Purpose: To quantify the magnitude of dosimetric differences relevant to plan quality evaluation between strut-adjusted volume implant (SAVI) plans computed using TG-43 formalism and a commercial model-based dose calculation algorithm (MBDCA).
Methods: Plans for 40 breast patients treated via APBI using the SAVI device were collected. Prescribed dose in all cases was 34 Gy in ten fractions over five treatment days to a PTV_EVAL which contained all breast tissue within 1 cm of the SAVI applicator and more than 5 mm from skin. Differences between dose distributions were evaluated with MBDCA results subtracted from TG-43 calculations. PTV_EVAL DVH metrics with previously-established target and ideal values, including V90, V95 and V100 (%), were compared. Instances where re-calculation with the MBDCA might trigger plan adjustment based on quality metrics were tallied. Wilcoxon rank-sum tests were performed with dose metrics stratified by features that were hypothesized to be predictive of greatest deviations between TG-43 and MBDCA. Features tested included 1) mean CT number in the breast cavity, 2) ratio of PTV_EVAL to contoured SAVI+1cm, 3) applicator size and 4) relative dwell times between central/peripheral catheters.
Results: The minimum and maximum mean dose differences within the PTV_EVAL across the cohort were 3.0% and 14.1% respectively (specified relative to prescription dose). Dose discrepancies increased in closer proximity to catheters, with minimum and maximum values for the 90th percentile in differences of 4.2% and 16.9%, respectively. In 39% (V90), 34% (V95), and 54% (V100) of cases, DVH parameters crossed thresholds distinguishing either â€œidealâ€? vs. â€œexpectedâ€? or â€œexpectedâ€? vs. â€œbelow expectedâ€? plan quality. DVH metrics stratified by features 1, 2 and 3 were observed to be significantly different (p<0.05).
Conclusion: Dose differences between TG-43 and MBDCA calculations for SAVI plans can be large enough to impact plan quality, and hence acceptability, based on previously established DVH metrics.
Not Applicable / None Entered.