Single-system outcomes after adopting yttrium-90 radioembolization with personalized dosimetry as the primary treatment approach for unresectable, solitary hepatocellular carcinoma.

Introduction
Hepatocellular carcinoma (HCC) is the most common primary liver malignancy, accounting for >75% of all liver cancer cases worldwide.1 It ranks among the top 10 most prevalent cancers and is one of the top 5 causes of cancer-related mortality.2,3 Over the past two decades, both the incidence and mortality rates of HCC have steadily increased.4 In the United States, rates have increased by 4.5% each year from 2000 to 2009, with mortality rising by 2.1% between 1999 and 2016.4,5 HCC most often develops against a background of cirrhosis due to chronic inflammation from a variety of etiologies with the overall incidence of HCC continuing to rise.
Currently, liver transplantation remains the best curative option for unresectable HCC. For patients with early- to intermediate-stage disease—classified as Barcelona Clinic Liver Cancer (BCLC) stages A and B—liver-directed therapies (LDTs) serve either as definitive treatment or as bridging/downstaging strategies to enable liver transplant eligibility. LDT modalities include microwave ablation, transarterial chemoembolization (TACE), and yttrium-90 (90Y) radioembolization with treatment selection determined based on tumor size and location, along with provider experience. While providers and treatment centers have individual preferences, there are limited data to suggest an optimal treatment strategy, particularly in solitary disease >3 cm and multifocal disease.
90Y uses microspheres loaded with beta-emitting 90Y particles to deliver targeted ionizing radiation to ablate the tumor tissue. In recent years, the treatment paradigm for 90Y has evolved significantly, with several trials demonstrating safety and efficacy across BCLC A-C.6,7 Personalized dosimetry has emerged as a strategy to optimize 90Y therapy for HCC treatment by maximizing the radiation dose to the targeted tumor while minimizing radiation exposure to surrounding healthy liver tissue, as reported in the DOSISPHERE-01 trial. This trial demonstrated that with personalized dosimetry, median tumor doses exceeding 205 Gy led to improved tumor response, longer duration of response, and better overall outcomes compared with standardized dosimetry (120 Gy).8, 9, 10 These findings were further supported by the TARGET trial, which demonstrated that tumor-absorbed doses >200 Gy were associated with higher complete response (CR) rates and prolonged overall survival (OS) compared with patients who received <200 Gy.9
Subsequent studies have explored the effects of even higher tumor-absorbed doses, with the goal of defining an optimal dose to achieve a CR.9, 10, 11, 12, 13 The LEGACY trial demonstrated that tumor-absorbed doses >400 Gy correlated with complete pathological necrosis in patients with BCLC-A disease.12 Greater tumor-absorbed doses were achieved in the RASER trial (>532 Gy, median 1005 Gy) in BCLC-A disease with limited grade 3 adverse events, demonstrating the safety of 90Y at high doses.13 Other studies have similarly reported CRs in patients treated with doses >400 Gy, thus establishing the current recommendation of a 400-Gy minimum threshold dose to achieve a complete ablative effect when treating HCC.11,14
90Y radioembolization is now utilized to downstage/bridge liver transplant candidates to surgery in which complete-to-extensive tumor necrosis was achieved.15,16
Despite these promising findings that have shaped current treatment recommendations for 90Y, most supporting data originate from institutions with extensive experience in 90Y radioembolization.10 Therefore, in this study, we evaluated response rates and patient outcomes following first-cycle 90Y (FC-90Y) in patients with solitary, unresectable HCC from a multicenter, single system with <5 years’ experience carrying out 90Y with personalized dosimetry.

Materials and methods

Study design
A retrospective study was conducted by incorporating patients from a single regional health care system and including multiple treatment centers with center-independent HCC tumor boards. All patients received FC-90Y radioembolization with personalized dosimetry (TheraSphere; Boston Scientific, Marlborough, MA) via segmental delivery from May 2018 to November 2024. The study was conducted following institutional review board approval at Ochsner Health (IRB# 2019.308).
The study inclusion criteria included: (i) age ≥18 years, (ii) confirmed HCC diagnosis by triple-phase imaging according to the Liver Imaging Reporting and Data System criteria (v2018 American College of Radiology) or biopsy, (iii) unresectable HCC determined by surgical oncologists, (iv) BCLC-A, (v) Child–Pugh A5-B7, (vi) Eastern Cooperative Oncology Group (ECOG) score of 0-1, and (vii) scheduled to undergo FC-90Y radioembolization. Exclusion criteria were: (i) prior treatment with LDT, (ii) prior systemic therapy, (iii) prior liver resection or transplantation, and (iv) incomplete baseline medical records. The study complied with the Health Insurance Portability and Accountability Act (HIPAA) and the ethical principles outlined in the Declaration of Helsinki. This study was approved by the Institutional Review Board of Ochsner Health (IRB protocol #2019.308).

