We retrospectively identified 12 patients with cancers involving the liver and associated PVTT who received both ⁹⁰Y radioembolization and SBRT to the PVTT between May 2015 and August 2020. The Institutional Review Board of City of Hope National Medical Center (No. 21240) approval was obtained for this study. The informed consent was waived as this was a retrospective study. Demographic, clinical, and treatment factors were recorded. Primary endpoints were treatment complication rate and toxicity rate by the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. All toxicities ≤3 months (90 days) after treatment completion were considered acute, and all others were considered late. Toxicities were reported during weekly on treatment visits and 1–3 month follow–up visits by the attending or resident physician. When quantifying hepatic impairment, the following criteria were used based on the CTCAE v5.0: grade 1, alanine aminotransferase/aspartate aminotransferase (ALT/AST) 1–3.0 times the upper limit of normal (ULN) or bilirubin 1–1.5 times ULN; grade 2, ALT/AST 3–5 times ULN or bilirubin 1.5–3 times ULN; grade 3, ALT/AST 5–20 times ULN or bilirubin 3–10 times ULN; grade 4, ALT/AST >20 times ULN or bilirubin >10 times ULN. For the purposes of our analysis, we chose to use the term “elevated liver function tests (LFTs)” to capture this toxicity. Toxicity data is recorded for all patients using a “toxicity assessment” smartsheet on the Epic electronic medical record system. This is a regular part of our workflow and is almost always filled out.

Secondary endpoints were the best response rate, PV patency rate, 1-year OS rate, and median OS. Best response rates were determined based on the Response Evaluation Criteria in Solid Tumors criteria (RECIST v1.1). Treatment response was evaluated based on the vascular component of treatment. A complete response (CR) was defined as no tumor within the PV, a partial response (PR) was defined as decreased tumor within the PV, stable disease (SD) was defined as stable tumor within the PV, and progressive disease (PD) was defined as tumor growing within the PV. PV patency was determined based on radiologist report and viewing the scan for confirmation. 1-year OS rate was based on the time from start of SBRT to the time of death or most recent follow-up.

All patients were reviewed in a weekly multidisciplinary tumor board dedicated to upper gastrointestinal malignancies. It is our general principle to use ⁹⁰Y radioembolization followed by SBRT to treat patients all patients who are not surgical candidates with HCC with PVTT. Reasons that may preclude a patient from surgery include co-morbidities such as cirrhosis and diffuse or multifocal disease. Our approach has been to deliver ⁹⁰Y radioembolization to treat the parenchymal component of the tumor followed by SBRT to the vascular invasion component of the tumor. SBRT was typically planned from the start. When we first started this combination treatment modality, we waited until after 90 days after ⁹⁰Y radioembolization to initiate SBRT because we typically do not see the maximum effect of ⁹⁰Y until after 90 days. However, our experience has progressively shown that PVTT does not respond to ⁹⁰Y alone, so we have been more routinely offering SBRT sooner within 1–3 months and not wait 90 days to evaluate response. Fig. 1 illustrates the overall timeline for ⁹⁰Y radioembolization and SBRT treatment planning and evaluation.

For ⁹⁰Y radioembolization, all patients underwent a standardized pretreatment workup consisting of clinical evaluation, laboratory and imaging assessment, a mapping angiographic procedure, and technetium-99 macroaggregated albumin scintigraphy. ⁹⁰Y-microspheres (Sirtex, Woburn, MA, USA) were administered to the appropriate segmental or lobar feeding arteries. Our institutional practice is to give the ⁹⁰Y activity that will result in the intended volume of the liver receiving between 80 to 150 Gy. There were no whole liver ⁹⁰Y treatments in this cohort. After treatment, patients routinely underwent a limited positron emission tomography/computed tomography (PET/CT) scan to visualize areas of inflammation to approximate ⁹⁰Y uptake.

For SBRT treatments, all patients underwent a four-dimensional CT simulation scan. Patients received IV contrast and were immobilized using a VacLoc. In patients with adequate lung function, end expiratory breath holding was used to minimize breathing-related liver motion, otherwise respiratory gating was used. The gross tumor volume (GTV) defined as the primary lesion in the PV was delineated using the CT simulation image and/or fused magnetic resonance imaging (MRI) scan. For patients undergoing respiratory gating, the internal target volume (ITV) was delineated similarly, making sure to include the primary lesion in all phases of breathing during which the patient will be treated. The planning target volume (PTV) was defined as the GTV or ITV plus a 5-mm radial margin. Dose was prescribed to 95% of the PTV. For treatment planning for 5-fraction SBRT, dose constraints commonly used were liver V₁₅ <700 mL, duodenum D_max <32 Gy, esophagus D_max <35 Gy, and spinal cord D_max <30 Gy. To limit hot spots, max dose within the target was limited to <110%. Unfortunately, there was no method to account for the prior radiation dose delivered to the normal liver via ⁹⁰Y radioembolization. Generally, SBRT to the vascular invasion component of the tumor contributed very little radiation to the normal liver given its focused target area. However, if there is concern for normal liver receiving too much radiation from both treatments, the limited PET/CT scan obtained after ⁹⁰Y can be used to crop the SBRT treatment volume off of overlapping liver volumes. For treatment delivery, image guidance with cone beam CT was used to verify positioning before each treatment. Fiducial markers were used in select patients. SBRT was delivered every other day up to a total of 5 fractions.

