Adaptive Strategy for External Beam Radiation Therapy in Prostate Cancer: Management of the Geometrical Uncertainties With Robust Optimization

Abstract

PurposeWe aim to develop and validate a new adaptive method for prostate cancer radiation therapy (RT), using an offline strategy to improve treatment personalization by modeling the internal target volume on individual basis and account for the residual set-up uncertainties by robust optimization. Methods and MaterialsTwenty patients with intermediate-high prostate cancer treated with radical radiation therapy were enrolled. The first step of the offline adaptive RT strategy is the identification of a patient-specific internal target volume based on the kV cone beam computed tomography (kV-CBCT) data sets acquired during the first 5 fractions. The deformable image registration algorithm ANACONDA was used to propagate the clinical target volumes (CTVs) from the reference-planning computed tomography to the CBCTs; these contours were assessed by a radiation oncologist. In the second step, the internal target volume was used to replan the treatment using a min-max robust algorithm based on the worst scenario optimization. The CTV coverage and organs-at-risk sparing achieved with the robust plan (RP) were analyzed and compared with the original standard plan, calculating the dose distributions on the residual CBCTs. ResultsThe RP was shown to achieve optimal coverage of the CTV even in the worst scenario, with significantly lower doses to the rectum and bladder. CTV coverage of the RP was statistically better than the standard plan in terms of D99 (P = .008) and D98 (P = .02). Statistically significant mean dose reduction and D2 reduction were noted for the rectum (P < .05) and bladder (P < .009). Moreover, the RP appeared to be less sensitive to bladder and rectal filling. ConclusionsThis adaptive strategy in prostate cancer radiation therapy is feasible and safe; it may be used to adapt the treatment with better target coverage and organs-at-risk sparing than standard planning target volume–based planning.