Data source and study variables
Clinical variables including demographics, cirrhosis history, complete metabolic liver panels, and HCC characteristics at the time of diagnosis were all extracted from the electronic medical records. In patients bridged to liver transplantation, complete pathological necrosis was defined as having 99%-100% necrosis in the tumor or tumors treated with 90Y as assessed by a blinded hepatobiliary pathologist.

Treatment selection criteria
All study patients underwent FC-90Y radioembolization with TheraSphere glass microspheres (Boston Scientific). Institutional criteria for FC-90Y included: (i) Child–Pugh A-B, (ii) absence of portal vein thrombus, (iii) absence of extrahepatic metastasis, (iv) total bilirubin <4 mg/dl, (v) serum creatinine concentration <1.5 mg/dl, and (vi) absence of gross ascites. The institutional 90Y algorithm for solitary disease required the tumor to be (i) deemed unresectable by the tumor board and (ii) size >3 cm or ≤ 3 cm and in a location not amendable to thermal ablation. Patients with an incomplete response to FC-90Y were treated with additional LDT with modality selection based on the recommendations of the institution’s multidisciplinary tumor board.

90Y radioembolization procedure
90Y treatment was carried out in two stages. Firstly, a pre-radioembolization mapping hepatic angiogram was conducted to delineate liver vascular anatomy and assess tumor perfusion. Contrast-enhanced cone-beam computed tomography (CT) was also utilized during this mapping procedure to evaluate perfused liver volume. Subsequently, technetium-99m-labeled macroaggregated albumin was administered to calculate lung shunt fraction. In the second stage, TheraSphere glass 90Y microspheres were selectively administered into the tumor-feeding arteries, targeting a radiation dose exceeding 200 Gy to the perfused volume. Dosimetry was evaluated using Mirada DBx Build 1.2.0 Simplicit90Y dosimetry software (Mirada Medical, Denver, CO). The multicompartment dosimetry model was used to estimate the dose to tumor and normal tissue.
Institutional LDT approach was primarily TACE and microwave ablation before 2018. Collectively, providers had <2 years’ experience with standardized dosimetry (<2017) before institution’s translation to personalized dosimetry for 90Y radioembolization (2018).

Study endpoints
The primary study endpoint was target CR rate following FC-90Y. This was defined using either triple-phase, paired CT or magnetic resonance imaging, according to the modified RECIST (mRECIST),17 and was determined by board-certified, fellowship-trained interventional radiologists.
The secondary endpoints were target/overall objective response rates (ORRs) and time-to-event measures including OS, progression-free survival (PFS), time to progression (TTP), target time to progression (tTTP), and target time to retreatment (tTTR). Overall OR included CR or partial response (PR) according to mRECIST. Target OR included CR and PR to the targeted lesion. Incomplete responses refer to responses that obtained either PR, stable disease, or disease progression according to mRECIST. OS was defined as the time from FC-90Y to death due to any cause. For time-to-event measures involving disease progression, progression was defined as the development of either vascular invasion or extrahepatic disease and characterized as progression within or outside of the treatment field. PFS was defined as the time from FC-90Y to stage progression or death. TTP was defined as the time from FC-90Y to stage progression. tTTP was defined as the time from FC-90Y to stage progression attributable only to the initially targeted tumor with 90Y treatment. TTR was defined as the time from FC-90Y until the initially targeted tumor required retreatment due to localized recurrence and was assessed only in those patients who initially achieved a target CR in the first treatment cycle.

Statistical methods
Statistical analysis was conducted using JMP Pro 17.0.0 (SAS Institute Inc., Cary, NC). Both categorical and continuous variables were reported, with categorical variables presented as the number and percentage of the cohort, and continuous variables summarized as the median with interquartile ranges (IQRs). Time-to-event outcomes were analyzed using the Kaplan–Meier method, with graphical output generated in GraphPad Prism version 10.2.0 (GraphPad Software Inc., Boston, MA).

