Radiation treatment outcomes for unresectable hepatocellular carcinoma

Abstract

Background. Hepatocellular carcinoma is one of the most common cancers worldwide. Data regarding the use of radiotherapy is limited in patients from populations without endemic viral hepatitis. We examine the outcomes for patients treated with radiotherapy in the modern era at a single institution. Material and methods. A total of 29 patients with localized hepatocellular carcinoma treated from 2000–2010 were reviewed. Patients with metastatic disease at the time of radiation were excluded. Median radiation dose was 50 Gy (range 30 to 75 Gy) with a median biologically effective dose of 80.6 (range 60 to 138.6). Median tumor size at the time of radiation was 5.2 cm (range 2 to 25 cm). Results. Eighty three percent of all patients had either stable disease or a partial response to radiation, based on RECIST criteria. Median change in tumor size following radiation was −17% (range −73.5 to 177.8%). Estimated one-year overall survival and in-field progression-free survival rates for the study population were 56% and 79%, respectively. One-year overall survival in patients treated to a biologically effective dose <75 was significantly lower than in patients treated to a biologically effective dose ≥75 (18% vs. 69%). One-year in-field progression-free survival rate (60% vs. 88%) and biochemical progression-free survival duration (median 6.5 vs. 1.6 months) were also significantly improved in patients treated to a biologically effective dose ≥75. Grade 3 toxicity was seen in 13.8% of patients. Discussion. In a population without endemic viral hepatitis, unresectable hepatocellular carcinoma demonstrates significant response to radiotherapy with minimal toxicity. Furthermore, our findings suggest that increased biologically effective dose is associated with improved survival and local tumor control.

Hepatocellular carcinoma (HCC) is one of the most common malignancies worldwide, and is the third most common cause of cancer death [1]. The majority of cases are found in countries with endemic hepatitis B, which dramatically increases the predisposition to develop HCC. In the USA and many western countries, HCC is more commonly related to alcoholism, with anywhere between 32–45% of all cases being related to alcohol consumption [2]. The most common curative approach for HCC includes surgical resection of the tumor or liver transplantation in appropriate candidates. Unfortunately, many patients have unresectable HCC and are not candidates for transplantation [3]. These patients are currently treated with liver-directed interventional approaches, such as radiofrequency ablation (RFA), transarterial chemoembolization (TACE), hepatic arterial infusion (HAI) or Yttrium microsphere therapy. Many tumors are not amenable to RFA however, secondary to large size, multiple tumor nodules or proximity to large vessels, the diaphragm or the gastrointestinal (GI) mucosa [4]. Furthermore, other than RFA for highly selected small peripheral lesions, non-surgical approaches are not generally thought to be curative in the majority of patients and are associated with significant toxicity.

Historically, external beam radiotherapy has been limited in its application in HCC due to concerns regarding toxicity, particularly the development of radiation induced liver disease (RILD) [5]. However, these concerns have been minimized by multiple technologic advances that permit partial liver radiation to very high doses while limiting dose to uninvolved liver. These advances include computed tomography (CT) and magnetic resonance (MR) guided target delineation, the use of fiducial markers and cone-beam CT for target localization, and respiratory gating and four dimensional (4D)-CT imaging to account for respiratory motion. These techniques allow for targeted dose escalation without increased toxicity. Furthermore, the increasing use of charged particle therapy, such as proton radiotherapy, allows for further conformality and sparing of critical surrounding structures.

