Safety and Efficacy of External Beam Radiation Therapy in Hepatocellular Carcinoma Patients Previously Treated With Yttrium 90 Transarterial Radioembolization.

Introduction
The incidence of hepatocellular carcinoma (HCC) has been increasing over the past few decades and is now a leading cause of cancer-related mortality worldwide.1 Liver-directed therapies (LDTs) are the standard treatment for early-stage to locally advanced HCC and are even considered for cases with limited metastatic spread. Among various LDT options, yttrium 90 (Y-90) transarterial radioembolization (TARE) and external beam radiation therapy (EBRT) are recognized as effective treatments.2,3 These modalities, although both use radiation, differ fundamentally in their source of radiation, delivery methods, and radiobiological effects.
TARE, considered a form of brachytherapy, uses microspheres to directly deliver beta-emitting radiation to the tumor through the hepatic arterial vasculature that supplies it. The physical characteristics of beta-emitting Y-90 ensure that the radiation dose diminishes rapidly within a few millimeters from the source, thereby minimizing exposure to the surrounding liver tissue and enhancing the therapeutic ratio.4 Despite its advantages, TARE has limitations. For example, depending on the tumor's vasculature, there is the challenge of achieving the appropriate dose in hypovascular tumor vascular beds. Conversely, EBRT offers more predictable, homogeneous dosing and is less invasive.5
As TARE becomes more widely used, radiation oncologists increasingly encounter patients requiring subsequent EBRT. The safety of such sequential treatments, particularly the risk of reirradiation, is a clinically relevant and important issue. Prior investigations into post-TARE EBRT have been limited. Most studies included a limited number of patients, with most patients receiving only a single course of TARE.6, 7, 8, 9 Reported liver complications, including radiation-induced liver disease (RILD), have varied across studies.6, 7, 8, 9 Moreover, biliary complications, a significant concern hypothesized due to high radiation doses in patients with tumors near the central bile duct, have largely been understudied. Finally, the efficacy of EBRT for tumors previously treated with TARE remains unclear.
We conducted a retrospective study on HCC patients who underwent EBRT following TARE at our institution. Our study focuses on the safety of this approach, particularly on liver and biliary complications, in addition to assessing the efficacy of EBRT in this setting.

Methods and Materials

Patient selection
We conducted a retrospective data collection of HCC patients who underwent EBRT with a history of prior TARE treatment. From 2014 to 2023, 186 HCC patients received EBRT targeting intrahepatic lesions at our institution. Of these, 31 had previously undergone TARE, and all were included in our analysis. These data were collected after receiving institutional review board approval.
In our practice, the local treatment options for each HCC patient were determined by a multidisciplinary tumor board. Typically, TARE is often given for tumors unsuitable for surgery or thermal ablation. EBRT was used as salvage therapy for recurrences when additional TARE was unsuitable due to factors such as significant lung shunt fraction, unfavorable vascular anatomy, or other contraindications to TARE.

Y-90 radioembolization treatment
The TARE procedure followed the standard protocol using Y-90 glass microspheres. Radiation segmentectomy (superselective techniques) with ablative dosimetry (>400 Gy) was the preferred institutional approach.10 Prior to Y-90 delivery, treatment planning included mapping an angiogram to identify the vessels supplying the tumor, followed by a technetium-99m macroaggregated albumin injection and imaging for lung shunt evaluation and treatment planning. Careful assessment of any possibility of delivery of Y-90 microspheres to adjacent bowel or gallbladder is made during the mapping angiography and assessment of technetium-99m macroaggregated albumin distribution. During treatment, Y-90 microspheres were injected as planned into the tumor-bearing vascular segmental arteries. Treatment verification was done after the procedure using single-photon emission computed tomography (CT)/CT.

