Abstract
Background/Aim: This study aimed to compare the oncological outcomes of proton beam radiotherapy (PBT) with those of radiofrequency ablation (RFA) for newly diagnosed hepatocellular carcinoma (HCC). Patients and Methods: This study included 323 patients who underwent PBT (n=40) or RFA (n=283) as a curative treatment for previously untreated HCC between October 2016 and June 2021. The primary endpoints were local progression and toxicity. Results: The median follow-up was 3.4 years (range=1.1-5.7 years). In terms of portal vein tumor thrombosis, tumor size, alpha-fetoprotein, and prothrombin-induced by vitamin K absence-II, the PBT group had significantly more severe tumor burdens than those of the RFA group (p<0.0001, p<0.0001, p=0.0004, and p<0.0001, respectively). No significant difference was observed in cumulative local progression rate (10.4% in PBT vs. 7.8% in RFA at 3-years, p=0.895). Grade 3 or higher toxicity was reported in only one patient (0.4%) after RFA. Multivariable analysis demonstrated that treatment modality was not a significant prognostic factor for local progression (hazard ratio=1.05; 95% confidence interval=0.32-3.48; p=0.934). Conclusion: PBT demonstrated comparable local control with acceptable toxicity to RFA in newly diagnosed HCC. Therefore, PBT may be a valid alternative.
Surgical resection, liver transplantation, and radiofrequency ablation (RFA) are recommended as curative treatment options for hepatocellular carcinoma (HCC), according to the Barcelona Clinic Liver Cancer (BCLC) staging system (1). However, some patients with HCC are not amenable to initial definitive treatments. RFA is usually limited or technically infeasible because of the tumor size or location of the lesion (s) (2-5). Furthermore, these factors can affect disease progression (3). When initial treatment modalities are not feasible, other locoregional treatments, such as transcatheter arterial chemoembolization (TACE) or transarterial radioembolization, have been suggested as alternative treatment options in the current version of the BCLC guidelines (1). Although the efficacy of radiotherapy against HCC has been demonstrated in several prospective studies (6-9), the BCLC system has not yet recommended it (1).
Proton beam radiotherapy (PBT) enables the targeting of cancer cells with minimal damage to neighboring tissues, a property known as the Bragg peak. Recent advancements in image guidance with respiration motion control have further improved the accuracy of PBT. Due to these benefits, PBT has been recommended for the enhanced delivery of higher radiation doses to the target. For patients with advanced HCC, PBT showed both efficacy and safety when combined with lenvatinib (10). A recent phase III randomized controlled trial (RCT) demonstrated that PBT achieved a non-inferior local progression-free survival (LPFS) rate comparable to that of RFA and was tolerable in patients with recurrent HCC (11). Based on the physical characteristics of PBT and the results of phase III RCT, PBT may be considered as an acceptable alternative for achieving local control in the treatment of treatment-naïve HCC, particularly in cases where RFA is not feasible. However, only a few studies have compared the efficacy and safety of PBT and RFA in patients with newly diagnosed HCC. In this study, we aimed to identify the oncological outcomes and toxicity of PBT in patients ineligible for standard definitive treatments.
Patients and Methods
Study cohort. This retrospective study was performed at Samsung Medical Center. The eligibility criteria for the present study were as follows: 1) treatment initiation since October 2016, when PBT was initially performed as the first-line therapy for patients with treatment-naïve HCC; 2) PBT or RFA as the initial treatment; 3) treatment not combined with other treatment modalities; and 4) no prior therapy against targeted lesion(s). Twenty-seven patients were excluded due to missing data (n=23), baseline Child Pugh score ≥10 (n=1), modified Union for International Cancer Control (mUICC) stage IV (n=1), or baseline albumin-bilirubin (ALBI) grade 3 (n=2). Consequently, we included 323 patients who met the inclusion criteria.
The present study was approved by the Institutional Review Board of our institution (IRB No. 2022-11-023). The requirement for informed consent was waived owing to the retrospective nature of this study. The study was performed in accordance with the Declaration of Helsinki.
Study design and treatment. The treatment strategy was determined through a multidisciplinary tumor board meeting with hepatologists, surgeons, radiologists, interventional radiologists, and radiation oncologists.
