Clinical results and risk factors of proton and carbon ion therapy for hepatocellular carcinoma

Authors


Abstract

BACKGROUND:

The objective of this study was to evaluate the clinical outcome of proton and carbon ion therapy for hepatocellular carcinoma (HCC).

METHODS:

In total, 343 consecutive patients with 386 tumors, including 242 patients (with 278 tumors) who received proton therapy and 101 patients (with 108 tumors) who received carbon ion therapy, were treated on 8 different protocols of proton therapy (52.8-84.0 gray equivalents [GyE] in 4-38 fractions) and on 4 different protocols of carbon ion therapy (52.8-76.0 GyE in 4-20 fractions).

RESULTS:

The 5-year local control and overall survival rates for all patients were 90.8% and 38.2%, respectively. Regarding proton and carbon ion therapy, the 5-year local control rates were 90.2% and 93%, respectively, and the 5-year overall survival rates were 38% and 36.3%, respectively. These rates did not differ significantly between the 2 therapies. Univariate analysis identified tumor size as an independent risk factor for local recurrence in proton therapy, carbon ion therapy, and in all patients. Multivariate analysis identified tumor size as the only independent risk factor for local recurrence in proton therapy and in all patients. Child-Pugh classification was the only independent risk factor for overall survival in proton therapy, in carbon ion therapy, and in all patients according to both univariate and multivariate analyses. No patients died of treatment-related toxicities.

CONCLUSIONS:

Proton and carbon ion therapies for HCC were comparable in terms of local control and overall survival rates. These therapies may represent innovative alternatives to conventional local therapies for HCC. Cancer 2011;. © 2011 American Cancer Society.

Hepatocellular carcinoma (HCC) is the fifth leading cause of cancer death worldwide, and the majority of patients with HCC reside in Asian countries.1, 2 HCC is well suited to local therapy, because it has a tendency to stay within the liver, and distant metastasis generally occurs late. This implies that curative local therapy, as represented by hepatectomy and liver transplantation, has a great impact on the disease course and also offers the best chance of long-term survival for patients with HCC.3, 4 However, only 5% to 40% of patients with HCC are amenable to a hepatectomy because of either advanced tumors or coexisting cirrhosis,5, 6 and a shortage of liver grafts limits the applicability of liver transplantation. Although local ablative therapies, such as radiofrequency ablation (RFA), recently have gained widespread clinical acceptance, there is growing evidence of a high local recurrence rate after RFA that reaches up to 36%.7, 8 In addition, local ablative therapies also are unsuitable for patients who have bleeding tendencies, unfavorable anatomic tumor locations, or large tumors.8, 9 Patients who are not eligible for local ablative therapies usually receive noncurative modalities, such as transarterial chemoembolization (TACE) or systemic chemotherapy.

Radiotherapy also is a local therapy but historically has played a limited role in the treatment of HCC, because the hepatic tolerance dose is lower than the tumoricidal dose, especially when liver function is impaired by chronic liver disease.10-12 Particle beams, such as proton and carbon ion beams, have demonstrated an increase in energy deposition with a penetration depth up to a sharp maximum at the end of their range: the so-called Bragg peak phenomenon.13 Therefore, higher tumor doses can be delivered without increasing toxicity to the surrounding noncancerous tissues and organs. Particle therapy results for HCC have been reported in several case series, all of which have reported good overall survival and encouraging local control rates.14-19 However, most of those studies were conducted at proton treatment centers, and few studies have reported results of carbon ion therapy for HCC. To our knowledge, no reports have focused on the differences in treatment results between the 2 types of particle beams.

The Hyogo Ion Beam Medical Center (HIBMC) is the only facility in the world that provides both proton and carbon ion therapies.20 In the current study, we analyzed the efficacy and safety of proton and carbon ion therapy for HCC at the HIBMC.

MATERIALS AND METHODS

Patient and Tumor Characteristics

The current study was conducted according to the Helsinki Declaration, and written informed consent was obtained from all patients. From May 2001 to January 2009, 343 consecutive patients with 400 HCCs were treated at the HIBMC (excluding 6 patients who discontinued treatment). Patients who met the following conditions were ineligible for treatment: 1) uncontrolled ascites and 2) tumors that measured >15 cm in greatest dimension (the upper limit of the irradiation field). No patients were lost to follow-up, although we could not evaluate the post-treatment imaging findings from 12 patients with 14 tumors. Thus, overall survival rates were determined for all 343 patients, and local control rates were determined for 386 tumors. In total, 242 patients with 278 tumors received proton therapy, and 101 patients with 108 tumors received carbon ion therapy. For all patients, HCC was diagnosed on the basis of the results from imaging studies, which usually included a combination of contrast-enhanced computed tomography (CT) and magnetic resonance imaging (MRI) studies. Tumor markers, including serum α-fetoprotein (AFP) and serum protein induced by vitamin K absence or antagonist II (PIVKAII), also were measured before and after treatment. Chest CT scans, bone scintigrams, and positron-emission tomography studies, if necessary, were obtained to exclude the possibility of distant metastasis.