EBRT
The selection of EBRT modalities was determined by radiation oncologists, considering various risk factors, including baseline liver function, tumor size, and normal liver volume, with a preference for proton beam therapy (PBT) in higher-risk patients, including those who have received prior TARE or EBRT.
Prior to simulation, fiducial marker implantation was performed unless contraindicated. During simulation, a 4-dimensional CT scan was done for respiratory motion assessment; if tumor motion was ≥1 cm, abdominal compression or end-exhalation breath-hold techniques were employed. CT simulation included multiphase contrast imaging. In selected cases, technetium-99m sulfur colloid single-photon emission CT/CT was used to assess regions of the functional liver, particularly for patients with prior radiation treatment.
The gross tumor volume (GTV) included the viable tumor, typically without additional clinical target volume (CTV) margins. When microscopic disease was suspected, a 5 mm margin to the GTV was considered to generate a CTV and adjusted based on clinical judgment. The internal target volume accounted for tumor motion identified via 4-dimensional CT, with a 5 to 8 mm margin for the planning target volume.
For photon treatments, stereotactic body radiation therapy (SBRT) was preferred with a dose of 40 to 50 Gy in 5 fractions. For PBT, hypofractionation was used with a dose of 60 to 67.5 Gy in 15 fractions, assuming a relative biological effectiveness of 1.1. Robust optimization was performed using a 3-mm setup margin and 3.5% range uncertainty applied to the CTV. Detailed contouring and planning methodology for PBT have been reported previously.11
A summary of dose constraints by fractionation schedule for both SBRT and PBT is provided in Table E1. Daily image guidance used cone beam CT for SBRT and KV imaging for PBT, with alignment to liver contours or fiducial markers when indicated. Because PBT image guidance relied on KV rather than volumetric imaging, a 5-mm planning risk volume (OAR + 5 mm) was applied to PBT treatment planning. Given limited data and a lack of consensus regarding dose tolerance of the central biliary tree (CBT), defined as a 15 mm expansion of the portal vein from the bifurcation to the splenic confluence, no specific CBT constraints were routinely applied beyond limiting dose hot spots within the CBT. Radiation doses from TARE to EBRT were not taken into account due to the limited understanding of TARE dosimetry and accurately extrapolating it to radiation treatment planning and liver recovery and regeneration after TARE.

Outcomes and statistical analysis
Our primary outcomes were toxicities associated with EBRT. We evaluated hepatotoxicity as nonclassic radiation-induced liver disease (ncRILD), defined as an increase of at least 2 points in the Child-Pugh (CP) score within 6 months post-EBRT in the absence of intrahepatic tumor progression or further LDTs. Other toxicities, including biliary complications, gastrointestinal, lung, skin, and general adverse effects, were assessed using the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. Descriptive statistics were used for the analysis of toxicities. Because the incidence of ncRILD was low (n = 4), multivariable logistic regression was not statistically feasible. Consequently, comparisons of ncRILD incidence across TARE-related groups were performed using Fisher’s exact test for categorical variables and the Mann-Whitney U test for continuous variables.
Secondary outcomes included treatment efficacies: local control (LC), progression-free survival (PFS), and overall survival (OS). Competing risk analysis determined the cumulative incidence of local failure. Gray's test was used to compare local failure rates between patients with prior TARE in the EBRT area. PFS and OS were analyzed using the Kaplan-Meier survival method. A P value of less than .05 was considered statistically significant. All statistical analyses were conducted using STATA (version 17.0; StataCorp).

Results

Patient characteristics
Prior to receiving EBRT, the majority of patients had well-compensated baseline liver function (CP-A, n = 27, 87%), favorable performance status (Eastern Cooperative Oncology Group 0-1, n = 27, 87%), and disease confined to the liver (n = 27, 87%) (Table 1). EBRT was selected over additional TARE treatment for various reasons: technical limitations in 25 patients (81%), poor response to TARE in 2 patients (6%), patient preference in 2 patients (6%), and unknown reasons in 2 patients (6%). The technical limitations included limited vascular supply in 19 patients (61%) and high lung shunt fraction in 6 patients (19%). The median interval between the last TARE and EBRT was 12 months (range, 2-85 months).
Ten patients (32%) had prior TARE treatment confined to the area subsequently targeted by EBRT. Thirteen patients (42%) had prior TARE treatment solely outside the EBRT area, and 8 (26%) had prior TARE treatment both inside and outside the EBRT target area.
EBRT was administered as salvage therapy for locally recurrent tumors following prior LDTs in 21 patients (68%). Eighteen of these (86%) had received previous TARE to the lesion targeted by EBRT, and 3 (14%) had recurrences following other LDTs, including transarterial chemoembolization in 2 patients and ablation in 1 patient. The median number of prior TARE treatments targeting the EBRT lesion was 1 (range, 1-4): 1 course for 11 patients (61%), 2 courses for 4 patients (22%), and 3 or more courses for 3 patients (17%). Additionally, 6 of these patients (33%) also received other LDTs combined with TARE targeting the EBRT lesion. Treatment approaches included lobar TARE in 4 patients (22%) and segmentectomy in 14 patients (78%). The median dose of prior TARE targeted at the EBRT-targeted lesion was 72 mCi (range, 8-273 mCi).
With regard to prior LDTs outside the EBRT treatment zone, 21 patients (68%) underwent prior TARE treatment outside the EBRT zone, with 7 receiving TARE only and 14 receiving both TARE and other LDTs in areas not covered by EBRT. Nine patients (43%) had 1 course, 6 (29%) had 2 courses, and 6 (29%) had 3 or more courses of TARE outside the EBRT treatment area. Four of these patients (19%) underwent lobar treatment, and 3 of them also had combined lobar treatment and segmentectomy outside the EBRT target zone. The median total dose to the area outside the EBRT treatment zone for those with prior lobar treatment was 102 mCi (range, 49-330 mCi). Seventeen patients (81%) underwent a segmentectomy approach outside the EBRT target zone, receiving a median total dose of 80 mCi (range, 21-216 mCi).
Photon therapy was administered to 18 patients (58%) and PBT to 13 patients (42%). The most common photon therapy regimen was SBRT with doses of 40 to 50 Gy in 5 fractions, administered to 16 patients (89%). For PBT, the typical fractionation regimen was 60 to 67.5 Gy in 15 fractions, used in 11 patients (85%). The median liver-GTV volume was 1340 cm3(range, 808-2204). The median mean liver dose for SBRT was 7.5 Gy (range, 3.6-12.6) and 11.8 Gy (range, 2.5-18.4) for hypofractionation.