Much of the data regarding modern use of radiotherapy in HCC is derived from populations where hepatitis B virus (HBV) infection is endemic, with one year overall survival rates ranging anywhere from 50–90% depending upon the number of lesions treated, Child-Pugh class and the method by which survival rates were measured (i.e. from the time of diagnosis vs. radiotherapy treatment) [6–10]. Reported outcomes from non-HBV-endemic populations with HCC treated with external beam radiotherapy is somewhat scarce. In the initial University of Michigan experience, patients with Child-Pugh class A were treated twice daily to doses between 40–90 Gy, with a median survival of 15.2 months and higher doses of radiation correlating with improved outcomes [11]. Also, in a European phase II trial of external beam radiotherapy in patients with small foci of HCC, of 23 evaluable patients, in-field recurrence free survival was approximately 87% at one year [12]. Finally, a phase I study conducted by Dawson and colleagues, showed a median survival of 11.7 months in this patient population after individualized hypofractionated stereotactic body radiation therapy using daily image-guidance and doses ranging from 24–54 Gy depending upon the size of the tumor and predicted toxicity [13]. Because of the relatively few studies examining external beam radiotherapy in unresectable HCC in non-HBV-endemic populations, our goal was to retrospectively examine our single institution experience. Specifically, in this study, we sought to determine the outcomes of patients treated in the modern era with external beam radiotherapy for unresectable HCC with attention to possible dose escalation and toxicity.

Material and methods

Patients treated at MD Anderson Cancer Center from 2000–2010 for unresectable HCC with external beam radiotherapy were included in the study. Patients with metastatic disease at presentation at the time of radiotherapy were excluded from the study, with a total of 29 patients meeting the study criteria. Patient characteristics at the time of diagnosis are listed in Table I. Median age at diagnosis was 65 years, with a median follow-up time from diagnosis of 19 months. Patients were generally in good health at the time of diagnosis, with a median Charleston Comorbidity Index (CCI) score of 6, and nearly all patients having an ECOG performance status of 2 or better. The majority of patients were negative for hepatitis C (51.7%) or hepatitis B (69%) at the time of diagnosis, in keeping with the decreased incidence of HBV infections in western populations compared to similar studies in Asia and elsewhere. At the time of diagnosis most patients were found to have a degree of cirrhosis (65.5%), however the majority of these patients were Child-Pugh class A (58.6%). Only 21% of all patients had nodal enlargement at diagnosis.

Table I. Patient characteristics at diagnosis.

	Median	S.D.	Min	Max
Age at diagnosis (years)	65.0	13.9	20.0	81.0
Follow-up (months)	7	16.6	0	85
Age adjusted CCI	6.0	2.4	3.0	13.0
Tumor size	6.4	5.3	1.8	25.0

Patient Characteristics	N	%
Race
Caucasian	13	44.8
African-American	4	13.8
Hispanic	7	24.1
Asian	2	6.9
Arabic	2	6.9
Other	1	3.4
Gender
Male	17	58.6
Female	12	41.4
Hepatitis serology
Hepatitis B	6	20.7
Hepatitis C	11	37.9
Unknown	3	10.3
Negative	15	51.7
Alcohol use
Current	8	27.6
History of use	9	31.0
Never	11	37.9
Unknown	1	3.4
Cirrhosis
Absent	8	27.6
Present	19	65.5
Unknown	2	6.9
Nodal stage
N0	23	79.3
N1	6	20.7
Tumor stage
T1	13	44.8
T2	6	20.7
T3	8	27.6
T4	1	3.4
Tx	1	3.4
Portal vein thrombosis
Absent	20	69.0
Present	9	31.0
Child-Pugh Class
A	17	58.6
B	5	17.2
Unknown	7	24.1
ECOG PS
0	3	10.3
1	19	65.5
2	6	20.7
3	1	3.4
Number of lesions
Single	14	48.3
Multiple	15	51.7

CCI, Charleston Comorbidity Index; ECOG PS, Eastern Cooperative Oncology Group Performance Status.

