary HCC and observed high rates of local control (LC) and overall survival (OS) duration and a low rate of treatment-related severe toxicity.11 Because of those results, we undertook the current phase 2 trial of SBRT as a local salvage treatment after incomplete TACE in patients with HCC, and we focused on tumor response and LC rates.


Patient Eligibility

This phase 2, nonrandomized study was approved by the Institutional Review Board of the Korea Cancer Center Hospital. All patients provided written informed consent before participating in the study. During the study design stage, we presumed a response rate improvement from 40% to 60%. Sample size was calculated with a significance level of .05 and a power of 80% by using the Simon optimal 2-stage design. With this design, the study treatment would be acceptable if >23 of 46 patients responded to treatment. Assuming loss to follow-up and potential ineligibility, the final sample size required was 50 patients. Accordingly, 50 patients with inoperable HCC who were treated between May 2008 to February 2011 were enrolled in this study.

The study eligibility criteria were: 1) age >18 years; 2) an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1; 3) no previous abdominal radiation therapy (RT); 4) an initial diagnosis of primary HCC or recurrence; 5) an inoperable disease status or refusal to undergo surgery; 6) unsuitability for RFA or PEI; 7) leukocyte count ≥2000/μL, hemoglobin level ≥8.0 g/dL, platelets >50,000/μL, absolute neutrophil count ≥1500/μL, aspartate and alanine aminotransferase levels <5 times upper normal limit, bilirubin ≤3.0 mg/dL, and prothrombin time-international normalized ratio ≤2.3; 8) Child-Turcott-Pugh (CTP) class A or B7 disease; 9) an incomplete response after TACE after 1 to 5 attempts; 10) a single lesion or multiple lesions, including portal vein tumor thrombosis, so attached that they could be viewed as 1 lesion for treatment purposes; 11) a greatest tumor dimension (LD) <10 cm; and 12) no evidence of an uncontrolled lesion at any other site.

Stereotactic Body Radiation Therapy Technique

All patients underwent TACE, and TACE could be repeated up to 5 times with a 1-month interval if the TACE produced a response; however, patients were considered for SBRT when it was determined that TACE was no longer effective, which was defined as incomplete tumor filling by the lipiodol-doxorubicin mixture used by either angiography or computed tomography (CT) 1 month after TACE or by an increasing alpha-fetoprotein level. The patients were registered post-TACE, and the interval from TACE to SBRT was from 1 to 2 months.

Six gold fiducials (4 mm long and 0.8 mm in diameter) were inserted into the periosteum percutaneously at the transverse processes of vertebral bodies nearest the lesions. Three or 4 days later, a CT scan was performed with the patient supine and immobilized by an alpha cradle (Smithers Medical Products, North Canton, Ohio); 4 belts were used to compress the abdomen. A thin-slice CT was performed with a slice thickness of 2 mm at 3 seconds per slice, and the data were entered into the CyberKnife planning system (Accuray Inc., Sunnyvale, Calif). These relatively slow CT images included respiratory movement of the target; therefore, the tumor volumes used during planning were larger than the gross tumor volumes (GTVs) and are referred to as internal target volumes (ITVs).12, 13 The planning target volume (PTV) was defined as ITV + 4 mm in the longitudinal direction and ITV + 2 mm in all other directions. These margin sizes were determined in a previous study that was undertaken to determine the margin sizes required to compensate for respiratory movement.14 Margin modification was allowed sometimes because of constraints. SBRT doses were prescribed at an isodose line (70%-80% of the maximum dose) that covered at least 97% of the PTV.

Prescribed doses were up to 60 gray (Gy) in 3 fractions in the absence of a no dose limitation to normal tissues near the SBRT site, but dosages were reduced by 0.5 Gy or 1 Gy per fraction until normal tissue constraints were allowed. We adopted the constraint that, regardless of whether the patient had CTP class A or B disease, at least 700 mL of normal liver (entire liver minus cumulative GTV) should not receive a total dose ≥17 Gy in 3 fractions, and, for the spinal cord, the maximum dose should not exceed 22 Gy in 3 fractions and 0.25 mL or less of irradiated volume of spinal cord >18 Gy in 3 fractions. In addition, although other normal tissue constraints were not considered, dosages to the kidney, intestine, and stomach were restricted to the lowest levels possible.

