| | An anthropomorphic phantom study of visualisation of surgical clips for partial breast irradiation (PBI) setup verification☆Received 29 November 2007; received in revised form 16 March 2008; accepted 16 March 2008. published online 17 April 2008. Abstract Surgical clips were investigated for partial breast image-guided radiotherapy (IGRT). Small titanium clips were insufficiently well visualised. Medium tantalum clips were best for megavoltage IGRT and small tantalum clips were best for floor mounted kilovoltage IGRT (ExacTrac™). Both small tantalum and medium titanium clips were suitable for isocentric kilovoltage IGRT. External beam partial breast irradiation (PBI) is currently being investigated in a number of randomised clinical trials (USA – NSABP B-39 [1], Canada – RAPID [1], UK – IMPORT LOW [2]). With the accelerated, hypofractionated conformal breast radiotherapy, accurate localisation of the target volume is important. Firstly, the surgical cavity, or seroma, must be accurately contoured. Surgical clips aid contouring because they are strong radiographic surrogates for the seroma [3]. Planning of breast radiotherapy with clips has been shown to improve the coverage of partial breast boost volumes [4], [5]. Secondly, using image guidance to minimize interfraction setup uncertainty for PBI minimizes interfraction setup variation. The planning target volume (PTV) expansion can be reduced from the usual 10 to 6 mm, thereby minimizing normal tissue toxicity and increasing the number of patients eligible for PBI [6]. The optimal clip size and material have not been determined for partial breast image-guided radiotherapy (IGRT). With the increasing use of magnetic resonance imaging (MRI) in breast cancer care, we chose to investigate titanium and tantalum clips, which do not produce significant artifact on MRI scans [7], [8]. Metallic surgical clips, however, cause X-ray attenuation in the axial plane on computed tomography (CT), leading to image reconstruction artifact [9]. Surgical clip materials with lower atomic number and, therefore, lower X-ray attenuation coefficient at kV energies, produce less CT artifact [10]. Titanium has an atomic number of 22 and tantalum has an atomic number of 73. These materials produce different amounts of artifact for clips of the same size [8], [11]. Different sizes of surgical clips and different CT scanning parameters create varying amounts of artifact [11]. The CT artifact caused by surgical clips may compromise seroma contouring, so an ideal clip type would cause minimal artifact. This study was designed to identify the ideal material and size of surgical clips for daily on-line setup verification with commercially available floor mounted and gantry mounted kV imaging systems, as well as with an electronic portal imager. We aimed to determine the best surgical clips for partial breast IGRT by investigating, firstly, the visibility of small and medium sized clips made of titanium and tantalum in all four breast quadrants of an anthropomorphic phantom and, secondly, the amount of CT artifact generated by the different surgical clips. Materials and methods  The four types of surgical clips studied were medium tantalum (MTa), small tantalum (STa), medium titanium (MTi), and small titanium (STi) (Edward Weck and Co. Inc., Research Triangle Park, NC, USA). The small clips were 3.5×1.0×0.5mm when closed and the medium clips were 5.5×1.0×0.8mm when closed. Four clips of one type were orientated at 0°, 30°, 60°, and 90° to each other in the plane tangential to the chest wall and imbedded in 0.5cm of tissue-equivalent wax bolus. The clip-imbedded bolus was then placed between the chest wall and the right breast attachment of an anthropomorphic, tissue-equivalent phantom (Rando, Harpell, and Associates Inc., Oakville, Canada). X-ray images were obtained as the four clips were placed in each of the four breast quadrants. This process was repeated for each of the four clip types and for each of the three imaging modalities. Clip visibility The visibility of each clip type in each breast quadrant was examined using the ExacTrac™ (BrainLAB AG, Munich, Germany) floor mounted kV system, isocentric kV and isocentric MV X-rays. We simulated the geometry of commercially available isocentric kV imaging systems {On-Board Imager™ (Clinac® 21EX linear accelerator, Varian Inc., Palo Alto, CA, USA), and the PlanarView™ (Synergy® linear accelerator, Elekta AB, Stockholm, Sweden)}. For isocentric MV portal imaging, AP, lateral and tangential images were taken with the amorphous silicon Portal Vision 1000™ imager of a Varian Clinac® 2100 EX linear accelerator. Portal images were captured using 2MU at a rate of 300MU/min. The visibility of