Introduction
Approximately 35%-40% of patients diagnosed with breast cancer opt to have a mastectomy as a component of curative treatment.Some patients opt for an immediate reconstruction due to associated psychological benefits and favorable aesthetic results.Reconstructive surgery can be
performed at the same time as mastectomy (immediate reconstruction) or later, after wound healing is complete (delayed reconstruction). Reconstruction of the breast with prosthetic implants is often performed as a 2-stage procedure; first, removal of the breast followed by
implantation of a tissue expander. Following expansion of the device and adequate development of tissue, the tissue expander is replaced with permanent implants. The procedure involves the placement of a saline tissue expander under the pectoralis major muscle and the remaining skin
after the breast is removed. Conventional treatment has employed the use of saline tissue expanders that are gradually inflated over several months by periodic injections of saline, causing the overlying skin and muscle to stretch. The injection is often performed via a metallic port in the expander that is just deep to the skin and can be identified with a magnet. Once adequate tissue is developed, the expander is removed and replaced by a permanent breast
implant. While this technique offers many advantages such as minimal additional surgical dissection, the outpatient process can be onerous and time consuming, involving months of office visits and discomfort from the bolus saline inflations. Many patients then require postmastectomy radiation therapy to ensure locoregional control and possibly improve survival.
Although the protocols for radiation vary among patients and institutions, in certain cases of
immediate reconstruction radiation may be administered during the reconstruction process while the tissue expander is in place. Typically, the tissue expansion is completed and the patient begins a 5- to 6-week course of radiation therapy. Several studies have documented the dosimetric effect of the metallic port of the traditional saline expanders. Two of these studies found that the presence of the metallic port did not have a clinically significant impact on the dose distribution. The third Study commented on the importance of using heterogeneity corrections in treatment planning to account for the metallic port. A new alternative to traditional saline expanders has recently been developed by AirXpanders (Palo Alto, CA), and is designed to improve the process of breast tissue expansion. The AeroForm tissue expander consists of a metallic reservoir of compressed carbon dioxide (CO₂
) embedded in a textured silicone shell. A hand-held, radiofrequency dosage controller communicates with the expander and allows the patient to administer 10-cc doses of gaseous CO₂ from a reservoir within the expander, resulting in the expansion of the silicone shell. The AeroForm thus provides a needle-free alternative to saline expanders and allows women to control the expansion process with wireless remote control and eliminates the need for periodic in-office bolus injections as is the case with saline expanders. In light of the fact that AeroForm has a large metallic reservoir in the expander and uses an air equivalent

material (CO₂) rather than a tissue equivalent material such as saline, the experiments described in this paper were undertaken to determine if either of these device characteristics lead to clinically relevant alterations in dose distribution.

Methods and materials

An intact AeroForm device (volume of 400 cc) and an unsealed metallic reservoir were used for measurements.Experiments were conducted to determine the effect on the dose distribution of the metallic reservoir alone and of the entire AeroForm setup on a RANDO phantom (The Phantom Laboratory, Salem, NY).
Metallic reservoir testing
The metallic reservoir was placed on solid water at a distance of 100 cm and irradiated to 100 monitor units using 6-MV photons and a 15 cm × 15 cm field. The xperiment was conducted with the length of the reservoir placed perpendicular to the film (Fig 1A), the length of the reservoir placed parallel to the film (Fig 1B), and with no reservoir present (to serve as a baseline).
For each experiment, Kodak EDR2 film (Lot 198-016; Eastman Kodak Company, Stanford, CT) was placed at various depths ranging from 0 to 8.2 cm from the surface of the solid water. The films were processed and the dose profile at each depth was evaluated after normalizing dose measurements to the average dose at 2 points of exposure on each respective film. These normalization points were chosen at the periphery of each field, away from the reservoir, and at the same location on each film. This procedure was followed in order to eliminate any film, processing, or scanning variability.
Intact tissue expander testing
The intact AeroForm device was inflated to full capacity nd positioned on a RANDO anthropomorphic phantom. Twelve optically stimulated luminescent dosimeters (OSLDs), supplied by LANDAUER (Glenwood, IL),were placed on the expander, 5 at the expander-phantom interface (positions 1-5), and 7 at the expander-bolus interface (positions 6-12). A bolus thickness of 1.0 cm was placed over and around the expander in order to replicate the in vivo position anticipated for clinical use (seeFig 2). A computed tomography (CT) scan of this setup was
performed and used for treatment planning. A standard

