Data from patients with HCC who had undergone C-ion RT at Gunma University Heavy Ion Medical Center between June 2010 and November 2018 were retrospectively collected. A total of 210 HCC patients received C-ion RT during the study period; 10 patients with LAHCC (i.e., HCC involving a major branch of the portal or hepatic vein, including the PVTT or IVCTT) were investigated. The diagnoses of all patients were confirmed by the presence of typical hallmarks of HCC using four-phase multi-detector-row computed tomography (CT) or dynamic contrast-enhanced magnetic resonance imaging. This study complied with the standards set forth in the Declaration of Helsinki and current ethical guidelines, and was reviewed and approved by our Institutional Review Board.
Treatment-planning CT with respiratory-gated CT, as well as four-dimensional CT images were acquired. Treatment-planning and contrast-enhanced CT images were merged to precisely delineate the gross tumor volume (GTV). The clinical target volume (CTV) was defined as the GTV plus least a 5 mm margin in all directions to encompass any microscopic disease progression. The internal target volume (ITV) was defined as the sum of CTV from the four-dimensional CT images. The planning target volume (PTV) was defined as the sum of the ITV and an additional margin for patient setup discrepancies .
At our facility, the C-ion RT dose distribution was calculated using the XiO-N, software, which is based on the XiO treatment planning system (Elekta, Stockholm, Sweden) and incorporates an ion beam RT dose engine (K2 dose) developed by the National Institute of Radiological Sciences  as well as a source management tool developed by Mitsubishi Electric (Tokyo, Japan). C-ion RT doses were expressed in Gy (relative biological effectiveness [RBE]), which was defined as the physical dose multiplied by the RBE of C-ions . The heavy ion accelerator at our facility generated C-ion beams; the passive scattering technique was applied for the treatment of HCC. The beam energy selections were 290 MeV/u, 380 MeV/u, or 400 MeV/u depending on the depth of the tumor. Prescribed doses were 60 Gy (RBE), with the aim of covering the PTV with at least 95% of this dose. C-ion RT plans used two or three coplanar ports to avoid the gastrointestinal (GI) tract. The maximum dose to 1 mL (D1cc) was <40 Gy (RBE) administered to the GI tract . Figure 1 shows the typical dose distribution of C-ion RT; all treatment plans were made by medical physicists.
IMRT treatment plans were calculated using the X-ray RT treatment-planning system (Eclipse, Varian Medical Systems, Inc. California, USA). All IMRT treatment plans were simulated using the treatment-planning CT images of patients who received C-ion RT; moreover, all the contours were the same as those used for C-ion RT. The prescribed doses were 60 and 50 Gy; the aim was to cover the PTV with at least 95% of the prescribed dose. IMRT plans used five or six coplanar 6 MV X-rays with gantry angles to avoid the GI tract and spinal cord. The dose constraints were D1cc <40 Gy administered to the GI tract. The weights of each field were also arranged to cover the PTV with at least 95% of the prescribed dose while minimizing the dose to the liver and GI tract. Figure 2 shows the typical IMRT dose distribution in the same patient as that shown in Figure 1; again, all IMRT treatment plans were made by medical physicists.
Dose-volume histogram (DVH) parameters and statistical analyses
We assessed the percentage of the minimum dose that covered 95% of the target volume (D95) based on the DVH for the PTV. To compare normal tissue sparing achieved with C-ion RT versus IMRT, we assessed the mean liver dose (MLD) as well as the percentage of the normal liver volume that received at least 5, 10, 20, 30, 40, and 50 Gy using both C-ion RT (RBE) and IMRT (the V5, V10, V20, V30, V40, and V50, respectively) for the normal liver (i.e., total liver volume minus GTV). Additionally, we defined patients at high risk for developing radiation-induced liver disease (RILD) as those with MLD >23 Gy and V30 values >28% .
The DVH parameters when using both modalities were analyzed. Normal distributions were compared using t-tests (the V5, V10, V20, V30, V40, V50, and MLD), and non-normal distributions were compared using Wilcoxon’s test (PTV D95), after testing for normality with the Shapiro–Wilk test. Statistical significance was defined as a P-value less than 0.05. All statistical analyses were performed using the SPSS 22.2.0 software (IBM Corp, Armonk, NY, USA).