Response and Toxicity Evaluations

Treatment responses were defined as the best response observed among abdominal CT scans obtained 1 month, 3 months, and 6 months after SBRT by measuring LD. Complete remission (CR) was defined as the disappearance of contrast enhancement in the tumor during the arterial phase according to modified Response Evaluation Criteria for Solid Tumors.15, 16 For tumors with dense lipiodol deposits, CR was defined as persistent lipiodol deposition without contrast enhancement. A partial response (PR) was defined as a decrease in volume of an enhancing area by >30% of the initial tumor size. Stable disease (SD) was defined as a decrease in initial tumor volume <30%, and progressive disease (PD) as defined as an increase in tumor volume ≥20%. Toxicities were evaluated using the Common Terminology Criteria of Adverse Events (CTCAE) version 3.0.

Endpoints and Statistics

The primary endpoint was the objective response rate, which was determined by CT studies obtained 1 month, 3 months, and 6 months after SBRT. The objective response rate was defined as the sum of CRs and PRs. Secondary endpoints were 2-year LC, OS rate, progression-free survival rate (PFS), pattern of failure, treatment-associated toxicity, and prognostic factors. LC, OS, and PFS were measured from the date SBRT was started using the Kaplan-Meier method. Intergroup comparisons were performed using the log-rank test. Any factor that influenced prognosis in univariate analysis was subjected to multivariate analysis using a Cox proportional hazards regression model with a backward, conditional, stepwise procedure to determine whether factors acted independently. All calculations were performed using SPSS statistical software (version 13.0; SPSS, Inc., Chicago, Ill), and P values < .05 were considered statistically significant.


Fifty patients with inoperable HCC who were treated between May 2008 and February 2011 were enrolled in this phase 2 trial. Two of the 50 enrolled patients were lost to follow-up <6 months after SBRT, and 1 patient achieved a complete response after TACE. These 3 patients were deemed nonevaluable. Finally, data from 47 patients were analyzed. Table 1 summarizes the demographic characteristics of the 47 evaluable patients.

Table 1. Patient Characteristics (n = 47)
CharacteristicNo. of Patients
  1. Abbreviations: AJCC American Joint Committee on Cancer; BCLC, Barcelona Clinic Liver Cancer; CLIP, Cancer of the Liver Italian Program; SBRT, stereotactic body radiation therapy; TACE, transarterial chemoembolization.

Diagnosis history at SBRT 
 Initially diagnosed20
 Diagnosed as recurrence27
 Hepatitis B virus32
 Hepatitis C virus6
Child-Turcott-Pugh class 
Pre-SBRT platelet count: Median (range), K/μL103 (55-299)
AJCC tumor classification 
BCLC stage 
Okuda stage 
CLIP score 
Sum of greatest tumor dimension: Median (range), mm29 (13-8)
Internal target volume: Median (range), cm314.9 (2.4-213.8)
No. of tumors 
Portal vein tumor thrombosis 
No. of prior TACE: Median (range)2 (1-5)

The median follow-up duration was 17 months (range, 6-38 months). For the patients who remained alive (n = 35), the median follow-up was 22 months. Of the 47 patients, 37 (78.7%) were men, and 10 (21.3%) were women. Twenty patients (42.6%) were diagnosed initially with inoperable HCC, and 27 patients (57.4%) were diagnosed with recurrent disease. The majority of patients had CTP class A disease (87.2%), and the median LD sum was 29 mm (range, 13-78 mm). SBRT doses ranged from 42 to 60 Gy in 3 fractions. Twenty-one patients received 60 Gy in 3 fractions, 18 patients received from 51 to 57 Gy in 3 fractions, and the other 8 patients received from 42 to 48 Gy in 3 fractions (Table 2). The equivalent dose range of 42 to 60 Gy delivered in 3 fractions was calculated as 84 to 150 Gy using an α/β ratio to 10 Gy in a linear-quadratic model.

Table 2. Prescribed Stereotactic Body Radiation Therapy Doses According to the Sum of Greatest Tumor Dimensions
 SBRT Dose, Gy 
Sum of Greatest Tumor Dimensions, mm42-4851-5760Total No. of Patients
  1. Abbreviations: SBRT; stereotactic body radiation therapy.

