Evaluation of response after stereotactic body radiotherapy for hepatocellular carcinoma




Hepatocellular carcinoma (HCC) is increasing in incidence due to hepatitis C. Stereotactic body radiotherapy (SBRT) is a noninvasive, effective therapy in the management of liver malignancies. The authors evaluated radiological response in 26 patients with HCC treated with SBRT at Indiana University.


Between March 2005 and June 2008, 26 patients with HCC who were not surgical candidates were enrolled in a phase 1 to 2 trial. Eligibility criteria included solitary tumors ≤ 6 cm or up to 3 lesions with sum diameters ≤ 6 cm, and well-compensated cirrhosis. All patients had imaging before, at 1 to 3 months, and every 3 to 6 months after SBRT.


Patients received 3 to 5 fractions of SBRT. Median SBRT dose was 42 Gray (Gy) (range: 24-48 Gy). Median follow-up was 13 months. Per Response Evaluation Criteria in Solid Tumors (RECIST), 4 patients had a complete response (CR), 15 had a partial response (PR), and 7 achieved stable disease (SD) at 12 months. One patient with SD experienced progression marginal to the treated area. The overall best response rate (CR + PR) was 73%. In comparison, by European Association for the Study of the Liver (EASL) criteria, 18 of 26 patients had ≥ 50% nonenhancement at 12 months. Thirteen of 18 demonstrated 100% nonenhancement, being > 50% in 5 patients. Kaplan-Meier 1- and 2-year survival estimates were 77% and 60%, respectively.


SBRT is effective therapy for patients with HCC with an overall best response rate (CR + PR) of 73%. Nonenhancement on imaging, a surrogate for ablation, may be a more useful indicator than size reduction in evaluating HCC response to SBRT in the first 6 to 12 months, supporting EASL criteria. Cancer 2012;118: 3191–98. © 2011 American Cancer Society.


Hepatocellular carcinoma (HCC) is the most frequent primary liver tumor in adults, representing the fifth most common cancer in the world with over 500,000 cases diagnosed annually.1 Chronic hepatitis C with cirrhosis is the leading cause of HCC in Western countries.2 Until more effective measures to cure viral infection are developed, early detection of HCC remains the best strategy for reducing tumor-related mortality. Ultrasound or dual-phase computed tomography (CT) scanning in combination with α-fetoprotein every 6 months is current standard of care for screening high-risk patients.3 Surveillance allows for early HCC diagnosis in up to 60% to 70% of cases when potentially curative therapies can be used. However, only approximately one-half of these patients benefit from radical interventions.

Liver transplantation offers the best chance for cure with a 5-year survival > 70% and low recurrence rates.4 Surgical resection, the only other potentially curative option, results in 5-year survival of approximately 50% to 60%, but the recurrence rate is very high (generally noted to be > 50%).4, 5 It has been estimated that only a very small (< 5%) percentage of cirrhotic patients with HCC are candidates for surgical resection.5 Thus, minimally invasive therapies are gaining increasing attention in the treatment of patients with HCC who are not surgical candidates.6-10

Stereotactic body radiotherapy (SBRT) refers to the management and delivery of image-guided high-dose RT with tumor-ablative intent within an abbreviated course of treatment.11 In the late 1990s, investigators at the Indiana University (IU) School of Medicine initiated several trials evaluating efficacy and outcomes of SBRT for lesions in various sites.12, 13 These are widely considered seminal works, and serve to define a foundation for this growing aspect of clinical radiation oncology.

To our knowledge, there are few data on the radiologic correlates of “response” for HCC. We recently reported results of a phase 1/2 trial designed to evaluate the maximum tolerated dose and dose-limiting toxicities of SBRT in selected patients with primary HCC in an effort to achieve a biologically potent, noninvasive cancer therapy for this population.14 Since 2001, the European Association for the Study of the Liver (EASL) has suggested that tumor necrosis, as recognized by nonenhanced areas by spiral CT, should be considered the optimal method to assess local response.15 A recent report assessed the concordance between Response Evaluation Criteria In Solid Tumors (RECIST) and EASL guidelines for evaluating response to local therapies. It concluded that RECIST missed all complete responses (CRs) and underestimated the extent of partial tumor response (PR).16 Here we evaluate 6-month and 12-month local in-field control directly comparing RECIST and EASL criteria as early indicators of response to treatment.