Patient and tumor characteristics are summarized in Tables 1 and 2, respectively, for the proton and carbon ion therapy groups and for all patients. All the patients were staged and categorized as either operable (operable group) or inoperable (inoperable group) according to Barcelona Clinic Liver Cancer (BCLC) classification criteria.21 Tumors were classified into 3 groups according to tumor size (<50 mm, 50-100 mm, and >100 mm). All tumors were divided grossly into 4 types according to Liver Cancer Study Group of Japan criteria22: 1) single nodular type, 2) single nodular type with extranodular growth, 3) confluent multinodular type, and 4) infiltrative type. Studies have indicated that single nodular type tumors have a better prognosis than the other tumor types;23 therefore, all tumors were categorized further as either single nodular type or nonsingle nodular type. Macroscopic vascular invasion was defined as gross tumor vascular invasion into the portal or hepatic veins identified by pretreatment imaging. Perivascular location was defined as a situation in which the tumor invaded or abutted the main portal trunk and/or inferior vena cava. Among the 181 tumors that had received treatment before particle therapy, 2 tumors were classified as local recurrences after hepatectomy, and 60 tumors were classified as local recurrences after percutaneous local therapy. In addition, 169 target tumors had undergone TACE before particle therapy. All data were analyzed retrospectively for proton and carbon ion therapy, and all patients were considered with regard to local tumor control rates, overall patient survival rates, and treatment-related toxicities.

Table 1. Patient Characteristics
 No. of Patients (%)
CharacteristicProton Therapy, n=242Carbon Ion Therapy, n=101All Patients, n=343
  1. Abbreviations: BCLC, Barcelona Clinic Liver Cancer.

Age, y   
 <70115 (48)55 (54)170 (50)
 ≥70127 (52)46 (46)173 (50)
Sex   
 Men182 (75)73 (72)255 (74)
 Women60 (25)28 (28)88 (26)
Positive viral marker   
 Hepatitis B virus27 (11)19 (19)46 (13)
 Hepatitis C virus159 (66)60 (59)219 (64)
 None54 (22)21 (21)75 (22)
 Both2 (1)1 (1)3 (1)
Performance status   
 0172 (71)73 (72)245 (71)
 157 (24)18 (18)75 (22)
 210 (4)9 (9)19 (6)
 33 (1)1 (1)4 (1)
Child-Pugh classification   
 A184 (76)78 (77)262 (76)
 B55 (23)20 (20)75 (22)
 C3 (1)3 (3)6 (2)
BCLC stage   
 09 (4)9 (9)18 (5)
 A82 (34)36 (36)118 (34)
 B32 (13)15 (15)47 (14)
 C113 (47)37 (36)150 (44)
 D6 (2)4 (4)10 (3)
Recommended treatment according to BCLC stage   
  Resection: Operable group49 (20)29 (29)78 (23)
  Others: Inoperable group193 (80)72 (71)265 (77)
No. of tumors   
 Single213 (88)81 (80)294 (86)
 Multiple29 (12)20 (20)49 (14)
Table 2. Tumor Characteristics
 No. of Tumors (%)
CharacteristicProton Therapy, n=278Carbon Ion Therapy, n=108All Patients, n=386
  1. Abbreviations: AFP, α-fetoprotein; PIVKAII, protein induced by vitamin K absence or antagonist II.

Tumor size, mm   
 <50196 (71)81 (75)277 (72)
 50-10065 (23)22 (20)87 (22)
 >10017 (6)5 (5)22 (6)
Gross classification   
 Single nodular type153 (55)54 (50)207 (53)
 Single nodular with extranodular growth type85 (30)41 (38)126 (33)
 Confluent multinodular type13 (5)6 (6)19 (5)
 Infiltrative type27 (10)7 (6)34 (9)
Macroscopic vascular invasion   
 Yes73 (26)19 (18)92 (24)
 No205 (74)89 (82)294 (76)
Perivascular location   
 Yes121 (44)32 (30)153 (40)
 No157 (56)76 (70)233 (60)
Prior treatment history to the target tumor   
 Yes132 (47)49 (45)181 (47)
 No146 (53)59 (55)205 (53)
Serum AFP, ng/mL   
 <100184 (66)72 (67)256 (66)
 ≥10094 (34)36 (33)130 (34)
Serum PIVKAII, mAU/mL   
 <100129 (46)58 (54)187 (48)
 ≥100149 (54)50 (46)199 (52)

Treatment Protocol

The biologic effects of both proton and carbon ion therapy at the HIBMC were evaluated in vitro and in vivo, and the relative biologic effectiveness (RBE) values of these therapies were determined as 1.1 and 2,0 to 3.7, respectively (depending on the depth of the spread-out Bragg peaks).24 Because we assumed that all tissues had almost the same RBE for protons or carbon ions, doses expressed in gray equivalents (GyE), were directly comparable to photon doses.