Prior to RFA, planning ultrasonography was performed to evaluate the feasibility of ablation using the fusion imaging technique (Volume Navigation; GE Healthcare, Chicago, IL, USA), which combined real-time ultrasonography and computed tomography (CT)/magnetic resonance imaging. We aimed to achieve at least 5-10 mm ablative margins around the tumor. After the ablative procedure, tract cauterization was performed to avoid tumor seeding or tract bleeding.
Immediately after RFA, patients underwent contrast-enhanced liver CT to assess complete ablation or complications. When the index tumor was believed to be completely covered by the ablative zone on the immediate follow-up image, treatment was regarded as a technical success (12). One month after the treatment, follow-up radiological imaging and laboratory tests, including tumor markers, were performed. Follow-up examinations were performed every 3-4 months for the first two years, and every 4-6 months thereafter.
PBT has been suggested as a definitive treatment option for cases where other standard treatments are unsuitable. All patients considered candidates for PBT underwent respiration training and were assessed for technical eligibility before the simulation process and treatment. Clinicians evaluated the suitability of patient’s respiratory patterns for PBT and determined the optimal techniques for respiratory management based on the results of the training test (13). The gross tumor volume was delineated on the six sets of CT images using the breath-hold technique or on the 4D-CT images covering the amplitude of the gating window or whole respiratory phases (13). The clinical target volume (CTV) was defined as expansion up to 0.5 cm from the union of all delineated gross tumor volumes and was modified in accordance with the adjacent organs, if necessary (13). An additional margin of 0.3-0.5 cm was applied to the CTV to delineate the planning target volume (PTV) (13). A biologically effective dose for the prescribed dose was calculated using the standard linear-quadratic model with an α/β of 10 Gy for HCC, which is a commonly used value. Follow-up examinations, including enhanced abdominal CT scans, were performed at 1, 3, and 6 months after PBT. All patients were assessed at least every six months thereafter.
Outcomes and assessments. The patterns of failure following PBT or RFA were evaluated and classified as local, intrahepatic, or extrahepatic progression. Local progression was defined as progression within 1 cm off the margin of the RFA zone or PTV margin. Intrahepatic progression was defined as progression outside the local progression area within the liver. Treatment-related toxicities were graded according to the Common Terminology Criteria for Adverse Events version 5.0.
Subcapsular HCC located beneath the diaphragm was referred to as subphrenic HCC (12). Perivascular HCC was defined when abutting the portal or hepatic vein with a lumen caliber of 3 mm or wider (12).
The primary endpoints were local progression, which was calculated as the interval from the start date of each treatment to the date of local progression, censored at the date of the last follow-up when there was no evidence of local progression, and toxicity, defined as the presentation of treatment-related adverse events. Secondary endpoints were progression, calculated as the period from the start date of each treatment to the date of progression, censored at the date of the last follow-up when there was no evidence of progression, and cancer-specific death, calculated as the period from the start date of each treatment to the date of death due to liver cancer, censored at the date of the last follow-up when there was no evidence of the event. In addition, we identified the ALBI grades at baseline and at each follow-up time point to analyze the liver function after each treatment.
Statistical analysis. The chi-square test or Fisher’s exact test was used for categorical variables and the Mann–Whitney U-test or Student’s t-test was used for continuous variables to compare differences in characteristics between the PBT and RFA groups. Propensity score matching (PSM) analysis was performed to reduce selection bias and differences in potential confounders between the two groups. The PBT and RFA groups were matched (1:3 ratio) based on propensity scores using greedy nearest neighbor matching with a caliper of 0.2. Propensity scores were calculated using five covariables including ALBI grade, pretreatment alpha-fetoprotein (AFP) level, pretreatment prothrombin-induced by vitamin K absence-II (PIVKA-II) level, tumor size, and mUICC stage. Standardized mean differences were compared before and after PSM to evaluate the matching of balanced potential confounders. The multivariable analysis for local progression was performed using Cox regression analysis. We estimated the cumulative incidence rate of local progression, progression, and cancer-specific death, and used Gray’s test to test the difference in cumulative incidence curves between the two groups. The Fine-Gray subdistribution hazard model accounting for competing risk events was applied to both matched and unmatched datasets. Firth’s penalized likelihood approach was applied when there were no events. For the forest plots, hazard ratio (HR) with 95% confidence interval (CI) were obtained from a Cox model. The chi-square test was used to compare the failure patterns based on the treatment modality. The difference in ALBI grades between the two groups during the 12-month follow-up period was examined using generalized estimating equations. A two-sided p≤0.05 was considered statistically significant. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA) and R 4.1.0 (Vienna, Austria).
