361 - Outlining Variations and Limitations in 90Y-SIRT Dosimetry: Impacts on Dose Thresholds for Response and Toxicity

Purpose: Further studies linking dosimetry with response and toxicity remain highly relevant in ⁹⁰Y-SIRT, but a primary issue is the lack of standardization in dosimetry methods and reporting. Previous guidance has identified different planning dose thresholds between glass {1} and resin {2} devices to increase response probability and limit toxicity. Additionally, glass and resin have minimal overlap with respect to specific activity, affecting absorbed dose/microsphere concentration relationships. Recent research has highlighted systematic vendor activity calibration biases (VAC-bias) for glass (true activity ~20% lower) {3} and resin (~25% greater) {4}, yielding an absolute difference of ~50% between devices. Here, the ⁹⁰Y-SIRT dosimetry literature was critically examined for both glass and resin to explore variations in methodology and their impacts on dose-response assessments.

Materials and Methods: A search of the PubMed database was conducted from 2014 onwards for ⁹⁰Y-SIRT dosimetry and HCC tumor response studies. Mean/median tumor absorbed dose and dose thresholds associated with objective response (OR) were collected. Variations in dosimetry imaging modality and computational details such as activity normalization (self-calibration to VOI, calibration factor yielding Bq/ml) were identified. The impact of VAC-biases on dose thresholds for response and toxicity was also assessed.

Results: Twenty-nine studies were included, with roughly equal splits for device (19 glass, 14 resin) and dosimetry approach (12 pre-therapy, 19 post-therapy). Most studies utilized voxel dosimetry, but other details like activity normalization were highly varied or ambiguous. Mean/median doses for OR ranged from 194 – 564 Gy for glass and 102 – 263 Gy for resin. In one head-to-head glass vs resin study, threshold doses for response between devices were < 1% different before correcting for VAC-bias. In a similar study, an initial 25% difference was made roughly equal after accounting for VAC-bias. Differences in mean lung dose associated with radiation pneumonitis {5} were decreased after bias corrections, with glass vs resin changing from (41 vs 21 Gy) to (33 vs 26 Gy) {6}.

Conclusion: Determining the optimal therapeutic index in ⁹⁰Y-SIRT requires accurate dose-response and dose-toxicity thresholds. Aggregating accurate dosimetry and outcome data within and across devices will facilitate improvement of response/toxicity models. To maximize the impact of ⁹⁰Y-SIRT patient studies, standardized dosimetry approaches including an absolute dosimetry scale without VAC-biases are crucial.