Median time from diagnosis to radiotherapy was seven months. Patients were treated locally with a combination of surgical resection, RFA, TACE or HAI prior to radiotherapy, with a median number of one treatment each (Table II). At the time of radiotherapy all patients had unresectable disease, with most patients being Child-Pugh class A (Table III). Median tumor size at the time of radiation was 5.2 cm, with a median alpha-fetoprotein (AFP) of 1114 ng/ml. Nine patients (20%) had portal venous thrombosis prior to radiation treatment. Median total radiation dose was 50 Gy (range 30–75 Gy) with a median dose per fraction of 2.5 Gy (range 1.8–7 Gy) (Table IV). Biologically equivalent dose (BED) was calculated using the linear-quadratic formula described elsewhere [14], with an assumed α/β of 10. Median BED for the study population was 80.6 (range 60–138.6). Radiotherapy was generally given to gross disease only, with no elective nodal radiation. All radiotherapy plans were generated using CT-guidance. Around half of all patients (51.7%) were treated with three dimensional (3D)-conformal radiotherapy, most commonly with 18 MV photons. The remainder of patients in the study population was treated using either intensity modulated radiotherapy (IMRT) (31%) or proton radiotherapy (17.2%). Median follow-up from radiotherapy was 12.2 months (range 1–85 months). Following completion of radiotherapy, the majority of patients were followed with serial abdominal CT imaging and laboratory studies every two to three months. Response to radiotherapy was determined by size on abdominal CT using RECIST criteria, which have been described elsewhere [15].

Table II. Treatment prior to radiation. Median represents median number of treatments per patient.

Treatment prior to XRT	N	Median	95% CI	Min	Max
RFA	3	1	0.6	1	2
Alcohol ablation	1	1	0	1	1
Resection	6	1	0.5	1	2
TACE/HAI	13	1	0.7	1	3
Systemic chemotherapy
Erlotinib/bevacizumab	2
FOLFOX→Taxotere	1
5FU/Gemcitabine	1
5FU/Gemcitabine/Interferon	3
Capecitabine→Thalidomide	1
Sorafenib	1
Multiple	3
None	17

CI, confidence interval; 5FU, 5fluorouracil; FOLFOX, 5fluorouracil leucovorin and oxaliplatin chemotherapy; HAI, hepatic arterial infusion; RFA, radiofrequency ablation; TACE, transarterial chemoembolization; XRT, radiation therapy.

Table III. Patent characteristics at radiation.

	N	%
Ascites
Absent	17	58.6
Present	12	41.4
Child-Pugh Class
A	20	69.0
B	8	27.6
Unknown	1	3.4

	Median	95% CI	Min	Max
Tumor length prior to XRT (cm)	5.2	±2.2	2.0	25.0
AFP at XRT	1114.3	±87603.5	1.5	770479.0
Tumor length at maximal response to XRT (cm)	4.3	±2.4	0	27
AFP nadir after XRT	450.3	±109179.4	1.8	1191470

AFP, alpha fetoprotein; CI, confidence interval; XRT, radiation therapy.

Table IV. Radiation treatment.

	Median	95% CI	Min	Max
Total dose	50.0	±12.4	30.0	75.0
Dose per fraction	2.5	±1.1	1.8	7.0
BED	80.6	±27.7	60.0	138.6
Fractionation	N	%
Conventional	9	31.0
Hypofractionation	16	13.8
Hyperfractionation	4	55.2
Radiotherapy
Conventional	15	51.7
IMRT	9	31.0
Protons	5	17.2

CI, confidence interval; IMRT, intensity modulated radiation therapy.

Biologically effective dose (BED) was calculated using the equation BED = nd (1 + d/α/β), with n = number of fractions, d = dose per fraction and an α/β = 10.

Primary outcomes for the study include overall survival (OS) and survival time from radiotherapy to disease progression in the radiation field (in-field PFS), elevation of AFP from nadir (biochemical PFS), and development of metastatic disease (DMFS). SPSS software (v16.0) was used to perform the statistical analysis. Comparisons between groups were performed using the χ² statistic. The Kaplan-Meier method was used to determine the probabilities of in-field PFS, biochemical PFS, DMFS and OS, with comparisons between groups determined using log rank statistics. Univariate analysis was performed using forward step-wise Cox regression. Two-sided p-values less than 0.05 were considered significant.