Liver toxicity
Four patients (13%) developed ncRILD. All demonstrated clinical signs of hepatic decompensation, including worsening encephalopathy and/or new or progressive ascites, accompanied by decreased albumin and elevated total bilirubin levels. Transaminase changes were variable and did not follow a consistent pattern. Of these, 3 had EBRT targeted to lesions previously treated with TARE. Two patients had a prior history of TARE at other locations in the liver, whereas 2 did not. There was no significant difference in the incidence of RILD between patients who had EBRT administered within or outside the TARE treatment zone (P = .62), nor between patients with a history of TARE at nontarget sites compared to those without any TARE at nontarget sites (P = .58). In univariable analysis, mean liver-GTV dose (OR, 1.08; 95% CI, 0.81-1.44; P = .59), liver-GTV volume (OR, 1.00; 95% CI, 1.00-1.003; P = .98), baseline CP score (OR, 2.71; 95% CI, 0.98-7.49; P = .055), and tumor size (OR, 1.17 per cm; 95% CI, 0.94-1.46; P = .17) were not significantly associated with ncRILD. Given the small number of events, multivariable analysis was not performed. Baseline liver function, tumor burden, and liver dosimetry were well balanced across TARE-related subgroups (Tables E2 and E3). The median interval between the end of EBRT and the date of RILD was 1.4 months (range, 0.9-4.4 months).
Three patients (10%) experienced RILD-related deaths, occurring shortly after EBRT, at 4 (1 patient) and 5 months (2 patients) after the end of the treatment (Table 2). The remaining patient, despite a CP score increase from 5 at baseline to 7 at 4 months post-EBRT, experienced liver function recovery at 6 months and was still alive at 22-month follow-up.

Other toxicities according to CTCAE version 5.0
Three patients (10%) experienced acute or late biliary complications (Table 3). One patient (3%) had an acute grade 4 biliary complication, presenting with an infected biloma within the EBRT-treated area 17 days after completing PBT. Notably, this patient had risk factors for biliary tract complications, which included the placement of a biliary duct stent prior to TARE treatment and the patient had also received bevacizumab prior to EBRT. Among 26 patients with a follow-up of over 90 days after EBRT, 2 patients (12%) developed late biliary complications, including a biloma in one other patient, and one case of biliary stenosis. The biliary leakage in the biloma patient occurred at 3 months, and the patient with biliary stenosis showed symptoms 41 months post-EBRT that led to a biliary stenosis–related death. Notably, this patient received additional TARE treatment close to the area receiving EBRT and near the bile duct. All 3 patients with biliary complications had received 2 TARE treatment courses overlapping with the EBRT area delivered with PBT and had tumors located near the central bile duct (Fig. 1). No clear trends in laboratory values (eg, transaminases) were observed among these patients. The dose to the CBT in these patients ranged from 101% to 105% of the 67.5 Gy prescription dose.
No gastrointestinal hemorrhage or ulceration was observed. There were no other acute complications of grade 3 or higher. Other acute toxicities were mild and self-limiting, with nausea found in 13 patients (42%), followed by fatigue in 11 patients (35%), and dermatitis in 7 patients (23%). No other late complications were noted. Details of toxicities are presented in Table E4.