Eight protocols for proton therapy (52.8-84 GyE in 4-38 fractions using 150-megaelectron volt [MeV], 190-MeV, 210-MeV, or 230-MeV proton beams) and 4 protocols for carbon ion therapy (52.8-76 GyE in 4-20 fractions using 250-MeV or 320-MeV carbon ion beams) were used during the study period (Table 3). The radiobiologic equivalent dose for acute-reacting tissues (BED10) was calculated for each protocol. The protocols for proton and carbon ion therapy were set first on the basis of earlier experience at the National Cancer Center East (Kashiwa, Japan), the Proton Medical Research Center (Tsukuba, Japan), and the National Institute of Radiological Sciences (Chiba, Japan). Thereafter, we adopted dose-escalation or hypofractionation protocols, depending on patient and tumor factors.

Table 3. Treatment Protocols
BED10aProtocol (BED10)No. of Patients [%]
  • Abbreviations: BED10, biologic effective dose for acute-reacting tissues; Fr, fractions; GyE, gray equivalents.

  • a

    The BED10 was calculated by linear-quadratic formalism assuming an α/β ratio of 10 GyE.

Proton therapy  
 <10076 GyE/38 Fr (91.2)11 [4]
 56 GyE/8 Fr (95.2)4 [2]
 60 GyE/10 Fr (96.0)89 [37]
 ≥10076 GyE/20 Fr (104.88)70 [29]
 66 GyE/10 Fr (109.56)53 [22]
 80 GyE/20 Fr (112)3 [1]
 84 GyE/20 Fr (119.28)3 [1]
 52.8 GyE/4 Fr (122.496)9 [4]
Carbon ion therapy  
 <10052.8 GyE/8 Fr (87.648)23 [23]
 ≥10076 GyE/20 Fr (104.88)3 [3]
 66 GyE/10 Fr (109.56)16 [16]
 52.8 GyE/4 Fr (122.496)59 [58]

The policy for the selection of beam type was determined by the following: 1) from May to October 2001 and from April 2003 to March 2005, only proton therapy was available (52 patients with 57 tumors); 2) from February to June 2002, only carbon ion therapy was available (6 patients with 6 tumors); and 3) since April 2005, treatment plans for both proton and carbon ion therapy were made for all patients, and a better suited beam was selected on the basis of the treatment plans (285 patients with 323 tumors). Regarding the choice of either proton beam or carbon ion beam therapy, the following factors were considered: 1) the values for the percentage prescription dose received by at least 95% volume (D95) of the gross tumor volume (GTV), 2) D95 of the clinical target volume (CTV), 3) D95 of the planning target volume (PTV), 4) the percentage of the volumes of hepatic noncancerous portions (entire liver volume − GTV) receiving ≥30 GyE (Liver V30), 5) the maximum exposure doses of the adjacent gut (Gut Dmax), 6) the percentage of the volumes of the adjacent gut receiving ≥40 GyE (Gut V40), 7) the maximum exposure doses to the skin, and 8) the maximum exposure doses to the ribs. D95 of the PTV and Liver V30 values have always been high-priority factors. Among these factors, Liver V30 is used as the most important factor for patients whose liver function already has deteriorated, and Gut Dmax and/or Gut V40 values have become secondary major concerning factors in patients who have tumors located close to the gut.

A representative case presentation of treatment plans for both proton therapy and carbon ion therapy is provided in Figure 1. The D95 of PTV was equal for proton and carbon ion therapy. Conversely, Gut Dmax and Gut V40 were significantly higher for the proton treatment plan than for the carbon ion treatment plan. Therefore, carbon ion therapy was selected in this representative case.

Figure 1.

These images are from a representative case presentation of treatment plans for both (A) proton therapy and (B) carbon ion therapy. D95 indicates the dose received by at least 95% volume; PTV, planning target volume; Liver V30, percentage volumes of noncancerous hepatic portions (entire liver volume − gross tumor volume) that received ≥30 gray equivalents (GyE); Gut Dmax, the maximum exposure doses of the adjacent gut; Gut V40, percentage volumes of the adjacent gut that received ≥40 GyE.

Treatment Planning

The radiation treatments were designed to use a CT-based, 3-dimensional treatment planning system (FOCUS-M; CMS, Tokyo, Japan; and Mitsubishi Electric, Kobe, Japan). CT images were obtained at the phase of expiration using a respiratory gating system. A respiratory gating irradiation system that was developed at the National Institute of Radiological Sciences in Chiba25 was used for irradiation of the beam during the exhalation phase for all patients. The GTV and the organs at a risk of irradiation, such as the liver and intestines, were delineated according to fusion images that were constructed from contrast-enhanced CT and MRI studies. Treatment planning was defined as follows: CTV = GTV + 5 mm, PTV = CTV + 5 mm. In addition, another 5-mm to 10-mm margin was included in the caudal axis to compensate for uncertainty caused by respiration-induced hepatic movements. Doses were calculated on the basis of the pencil beam algorithm. Beam parameters, including energy level, the width of the spread-out Bragg peak, and degrader thickness, were selected adequately using FOCUS-M. Dose-volume histograms were calculated for all patients to evaluate the risk of radiation-induced liver disease.

