Ultra-selective radiation segmentectomy for early-stage hepatocellular carcinoma

Abstract

Background & aims: Radiation segmentectomy (RS) is an emerging curative-intent therapy for early-stage hepatocellular carcinoma (HCC) when resection or ablation is not feasible. In this study, we evaluated the safety, efficacy, and dosimetric correlates of ultra-selective RS, defined as glass ⁹⁰Y-radioembolization delivered to vessels at least one order beyond parent segmental arteries, targeting <1 Couinaud segment.

Methods: This retrospective study included 38 patients with 42 early-stage (BCLC 0-A) HCCs treated with ultra-selective RS from December 2022 to July 2024. All treatments used glass ⁹⁰Y-microspheres with perfused treatment volumes assessed by cone-beam CT. Post-treatment voxel-based dosimetry was conducted using ⁹⁰Y single-photon emission CT (SPECT)/CT. Tumor response and progression-free survival were assessed by modified RECIST. Explant pathology was used to evaluate treatment effect in transplant recipients, and albumin-bilirubin (ALBI) scores were tracked longitudinally.

Results: The median tumor size was 2.4 cm with a median perfused treatment volume of 66.3 cc (4.5% of total liver volume). The median administered activity was 1.36 GBq (median absorbed dose 837 Gy). Complete response (CR) was achieved in 87% (n = 33), with only one local progression. Median local progression-free survival was not reached. Among 16 tumors with explant data, 69% showed complete necrosis and 25% extensive necrosis (median 88%). Tumor D₉₅ >300 Gy predicted CR (97% CR vs. 0% CR with D₉₅ <300 Gy; p <0.001), with logistic regression yielding an AUC of 0.98. Models incorporating tumor alignment within high-activity ⁹⁰Y-SPECT regions improved predictive accuracy. Over 82% of patients retained or improved ALBI grade during follow-up, with only 2 of 15 patients with baseline ALBI 2b declining to grade 3.

Conclusions: Ultra-selective RS is a feasible and liver-sparing therapy for early-stage HCC. Voxel-based dosimetry confirms dose-response relationships and underscores the importance of tumor coverage.

Impact and implications: This study demonstrates that ultra-selective radiation segmentectomy (uRS) with glass ⁹⁰Y microspheres can achieve high rates of complete imaging and pathologic response in early-stage HCC while treating very small liver volumes. These results replicate and extend prior radiation segmentectomy studies, underscoring that the ablative potential of ⁹⁰Y can be maintained with even greater liver parenchymal preservation. These findings are important for patients who are not candidates for surgery or ablation, especially given low rates of liver function decline after uRS. In practice, uRS may serve as a definitive therapy or as a bridge to liver transplantation, while voxel-based dosimetry provides a framework for ensuring adequate tumor coverage and identifying incomplete responders.

Keywords: Yttrium-90 radioembolization; early-stage hepatocellular carcinoma; liver transplant; pathologic response; radiation segmentectomy; voxel-based dosimetry.

Graphical abstract

Fig. 1

Local PFS, censored to last follow-up, liver transplant, or additional local treatment (if applicable). Median local PFS was not reached. PFS, progression-free survival.

Fig. 2

Voxel-based dosimetry data for the tumors in this study, separated by CR and PR/PD. (A) Vy is percent of tumor volume receiving y dose and (B) Dx is minimum dose received by x percentage of tumor volume. Clear distinctions in the dose metric distributions between CR and PR/PD tumors, but regions of overlap remain where the same range of values led to either CR or PR/PD. The overlapped regions shrink as the minimum dose (y) decreases in Vy and as the volume receiving a minimum dose (x) increases in Dx. (C) Tumor D₉₅ is plotted vs. MIRD_Rx, with CR and PR/PD tumors differentiated in the plot. A linear fit of Tumor D₉₅ = 0.99∗MIRD_Rx (R2 = 0.50) for CR tumors is also plotted. A D₉₅ threshold of 300 Gy (dotted line) is 33/34 predictive for CR and 4/4 predictive for PR/PD. CR, complete response; MIRD_Rx, prescribed dose to the PV (MIRD); PD, progressive disease; PR, partial response.

Fig. 3

Liver function after uRS. (A) Changes in ALBI score plotted over time after uRS treatment (day 0). Patients are grouped based on their baseline ALBI score just prior to treatment (blue – ALBI 1, green – ALBI 2a, orange – ALBI 2b). (B) Sankey plot illustrating transitions of Child-Pugh class from baseline to last follow-up. Bar height is proportional to number of patients in that category. Ribbon width reflects number transitioning between categories. Ribbons colored by class at baseline (green = Child-Pugh A, orange = Child-Pugh B). ALBI, albumin-bilirubin; uRS, ultra-selective radiation segmentectomy.

Fig. 4

Fused axial ⁹⁰Y-SPECT/CT images. Fused axial ⁹⁰Y-SPECT/CT images for four tumors with PR (A-D), one with PD (E), and an example case with two CR tumors (F). The liver, PV, and tumor contours are shown in each image, and the MIRD_Rx and tumor D₉₅ dose values are also included. ⁹⁰Y-SPECT, ⁹⁰Y single-photon emission CT; CR, complete response; MIRD_Rx, prescribed dose to the PV (MIRD); PD, progressive disease; PR, partial response; PV, perfused volume.