Treatment efficacy
Median follow-up duration was 30 months (range, 0.8-100.3). The median OS was 22 months (95% CI, 7.1-39 months). The 1-year and 2-year OS rate was 61% (95% CI, 41%-77%) and 43% (95% CI, 24%-60%), respectively. The median PFS was 6 months (95% CI, 2.8-14.3 months), with 1- and 2-year PFS rates of 33% (95% CI, 17%-51%) and 18% (95% CI, 6%-34%), respectively. There were 21 deaths; disease progression was the most common cause of death, occurring in 10 patients (48%). Treatment complication-related death was found in 4 patients. Other causes of death occurred in 6 patients (29%), and 1 death (5%) was of unknown etiology.
Two patients experienced local failures. The cumulative incidence of local failure at 1 and 2 years was 7%. One patient received EBRT within a previously TARE-treated area, whereas another underwent EBRT for a lesion without prior treatment. There was no significant difference in the 1- and 2-year cumulative incidence of local failure for patients who had a history of TARE to the EBRT lesion versus those who did not (6% vs 9%; subdistribution hazard ratio [SHR], 0.6; 95% CI, 0.04-9.38; P = .71). Local failure was also not significantly associated with tumor size (SHR, 1.01; 95% CI, 0.86-1.18; P = .92), macrovascular invasion (SHR, 1.66; 95% CI, 0.10-26.62; P = .72) or pretreatment alpha-fetoprotein level (SHR, 1.00; 95% CI, 1.00-1.00; P = .55). Survival outcomes are presented in Fig. 2.

Systemic therapy and outcomes
No patients received concurrent systemic therapy. Four patients had received systemic therapy prior to EBRT (1 with a tyrosine kinase inhibitor and 3 with immunotherapy); among the immunotherapy-treated patients, one resumed systemic therapy after completing EBRT. One additional patient initiated systemic therapy sequentially after EBRT as part of planned management.
With respect to biliary toxicity, 1 of the 3 patients who developed biliary complications had received atezolizumab/bevacizumab, whereas the other 2 had not received any systemic treatment. Liver decompensation occurred in one patient who had multiple high-risk features, including a large tumor burden, relatively high mean liver dose, and prior exposure to multiple systemic agents. The remaining patients who experienced liver-related complications had not received systemic therapy either before or after EBRT. Sequential systemic therapy was not associated with a significant difference in PFS (HR, 2.70; 95% CI, 0.59-12.37; P = .20). Given the small sample size and the presence of multiple clinical and dosimetric confounders, no meaningful association between systemic therapy and toxicity or disease control could be established.