Follow-Up and Evaluation Criteria

Patients underwent a complete blood count, biochemical profile, detection of tumor markers (including serum AFP and PIVKAII), and abdominal imaging studies (CT or MRI) every 3 months for 3 years after treatment and every 6 months thereafter. In general, for patients with HCC, the objective of all effective locoregional therapies is to obtain necrosis of the tumor regardless of the shrinkage of the lesion. Even if extensive tumor necrosis is achieved, this may not be accompanied by a reduction in the greatest dimension of the lesion. Consequently, several studies have indicated that World Health Organization and Response Evaluation Criteria in Solid Tumors criteria have no value in the assessment of tumor response after locoregional therapies in patients with HCC.26, 27 It has been reported that such tumors, even after complete response, tend to persist for a long period after the completion of particle therapy.19 Therefore, local recurrence was defined either as the growth of an irradiated tumor or as the appearance of new tumors within the PTV based on criteria established in previous reports.16, 17, 19, 28 Acute and late toxicities were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (version 2.0; National Cancer Institute, Bethesda, Md).

Statistical Analyses

The statistical significance of differences in each classification for both local control and overall survival rates was estimated by the Kaplan-Meier method and was compared using the log-rank test. Univariate and multivariate analyses using Cox proportional hazards regression models were used to identify independent risk factors that predicted local control and overall survival rates. Differences of P < .05 were considered statistically significant, and variables with P < .10 were entered into a multivariate analysis using a Cox proportional hazards model. All statistical analyses were performed using SPSS statistical software (version 17.0 for Windows; SPSS, Inc., Chicago, Ill).

RESULTS

Local Control Rates After Proton and Carbon Ion Therapies

Patients were followed either until death or to March 2010 (median follow-up, 31.0 months). Among 343 patients with 386 tumors, 223 patients developed recurrences after treatment. Nineteen patients developed extrahepatic metastasis, and 210 patients developed intrahepatic recurrences, including 23 local recurrences (proton therapy, 18 patients; carbon ion therapy, 5 patients). The longest interval to local recurrence was 27.1 months, and all local recurrences developed within 3 years. The 5-year local control rates for patients who received proton therapy and carbon ion therapy were 90.2% and 93%, respectively (Fig. 2A). The effective 3-year and 5-year local control rates for all 386 tumors were both 90.8% (Fig. 2B). An analysis of the local control rates according to the tumor factors identified above (see Treatment Protocols) is listed in Table 4. Univariate analysis revealed that tumor size was a significant risk factor for local recurrence in the proton therapy group, the carbon ion therapy group, and all patients. In multivariate analysis, tumor size was identified as an independent risk factor for local recurrence in the proton therapy group and in all patients (Table 5). In addition, the local control rates for all 386 tumors that measured <50 mm, 50 to 100 mm, and >100 mm were 95.3%, 84.4%, and 42.2%, respectively (Fig. 2C). In contrast, other tumor factors, including gross classification, macroscopic vascular invasion, perivascular location, treatment history, serum AFP level, and serum PIVKAII level, did not affect the local control rate in any tumor subset in multivariate analysis.

Figure 2.

Local control rates after treatment are illustrated for (A) tumors that were treated with proton and carbon ion therapy, (B) all 386 tumors, and (C) all 386 tumors according to tumor size.

Table 4. Univariate Analysis of Prognostic Factors for Local Control Rate
 Proton Therapy, n=278Carbon Ion Therapy, n=108All Patients, n=386
FactorLC Rate at 5 Years, %PLC Rate at 5 Years, %PLC Rate at 5 Years, %P
  1. Abbreviations: AFP, α-fetoprotein; LC, local control; PIVKAII, protein induced by vitamin K absence or antagonist II.

Tumor size, mm <.0001 .0062 <.0001
 <5095.5 94.5 95.3 
 50-10084.1 90.9 84.4 
 >10043.4 80 42.2 
Gross classification .0901 .0943 .0219
 Single nodular type93.3 96 94 
 Nonsingle nodular type86.2 89.4 86.7 
Macroscopic vascular invasion .2544 .0292 .0535
 Yes83.9 80.4 82.8 
 No92 94.8 92.8 
Perivascular location .0704 .4267 .0403
 Yes85.5 86.8 85.7 
 No93.5 95.1 93.8 
Prior treatment history .7332 .9000 .7629
 Yes91.5 95 92 
 No89.2 91.9 89.9 
Serum AFP, ng/mL .5352 .6111 .4310
 <10090.9 95.1 91.8 
 ≥10089 86.8 88.6 
Serum PIVKAII, mAU/mL .0997 .3468 .2976
 <10094.5 90.1 93.4 
 ≥10085.5 97.9 87.8 
Table 5. Independent Risk Factors Related to the Local Control Rate: Multivariate Analysis
FactorSEChi-Square StatisticRR95% CIP
  1. Abbreviations: CI, confidence interval; PIVKAII, protein induced by vitamin K absence or antagonist II; RR, relative risk; SE, standard error.

