Can Post-Treatment Dosimetry Predict Clinical Outcomes in Segmental Transarterial Radioembolization of Hepatocellular Carcinoma?

Abstract

<h3>Purpose/Objective(s)</h3> Prognosis for patients with hepatocellular carcinoma (HCC) is poor but segmental transarterial radioembolization (STARE) using Yttrium-90 (Y90) can achieve high rates of objective response (OR) in early-stage disease. Advances in 3D computational treatment planning systems (TPS) and nuclear imaging platforms allow for accurate calculation of Y90 activity distribution after STARE. However, post-STARE dosimetry is not used to guide current clinical practice. In this single center, retrospective study of Y90 STARE, we hypothesized that 3D post-treatment dosimetry derived from bremsstrahlung single photon emission computed tomography (SPECT/CT) correlates with meaningful clinical outcomes. <h3>Materials/Methods</h3> We reviewed 24 consecutive STARE of 23 HCC patients treated with glass microspheres from 2019 to 2021. Baseline characteristics and outcomes were collected by chart and imaging review. Commercial software was used to contour gross tumor volume (GTV) and liver on post-Y90 SPECT/CT aided by deformable fusion with prior diagnostic scans. Dose distribution was assessed using the Local Deposition Model based on SPECT/CT activity maps. Mean, minimum, and maximum GTV dose, volume of GTV receiving <100 Gy, < 70 Gy and < 50 Gy, minimum isodose encompassing the hottest 2% of the GTV (D2), D5, D30, D50, D70, D95, D98, and doses to nontumorous liver were captured. Kaplan-Meier (KM) survival, multivariate regression, and areas under the receiver operating characteristic curve (AUC) were estimated using STATA. <h3>Results</h3> Most patients were Child-Pugh A (19/23) and Barcelona Clinic Liver Cancer Stage B (14/23) at baseline. Median follow up time was 12 mo (range 1-24), median prescription 185 Gy (range 120-450), and median time to best mRECIST response was 5 mo (range 1 to 12). 92% of treatments (22/24) achieved OR and 62.5% (15/24) achieved complete response (CR). Solitary lesions with long axis < 4 cm were most likely to achieve CR. CR correlated with PFS (p=0.05, median PFS for non-CR 4 months, for CR not reached) and trended to significance in OS (p=0.10, median OS for non-CR 10 mo, for CR not reached). Mean GTV dose correlated with CR only on univariate analysis while GTV volume receiving <100 Gy was highly predictive of CR (AUC 0.92) with inflection point at 3 cc. Grade 2 and 3 toxicity rates were 16.6% and 8.3% with baseline bilirubin and albumin predictive of G2+ (AUC 0.84). <h3>Conclusion</h3> In this cohort of patients with HCC treated with STARE Y90, CR correlated with meaningful clinical outcomes (PFS and OS) and a high OR rate. Our findings suggest that (1) less than CR can be identified immediately using same-day post-treatment dosimetry with suboptimal target coverage at threshold volumes most predictive and (2) concordance between suboptimal dosimetry and early post-treatment diagnostic imaging can guide recommendations for additional liver-directed therapy, with intent to convert patients to CR and improve PFS and OS. This paradigm will be tested prospectively at our institution.