Between March 2005 and June 2008, 26 patients with HCC who were not candidates for surgical resection or liver transplantation at the time of evaluation were enrolled in a phase 1 to 2 trial using SBRT. All IU protocols were, and continue to be, performed with full approval of the IU Institutional Review Board. All patients were discussed in our Multidisciplinary Liver Tumor Conference at IU. Different therapeutic options were discussed, and providing the patients were candidates for SBRT, this option was preferred over other invasive techniques, based on the data from our phase I trial.14

Eligibility criteria included solitary tumors measuring ≤ 6 cm or up to 3 discrete lesions with the sum of diameters ≤ 6 cm, no ascites, and fairly well-compensated liver cirrhosis, Child Pugh class (CPC) A or B. Initially, all CPC B patients were eligible. However, an interim analysis showed increased toxicity in patients with CPC B, scores 8 or 9. After the interim analysis, only CPC B, score 7 patients were eligible to participate. Patient and tumor characteristics are described in Table 1. None of the treated lesions were within 0.5 cm from the wall of hollow gastrointestional organs including the stomach, duodenum, or hepatic flexure of the colon.

Table 1. Patient and Tumor Characteristics
Patient/Tumor CharacteristicsNo. of Patients (%)
  1. Abbreviations: AFP, α-fetoprotein; cGy, centigray; GTV, gross tumor volume; NASH, nonalcoholic steatohepatitis; PTV, planning tumor volume.

 Females3 (12%)
 Males23 (88%)
Median age (range), y60 (24-83)
Child-Pugh score 
 A14 (54%)
 B = 77 (27%)
 B ≥85 (19%)
Hepatitis etiology 
 B3 (12%)
 C15 (58%)
 Other (alcohol, hemochromatosis, A1 antitrypsin, NASH, heart transplant)7 (27%)
 No hepatitis1 (3%)
Prior therapies 
 Yes7 (27%)
 No19 (73%)
AFP median (range), ng/ml19.0 (1.4-16,269.0)
Portal vein invasion 
 Yes3 (12%)
 No23 (88%)
No. of lesions treated 
 123 (88%)
 23 (12%)
TNM staging 
 T1N018 (69%)
 T1N11 (4%)
 T1N41 (4%)
 T2N05 (19%)
 T3N01 (4%)
Median GTV volume (range), cc33.9 (2.0-95.0)
Medican PTV volume (range), cc97.6 (20.8-252.9)
Median uninvolved liver volume (range), cc1606 (973-3083)
Dose per fraction, cGy 
 8007 (27%)
 12003 (12%)
 14006 (23%)
 160010 (38%)
No. of fractions 
 319 (73%)
 57 (27%)


Particulars of the phase 1/2 trial have been described previously.14 Dose level 1 delivered 3600 centigray (cGy) in 3 fractions (fx) (1200 cGy/fx in 1-3 weeks), with escalation in 6-gray (Gy) increments planned up to a maximum dose of 48 Gy in CPC A patients. Interim analysis after 2 dose levels demonstrated increased toxicity in the CPC B patients. CPC B, score 7 patients continued phase 1 with a modified fractionation schedule; 4000 cGy delivered over 5 fx (800 cGy/fx in 3-6 weeks).14 The current phase 2 trial delivers 16 Gy × 3 fx (48 Gy total) for CPC A patients and 8 Gy × 5 fx to eligible patients with CPC B. Patients were treated in the Stereotactic Body Frame (SBF) (Elekta, Stockholm, Sweden).

For planning purposes, a dual-phase CT scan was obtained with the patient in the stereotactic body frame to fuse the arterial phase enhancing images with the axial images from the CT simulation for improved delineation of the gross tumor volume (GTV). No additional margin was added for clinical tumor volume (CTV). The planning tumor volume (PTV) included a 0.5-cm to 1.0-cm radial and 1.0-cm to 1.5-cm superior-inferior expansion. This was verified under fluoroscopy. Cone-beam image-guided RT (IGRT) was used before each fx of SBRT (Fig. 1) for isocenter verification.