Proton therapy     
 Tumor size, mm    .0030
  50-100 (vs <50)0.6661.1752.0580.558-7.590 
  >100 (vs <50)0.70310.4639.7252.450-38.596 
 Single nodular type (vs nonsingle nodular type)0.5380.1871.2620.440-3.623.6652
 Perivascular location: Yes (vs no)0.5430.1470.8120.280-2.354.7011
 Serum PIVKAII ≥100 mAU/mL (vs <100 mAU/mL)0.5300.3891.3920.492-3.937.5327
Carbon ion therapy     
 Tumor size, mm    .4703
  50-100 (vs <50)1.5690.0690.6620.031-14.322 
  >100 (vs <50)1.9050.5754.2390.101-177.314 
 Single nodular type (vs nonsingle nodular type)1.2311.1103.6580.328-40.853.2921
 Macroscopic vascular invasion: Yes (vs no)1.5850.3472.5440.114-56.848.5557
All patients     
 Tumor size, mm    .0002
  50-100 (vs <50)0.64610.5278.1222.291-28.789 
  >100 (vs <50)0.5622.1460.4390.146-1.321 
 Single nodular type (vs nonsingle nodular type)0.5192.5442.2880.827-6.327.1107
 Macroscopic vascular invasion: Yes (vs no)0.6512.6012.8600.798-10.253.1068
 Perivascular location: Yes (vs no)0.5060.7380.6470.240-1.745.3902

Overall Survival Rates of Proton and Carbon Ion Therapies

The 5-year overall survival rates for patients who received proton therapy and carbon ion therapy were 38% and 36.3%, respectively (Fig. 3A). The overall survival rates for all 343 patients at 3 years and 5 years were 59% and 38.2%, respectively (Fig. 3B). Univariate and multivariate analyses of the overall survival rates according to the 8 relevant tumor factors are provided in Tables 6 and 7, respectively. According to the univariate analysis, Child-Pugh classification, macroscopic vascular invasion, and serum AFP levels were the only factors that significantly affected the overall survival rates in all groups (proton therapy, carbon ion therapy, and all patients) (Table 6). The Child-Pugh classification was the only independent factor for overall survival in proton therapy, carbon ion therapy, and all patients according to the multivariate analysis (Table 7). The 5-year overall survival rates for Child-Pugh classifications A, B, and C were 46.6%, 8.7%, and 0%, respectively (Fig. 3C).

Figure 3.

Overall survival rates after treatment are illustrated for (A) patients who received with proton and carbon ion therapy, (B) all 343 patients, and (C) all 343 patients according to Child-Pugh classification.

Table 6. Univariate Analysis of Prognostic Factors for Overall Survival Rate
 Proton Therapy, n=242Carbon Ion Therapy, n=101All Patients, n=343
FactorOS Rate at 5 Years, %POS Rate at 5 Years, %POS Rate at 5 Years, %P
  1. Abbreviations: AFP, α-fetoprotein; OS, overall survival; PIVKAII, protein induced by vitamin K absence or antagonist II.

Age, y .7986 .6769 .6448
 <7037.3 43.9 39.4 
 ≥7038.2 26.9 36.2 
Positive viral marker .9754 .1805 .8586
 Hepatitis B virus34.9 44.6 32.4 
 Hepatitis C virus35.8 40.8 36.3 
 None46.7 33.9 46.6 
Performance status <.0001 .2295 <.0001
 043.5 43.7 43.6 
 1 or 2 or 324.8 26.8 24.1 
Child-Pugh classification <.0001 <.0001 <.0001
 A46.8 41.2 46.6 
 B or C8.2 33.3 8 
Tumor size, mm .1438 .0003 .0038
 <5037.8 53.5 39.2 
 50-10037.4 17.9 33.8 
 >10041.1 0 39.8 
Macroscopic vascular invasion .0003 .0055 <.0001
 Yes33.2 22 31.5 
 No40.3 47.8 40.2 
Serum AFP, ng/mL .0026 .0024 <.0001
 <10042 30.9 42.6 
 ≥10029.5 23.1 28.9 
Serum PIVKAII, mAU/mL .0109 .4041 .0082
 <10040.4 58.4 41.8 
 ≥10035.5 16.5 33.8 
Table 7. Independent Risk Factors Related to the Overall Survival Rate: Multivariate Analysis
FactorSEChi-Square StatisticRR95% CIP
  1. Abbreviations: AFP, α-fetoprotein; CI, confidence interval; PIVKAII, protein induced by vitamin K absence or antagonist II; RR, relative risk; SE, standard error.