<25 to ≤5009817

Eighteen patients (38.3%) achieved CR within 6 months after completing SBRT. Eighteen patients (38.3%) had a PR, 10 patients (21.3%) had SD, and 1 patient (2.1%) had PD. Therefore, the objective response rate (defined as CRs plus PRs) was 76.6%. Local recurrence was observed in 2 patients, and the 2-year LC rate was 94.6% (Fig. 1). The 2-year OS rate was 68.7%. Twelve patients had died at the time of analysis, and the median survival duration was not reached at a median follow-up duration of 17 months. The 2-year PFS rate was 33.8%. Regional failure (defined as recurrence in the liver beyond the PTV) occurred in 22 patients (46.8%). Ten patients (21.3%) experienced distant failure, and the lung was most the common failure site, followed by bone and a lymph node. Of the 10 patients who experienced distant failure, 4 patients had regional failure, and 2 patients had local and regional failure (Fig. 2).

Figure 1.

Kaplan-Meier survival outcomes after stereotactic body radiation therapy (SBRT) are illustrated

Figure 2.

Patterns of failure for the study cohort are illustrated.

In univariate analysis, recurrent HCC and SBRT dose were identified as significant prognostic factors for PFS, and LD sum and SBRT dose were identified as significant prognostic factors for LC, but no factor was associated significantly with OS. In multivariate analysis, recurrent HCC and response were identified as independent prognostic factors for PFS, but no significant independent prognostic factor was associated with LC (Table 3).

Table 3. Prognostic Factors: Univariate Analysis
Prognostic Factor2-Year OS Rate, %P2-Year PFS RateP2-Year LC Rate, %P
  • Abbreviations: CTP, Child-Turcott-Pugh; Gy, grays; LC, local control; NS, nonsignificant; OS, overall survival; PFS, progression-free survival; PVTT, portal vein tumor thrombosis; SBRT, stereotactic body radiation therapy.

  • a

    In multivariate analyses of PFS, P = .032 for diagnosis history; P = .028 for SBRT dose.

  • b

    In multivariate analyses of LC, the P value was NS for both SBRT dose and sum of greatest tumor dimensions.

  • cGroups were divided into those who received 60 Gy and those who received <60 Gy.

 Women74.1 48 100 
Age, y      
 >5563.1 24.9 95 
CTP class      
 B83.3 66.7 100 
Diagnosis history      
 Recurrence59.6 18.8 96.2 
Sum of greatest tumor dimensions, mm      
 >5081.8 40.9 78.8 
 Yes66.7 20 100 
SBRT dose, Gy      
 ≤54 Gy70.6NS17.3.024a81.7.029
 >54 Gy61.5 52.5 100 
 Nonresponder50.5 35.1 88.9 

A summary of treatment-related toxicities is provided in Table 4. Three patients (6.4%) experienced grade 3 gastrointestinal (GI) toxicity (2 gastroduodenal ulcers and 1 colonic ulcer) between 1 month and 3 months after SBRT, and 2 patients (4.3%) experienced grade 4 gastric ulcer perforation at 7 months and 10 months after SBRT (Table 5). We retrospectively analyzed whether or not the enrolled patients had pre-existing GI disease before SBRT. Thirty-one of 47 patients underwent esophagogastroduodenoscopy before SBRT, and it was determined that 5 of those 31 patients had pre-existing gastroduodenal ulcers (GDUs). Furthermore, no severe GI complications developed in the 26 patients who did not have pre-existing GDUs, and 3 of the 5 patients with pre-existing GDUs experienced severe GI toxicities after SBRT and developed new ulcers after SBRT (Fig. 3). Moreover, this difference between the 26 patients without pre-existing GDUs and the 5 patients with pre-existing GDUs was significant in a Fisher exact test (P = .002). Within 3 months of SBRT, the CTP class of 6 patients (12.8%) increased from class A to class B. Conversely, in 1 patient (2.1%), the CTP class reduced from class B to class A. No patient experienced progression to CTP class C. Although no classic radiation-induced liver disease was observed, 4 patients (8.5%) who and no evidence of disease progression developed ascites with normal alkaline phosphatase levels. In 2 of these 4 patients, the ascites resolved spontaneously or on diuretics, but the other 2 patients required abdominal paracentesis continuously for refractory ascites. Two patients (4.3%) experienced grade 3 hyperbilirubinemia, and 5 patients (10.6%) experienced grade 3 thrombocytopenia. However, these 7 patients had pre-existing grade 1 or 2 hyperbilirubinemia and/or thrombocytopenia.

