Is 400 Gy enough for every tumor? The 3 cm cutoff and a modified BCLC classification for predicting complete response after radiation segmentectomy in hepatocellular carcinoma

Abstract

To evaluate the relative contribution of the 400 Gy perfused liver absorbed dose (PLAD) threshold and tumor size in predicting complete response (CR) after yttrium-90 (Y-90) radiation segmentectomy (RS) in hepatocellular carcinoma (HCC), and to investigate the prognostic discrimination of a modified Barcelona Clinic Liver Cancer (BCLC) classification based on a 3 cm cutoff in solitary lesions compared with the standard BCLC 2026 staging system. Fifty-five patients (67 target lesions) who underwent RS with Y-90 glass microspheres between June 2022 and June 2025, met BCLC 2026 stage A/B criteria, and had a target lesion diameter ≤ 8 cm were retrospectively analyzed. Patients with solitary lesions included within standard BCLC-A were reclassified according to tumor diameter: those with lesions ≤ 3 cm as modified stage A, and those with lesions > 3 cm—together with existing multinodular BCLC-B cases—as modified stage B. The primary endpoint was CR at 6 months according to modified Response Evaluation Criteria in Solid Tumors (mRECIST). Overall mRECIST response at the patient level and localized mRECIST response at the lesion level were evaluated. Predictors of CR were assessed by univariate and multivariate logistic regression. The mean age was 65 ± 7.8 years, and 72.7% of patients were male. Of the target lesions, 31.3% were ≤ 3 cm and 68.7% were > 3 cm; median lesion diameter was 4.2 cm (IQR 2.8–5.2), and mean PLAD was 494 ± 175 Gy. At 6 months, CR was achieved in 49.1% (27/55) of patients and 53.7% (36/67) of lesions. CR rate in the modified BCLC-A group was 92.3%, compared with 35.7% in the modified BCLC-B group ( p < 0.001); in the BCLC 2026 system, this difference was more modest (60% vs. 30%; p = 0.032). In multivariate analysis, standard BCLC 2026 classification lost significance ( p = 0.463), whereas the modified BCLC-A classification remained an independent predictor of CR (OR: 17.3; 95% CI: 1.9–158; p = 0.011). At the lesion level, CR rate was 90.5% for lesions ≤ 3 cm and 37.0% for lesions > 3 cm ( p < 0.001), with no significant difference in mean PLAD between the two groups (499 vs. 491 Gy; p = 0.864). In multivariate analysis, PLAD > 400 Gy lost statistical significance (OR 2.5; p = 0.124), whereas lesion diameter ≤ 3 cm remained a strong and independent predictor of CR (OR 14.4; 95% CI 2.9–70.8; p = 0.001). In patients undergoing RS for HCC, tumor size emerges as an important parameter to consider alongside the 400 Gy PLAD threshold when predicting complete response. The modified BCLC classification based on a 3 cm cutoff for solitary lesions may contribute to the discrimination of treatment response compared with the standard BCLC 2026 staging system.