Proton therapy     
 Performance status 1-3 (vs 0)0.2009.2830.5440.368-0.805.0023
 Child-Pugh classification B or C (vs A)0.20429.7310.3290.220-0.490<.0001
 Macroscopic vascular invasion: Yes (vs no)0.2039.4100.5360.360-0.799.0022
 Serum AFP ≥100 ng/mL (vs <100 ng/mL)0.1982.2811.3490.915-1.990.1310
 Serum PIVKAII ≥100 mAU/mL (vs <100 mAU/mL)0.1991.2311.2480.844-1.844.2672
Carbon ion therapy     
 Child-Pugh classification B or C (vs A)0.51917.6420.1130.041-0.313<.0001
 Tumor size, mm    .0297
  50-100 (vs <50)0.5696.7954.4121.445-13.468 
  >100 (vs <50)1.0403.2176.4540.841-49.524 
 Macroscopic vascular invasion: Yes (vs no)0.6250.6471.6540.486-5.631.4211
 Serum AFP ≥100 ng/mL (vs <100 ng/mL)0.3965.4062.5131.156-5.465.0201
All patients     
 Performance status 1-3 (vs 0)0.18010.8520.5540.389-0.787.0010
 Child-Pugh classification B or C (vs A)0.18245.6630.2920.204-0.417<.0001
 Tumor size, mm    .5976
  50-100 (vs <50)0.2200.0441.0470.680-1.613 
  >100 (vs <50)0.3750.6560.7380.354-1.539 
 Macroscopic vascular invasion: Yes (vs no)0.21610.9600.4890.320-0.747.0009
 Serum AFP ≥100 ng/mL (vs <100 ng/mL)0.1764.8481.4741.044-2.083.0277
 Serum PIVKAII ≥100 mAU/mL (vs <100 mAU/mL)0.1860.9221.1960.830-1.724.3371

The 5-year overall survival rates for BCLC stages 0, A, B, C, and D were 80.8%, 52.7%, 23.7%, 30.6%, and 0%, respectively (Fig. 4A). According to the BCLC classification, hepatic resection was categorized as stage 0 and part of stage A. In total, 78 patients were categorized into the hepatic resection group. The 5-year overall survival rates for patients classified into groups according to whether they underwent hepatic resection (operable group) or received treatments (inoperable group) were 67.6% and 29.4%, respectively (P < .0001) (Fig. 4B).

Figure 4.

(A) Overall survival rates are illustrated for all 343 patients according to disease stage classified by the Barcelona Clinic Liver Cancer classification. (B) Overall survival rates are illustrated for all 343 patients according to the operative indication based on the Barcelona Clinic Liver Cancer classification.

Local Control and Overall Survival Rates According to the BED10

We also analyzed the local control and overall survival rates after both proton and carbon ion therapies according to the BED10 using a cutoff score of 100 (Fig. 5). The 5-year local control rates for tumors that were treated on the protocols characterized by BED10 values <100 and ≥100 were 93.3% and 87.4%, respectively, for proton therapy and 80.7% and 95.7%, respectively, for carbon ion therapy. The 5-year overall survival rates for patients who were treated on the protocols characterized by BED10 values <100 and ≥100 were 31.7% and 43.9%, respectively, for proton therapy and 32.3% and 48.4%, respectively, for carbon ion therapy. There was no significant difference in local control and overall survival rates, irrespective of the BED10 score, between proton therapy and carbon ion therapy.

Figure 5.

Local control rates are illustrated according to the radiobiologic equivalent dose for acute-reacting tissues (BED10) for (A) proton therapy and (B) carbon ion therapy. Overall survival rates are illustrated according to the BED10 for (C) proton therapy and (D) carbon ion therapy.

Toxicities

All acute toxicities that occurred during treatment were transient, easily managed, and acceptable. However, grade ≥3 late toxicities were observed in 8 patients on proton therapy and in 4 patients on carbon ion therapy, and 4 of 12 patients were diagnosed with radiation-induced liver disease (Table 8). However, all of these patients with hematologic disorders were asymptomatic and required no further treatment. In addition, upper gastrointestinal ulcer, pneumonitis, and subcutaneous panniculitis healed with conservative management. Five patients who received proton therapy developed refractory skin ulcers, and 1 patient required skin transplantation. A salvage drainage operation also was required by 1 patient who developed infectious biloma 10 months after irradiation. No patients died of treatment-related toxicity.

Table 8. Late Toxicities After Proton and Carbon Ion Therapy
 No. of Patients (%)
 Grade 2Grade 3Grade 4
ToxicityProton TherapyCarbon Ion TherapyAll PatientsProton TherapyCarbon Ion TherapyAll PatientsProton TherapyCarbon Ion TherapyAll Patients
Dermatitis12 (5)5 (5)17 (5)4 (2)04 (1)1 (1)0 (0)1 (1)
Elevation of transaminase level5 (2)3 (3)8 (2)1 (1)3 (3)4 (1)0 (0)0 (0)0 (0)
Upper gastrointestinal ulcer3 (1)1 (1)4 (1)1 (1)0 (0)1 (1)0 (0)0 (0)0 (0)
Rib fracture8 (3)3 (3)11 (3)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
Pneumonitis4 (2)2 (2)6 (2)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
Subcutaneous panniculitis6 (2)2 (2)8 (2)0 (0)1 (1)1 (1)0 (0)0 (0)0 (0)
Biloma0 (0)0 (0)0 (0)1 (1)0 (0)1 (1)0 (0)0 (0)0 (0)
Low albuminemia1 (1)0 (0)1 (1)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
Nausea/anorexia/pain/ascites4 (2)2 (2)6 (2)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)