Results

Study cohort
The cohort included 171 patients with treatment-naive, unresectable, solitary HCC (Figure 1). Cohort demographics are summarized in Table 1. Briefly, the cohort had a median age of 68 years, with the majority being Caucasian (107/171, 62%) and male (117/171, 68%). Hepatitis C infection was the predominant cause of cirrhosis (72/171, 42%). All patients were BCLC-A with mostly Child–Pugh A score (148/171, 87%) and ECOG 0 (141/171, 82%). Median tumor size was 3.4 cm, with most tumor sizes <5 cm (82%). Median α-fetoprotein (AFP) level was 8.4 ng/ml, with most patients having AFP ≤50 ng/ml (134/171, 80%).

FC-90Y treatment characteristics and response rates
All FC-90Y treatments were technically successful with treatment characteristics displayed in Table 2. The median dose to volume achieved for the cohort was 506 Gy, with a median lung shunt fraction of 5.2%. Multi-vessel 90Y, defined as sequential treatment of multiple feeder vessels in a single treatment session, was carried out in 35% (59/171) of the cohort. Post-90Y treatment response was assessed in 97% (166/171) of patients, with a median imaging follow-up time of 46.5 days. Overall OR was achieved in 98% (163/166) of patients. Target OR rates were 99% (165/166), with 71% (118/166) achieving a target CR (Table 2).

Outcomes following FC-90Y
The median overall follow-up for the cohort was 21 months. The cohort did not reach median OS, and 1-, 3-, and 5-year OS rates were 94%, 73%, and 60%, respectively (Figure 2A). The median overall TTP was not met, with 1- and 3-year TTP rates of 6% and 33% (Figure 2B). Similarly, median PFS was also not met (Figure 2C), with 1- and 3-year PFS rates of 89% and 61%, respectively. The tTTP rates at 1 year and 3 years were 5% and 22% (Figure 3A). Patients who achieved a target CR had reduced progression rates at 1 year (2% versus 16%) and 3 years (21% versus 62%) compared with those without an initial target CR. Patients achieving an initial CR did not reach median tTTP, while those with an initial incomplete response had a median tTTP of 24 months (95% confidence interval 20-74 months) (Figure 3B). In patients who achieved a target CR, the median tTTR was 48 months, with a 1-year retreatment rate of only 9% (Figure 2C). The median tumor size was smaller compared with patients with a non-CR (3.0 cm versus 4.2 cm); this was also reflected in higher median dose to volume (534 Gy versus 465 Gy) (Supplementary Table S1, available at https://doi.org/10.1016/j.esmogo.2026.100309). Most patients who experienced progression of disease were attributed to out-of-field progression (Supplementary Table S2, available at https://doi.org/10.1016/j.esmogo.2026.100309).

Pathological responses
The cohort comprised 36 patients (36/171, 21%) deemed to be transplant candidates, of whom 67% (24/36) were successfully managed to liver transplantation. The median time from FC-90Y until liver transplantation was 222 days (IQR 118-302 days). Most patients (18/24) were treated with a single 90Y to the target lesion, while 25% (6/24) received an additional LDT before transplant (5/6 received additional 90Y; 1/6 received additional TACE). Imaging was assessed 35 days (median) before transplant, with 75% (18/24) of patients achieving a complete radiographic response to the target lesion (Table 3). The median target necrosis rate was 99% (IQR 90%-100%), with 67% (16/24) having complete pathological necrosis.