Results

Response to radiotherapy

Maximal response to radiotherapy for each evaluable patient is shown in Figure 1. Tumor size immediately prior to radiotherapy was compared to tumor size at the time of maximal imaging response. Only one evaluable patient developed progressive disease (PD) in-field after treatment, the remaining patients either had stable disease (37.9%) or a partial response to radiotherapy (41.4%). Median tumor size after radiotherapy was 4.3 cm (range 1–27 cm), with a median decrease in size of about 20%. In the seven patients with a tumor size of 10 cm or greater at the time of radiation, six had follow-up imaging to determine response. In these patients, an average 15% decrease in size following radiation was seen.

Figure 1. Tumor response to radiotherapy. Percent change in tumor size at time of maximal response to radiotherapy for all evaluable patients (n=24). RECIST category response to radiotherapy is also shown. PD – progressive disease, SD – stable disease, PR – partial response.

Local in-field progression

Median in-field PFS duration after radiotherapy was not reached, with a one year in-field PFS rate of 79% (Figure 2). On univariate analysis, age at diagnosis, gender, race, ECOG PS, CCI, therapy prior to radiation, tumor or nodal stage, number of lesions, tumor size or AFP at radiation and Child-Pugh class at radiation were not associated with in-field PFS rate after radiotherapy. However, BED (<75 vs. ≥75) was associated with in-field PFS (p = 0.032). Specifically, one year in-field PFS rate was 60% in patients treated to a BED < 75 compared to 88% in patients treated to a BED ≥ 75 (p = 0.023) (Figure 3). Also, the absence of portal vein thrombosis (PVT) was associated with improved in-field PFS (p = 0.031). Finally, RECIST category (PR vs. SD/PD) was associated with in-field PFS, however this did not reach statistical significance (p = 0.076).

Figure 2. Outcomes for the study population. A. Overall survival after XRT (OS). B. In field progression free survival after XRT (PFS). C. Biochemical failure free survival after XRT (Biochemical PFS). D. Distant metastasis free survival after XRT (DMFS). E. Median and one year outcomes. NR – Not reached.

Figure 3. The effect of BED on outcomes in unresectable HCC. A. Overall survival after XRT (OS) stratified by BED <75 vs. BED ≥75. B. In field progression free survival after XRT (PFS) stratified by BED <75 vs. BED >75. C. Biochemical recurrence free survival after XRT (Biochemical PFS) stratified by BED <75 vs. BED >75. D. Distant metastasis free survival after XRT (DMFS) stratified by BED <75 vs. BED ≥75. NR – Not reached.

Biochemical progession-free survival (PFS)

Median time to biochemical failure for the study population was 5.6 months, with a one year biochemical PFS of 30% (Figure 2). On univariate analysis age at diagnosis, gender, race, ECOG PS, systemic chemotherapy, tumor or nodal stage, number of lesions, tumor size or AFP at radiation, and Child-Pugh class were not associated with biochemical PFS after radiotherapy. Treatment with chemotherapy prior to radiation (no vs. yes, p = 0.029) and higher BED (≥75 vs. <75, p = 0.004) were associated with improved biochemical PFS. Specifically, median biochemical PFS in patients treated to a BED <75 was 1.6 months, compared to 6.5 months in patients treated to a BED ≥75 (p = 0.001). All evaluable patients treated to a BED <75 developed a biochemical failure by two months following XRT compared to a one year biochemical PFS of 40% in patients treated to a BED ≥75 (Figure 3).

Distant metastases

Median time to development of distant metastases for the study population was not reached, however one year DMFS was 66% (Figure 2). On univariate analysis age at diagnosis, gender, race, ECOG PS, CCI, therapy prior to radiation, tumor or nodal stage, number of lesions, tumor size, BED, dose and Child-Pugh class were not associated with DMFS after radiotherapy. However, RECIST category of response (PR vs. SD/PD) was associated with DMFS, as was AFP at the time of radiation (p = 0.003). One year DMFS was 100% in patients with a PR compared to 46% in patients with SD or PD (p = 0.036). Eight patients (27.6%) developed metastatic disease. The most common site of metastasis was the lung (five patients), with the remainder seen in the distant lymph nodes (one patient), peritoneal cavity (one patient) and bone (one patient).