Figure 1.

Cone-beam image-guided radiotherapy (IGRT) is shown. The top row shows cone-beam computed tomography (CBCT) images taken after initial set up on the treatment table. The bottom row shows how corrections can be made in x, y, and z axes to align bony anatomy.

Follow-Up and Data Collection

Imaging technique

All patients had dual- or triple-phase contrast-enhanced CT and/or magnetic resonance imaging (MRI). CT was performed as dedicated dual-phase exanimation of the abdomen using 40- or 64-slice CT scanners (Brilliance-40 or Brillaince-64; Philips Medical Imaging, Cleveland, Ohio). Intravenous contrast of 100 mL of was given. No oral contrast was used. MRI examinations were performed using an 18-channel 1.5 T MRI (MAGNETOM Avanto; Siemens, Erlangen, Germany). The standard sequences were T1-weighted in- and out-of-phase axial, T2-weighed fast-spin echo, and 3-dimensional volumetric fat-suppressed pre- and postgadolinium sequences (VIBE). A total of 16- to 20-mL intravenous gadolinium contrast was used. The type of imaging modality was determined by the referring clinician. Dual-phase helical CT (DPCT) of the liver was obtained at least at the time of the first follow-up visit. However, for those patients at risk of developing hepatorenal failure secondary to the intravenous contrast administration, in subsequent follow-up visits, MRI was obtained.

Image review

A single experienced radiologist reviewed all scans. Tumor response was assessed by RECIST criteria.17 In addition to bidimensional size measurements, enhancement pattern and precontrast T1-weighted and T2-weighted signal of the tumor and peritumoral liver were measured. Imaging was performed before treatment and at 3 months, 6 months, 9 months, and 12 months after therapy. All lesions were visible before therapy on both CT and MRI scans.

Statistical Analysis

Descriptive statistics were reported for demographic, clinical, dosimetric, and response data, as well as Kaplan-Meier estimates for 1-year and 2-year survival. Ninety-five percent confidence intervals (95% CIs) were reported for response and survival. Patients were censored at the time of last follow-up. The statistical package used was SAS (version 9.1; SAS Institute, Inc, Cary, NC).


Patient Population

Fourteen CPC A patients with 15 treated lesions and 5 CPC B patients with 7 treated lesions received 3 fx of SBRT. Seven CPC B patients with single lesions received 5 fx of 8 Gy/fx. The median SBRT dose was 42 Gy (range, 24-48 Gy). Median follow-up was 13 months (range, 3-42 months). When the study was opened, eligibility for liver transplantation was defined before initiating treatment. There are several reasons why patients with tumors within transplantation criteria were not considered for transplantation and include age, comorbid conditions, noncompliance, or patient preference. As the study matured, patients otherwise eligible for transplantation were offered SBRT preoperatively if the waitlist time was > 3 months. This longer wait time was necessary for reasons that include, but are not limited to,: 1) abnormal cardiac catheterization requiring stent placement and 2) the need for outpatient drug and/or alcohol treatment.

Seventeen of the 26 treated patients were alive without evidence of progression at a median follow-up of 18 months (range, 5-42 months). Eight patients were successfully bridged to orthotopic liver transplantation. The Kaplan-Meier 1-year and 2-year survival estimates were 77% and 60%, respectively (Fig. 2).

Figure 2.

Kaplan-Meier survival estimates (with 95% confidence interval) are shown.

During the study period, approximately 95 patients were treated with other liver-directed therapies. Patients with large tumors (> 6 cm) or multifocal disease were treated with small-particle embolization. There was little to no overlap in patients considered for a catheter-based technique versus SBRT. Twelve patients were treated with radiofrequency ablation (RFA). There was considerable overlap in patients who might be considered for RFA versus SBRT. The reason for choosing RFA over SBRT was most often due to tumor size (> 3 cm) and/or an inability to tolerate the immobilization device required for SBRT.

