Proton therapy of prostate cancer by anterior-oblique beams: implications of setup and anatomy variations.

Proton therapy of prostate by anterior beams could offer an attractive option for treating patients with hip prosthesis and limiting the high-dose exposure to the rectum. We investigated the impact of setup and anatomy variations on the anterior-oblique (AO) proton plan dose, and strategies to manage these effects via range verification and adaptive delivery. Ten patients treated by bilateral (BL) passive-scattering proton therapy (79.2 Gy in 44 fractions) who underwent weekly verification CT scans were selected. Plans with AO beams were additionally created. To isolate the effect of daily variations, initial AO plans did not include range uncertainty margins. The use of fixed planning margins and adaptive range adjustments to manage these effects was investigated. For each case, the planned dose was recalculated on weekly CTs, and accumulated on the simulation CT using deformable registration to approximate the delivered dose. Planned and accumulated doses were compared for each scenario to quantify dose deviations induced by variations. The possibility of estimating the necessary range adjustments before each treatment was explored by simulating the procedure of a diode-based in vivo range verification technique, which would potentially be used clinically. The average planned rectum, penile bulb and femoral heads mean doses were smaller for initial AO compared to BL plans (by 8.3, 16.1 and 25.9 Gy, respectively). After considering interfractional variations in AO plans, the target coverage was substantially reduced. The maximum reduction of V 79.2/D 95/D mean/EUD for AO (without distal margins) (25.3%/10.7/1.6/4.9 Gy, respectively) was considerably larger than BL plans. The loss of coverage was mainly related to changes in water equivalent path length of the prostate after fiducial-based setup, caused by discrepancies in patient anterior surface and bony-anatomy alignment. Target coverage was recovered partially when using fixed planning margins, and fully when applying adaptive range adjustments. The accumulated organs-at-risk dose for AO beams after range adjustment demonstrated full sparing of femoral heads and superior sparing of penile bulb and rectum compared to the conventional BL cases. Our study indicates that using AO beams makes prostate treatment more susceptible to target underdose induced by interfractional variations. Adaptive range verification/adjustment may facilitate the use of anterior beam approaches, and ensure adequate target coverage in every fraction of the treatment.