Overall survival

Median OS for the patient population as a whole was 13 months, with a one year OS of 56%. (Figure 2). On univariate analysis, age at diagnosis, gender, race, ECOG PS, CCI, therapy prior to radiation, tumor or nodal stage, tumor size, and Child-Pugh class were not associated with OS after radiotherapy, however, number of lesions (single vs. multiple, p = 0.043), total radiation dose (≥50 Gy vs. <50 Gy, p = 0.01) and BED (≥75 vs. <75, p = 0.01) were associated with OS. Specifically, median survival in patients treated with a BED ≥ 75 was 18 months, with nearly 70% of patients alive at one year, compared to those patients treated to a BED <75 whose median survival was four months, with an 18% one year survival (p = 0.002) (Figure 3).

Toxicity

Acute and chronic toxicity due to radiotherapy were graded according to the CTCAE version 3.0 and are shown in Table V. One patient could have more than one adverse event. No grade 4 or greater toxicities were seen. Four patients (14%) developed grade 3 complications within six months following radiotherapy. Two patients developed a gastrointestinal (GI) bleed, one patient was found to have a symptomatic pleural effusion which required drainage after a small portion of the lower lobe of the lung was included in the radiation field and one patient developed markedly elevated liver function tests (LFTs), which eventually returned to baseline levels. Other common toxicities included ascites (10.3%), low grade elevation of LFTs above baseline (10.3%), nausea and vomiting (37.9%), pain (13.8%), pleural effusion (10.3%) and fatigue (24.1%). Median dose for patients who developed Grade 3 toxicity was 42.5 Gy (range: 35–66), while median BED was 74 (range: 64–139). In one patient with GI toxicity, maximum dose to the duodenum was 40 Gy. In the second patient with a GI bleed, dose to the duodenum/stomach was minimal (< 10 CGE); however the maximum dose to the esophagus was 57 CGE. On upper endoscopy, neither varices nor gastric ulceration were seen. Both platelet count and prothrombin time were within normal limits as well. In patients with a BED ≥ 75, Grade 2 or greater toxicity was seen in 56% of patients compared to 45% of patients in the BED < 75 group. Two instances of grade 3 toxicity were seen in each BED group (< 75 and ≥ 75).

Table 5. Complications and toxicity due to radiation.

	Grade
	1		2		3
Complication/toxicity	N	%	N	%	N	%
Anemia			1	3.4
Ascites			3	10.3
Dysphagia	1	3.4
Dermatitis	1	3.4
Edema			1	3.4
Elevated liver function tests	3	10.3			1	3.4
Fatigue	7	24.1
Gastrointestinal bleed					2	6.9
Nausea	7	24.1	4	13.8
Pain			4	13.8
Pleural effusion			2	6.9	1	3.4
Radiation therapy type
Conventional	7	24.1	10	34.4	0	0
Intensity modulated radiotherapy	7	24.1	2	6.9	3	10.3
Proton beam therapy	3	10.3	2	6.9	0	0

Discussion

In this study we examine the response to modern external beam radiotherapy in patients with unresectable HCCs at a single institution. We found that, in this patient population, the primary determining factor in both in-field progression and overall survival was the BED. Specifically, patients treated to higher BEDs were found to have improved in-field progression rates and overall survival. These findings are similar to those seen in the phase II trial from the University of Michigan using radiotherapy combined with hepatic infusion chemotherapy. In that trial, doses greater than 75 Gy were associated with a significantly improved overall and progression free survival [16]. In both studies, higher doses were associated with improved tumor response and survival, arguing for a need for dose-escalation in this patient population. Similarly, in patient populations with endemic HBV infection, higher radiation doses have been associated with improved outcomes [7,9,17–19].