Dimensional Changes Within the Tumor

Mean (standard error) tumor dimension decreased by 35% (6%), 37% (7%), 48% (9%), and 55% (10%), respectively, at 3 months, 6 months, 9 months, and 12 months compared with pretreatment size (Fig. 3). Per RECIST criteria, 4 patients had a best response of CR, 15 had a PR, and 7 had stable disease (SD). The overall best response rate (CR + PR) was 73% (95% exact binomial CI, 52%-88%). Only 1 patient with a best response of SD experienced disease progression marginal to the treated area before 12 months of follow-up, but this patient was successfully bridged to transplantation. Two patients underwent orthotopic liver transplantation (OLT) < 2 months after SBRT and therefore did not have follow-up imaging.

Figure 3.

Change in size compared with tumor necrosis at various time points is shown. The mean percentage reduction in size versus the mean percentage necrosis was plotted against time. Percentage necrosis was greater than percentage size reduction at each point in time and the magnitude of difference increased at each point. Sample sizes are 21, 19, 14, and 12 at 3 months, 6 months, 9 months, and 12 months, respectively.

Necrosis/Ablation Within the Tumor Volume

Abnormal enhancement of peritumoral liver was seen in all cases by 6 months (Fig. 4).

Figure 4.

Normal liver reaction after stereotactic body radiotherapy (SBRT) is shown. Abnormal enhancement of peritumoral liver was noted in all cases by 6 months. No correlation was noted between dose of radiation with time of appearance of reaction or enhancement pattern of reaction. Increased enhancement of surrounding tissue made measurement of lesion size difficult in the arterial phase.

Given the difficulty in obtaining dimensional data based on the peritumoral enhancement, data regarding necrotic/ablative changes within the tumor, based on central nonenhancement within the mass, were also obtained. Mean (standard error) percentage necrosis (nonenhancement) was 59% (8%), 69% (8%), 81% (8%), and 92% (7%), respectively, at 3 months, 6 months, 9 months, and 12 months (Fig. 3). Note the percentage necrosis was greater than percentage size reduction at each time point, and the magnitude of difference increased as well. Figure 5 displays the percentage reduction in tumor size versus percentage necrosis for each tumor at each time point to illustrate this point at the individual tumor level. Eighteen of 26 patients (69%) had > 50% nonenhancement by 12 months. Thirteen of 18 demonstrated 100% nonenhancement and 5 patients showed > 50% nonenhancement. Eight patients had < 50% nonenhancement; 2 of these patients died < 3 months after SBRT and 2 underwent OLT < 2 months after SBRT.

Figure 5.

Percentage reduction in tumor size versus percentage necrosis for each tumor plot for each individual tumor at 3 months, 6 months, 9 months, and 12 months is shown. Percentage necrosis tended to be larger than percentage reduction in tumor size at each time point.


RT for the treatment of unresectable HCC has been attempted for more than 4 decades. In early trials involving the use of whole-liver RT, generally in combination with intra-arterial and/or intravenous chemotherapy, the reported 2-year survival rate was < 10%.18 The use of 3-dimensional conformal RT treatment planning has allowed the delivery of higher doses of RT to limited liver volumes, which has led to higher response rates and survival with an improved toxicity profile.19, 20 In addition, the use of 3-dimensional conformal therapy has allowed a better understanding of the relationships between dose, volume, and risk of complications and making possible dose-escalation studies in this patient population, by using dose-volume histogram analysis.19-21

Evaluation of Response to Treatment

Post-SBRT assessment was performed at 3 months, 6 months, 9 months, and 12 months by means of dual-phase CT scanning or liver MRI. All imaging was evaluated by an independent radiologist who was well versed in the evaluation of liver lesions and peritumoral reactions. An example of a typical response to SBRT within a lesion is shown in Figure 6. The phenomenon of peritumoral hyperenhancement on arterial and venous phases of imaging was consistent with radiation-induced inflammation. We have documented that little change in size occurred in these tumors during the first 3 months after treatment. Evaluation of response by classic RECIST criteria has proved to be challenging. In this study, both CT and MRI studies were reviewed in the same patients at different time intervals. It is well accepted that MRI is superior for the detection of small hepatocellular cancer (< 1 cm) and in the characterization of benign hepatic nodules from HCC.22, 23 Conversely, many patients with severe cirrhosis are unable to hold their breath for prolonged periods, and the higher temporal resolution of CT often results in a more consistent image quality.24 In the current study, all lesions were well visible on both CT and MRI, and the choice of the scanning modality did not affect the measurements of size or enhancement.