DISCUSSION

We analyzed the safety and efficacy of particle therapy using proton and carbon ion beams for HCC in a single center. The key findings of this study are as follows: 1) particle therapy produced excellent local control and overall survival rates with acceptable adverse events, 2) the treatment results from carbon ion therapy appeared to be equivalent to those from proton therapy, and 3) tumor size was the only risk factor that affected the local control rate.

Local control rates for both proton therapy and carbon ion therapy exceeded 90% in the current study. These data are very similar to those related to particle therapy for HCC, whereas they are superior to data related to conformal radiotherapy.16, 18, 29, 30 Recent improvements in dose localization techniques, such as intensity-modulated radiotherapy, conformal 3-dimensional planning, and breathing motion management strategies, thus, have made it possible to irradiate smaller, well defined targets in the liver. However, these highly computer-assisted irradiation techniques using photon beams have achieved limited efficacy in treating patients with HCC. The local control rates produced by these conformal approaches remain in the 40% to 66% range for several reasons.29, 30 Radiation-induced liver disease still is observed frequently with conformal approaches when a sufficient dose is delivered to completely kill the cells of the entire tumor nodule. This is especially the true for large and centrally situated liver tumors.31 In this regard, particle beams can achieve an excellent dose distribution to these targets. The area of radiation dose deposition can be controlled well by the beam energy, because there is a rapid drop-off in energy deposition beyond the target area. Indeed, such theoretical advantages of particle therapy were proven in part by the impressively high local control rate of approximately 90% in the current study. Therefore, we believe that it is reasonable to say that the tumor-eliminating capability of particle therapy is closely equivalent to that of hepatectomy, an outcome that has not been achieved with other radiation therapies.

Experience in the treatment of HCC by particle therapy has been accumulated mainly in Japanese centers, but there is increasing interest in other countries as well. There were 26 active proton therapy facilities as of February 2009, whereas there were only 3 carbon ion therapy facilities.32 Until now, several proton treatment centers and 1 carbon ion treatment center have reported HCC treatments results.16-19 However, except for the HIBMC, no single facility can deliver both proton and carbon ion beams. Therefore, our facility has a distinct advantage over other institutes with regard to comparing the efficacy of the 2 beams. To select proton therapy or carbon ion therapy, we made treatment plans for both proton and carbon ion therapy. When dose distributions were compared, there were many instances in which low-dose areas had spread into the surrounding normal liver during proton therapy planning. This was apparently because of the relatively large penumbra of proton beams. Consequently, dose distribution in a single beam appears to be better in carbon ion therapy than in proton therapy. However, in terms of beam arrangement, carbon ions are emitted from 3 fixed ports, such as vertical, horizontal, or 45-degree oblique; whereas a 360-degree rotating gantry can be used for protons. The high positioning accuracy achieved by irradiating patients in a supine position also was an advantage of proton therapy. Currently, 360-degree rotating gantries for carbon ion beams are under construction in Japan and Germany, and it is expected that these will enable the delivery of highly precise carbon ion beam arrangements and, thus, will improve the effectiveness of carbon ion therapy for HCC.

In addition to dose distribution, there are evident differences in biologic properties between the 2 beams, ie, the RBE. The RBE for proton therapy is comparatively simple. The International Commission on Radiation Units and Measurements has recommended 1.1 as a generic RBE for proton therapy based on an analysis of the published RBE values determined from in vivo systems.33, 34 All proton therapy centers, including the HIBMC, have accepted this recommendation. Conversely, the RBE for carbon ion therapy is complex, because there is no common model for selecting the RBE of carbon ion beams. In addition, it may vary depending on tissue type and the depth of the spread-out Bragg peaks.32 Because of these differences, planning the physical dose distribution is substantially more complex for carbon ion beams than for proton beams; therefore, a direct comparison of proton therapy and carbon ion therapy is not feasible. Under these circumstances, we established that the treatment results of carbon ion therapy were equivalent to those of proton therapy at our institute. These results may prove the validity of our treatment planning system for carbon ion therapy by using a variable RBE.

The current study has established the equal effectiveness of proton and carbon ion therapies for HCC. With regard to this result, we speculate that the superior dose distribution compensates for the limitation of carbon ion beam arrangements at HIBMC. With the development of irradiation equipment, compared with proton therapy, carbon ion therapy will play a major role in the treatment of patients with HCC who have tumors adjacent to the gut and/or those whose liver function has deteriorated. However, carbon ion therapy requires huge economic resources, and this issue should be resolved in the future.