Our current practice is to administer a BED of 75 or more, wherever feasible, for all non-metastatic Child-Pugh A and B, good performance status HCC patients with tumors that can be encompassed within a radiation field. The choice of treatment modalities is driven by the ability to adequately spare residual uninvolved liver and/or bowel mucosa. Proton therapy is utilized when tumors are farther away from bowel mucosa and either tumor motion during breathing is small (< 1 cm) and patients can be treated during free breathing or implanted fiducials permit breath-hold image-guided therapy. Often, when implanting fiducials is not possible (patient preference and/or technical limitations) for image-guided breath-hold treatment with protons (where we do not possess CT-on-rails or cone beam CT options), breath-hold IMRT treatment with CT-on-rails for image-guided therapy achieves better outcomes. Whenever possible, proton therapy is administered on protocol. The preferred dose and fractionation are 67.5 Gy in 15 fractions for more peripheral tumors and 58 Gy in 15 fractions for central tumors or those closer to bowel mucosa. Given the significantly poorer survival noted in patients who received a BED < 75, it seems reasonable to conclude that there are patients who may not benefit from radiation therapy. These include patients with a predicted survival of less than four months, particularly those with poor performance status, very large tumor size or widespread metastatic disease. Given the low levels of toxicity with radiation therapy, however, such patients could still benefit from liver-directed therapy for palliation of obstructive symptoms or pain.

In this study we observed significant responses to radiotherapy based upon RECIST criteria. Maximal radiographic response was typically seen by three months following the completion of radiotherapy; however some patients had continued decrease in tumor size up to 11 months after finishing treatment. We also observed that RECIST category was associated with freedom from in-field progression as defined by an increase in size of tumor after completion of radiotherapy as well as freedom from distant metastasis Previously some authors have suggested that use of RECIST criteria is of limited benefit in monitoring response to radiotherapy in this patient population compared to other methods such as the European Association for Study of the Liver (EASL) criteria, however, at least in this study, the use of the RECIST criteria appears to be appropriate [20].

In contrast to some other series [9,21–23], the presence of portal vein thrombosis was not associated with poorer overall survival in this study, however we did observe decreased local control of disease, as well as a trend to increased rates of metastasis in these patients. In the nine patients with PVT in this study two went on to receive further local therapy, with the remainder receiving systemic chemotherapy. Although recognizing the limited number of patients and retrospective nature of the current study, this lack of a detrimental survival effect of PVT may be a reflection of the efficacy of radiation therapy in making previously ineligible patients now eligible for liver-directed and/or systemic therapy.

This study is limited by both small patient numbers as well as a lack of randomization of patients to treatment arm. This variability in radiation approaches, although a reflection of technical innovations over time, is also a source of potential confounding factors that could ultimately have influenced outcome. In particular, it is likely that intangible factors such as physician assessment of patient's ability to tolerate fractionated therapy (not entirely captured by performance status and/or Child-Pugh score) may have contributed to the unwillingness to perform radiotherapy with large doses per fraction. Similarly, patients with large tumors and especially those closer to bowel mucosa may have been poorer candidates for escalated radiation doses and/or larger doses per fraction. Lastly, despite the availability of protons in the latter part of this time frame, its use was limited to tumors at least 1 cm away from adjacent bowel mucosa. Similarly, patients in this study were treated with multiple different modalities prior to radiotherapy, including surgical resection and TACE. However, the patients in this cohort were homogenous in respect to histology, which is an important criterion in determining both tumor response to radiotherapy as well as liver tolerance. Patients with other tumor types included in similar studies, such as metastatic colorectal cancer and cholangiocarcinoma, may have better liver reserve compared to those patients with HCC, thus affecting the reported toxicity.

In summary, our results provide further credence to the mounting evidence that external beam radiotherapy is effective in unresectable HCC in this patient population, can be delivered safely to high doses, and that higher BEDs correlate with improved outcomes.

Acknowledgements

This work was supported, in part, by NIH CA16672 Cancer Center Support Grant to The University of Texas MD Anderson Cancer Center.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.