each of the individual clips, in each breast quadrant and at each image angle, was graded by three observers trained in fiducial marker identification. A three-point clip grading system was used: 2 – “clearly visible” (suitable for matching with contoured surgical clips in digital reconstructed radiographs), 1 – “some uncertainty”, and 0 – “not visible”. For analysis, the clips graded as clearly visible were compared to the percentage of clips that were graded as not visible or some uncertainty (unsuitable for matching). Observers were able to adjust the contrast and brightness of images to optimise clip visibility. Intra-observer variability was determined by repeating the clip visibility grading for each clip on two separate occasions at least two weeks apart. Artifact Four clips of each type were placed on graph paper separated by 1.5cm and orientated at 0°, 30°, 60°, and 90°. The sets of clips were placed between 5mm of bolus material and 5cm of Solid Water™ (Gammex rmi®, Middleton, WI, USA) and imaged using a PQ 5000 CT simulator (Philips Medical Systems Inc., Andover, MA, USA). Images were obtained by using slice thicknesses of 1.5mm, 2mm, 3mm, and 5mm with an indices of 1.5mm, 2mm, 3mm, and 5mm, respectively. The clip and its artifact, defined as the combination of the volumes with Hounsfield units >100 (white artifact) and Hounsfield units <−100 (black artifact) around the clip, were objectively contoured by a single observer using Somavision 6.5™ (Varian Inc., Palo Alto, CA, USA), which was also used to calculate the artifact volume. Statistical analysis The percentage of clips clearly visible for each clip type was determined for each imaging modality in the whole breast and in each breast quadrant. Binomial logistic regression and pairwise comparisons were used for the analysis of the clip observations. Intra-observer and inter-observer variability was assessed with the Kappa statistic. The mean volume of artifact generated by each clip type was compared using the Tukey test for the different CT slice thicknesses. All statistical analyses were conducted by the study statistician (JP), using R version 2.2.1 (Free Software Foundation Inc., Boston, MA, USA). All reported tests were two-sided and the alpha was 0.05. Results For the assessment of clip visibility, the intra-observer kappa showed good agreement, with the values of 0.67, 0.72, and 0.79. For the three imaging modalities, the intra-observer kappa for floor mounted kV imaging, isocentric kV imaging, and isocentric MV imaging was 0.68 (good agreement), 0.53 (moderate agreement), and 0.75 (good agreement), respectively. The inter-observer kappa for floor mounted kV imaging, isocentric kV imaging, and isocentric MV imaging was 0.59 (good agreement), 0.38 (moderate agreement), and 0.64 (good agreement), respectively. For floor mounted kV imaging, the percentages of MTa, STa, MTi, and STi clips clearly visible in the whole breast were 98.4%, 91.1%, 53.1%, and 10.4%, respectively (32 clips of each type imaged). For the whole breast, the visibility of MTa>STa>MTi>STi (p=0.002, p<0.0001, p<0.0001). The STi clips were never clearly visible in the left imaging orientation (through the phantom’s spine or mediastinum). In the right orientation, an example of clip X-ray is shown in Fig. 1. For isocentric kV imaging, the percentages of MTa, STa, MTi, and STi clips clearly visible in the whole breast were 99.0%, 93.8%, 88.9%, and 61.8%, respectively (48 clips of each type imaged). For the whole breast, the visibility of MTa>STa>MTi>STi (p<0.0001, p=0.038, p<0.0001). For isocentric MV imaging, the percentages of MTa, STa, MTi, and STi clips clearly visible in the whole breast were 94.8%, 36.1%, 0%, and 0%, respectively (48 clips of each type imaged). For the whole breast, the visibility of MTa>STa>MTi=STi (p<0.0001, p<0.0001, p>0.05). STi and MTi clips were never clearly seen with isocentric MV imaging. Overall MTa clips were best visualised, followed by STa, followed by MTi and followed by STi (p<0.0001). There was no significant difference in clip visibility by quadrant for isocentric kV or isocentric MV imaging. Discussion The ideal surgical clip for partial breast IGRT would have 100% clear visibility on kV and MV imaging, and generate no artifact on the planning CT. In reality, there is no clip type that meets these criteria, and there must be a trade-off between clip visibility and artifact severity. Artifact severity increased for medium clips with decreased CT slice thickness, but this was not observed for small clips. When using medium clips, we recommend the use of CT slice thicknesses of 3mm or larger to minimize artifact. Although MTa clips were significantly better visualised than the other clip types, they produced artifact that degraded the axial CT images of the phantom to such a degree that, if they were used in patients, the oncologist’s ability to contour a seroma would likely be impaired. Further investigation in a patient setting is suggested. The STi clips were not adequately visualised in any of the three imaging modalities and are unsuitable as fiducial markers. MTi clips are adequate for isocentric kV imaging (clearly visible=89%), but they are not adequate for floor mounted kV imaging (clearly visible=53%). For both kV imaging modalities, STa clips performed well: their visibility was high (clearly visible>90%) and the amount of artifact they generated, although slightly greater, was not significantly greater than that produced by MTi clips. However, cost is an additional practical consideration: tantalum clips are usually two to three times more expensive than titanium clips, depending upon the metal market prices of tantalum and titanium. For radiotherapy centres using isocentric kV imaging for PBI, MTi clips may be preferred over STa clips because of the difference in cost. Conclusions The ideal clip to be used as a fiducial marker for partial breast IGRT should have high clip visibility and cause minimal CT artifact. Based upon this anthropomorphic phantom study, we recommend the use of STa clips for floor mounted kV imaging, MTi or STa clips for isocentric kV imaging and MTa for use with MV imaging. Acknowledgements We thank Dr. Kaushik Bhagat, Karen Yurkovich, Cathie Laurie, Diane Heinrichs, and Susan Murtha of the Diagnostic Imaging Department at the BC Cancer Agency for their invaluable assistance. References [1]. [1]www.clinicaltrials.gov; 2008 [accessed 10.03.08]. [2]. [2]www.controlled-trials.com; 2008 [accessed 10.03.08]. [3]. [3]Weed DW, Yan D, Martinez AA, Vicini FA, Wilkinson TJ, Wong J. The validity of surgical clips as a radiographic surrogate for the lumpectomy cavity in image-guided accelerated partial breast irradiation. Int J Radiat Oncol Biol Phys. 2004;60:484–492. Abstract | Full Text |
Full-Text PDF (243 KB)
|
CrossRef
[4]. [4]Bates AT, Swift CL, Kwa W, Moravan V, Aquino-Parsons C. A computed tomography-based protocol vs conventional clinical mark-up for breast electron boost. Clin Oncol (R Coll Radiol). 2007;19:349–355. Abstract | Full Text |
Full-Text PDF (142 KB)
|
CrossRef
[5]. [5]Rabinovitch R, Finlayson C, Pan Z, et al. Radiographic evaluation of surgical clips is better than ultrasound for defining the lumpectomy cavity in breast boost treatment planning: a prospective clinical study. Int J Radiat Oncol Biol Phys. 2000;47:313–317. Abstract | Full Text |
Full-Text PDF (137 KB)
|
CrossRef
[6]. [6]Kim L, Wong J, Yan D. On-line localization of the lumpectomy cavity using surgical clips. Int J Radiat Oncol Biol Phys. 2007;69:1305–1309. Abstract | Full Text |
Full-Text PDF (364 KB)
|
CrossRef
[7]. [7]Black J. Biological performance of tantalum. Clin Mater. 1994;16:167–173. MEDLINE |
CrossRef
[8]. [8]Wang JC, Yu WD, Sandhu HS, Tam V, Delamarter RB. A comparison of magnetic resonance and computed tomographic image quality after the implantation of tantalum and titanium spinal instrumentation. Spine. 1998;23:1684–1688. MEDLINE |
CrossRef
[9]. [9]Marks WM, Callen PW. Computed tomography in the evaluation of patients with surgical clips. Surg Gynecol Obstet. 1980;151:557–558. MEDLINE [10]. [10]Glover GH, Pelc NJ. An algorithm for the reduction of metal clip artifacts in CT reconstructions. Med Phys. 1981;8:799–807. MEDLINE |
CrossRef
[11]. [11]von Holst H, Bergstrom M, Moller A, Steiner L, Ribbe T. Titanium clips in neurosurgery for elimination of artefacts in computer tomography (ct) a technical note. Acta Neurochir (Wien). 1977;38:101–109. MEDLINE |
CrossRef
a Department of Radiation Oncology, Vancouver Centre, Vancouver, BC, Canada b Department of Surveillance and Outcomes Unit, Vancouver Centre, Vancouver, BC, Canada c Department of Radiation Therapy, Vancouver Centre, Vancouver, BC, Canada d Department of Medical Physics, Vancouver Centre, Vancouver, BC, Canada  Corresponding author. Department of Radiation Oncology, Vancouver Centre, British Columbia Cancer Agency, Vancouver, BC, Canada V5Z 4E6.
☆ Meeting Presentation: Abstract 2051, ECCO 14, Barcelona, Spain, September 23–27, 2007. PII: S0167-8140(08)00158-8 doi:10.1016/j.radonc.2008.03.011 © 2008 Elsevier Ireland Ltd. All rights reserved. | |
|
FULL TEXT00158-8/journalimage?img=PIIS0167814008001588.gr1.lrg.gif&fig=fig1&kwhquery=&issn=0167-8140&ishighres=false&allhighres=false&free=yes','journalimage');)
00158-8/journalimage?img=PIIS0167814008001588.gr2.lrg.gif&fig=fig2&kwhquery=&issn=0167-8140&ishighres=false&allhighres=false&free=yes','journalimage');)