Descriptive statistics such as medians for continuous variables and frequencies for categorical variables were used to describe patient characteristics. Median follow-up time (from the completion of SBRT) was calculated with the reverse survival method. Survival analyses were performed using Kaplan-Meier analysis and Cox regression for univariate analysis. All analyses were done using the statistical software SPSS version 26 (IBM, Armonk, NY, USA). All tests used a significance level <0.05.

A total of 12 patients received the combination of ⁹⁰Y and SBRT for PVTT from May 2015 to August 2020. Baseline patient characteristics were gathered and shown in Table 1. None of the patients had any prior treatment such as surgical resection, liver transplant, or systemic therapy. Most patients were 65 years or older (68%), male (68%), had a primary diagnosis of HCC (75%), the Eastern Cooperative Oncology Group (ECOG) performance status of 0 (58%), Child–Pugh class A (75%), with PVTT in the left or right PV (75%).

Patients received a median ⁹⁰Y dose of 104.3 Gy (range, 83.3 to 131.7 Gy) to a median volume of 846 mL (range, 675 to 1,068 mL). Most patients had a 1- to 3-month interval between ⁹⁰Y and SBRT (75%). Patients received a median SBRT dose of 32.5 Gy in 5 fractions (range, 27.5 to 50 Gy in 5 fractions) to a median volume of 91.3 mL (range, 14.7 to 623.2 mL). The most common fractionation was 30 Gy in 5 fractions. There were three patients with progression of parenchymal disease in the time between ⁹⁰Y and SBRT. These three patients with disease progression had a larger PTV volume (range, 266.1 to 623.2 mL) than patients who did not have disease progression (range, 14.7 to 192.4 mL) to cover additional areas of parenchymal disease.

Most patients were treated using end expiration breath-hold (75%), and the remaining were treated using respiratory gating. Fifty-eight percent of patients received some form of maintenance therapy after local treatment such as sorafenib, lenvatinib, bevacizumab, or nivolumab.

A summary of SBRT related toxicities can be seen in Table 2. Of the 12 patients in this study, 17% developed acute grade 1 elevation of LFTs, 17% developed acute grade 1 nausea, 17% developed acute grade 1 fatigue, and 8% developed acute grade 1 esophagitis. The total rate of grade 1 toxicity was 59%. There were no grade 2+ acute toxicities or late toxicities or treatment complications reported. As mentioned above, the interval between ⁹⁰Y and SBRT was shortened from over 3 months to 1–3 months after our experience showed PVTT rarely responds to ⁹⁰Y alone. After shortening the interval, we observed no increase in acute toxicities in patients treated with a 1- to 3-month interval (5/8 patients) vs. patients treated with >3-month interval (2/4 patients), although sample sizes are too small to draw any conclusions. Individual patient level demographics and toxicities can be seen in Table 3.

Based on RECIST criteria v1.1, the CR rate was 8%, the PR rate was 42%, the SD rate was 33%, and the PD rate was 17%. After shortening the interval between ⁹⁰Y and SBRT from over 3 months to 1–3 months, we observed no difference in response rates in patients treated with a 1- to 3-month interval (7/8 patients with CR+PR+SD) vs. patients treated with >3-month interval (3/4 patients with CR+PR+SD), although sample sizes are too small to draw any conclusions. The local control rate (CR+PR+SD) was 83% (Fig. 2). On follow-up imaging, 58% patients had a patent PV. With a median follow-up of 29 months, 1-year OS was 55% and median OS was 14 months (Fig. 3). A summary of response rate, PV patency rate, and 1-year OS rate can be seen in Table 2. Individual patient level demographics, response rates, and 1-year OS rates can be seen in Table 3.

A univariate analysis was performed to determine predictors for improved OS (Table 4). The only variable to be significantly associated with improved OS was treatment response, dichotomized into CR+PR+SD versus PD (p < 0.001). Age, sex, diagnosis, performance status, Child–Pugh, location of PVTT, ⁹⁰Y dose, total SBRT dose, PTV volume, interval between ⁹⁰Y and SBRT, and receiving maintenance therapy after local treatment were not significantly associated with OS. Fig. 3 shows the Kaplan-Meier curve comparing OS for responders to ⁹⁰Y-SBRT versus non-responders. Median OS for treatment responders was 33 months versus 2 months for treatment non-responders (p < 0.001).