Results
Baseline characteristics. Patient and tumor characteristics according to the treatment modality are summarized in Table I. Compared with the RFA group, the PBT group was more frequently characterized by the presence of unfavorable factors, including advanced Eastern Cooperative Oncology Group (ECOG) performance status, elevated AFP level, elevated PIVKA-II level, large tumor size, bile duct invasion, advanced stage (mUICC stage, BCLC stage), macrovascular invasion, and portal vein tumor thrombus. Especially in the tumor size ≤3 cm population, tumor size (mean±standard deviation) was 2.1 cm (±0.5) and 1.5 cm (±0.5) in the PBT and RFA groups, respectively (p<0.0001). We subdivided ALBI grade 2 at baseline into 2a and 2b according to the modified ALBI definition. Before PSM, 7 (63.6%) and 34 (50.7%) patients in the PBT and RFA groups, respectively, were classified as grade 2a. After PSM, 4 (57.1%) and 10 (50.0%) patients in the PBT and RFA groups, respectively, were classified as grade 2a.
Patient and tumor characteristics according to the treatment modality before and after propensity score matching.
More than half (52.5%) of the 40 patients treated with PBT had their respiratory motion controlled solely by gating. In terms of the beam module, 22 (55.0%) patients were treated with passive scattering only, and 16 (40.0%) patients with pencil beam scanning only. The main (28/40) dose scheme was 66 Gy, delivered in ten fractions. The PBT characteristics are listed in Table II.
PBT characteristics (n=40).
Outcomes. The median follow-up duration was 3.4 years (range=1.1-5.7 years) in the total cohort (2.8 years in the PBT group and 3.7 years in the RFA group, respectively). Figure 1 shows cumulative oncological outcomes before and after PSM. After PSM, no significant difference was observed between the two groups in terms of local progression, progression, and cancer-specific death rate (p=0.979, 0.937, and 0.415, respectively).
Cumulative local progression, progression, and cancer-specific death incidence rate before and after propensity score matching. Before (p=0.895, p=0.021, p=0.063) and after (p=0.979, p=0.937, p=0.415) propensity score matching, respectively. PBT: Proton beam radiotherapy; RFA: radiofrequency ablation.
In terms of local progression and progression, we did subgroup analyses, and the forest plots results are depicted in Figure 2. Except for the subgroup with AFP at baseline ≥15 ng/ml, the PBT group was not significantly inferior to the RFA group in both local progression and progression rate.
Forest plots of local progression and progression in all patients and subgroups. (A) local progression, (B) progression. Favorable means neither subphrenic nor perivascular tumor location. Unfavorable means subphrenic or perivascular tumor location. RFA: Radiofrequency ablation; PBT: proton beam radiotherapy; HR: hazard ratio; AFP: alpha-fetoprotein; PIVKA-II: prothrombin-induced by vitamin K absence or antagonist-II; BCLC: Barcelona Clinic Liver Cancer.
The failure patterns in both groups are shown in Figure 3. Intrahepatic progression alone was the main failure pattern in both treatment groups (p=0.002 and 0.500 before and after PSM, respectively). Overall, no disease progression was observed in 19 (47.5%) and 170 (60.1%) patients in the PBT and RFA groups, respectively.
Patterns of failure based on the treatment modality before and after propensity score matching. Before propensity score matching (A), p=0.002 and after propensity score matching (B), p=0.500. PBT: Proton beam radiotherapy; RFA: radiofrequency ablation.