Figure 3.

The relation between the presence of a pre-existing gastrointestinal (GI) ulcer and stereotactic body radiation therapy (SBRT) related severe GI toxicity is illustrated. A preexisting GI ulcer significantly affected SBRT-related severe GI toxicity (Fisher exact test).

Table 4. Grade 3 or 4 Toxicities According to the Common Toxicity Criteria for Adverse Events Version 3.0 Grading Scalea
 Grade 3Grade 4
ToxicityNo. of Patients%No. of Patients%
  • a

    The Child-Turcott-Pugh class increased from A to B in 6 patients (12.8%) and reduced from B to A in 1 patient (2.1%).

  • b

    Gastroduodenal ulcer or colonic ulcer

  • c

    Gastric ulcer perforation.

Gastrointestinal ulcer3b6.42c4.3
Table 5. Characteristics of the Patients Who Developed Grade 3 or 4 Gastrointestinal Toxicity
   Characteristics Before SBRT       
Patient No.SexAge, yEtiologyLC or VaricesGI ProblemCTP ClassTumor Location (Segment)SBRT Dose, GyDmaxaD2ccbTime From Completion of SBRT, moToxicity GradeLesion
  • Abbreviations: CTP, Child-Turcott-Pugh; DU, duodenal ulcer; GI, gastrointestinal; GU, gastric ulcer; HBV, hepatitis B virus; LC, cirrhosis; NE, not evaluable; SBRT, stereotactic body radiation therapy.

  • a

    The maximum dose at which bowel toxicity (gastroduodenal or colonic) occurred.

  • b

    The minimum dose (up to 2 mL) at which bowel toxicity (gastroduodenal or colonic) occurred.

2Man55HBVLC with varicesGU, DUA651423033DU bleeding
3Man59HBVLC with varicesGUB2545539104GU perforation
4Man52HBVLC with varicesGUA751452874GU perforation
5Man56HBVNEA460605613Colonic ulcer


The current phase 2 trial was conducted to evaluate the efficacy and safety of SBRT as a local salvage treatment after incomplete TACE for inoperable HCC. In a meta-analysis of TACE alone in patients with inoperable HCC, Marelli et al reported a median response rate of 38%, and median CR and PR rates of 0% (range, 0%-35%) and 30% (range, 3%-62%), respectively.17 To overcome this low response rate, various modalities have been investigated in combination with TACE, such as RFA, PEI, drug-eluting beads, sorafenib, conventional RT, and SBRT. Although these strategies produced better response rates than TACE alone,18 more controlled trials are required to confirm these promising outcomes. In this situation, SBRT combined with TACE may be an attractive strategy for patients with inoperable HCC who are not suitable for RFA or PEI. Table 6 summarizes recently published data on SBRT for HCC and indicates that, using various doses and small fraction numbers, the 2-year LC rate reached >90%. These results for SBRT are more promising than those achieved by TACE alone.

Table 6. Comparison of the Current Study and Recent Published Data Concerning Stereotactic Body Radiation Therapy for Primary Hepatocellular Carcinoma
  Range [Median]    
ReferenceNo. of PatientsSBRT Dose, Gy/No. of FractionsTumor VolumeFollow-Up Duration, moRR, %2-Year LC Rate, %2-Year OS Rate, %Severe Toxicity
  1. Abbreviations: CR, complete response; CTP A/B, Child-Turcott-Pugh class A/B; DU; duodenal ulcer; GI, gastrointestinal; Gy, grays; LC, local control; NR; not reported; OS, overall survival; PR, partial response; PVT, portal vein tumor thrombosis; RILD; radiation-induced liver disease; RR, response rate; SBRT, stereotactic body radiation therapy.

