ABSTRACT
PURPOSE: The transarterial radioembolization (TARE) dose is traditionally calculated using the single-compartment Medical Internal Radiation Dose (MIRD) formula. This study utilized voxel-based dosimetry to correlate tumor dose with explant pathology in order to identify dose thresholds that predicted response. METHODS: All patients with HCC treated with TARE using yttrium-90 [90Y] glass microspheres at a single institution between January 2015 - June 2023 who underwent liver transplantation were eligible. The [90Y] distribution and dose-volume histograms were determined using Simplicity90 (Mirada Medical, Oxford UK) with a Bremsstrahlung SPECT/CT. A complete response was assigned if explant pathology showed complete necrosis and the patient had not undergone additional treatments to the same tumor after TARE. Logistic regression and receiver operator characteristic (ROC) curves were constructed to evaluate dose thresholds correlated with response. RESULTS: Forty-one patients were included. Twenty-six (63%) met criteria for complete response. Dose to 95% (D95), 70% (D70), and 50% (D50) of the tumor volume were associated with likelihood of complete response by logistic regression (all p < 0.05). For lesions with complete response versus without, the median D95 was 813 versus 232 Gy, D70 was 1052 versus 315 Gy, and D50 was 1181 versus 369 Gy (all p < 0.01). A D95 > 719 Gy had the highest accuracy at 68% (58% sensitivity, 87% specificity) for predicting complete response. Median percent of tumor volume receiving at least 100 Gy (V100), 200 Gy (V200), 300 Gy (V300), and 400 Gy (V400) also differed by pathologic response: the median V100, V200, V300, and V400 was 100% versus 99%, 100% versus 97%, 100% versus 74%, and 100% versus 43% in the complete response versus non-complete response groups, respectively (all p < 0.05). CONCLUSION: Voxel-based dosimetry was well-correlated with explant pathology. The D95 threshold had the highest accuracy, suggesting the D95 may be a relevant target for multi-compartment dosimetry.