The univariable and multivariable analyses for local progression in the unmatched and matched groups are presented in Table III. In univariable analysis, no statistical difference was observed between the two groups in the unmatched and matched sets (HR=1.05, 95%CI=0.32-3.48, p=0.934 and HR=0.96, 95%CI=0.30-3.08, p=0.950, respectively). In addition, the multivariable analysis revealed that treatment modality was not associated with local progression (HR=0.45, 95%CI=0.08-2.70, p=0.385 and HR=4.31, 95%CI=0.60-30.85, p=0.146, respectively) in the unmatched and matched sets.
Univariable and multivariable analyses for local progression in unmatched and matched sets.
Safety. The treatment-related toxicities are summarized in Table IV. Most patients in both groups experienced ≤ grade 2 toxicities, except for one (0.4%) patient in the RFA group. In the RFA group, the incidence of grade 2 toxicities was 1.1%, with three patients exhibiting pleural effusion, and 0.7%, with two patients exhibiting fever due to thermal injury and inflammation, respectively. One patient who experienced a grade 3 adverse event visited the emergency room 6 days after treatment because of high fever and right upper quadrant pain. Abdominal CT revealed a cavitary lesion with gas within the RFA zone and gallbladder perforation. Consequently, percutaneous transhepatic gallbladder drainage was performed, and intravenous antibiotics were administered. None of the patients in either group experienced grade 4 or higher toxicities.
Toxicities according to the treatment modality.
Regarding the change in ALBI grade over a year, no significant difference was observed with time in the probability of having ≥ grade 2 ALBI before and after PSM (p=0.173 and p=0.181, respectively) (Figure 4).
Probability for grade 2 or higher albumin-bilirubin (ALBI) at each time point. Before propensity score matching (A), p=0.173, and after propensity score matching (B), p=0.181. PBT: Proton beam radiotherapy; RFA: radiofrequency ablation.
Discussion
The present retrospective cohort study analyzed the clinical efficacy and safety of PBT compared with those of RFA in patients with newly diagnosed HCC. Although patients treated with PBT had relatively more severe tumor burdens than those treated with RFA, in terms of lesion size and tumor markers, PBT demonstrated oncological outcomes comparable with those of RFA, including cancer-specific death and local progression rate. In addition, no treatment-related toxicities of grade 3 or higher were observed in the PBT group.
According to the BCLC guidelines, which are widely used globally for HCC treatment strategies and prognosis, surgical resection, liver transplantation, or RFA are recommended as curative treatments in the very early or early stages of the disease (1). In patients with preserved liver function and performance status 0, resection and ablation are recommended for single lesion or each lesion ≤3 cm, respectively (1). Prospective studies, including RCTs, have shown comparable oncological outcomes regarding overall survival (OS) or recurrence between RFA and surgery (14-18). In the case of RFA, there was a study indicating a similar efficacy and safety compared to microwave ablation for early HCC (19). Previous studies, including systematic reviews, have demonstrated that TACE improves survival outcomes in patients with unresectable tumors (20, 21). However, radiation therapy is not recommended as a curative option, as reported by recently updated BCLC guidelines (1).
Based on recent technical advances in radiation therapy, stereotactic body radiation therapy (SBRT) has demonstrated promising oncological outcomes for liver tumors (22). Several prospective studies have reported that SBRT improves oncological outcomes in patients with HCC (6, 7, 23). According to the literature, the 3-year LPFS and OS rate was 73-95% and 76-78%, respectively (6, 7). Additionally, Kim et al. reported a better local control rate of SBRT than that of RFA in patients with unresectable HCC (24). However, due to its unique physical properties related to dosimetric advantages, PBT may result in more favorable outcomes than SBRT during HCC management. A recent study reported that PBT is independently associated with longer survival than SBRT in patients with HCC (25). Moreover, in a phase II multi-institutional study, approximately 94.8% of patients with unresectable HCC demonstrated a local control rate of >80% at two years after high-dose hypofractionated PBT (26). Although a recent RCT comparing PBT with RFA for recurrent or residual HCC included patients who were considered eligible, regardless of the assigned treatment, a crossover was permitted in cases of technical infeasibility (11). The rate of switching to the other treatment modality was significantly lower in the PBT group than in the RFA group (8.3% vs. 26.4%; p<0.05) (11). In addition, LPFS and OS rates were not significantly different between the intention-to-treat and per-protocol cohorts (11). Furthermore, PBT was tolerable even in patients who were ineligible for RFA (11).