Kwon 20101942: CTP A, 90%30-39/3 [33/3]3-82 cm3 [15.4 cm3]8-49 [29]8668 at 3 y59 at 3 yGrade 4 liver failure, 2%
Cardenes 20102017; 25 Lesions: CTP A/B, 6/11; PVT, 336-48/3-58-95 cm3 [34 cm3]10-42 [24]8110060RILD, 12%
Louis 20102125: CTP A/B, 22/3[45/3]18-100 mm [45 mm]1-24 [13]869552Acute grade 3, 8%; acute grade 3 DU, 4%
Andolino 20112260: CTP A/B, 36/2424-48/3-52-112 cm3 [29 cm3]2-52 [27]709067Increased CTP class, 20%
Facciuto 20112327; 39 Lesions24-36/2-42.01±0.78 cm32-48 [22]37NR82
Current study47: CTP A/B, 41/6; PVT, 442-60/3 [57/3]13-78 mm [29 mm]; 2-214 cm3 [14 cm3]6-38 [17]769569Ascites, 9%; GI toxicity, 11%; increased CTP class, 13%

Theoretically, a combination of RT and TACE has several advantages. Tumor shrinkage after TACE makes it easy to deliver a high radiation dose, and remaining lipiodol inside tumors after TACE simplifies determining tumor volumes and provides a target for image-guided RT. Several studies concerning conventional RT plus TACE have been performed, and they too produced better outcomes than TACE alone.24-27 However, conventionally fractionated RT does not provide satisfactory LC because of the limitations associated with the delivery of high tumoricidal doses.28 Conversely, SBRT improves on the unsatisfactory results of conventional RT, because it allows the delivery of higher doses to the tumor. In several recent studies, intensive hypofractionated schedules achieved outstanding results in lung and liver tumors. In a phase 2 multicenter trial, Timmerman et al demonstrated that high-dose SBRT (60 Gy in 3 fractions) can achieve LC rates that match surgery for medically inoperable, early stage lung cancer.29 Rusthoven et al reported a 2-year actuarial, in-field, LC rate for liver metastasis after SBRT of 92% when they used 60 Gy in 3 fractions.30 Furthermore, McCammon et al suggested that excellent LC rates can be achieved with 54 Gy in 3 fractions or greater in lung and liver tumors.31 In a phase 1 feasibility trial of SBRT in HCC, Cardenes et al achieved a maximum LC rate of 100% for doses from 36 to 48 Gy in 3 fractions.20 In addition, a phase 2 trial at Indiana University in patients with CTP class A disease who received 48 Gy in 3 fractions demonstrated that doses lower than 60 Gy may be enough to control primary HCC. However, in our previous study, we used dose escalation from 33 to 57 Gy in 3 fractions and achieved a 2-year LC rate of 66.4%.11 We experienced several local recurrences at high doses of 51 or 54 Gy in 3 fractions and formed the opinion that, if normal tissue constraints could be met, then higher intensity SBRT doses may be warranted in patients with inoperable HCC to achieve LC. In the current study, we achieved an LC rate of 95% and observed that it was statistically influenced by SBRT dose. In particular, whereas the 2-year LC rate in patients who received >54 Gy was 100%, the rate in those who received ≤54 Gy was 81.7% (P = .029 in univariate analysis). Thus, the current study demonstrates that a higher dose may be necessary for patients with inoperable HCC to achieve a better LC rate.

In terms of hepatic complications, the dosages used in this study were generally well tolerated, and severe hepatic toxicity was acceptable. However, we did not apply a clear constraint to GI tissues in the current study. Consequently, 5 patients experienced severe GI complications (2 GDUs, 1 colonic ulcer, and 2 gastric perforations). It is noteworthy that, of the 4 patients who experienced severe gastroduodenal complications, 3 had pre-existing GDUs and cirrhosis. Furthermore, among patients with pre-existing cirrhosis, the prevalence of GDU is greater than in the general population, because portal hypertension probably impairs the mucosal defense mechanism.32 Moreover, it has been suggested that cirrhosis increases GI toxicities after chemoradiation.33 Our results suggest that a pre-existing GDU with cirrhosis is a significant risk factor in GI toxicity. Therefore, esophagogastroduodenoscopy before SBRT to determine the presence of pre-existing GDU may be warranted if a tumor is located near GI organs; and, when it is determined that a patient had a pre-existing GDU, more careful strategies to reduce GI toxicity, such as a lower SBRT dose with smaller fractions or prolongation of treatment time with a longer interval between fractions, should be considered.

In conclusion, the current trial demonstrates that SBRT after incomplete TACE for inoperable HCC achieves promising response and LC rates. However, a longer follow-up will be required to confirm this finding. On the basis of the results from this study, we are planning a new multi-institutional phase 2 trial to reduce GI toxicities and achieve optimal outcomes.


This work was supported by the National Nuclear R&D Program of the Ministry of Education, Science, and Technology, Republic of Korea.


The authors made no disclosures.