From 1998 to 2011, 45 patients with unresectable HCC received liver directed second RT with high dose 3D-CRT in Samsung Medical Center. The eligibility criteria for repeated RT were as follows: 1) HCC not eligible for or refractory to other therapies; 2) persistent disease after initial RT. We included 43 patients by excluding two ineligible patients, one with immediate follow-up loss and the other who underwent liver transplantation after second RT. Radiation-related and medical records of the eligible patients were retrospectively reviewed.

Computed tomography (CT) scans for RT planning were done with each patient in a supine position, with both arms raised above the head. CT simulation data were transferred to a 3D-CRT planning system (from 1998 to 2003, PROWESS, Alliant Medical Technology, Chico, CA, USA; from 2004 to 2011, PINNACLE, The Philips Medical System, Madison, WI, USA). The tumor, normal liver, kidneys, spinal cord, and bowels were contoured and reconstructed to form a 3-demensional representation.

The clinical target volume (CTV) was regarded same as the gross tumor volume. To determine the cranial-caudal margins, the diaphragmatic excursion during respiration was visualized using fluoroscope, and CTV was expanded by the amplitude of this excursion (1.5-2.5 cm). RT was delivered with a 6- to 15-MV linear accelerator. Treatment planning for re-irradiation was performed with an emphasis on maximizing non-irradiated liver volume and minimizing the overlapped volume of irradiation in the skin using 3D-CRT technique. Total dose was designed under tentative guidelines so that the normal liver volume irradiated with over half of the prescribed dose should not exceed 50% of the total liver volume. The total dose was determined by considering the doses of the remnant functioning liver, gastro-duodenum, and large bowel by the discretion of the treating physician considering previous radiation dose to the critical structures, interval from the previous course of radiotherapy. The re-irradiation dose ranged from 30 to 55 Gy (median, 45 Gy) in daily 2 to 15 Gy (median, 3 Gy) fractions.

Tumor response was evaluated on serial CT scans 4 to 6 weeks after completion of second RT and then at 3 months intervals. Tumor response was determined using the criteria from the modified Response Evaluation Criteria in Solid Tumors for HCC [20] based on the reports of the diagnostic radiologists. Survival was calculated from the date that second RT has started.

Physical examinations and chemistry profiles including liver function were checked on every follow-up visit. Toxicities were assessed using the Common Terminology Criteria for Adverse Events (CTCAE) ver. 4.0. Radiation-induced liver disease (RILD) was separated into 'classic' and 'non-classic' RILD. Classic RILD was characterized by the presence of anicteric ascites and the elevation of alkaline phosphatase levels to at least a 2-fold increase over pretreatment values in the absence of tumor progression. Non-classic RILD was defined as the elevation of liver transaminases more than five times the upper limit of normal or CTCAE grade 4 levels in patients with baseline values more than five times the upper limit of normal within 3 months after completion of RT, or a decline in liver function (measured by worsening of the Child-Pugh score by 2 or more) [2122].

Tumor response according to RT dose was compared using the chi-square test. Survival was estimated using the Kaplan-Meier method and compared using the log-rank test to determine potential prognosticators. The Cox proportional hazard regression analysis was used for multivariate analysis. A p-value <0.05 was considered to be statistically significant. All aforementioned analyses were performed using IBM SPSS Statistics ver. 20 (IBM, Armonk, NY, USA).

Patient characteristics at the time of second RT are summarized in Table 1. The median age of all patients was 59 years (range, 29 to 71 years). Modified Union for International Cancer Control (UICC) T stage was determined based on the viable tumor evaluated before starting the second RT, and the group of 43 patients was comprised of 30 (69.8%) T1-T3 and 13 (30.2%) T4 patients. Vascular invasion was accompanied in 24 patients (55.8%) and 33 patients (76.7%) had multiple tumors. Majority of patients had a clinical liver function of Child-Pugh A classification (83.7%).

Median interval between first and second RT was 13.8 months (range, 1.0 to 58.1 months). Because different doses per fraction were used, the equivalent doses to 2 Gy fractions (EQD_2Gy) for α/β ratio of 10 were calculated. The dose of first RT ranged from 32.5 Gy₁₀ to 93.8 Gy₁₀ (median, 48.8 Gy₁₀). Second RT was delivered with palliative (n = 24) or salvage (n = 19) intent, and the RT dose ranged from 31.25 Gy₁₀ to 93.75 Gy₁₀ (median, 44 Gy₁₀).

Target volumes of second RT in the liver were overlapped with the first RT site in 13 patients (30.2%), adjacent with the volume of first RT in 21 patients (48.8%), and apart from the first RT site in 9 patients (20.9%). Cumulative RT dose in the overlapping target volume from two RT sessions ranged from 65.0 Gy₁₀ to 117.0 Gy₁₀ (median, 90.6 Gy₁₀). Gastro-duodenum was re-exposed in 16 patients (37.2%).

During the follow-up after second RT, additional transarterial chemoembolization (TACE) was performed in 11 patients, radiofrequency ablation in one patient, third RT in three patients, and fourth RT in one patient.

The radiographic in-field tumor responses were evaluable in all patients. An objective tumor response rate was 62.8% with complete response in 8 lesions (18.6%) and partial response in 19 (44.2%). The tumor response rates were 81.0% and 45.5% in patients receiving ≥45 Gy₁₀ and <45 Gy₁₀, respectively (p = 0.016) (Table 2). Local control rate was 84.0% at 1-year and 77.6% at 2-year (Fig. 1A).

The median follow-up duration calculated from the initiation of the second RT was 11.2 months (range, 4.1 to 58.3 months). During the follow-up period, eight patients remained alive and 35 died. The overall survival (OS) rates at 1- and 2-year were 57% and 38%, respectively (Fig. 1B). The median OS was 11.2 months, and it was significantly affected by the Child-Pugh class and the modified UICC T stage. The median OS were 14.2 months vs. 6.1 months (Child-Pugh A vs. B, p < 0.001) (Fig. 2A) and 15.6 months vs. 8.3 months (T1-3 vs. T4, p = 0.004) (Fig. 2B), respectively (Table 3). Higher RT dose (≥45 Gy₁₀) tended to be associated with better OS (14.2 months vs. 10.3 months, p = 0.199) (Fig. 2C). Median OS increased to 24 months when patients with poor prognostic factor (Child-Pugh Class B and modified UICC T-stage T4) were excluded (Fig. 2D). Overlap of RT volumes in the liver or gastro-duodenum between first and second RT was not significantly related to OS.

By multivariate analysis (Table 4), the Child-Pugh class and the modified UICC T stage were significant prognostic factors affecting OS (p = 0.001 and p = 0.009, respectively).

The profiles of treatment-related RT toxicity are displayed in Table 5. Overall CTCAE grade 1 or 2 toxicities were developed in 18 patients (41.9%). Nausea, anorexia, and gastrointestinal pain were the most common toxicities, but the acute symptoms were resolved spontaneously or with supportive care. Two patients developed grade 3 or higher toxicity during follow-up period, both of them received ≥50 Gy₁₀ at the time of second RT. One patient suffered from grade 3 duodenal perforation and the other from grade 3 pneumonitis. The patient with grade 3 duodenal perforation underwent three times of RT before the adverse event with 45 Gy in 15 fractions and 11 months after the first course with 54 Gy in 18 fractions followed by 48 Gy in 16 fractions after 6 months after the second course. The site of perforation seemed to correlate with the site repetitively exposed with 3 times of RT. Although it was difficult to differentiate RILD from deteriorating liver function associated with disease progression or the side effect of TACE, no definitive classic or non-classic RILD was observed during the follow-up period.