In our study, the clinical outcomes, including local progression and cancer-specific death rate, did not differ before and after PSM between the two groups (p>0.05). The PBT group demonstrated a 2- and 3-year cumulative incidence of local progression of 6.0% and 10.4%, respectively, which is comparable with the results in previous studies (8, 26-29). In addition, compared to SBRT, PBT may be considered a suitable option for selected patients with relatively large unresectable HCC lesion (s) to improve local control (23). In contrast, the PBT group demonstrated a higher progression rate than the RFA group before PSM (p=0.021). As these values included intrahepatic or extrahepatic progression rate, we assumed that several unfavorable factors, such as larger tumor size or higher tumor marker levels, would have affected the outcomes. After PSM, no statistically significant difference was observed in the progression rate (p=0.937).
Regarding safety, no difference in overall toxicities was observed between the two groups. We examined changes in ALBI after each treatment. Despite the PBT group having a larger tumor size than that in the RFA group before PSM (p<0.0001), no significant difference was observed in the probability of grade 2 or higher ALBI at each time point during the 1-year follow-up period (p=0.173). In addition, no significant difference was observed between the two treatment groups after PSM (p=0.181). In summary, PBT demonstrated comparable safety with RFA in terms of treatment-related morbidities and ALBI changes in the present study.
A recent retrospective study conducted at two hospitals in Japan compared PBT and RFA using PSM in treatment-naïve solitary HCC (30). Similar to our results, PBT did not differ from RFA in terms of local tumor progression, and no severe toxicity was identified in either group. However, the study focused on patients treated in the preceding years, and the majority of HCC cases were caused by hepatitis C infection. The results of both studies suggested that PBT may be performed for curative purposes in patients with newly diagnosed HCC.
Study limitations. First, there was a potential inherent selection bias owing to the retrospective nature of this study. The characteristics of the two groups were different and the treatment strategies were sequentially determined. Therefore, we performed PSM analysis to overcome these challenges. Second, due to the small sample size and single-institution design, the clinical outcomes may have been distorted. Third, although we performed PSM analysis, unobservable confounders may have persisted. Finally, clinical outcomes may have been overestimated because of the short follow-up duration. In particular, ALBI grade changes at the 1-year follow-up may not be sufficient to assess liver function changes after each treatment option. Despite these limitations, to the best of our knowledge, the present study is significant because it is the first to compare PBT and RFA as curative strategies, irrespective of tumor burden, in patients with newly diagnosed HCC.
Conclusion
In conclusion, we observed that PBT demonstrated no significant difference in local control compared to RFA in patients with initially diagnosed HCC. Both treatment modalities were safe and none of the patients in the PBT group reported any severe toxicity. Thus, our findings suggest that PBT could potentially serve as an acceptable alternative treatment option when RFA is not feasible due to the high risk of complications or treatment failure. However, further large-scale studies, including prospective RCTs, are required to validate our results.
Acknowledgements
The Authors thank the Samsung Medical Center HCC registry team for supplying the cohort data.
Footnotes
Authors’ Contributions
Study concept and design: Yu JI, Han S, and Park HC. Acquisition of data: Seo SH, Park HC, Yu JI, Yoo GS, Rhim H, Lee MW, Han S, and Choi MS. Analysis and interpretation of data: Seo SH, Park HC, Yu JI, Park B, and Shim JS. Drafting of the manuscript: Seo SH, Park B, and Yu JI. Critical revision of the manuscript for important intellectual content: All Authors. Statistical analysis: Seo SH, Yu JI, Park HC, Park B, and Shim JS. Funding acquirement: Yu JI. Administrative, technical, or material support: All Authors. Study supervision: Yu JI and Park HC.
Conflicts of Interest
The Authors declare that they have no conflicts of interest in relation to this study.
Funding
This research was supported by the Future Medicine 2030 Project of the Samsung Medical Center (SMX1220101) and the Basic Science Research Program through the National Research Foundation of Korea (NRF), which is funded by the Ministry of Education (NRF-2022R1C1C1005415).
- Received March 6, 2024.
- Revision received March 21, 2024.
- Accepted March 22, 2024.
- Copyright © 2024 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.