Figure 6.

Typical early response to stereotactic body radiotherapy (SBRT) is shown in a 47-year-old male with hepatitis C, Child Pugh class B liver cirrhosis, and a 2-cm hypervascular mass in segments 5 and 6. Arterial and venous phase postgadolinium magnetic resonance imaging (MRI) 1 month after SBRT showed reduced enhancement of the mass (black arrows) but no significant change in size. Note peritumoral hyperenhancement in both phases (white arrows). Arterial phase image at 3 months showed the mass was almost completely necrotic, smaller in size, and with a thin enhancing rim (arrow).

The EASL has suggested that tumor necrosis, as recognized by nonenhanced areas on spiral CT, should be considered the optimal method with which to assess local response.15, 16 In following the imaging for this group of patients over time, we also noted that size reduction may not accurately reflect response to treatment. The current analysis of these data suggests that tumor nonenhancement may be a better early indicator of response to treatment and this will continue to be evaluated in the phase 2 portion of the trial. To illustrate this point, per RECIST criteria we have a 15% CR rate compared with 50% complete necrosis at 12 months. These data includes 17 phase 1 patients and 9 patients enrolled on phase 2 for a total of 26 patients.

Table 2 shows a direct comparison of response rates among the patients using RECIST versus EASL criteria on imaging. As shown in the highlighted portion of the bottom row of this table, there are only 3 patients in whom the RECIST criteria demonstrated a better response (CR or PR) than the percentage ablation. Further research is being done at this institution to evaluate RECIST and EASL response as it correlates to pathologic response in explanted livers. This will provide more insight into which method results in a more accurate evaluation of response to therapy.

Table 2. Direct Comparison of Best Response by RECIST Criteria Versus Ablation at 12 Months
  1. Abbreviations: CR, complete response; EASL, European Association for the Study of the Liver; PR, partial response, RECIST, Response Evaluation Criteria In Solid Tumors; SD, standard deviation.


Successful Bridge to Transplantation

Now that the patient selection criteria and safety profile of this technique have been established in the phase 1 portion of the trial,14 the short-term and long-term outcomes continue to be evaluated in phase 2. This technique is unique from all other therapeutic options in that it is noninvasive. This is an important consideration because the population of patients with cirrhosis and HCC is an overall medically compromised population that is at high risk for complications from invasive procedures.

Although the study only enrolled patients who were ineligible for transplantation at the time of study entry, we successfully bridged 8 of the 26 patients to transplantation. Because of prolonged wait times on transplantation lists, the incidence of disease progression while listed for organ transplantation ranges from 10% to 23%.5, 25 With further study of outcomes in the phase 2 portion of the trial, we may be able to offer SBRT as a bridge to transplantation in transplantation-eligible patients as well.


SBRT is a very effective and well tolerated therapy for patients with HCC in the context of well-compensated liver disease. Axial bidimensional (RECIST) criteria are not useful in assessing early (1 month-3 months) response to therapy. This is partly because peritumoral hypervascular response to therapy makes exact tumor size measurements difficult in the initial months after SBRT. The overall best radiological response rate (CR + PR) is 73% at 12 months. Importantly, only 1 patient failed marginal to the treated area, but was successfully bridged to transplantation. Reduced vascularity or nonenhancement on imaging, a sign consistent with tumor necrosis, may be a more useful indicator than size reduction in evaluating HCC response to SBRT in the first 6 to 12 months, supporting recently confirmed EASL criteria.


We gratefully acknowledge Deborah S. MacDougall for her assistance in preparing this manuscript.


No specific funding was disclosed.


The authors made no disclosures.