Discussion
Given the limited published data, the safety of delivering EBRT after TARE remains a significant concern. Our cohort study of 31 patients who received additional EBRT (proton and photon) following definitive intent TARE treatment (median of 2 courses [range, 1-5], with 32% undergoing ≥3 courses) reveals that EBRT yielded excellent LC (1- and 2-year cumulative incidence of local failure 7%). However, cases of serious liver and biliary complications were observed: 3 cases (10%) of RILD-related death and 3 cases (10%) of biliary complications, which are typically rare complications seen with EBRT alone.
Theoretically, the addition of EBRT treatment in patients previously treated with TARE can lead to a higher risk of liver complications. Other investigators have observed a similar rate of liver complications in this setting. Our study observed RILD in 4 patients (13%), of whom 3 experienced possible RILD-related mortality. These RILD rates are comparable to those reported for EBRT treating unresectable HCC, which usually range from 5% to 15%.2 Other studies have reported similar findings for HCC patients treated with EBRT with prior TARE treatment. Hardy-Abeloos et al,7 evaluating 31 patients with less intensive prior TARE (only segmental TARE with a median of 1 course [range, 1-2]), found a liver decompensation and liver failure–related death rate of 9%. Consistent with our findings, they reported that the location of EBRT, whether within or outside the TARE treatment zone, was not associated with an increased risk of RILD. However, it is important to note that both our study and theirs are limited by the small number of patients. Liu et al8 found only grade 1 liver function changes according to CTCAE version 5.0 when combining SBRT treatment with TARE in HCC patients with portal vein tumor thrombosis. Conversely, Wang et al9 reported a higher incidence of grade 3+ liver toxicities according to CTCAE v.4.03 at 36%, with 23% of patients experiencing liver failure–related death. However, these numbers may be overestimated due to the lack of censoring for patients with intrahepatic tumor progression, which was present in all patients with liver failure–related death.
Although the RILD rates in our cohort appear acceptable, EBRT in this setting should still be approached with caution, given the potential morbidity. The standard practice is to limit normal liver exposure according to specific dose constraints; however, such constraints remain undefined for patients previously treated with TARE, in part due to the incomplete understanding of comparative dosimetry between the 2 modalities. TARE differs fundamentally from EBRT in dose rate, spatial dose distribution, and cytotoxic mechanism and, therefore, cannot be directly translated into equivalent EBRT dose metrics.12 For these reasons, our study, alongside several others,7,8,13 did not combine or sum doses across the 2 modalities. Some groups have explored composite dose modeling by converting the TARE dose to biologically equivalent dose and summing it with EBRT dose.9,12 Wang et al9 reported that V110 in the composite plan was the strongest predictor of hepatotoxicity. However, these findings should be interpreted cautiously, as oversimplified TARE dose-conversion models may not accurately reflect the complex and heterogeneous true dose distribution delivered by microspheres.
In contrast to other studies that found no biliary complications following combined treatment,6, 7, 8, 9 we observed a 10% incidence of biliary complications. This higher rate could be attributed to the fact that our cohort included patients who had undergone multiple prior TARE treatments, whereas most participants in previous studies received only one course of TARE. Although such complications are rare in HCC treated with EBRT,14 the risk of biliary complications increases with higher doses of radiation.15 This implies that the complication rate could increase with combined radiation treatments when the biliary system is exposed to higher radiation doses. Determining whether the biliary complications were primarily due to EBRT TARE, or both, is challenging. However, our data indicate that the incidence of biliary complications is higher than what has been reported in patients treated with EBRT alone.14,16,17 In 80 patients with perihilar HCC treated with PBT, none developed PBT-related biliary complications, despite 60% receiving biologically equivalent dose10 of 100 to 105 Gy.14 In our cohort, we observed that patients with biliary complications had received high radiation to the bile duct, having undergone at least 2 courses of TARE treatment overlapping with the EBRT treatment zone, and had tumors located near the central bile duct. However, these findings should be interpreted cautiously and warrant further studies for confirmation due to the limited number of patients with bile duct complications. It is also important to note that all patients with biliary complications had other significant risk factors: the patient with an infected biloma had prior biliary stent placement, a major known risk factor; the biloma patient had sterile cholecystitis with possible gallbladder necrosis on post-TARE CT imaging; and the patient with biliary stenosis underwent multiple LDTs after EBRT. Therefore, it is challenging to determine whether the complication is directly related to the combined treatment; additional investigations studying the biliary complication risks of EBRT and TARE in centrally located HCC tumors are needed.
Our study also demonstrates promising LC with 1- and 2-year cumulative incidence of local progression of 7%. This aligns with another study showing a 97% 1-year LC in patients receiving SBRT post-TARE.7 Despite concerns that radioresistant tumors remaining after TARE might respond less favorably to subsequent EBRT, our findings indicate no significant difference in LC between patients with or without prior TARE to the EBRT region. This suggests that EBRT remains an effective treatment modality in scenarios where further TARE or other LDTs are not feasible options.
Although our study includes a relatively large cohort compared with previously published data, the small number of local failures and complications makes it statistically difficult to determine the correlation between incidence and prior TARE treatment. Additionally, the retrospective nature of our methodology has resulted in missing data and incomplete records. We also lack detailed information on the actual doses received by the normal liver and bile duct when combining radiation doses from both modalities. As previously mentioned, this is due to significant uncertainties that still exist in the assessment of liver dosimetry when combining these modalities. For these reasons, additional research in this area is essential. The University of Michigan is currently investigating adjuvant SBRT delivery to underdosed regions from TARE.18,19 Further studies, including dosimetric dose comparison and dose conversion between TARE and EBRT, are needed to accurately determine the cumulative dose delivered to normal organs and ensure adherence to established safety constraints. Additionally, the integration of functional imaging for identifying areas of viable liver regions could potentially mitigate liver complications.20
In conclusion, EBRT in the setting of HCC patients previously treated with TARE offers excellent LC, even when patients require EBRT as a salvage treatment for tumors previously treated with TARE that may be radioresistant. However, liver and biliary complications remain issues of concern. Biliary complications, especially in patients who have already received at least 2 TARE treatments near the central bile ducts, should be consented for potential biliary complications and closely monitored. Complications should be weighed against the clinical merits of EBRT in this setting.

Disclosures
None.