Discussion
Unresectable, early- to intermediate-stage HCC is ideally treated with LDT as either a definitive treatment approach or a bridging/downstaging to liver transplantation. 90Y radioembolization has emerged as an effective treatment option for BCLC A-B disease, with recent clinical trials (ROWAN and EMERALDY90) combining 90Y with immunotherapy. While several landmark trials have demonstrated that 90Y radioembolization is an effective treatment for early- to advanced-stage HCC, these studies were primarily conducted at a small number of high-volume centers with extensive experience utilizing 90Y. To date, there has been limited evidence that these results could be reproduced at other centers driven to be early adopters based on the clinical trial evidence. This study addresses this gap by evaluating real-world outcomes in a multicenter setting including multiple independent treatment centers united under a common multidisciplinary tumor board with <5 years’ experience carrying out 90Y with personalized dosimetry.
The movement toward personalized dosimetry began with the DOSISPHERE-01 trial, which demonstrated that personalized dosimetry was able to achieve greater absorbed tumor doses compared with standardized dosimetry (331 Gy versus 221 Gy).8 The TARGET study further illustrated this point with greater absorbed tumor doses >300 Gy in some patients.9 In both studies, higher absorbed tumor doses correlated with higher ORRs (DOSISPHERE-01: 71%; TARGET: 61.7%). In this study, personalized dosimetry in solitary, unresectable HCC resulted in median absorbed tumor doses >500 Gy, which yielded high overall ORRs (98%, 163/166). Similarly, the LEGACY and RASER trials also treated solitary, unresectable HCC with personalized dosimetry, achieving absorbed doses >500 Gy and demonstrating high overall ORRs (LEGACY: 86.4%, 140/162; RASER: 100%, 29/29).12,13 The differences in response rates compared with LEGACY and RASER are likely attributable to the larger tumor sizes in this study and the well-described prognostic link between tumor size and initial response to LDT.18, 19, 20 While RASER focused on only tumor burden <3 cm, 62% of the LEGACY cohort treated lesions <3 cm compared with this study in which 59% of cohort had solitary lesion >3 cm. Larger tumors are prone to heterogeneity with multifocal tumor supply,21 potentially making them more difficult to treat. The results of this study continue to support that ablative 90Y dosing with personalized dosimetry achieves excellent response rates in BCLC-A disease.
While ORRs include those patients with CR and PR, complete radiographic responses remain the ideal target for optimal patient outcomes. Our results and those of others have shown that 90Y with personalized dosimetry can achieve high target CR rates. In this study, target CR rates were 71% (118/166), slightly lower than those reported in the LEGACY and RASER trials, which demonstrated a target CR rate of 84% and 83%, respectively.12,13 Patients who achieved a target CR had lower 1- and 2-year progression risks compared with patients with an incomplete response (CR: 2% and 12% versus non-CR: 16% and 54%). This finding confirms the prognostic significance of first cycle response and subsequent impacts on progression risk. Another potential benefit of 90Y based on the PREMIER and TRACE trials is a more durable CR compared with TACE. This study confirmed durable target TTR rates in patients achieving a CR, with a median of 4 years and 1-, 2-, and 3-year retreatment rates of 9%, 24%, and 46%, respectively. This is also supported by explant findings in the transplant subgroup, where most patients received a single 90Y treatment cycle which yielded a median necrosis rate of 99%.
Early-stage HCC has the highest expected survival (>5 years) with a plethora of treatment options.22 Even with high response rates, 25% of early-stage patients experience disease progression post-90Y.12 Despite this, mounting evidence supports a survival benefit in early-stage HCC when using 90Y with personalized dosimetry.8,9,12 In this study, OS remained high with 1- and 2-year rates of 94% and 73%, respectively, and is in line with the previously reported OS rates (LEGACY: 2-year OS 94.8%). We also report low PFS with 1- and 2-year rates of 89% and 71%, respectively. Comparison to LEGACY and RASER is difficult due to differences in PFS assessment (LEGACY: PFS based on mRECIST scores; RASER: PFS not assessed). However, this study adds to the growing literature on survival outcomes following FC-90Y.
This study is limited by its retrospective design. It includes outcomes from >12 interventional radiology providers with varying levels of experience in personalized dosimetry, ranging from 0 to 4 years with consistent outcomes independent of provider experience. Although it allows for a real-world analysis of a standardized approach to radioembolization, the analysis was not powered to control for provider experience, which may have contributed to lower response rates compared with those reported in the LEGACY and RASER studies, in addition to the tumor size discrepancy noted in the discussion. Because of the early institutional adoption of radiation segmentectomy following LEGACY, we were unable to compare 90Y with other treatment modalities during the same time frame.
In conclusion, this study adds to the growing body of literature demonstrating that 90Y with personalized dosimetry results in high initial and sustained CR rates that translate into decreased progression risk in early-stage HCC. These ongoing efforts aim to further refine our understanding of personalized 90Y radioembolization and expand its role as a definitive treatment modality in treating HCC.