Tumor size was the only significant risk factor for local recurrence after particle therapy (for proton therapy, carbon ion therapy, and all patients). Conversely, it is noteworthy that the 6 other tumor factors, including gross classification, macroscopic vascular invasion, perivascular location, prior treatment history, serum AFP levels, and serum PIVKAII levels, had no significant influence on the local control rate after either therapy. The application of local ablative therapies is contraindicated in tumors with vascular invasion,9, 35 and it has been reported by several studies that perivascular location significantly increased the local recurrence rate after RFA mainly because of the heat-sink effect.8, 9 In addition, hepatectomy frequently is abandoned to as a treatment for centrally situated tumors adjacent to the inferior vena cava and/or the main portal trunk in patients with cirrhosis, because these tumor locations tend to require major hepatectomy. In the current study, however, neither factor reduced the efficacy of proton therapy or carbon ion therapy in terms of the local control rate.

The local control rates achieved with proton therapy and carbon ion therapy for tumors <50 mm were 95.5%, and 94.5%, respectively. These data are similar or superior to those reported with local ablative therapies.36 At the same time, the local control rates achieved with proton therapy and carbon ion therapy for tumors that measured from 50 mm to 100 mm in greatest dimension were 84.1% and 90.9%, respectively (Table 4). Because the upper limit of tumor size is 50 mm for local ablative therapies, these results clearly demonstrate the distinct advantage of particle therapy over other local therapies for tumors ≥50 mm. Taken together, in our opinion, particle therapy would be the best therapeutic option for patients who have tumors that preclude currently available local therapies because of tumor size, macroscopic vascular invasion, or deep tumor location.

According to the BCLC classification, the 5-year overall survival rate of patients in the operable group was 67.6%. This survival rate is comparable to reported data associated with hepatic resection.21 It is noteworthy that the overall survival rate of patients classified with stage C disease at 5 years was 30.6% in the current study; this is far superior to other reported data.21 Patients in this stage have macroscopic vascular invasion and/or extrahepatic metastasis. According to the BCLC classification, these patients usually are excluded from curative treatments and receive either TACE or sorafenib. In the current study, most of patients with stage C disease had macroscopic vascular invasion without extrahepatic metastasis. They were received proton and carbon ion therapies with curative intent, and the local control rates for these patients exceeded 80% (Table 4). These results suggest that some of patients with BCLC stage C disease may benefit from more aggressive local therapies, such as particle therapy.

Most of the treatment-related toxicities in the current study were transient, easily managed, and acceptable. Rib fracture and dermatitis were observed frequently in patients who were treated during the early period at our center. Most of these patients, including 1 patient with grade 4 dermatitis, were treated with only 1 portal to obtain an adequate spread-out Bragg peak. Thereafter, we used 2 or more portals and rarely observed such complications. Regarding intrahepatic structure-related complications, no studies, including ours, have reported blood vessel-related complications. This is a distinct advantage of particle therapy over other local therapies and supports our proposal that tumors in perivascular locations are appropriate candidates for particle therapy. In contrast, although less common, bile duct complications, including biloma and stenosis, have been reported in several studies.16 In the current study, biloma formation was observed in 1 patient whose tumor was adjacent to the porta hepatis. The bile duct may stand as the single greatest obstacle of intrahepatic structures after particle therapy. It is almost impossible to predict bile duct complications before treatment; thus, tumors adjacent to the porta hepatis should be treated with caution.

Grade 2 or greater gastrointestinal ulceration was observed in 5 patients whose tumors were adjacent to the gut. To minimize toxicity in these patients, we reduced the fraction size and initiated proton pump inhibitors immediately after treatment; and, ultimately, we were able to prevent the development of severity. The proximity of the gut is an important consideration in selecting particle therapy for patients with HCC. We introduced operative placement of a spacer between the tumor and the gut before particle therapy as a countermeasure for this limitation to ensure safe irradiation.37, 38

To our knowledge, this is the first study to assess the clinical treatment results from both proton therapy and carbon ion therapy. However, our study has some important limitations: 1) the results of this study were achieved retrospectively and not through randomized or controlled trials; 2) during the study period, we used different treatment protocols for proton therapy and carbon ion therapy; and 3) the RBE of carbon ion beams for HCC has not been completely clarified. Although further investigation is required, our data can serve as a basis for future refinement of beam selection.

In conclusion, both proton therapy and carbon ion therapy produce favorable results as treatment for HCC. Both therapies have great advantages in treating HCC, a condition that is a contraindication for other local therapies. Randomized clinical trials are required to compare particle therapy with other local therapies and to clarify the roles played by particle therapy in the HCC treatment algorithm.

FUNDING SOURCES

This study was supported by grants-in aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology of Japan to Y.H. (C-21591773), Y.K. (C-20591611), and M.M. (B-22390234) and by grants for Global Center of Excellence Program for Education and Research on Signal Transduction Medicine in the Coming Generation “Bringing Up Clinician-Scientists in the Alliance Between Basic and Clinical Medicine” (to Y.K.).

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary