1133 Reduction of PTV Margin for Accelerated Partial Breast Irradiation Using On-Line Detection of SurgicalClips

L. Kim,1 F. Vicini,1 D. Yan,1 C. Vargas,1 A. Martinez,1 J. Wong1

1Radiation Oncology, William Beaumont Hospital, Royal Oak, MI

Purpose/Objective: External beam radiation therapy (EBRT) provides an alternate method for accelerated partial breastirradiation (APBI) when the location and shape of the lumpectomy cavity renders a patient unsuitable for brachytherapy.However, the EBRT requirement of a 1.5 cm expansion from the cavity to form the clinical target volume and an additional1.0 cm to form the planning target volume (PTV) results in increased volume of breast irradiated. Because of this, somecandidates for EBRT cannot meet the requirement that no more than half of the breast tissue receive more than 50% of theprescribed dose at 19.25 Gy. Reduction of PTV expansion by 5 mm would increase the number of patients eligible for EBRTby about 25% based on our experience with 40 APBI patients. This study investigates the PTV reduction possible whenlocalization of clips implanted during surgery is used to correct for patient setup.

Materials/Methods: In previous work, using repeat CT scans of 28 patients acquired 7 days to 3 months apart, we demonstratedthat surgical clips placed during lumpectomy are strong radiographic surrogates for the biopsy cavity. The 3D planning studybased on center of mass evaluation showed that, using clips to guide setup, a PTV margin reduction of 5 mm or more is possiblefor the majority of patients. For clinical application, the surgical clips need to be localized at treatment time, a requirement nowfeasible with kV-imaging capability onboard the medical accelerator. In the present study, we first mimic clinical practice usingthe same patient CT set as the first study. Clips are localized using a near-orthogonal pair of digitally reconstructed radiographs(DRRs). The shift in the clips’ center of mass obtained from the 2D DRRs is compared with that obtained from 3D CT. Thiscase represents an idealized situation without imaging noise. The next step is to validate the approach prospectively in the clinic.Cone-beam CT (CBCT) scans are taken of patients in treatment position. Displacement of the clips relative to the prescriptionis calculated in 3D using the CBCT and the planning CT scans. These values are taken as “true.” The same displacement iscalculated using planning DRRs and 2D projection images extracted from the CBCT scan.

Results: In the absence of imaging noise, the shifts in the clips’ center of mass based on 2D projection images and 3D CT dataagree to 0.3 � 0.1 mm. This analysis was ceased after 7 patients as the consistent agreement confirmed the theoreticalequivalence of the 2D and 3D localization algorithms. For the first 5 patients of the intended 30 patient clinical study, 3Danalysis using the planning CT and CBCT showed shifts of 6 � 2mm. When the clips were localized using 2D projected images,the measured shifts differed from the 3D “truth” by 1.3 � 0.5mm.

Conclusions: Clip localization is ideally performed in 3D space, using CBCT. However, efficient, integrated software for thispurpose is not yet available. In the interim, clip localization using projection images can be readily performed. This method hasthe benefits of speed and lower dose, but carries with it sources of added uncertainty. These include the qualitatively differentappearance of the clips in projection images versus DRRs, which can lead to greater user error, and the difficulty in visualizingthe clips against ribs or other high density structures in some projections. Our present study indicates that the added uncertaintyis of the order of 1 mm, making possible a reduction in PTV margin for APBI of about 5 mm.

S336 I. J. Radiation Oncology ● Biology ● Physics Volume 60, Number 1, Supplement, 2004