plan of opposed tangential fields using 6-MV photons with 15-degree wedges was created using the Varian Eclipse treatment planning system (Varian Medical Systems, Palo Alto, CA), AAA-8615 calculation algorithm. The reservoir in the expander created an artifact on the treatment planning CT

scan. The long axis of the reservoir appeared to be parallel to the central axis of the tangents as seen onFig 3. The artifact in this figure was most pronounced at the medial and lateral aspects of the reservoir. This artifact was manually corrected on each CT slice before treatment planning.
The phantom-expander setup was irradiated per plan to a dose of 2 Gy (seeFig 3A;B). The setup was disbanded and the dose to each OSLD was measured. The baseline dose to the OSLDs from the CT simulation process was measured and subtracted from these readings. The entire experiment was repeated 4 additional times. The OSLDs were identified on the treatment planning CT scan and the expected dose (Dplan) to each OSLD was determined as calculated by the treatment planning
system with heterogeneity corrections taken into account. Heterogeneity correction was then turned off in order to calculate the expected dose (Dtissue) in the setting of tissue equivalent material throughout; ie, without the presence of either metal or air. Prior to testing the intact expander, a control experiment was conducted by placing OSLDs on the RANDO phantom (under 1 cm of bolus) without the expander in place, to determine the baseline range of accuracy of the OSLDs.

Results
Metallic reservoir
The results of the metallic reservoir testing reveal that the reservoir causes a significant decrease in transmitted dose when placed perpendicular or parallel to the film (Fig 4).When the reservoir is placed perpendicular to the film, the results show that at a depth of 0.7 cm approximately 40% of the dose is transmitted relative to the open field with no reservoir. At a depth of 8.2 cm, the percentage of transmitted dose relative to the open field with no reservoir rises to 60%. When the reservoir is placed parallel to the film the results vary depending on which area under the reservoir is examined, indicating that the reservoir is not a uniformly
dense object.Figure 5illustrates the percentage depth dose along the central axis as a fraction of Dmax(maximum overall dose) in each of these scenarios. The Dmaxwas found to be at a depth of 1.2 cm when no reservoir was present. Intact tissue expander results The OSLDs have a reported accuracy of ± 5% at the 95% confidence level. Theresultsofacontrol experiment that was conducted on the RANDO phantom without the AeroForm are tabulated inTable 1. After correcting for CT dose to the OSLDs, the readings were found to be within the margin of error of ± 5% when compared with the dose predicted by Eclipse regardless of heterogeneity correction. In experiments using the intact expander on the phantom, the Eclipse calculated.

(expected) dose was compared to the measured OSLD dose at each point (seeTables 2 and 3). With heterogeneity correction on, the average percent difference (measured vs expected dose) with the expander on the phantom was 2.7%,σ= 6.2%. With heterogeneity correction off, the average percent difference (measured vs expected dose) was 3.7%,σ= 2.4%. The only position where the OSLD readings were consistentlyN5% higher than the calculated dose was at position 1, just deep to the canister at the expander-phantom interface. At this position, the readings varied from 5.2% to 14.5%, regardless of heterogeneity correction. Although we did have some readingsN5% at other OSLD positions, we were not able to replicate these readings each time we set up the experiment (see Tables 2 and 3).
Discussion
This study has attempted to address 2 concerns with respect to the dosimetric impact of using the AeroForm^®

tissue expander in patients receiving postmastectomy radiation therapy. First there is the impact of the metallic reservoir, which is somewhat larger than the metallic port in the traditional saline expander. The second concern is the dosimetric impact of air-equivalent material rather than tissue-equivalent material filling the expander. Film dosimetry demonstrated that the partially hollow metallic reservoir has varying impact on the dose
distribution, as seen in Fig 5. There was a definite decrease in dose transmission along the length of the reservoir but this difference decreased in magnitude with increasing depth. Irradiating the reservoir with multiple films in place at selected depths may have had a small
effect on the measured dose at depth but this did not affect the analysis. The dose measured under the reservoir was compared with the dose without the reservoir in place. The measurements were performed with the same film configuration so any effect due to the presence of the film was negated. We were unable to demonstrate a reproducible decrease in dose when irradiating the intact expander. Planning with the use of heterogeneity correction does indicate a region of decreased dose (“cold spot”) at the superficial aspect of the expander (representing the expander-skin interface), medially and laterally. Two possible explanations may be considered. (1) The OSLDs were not placed at the precise location of these cold areas. This is certainly a possibility as scan slices inferior and superior to this precise axial CT slice did not show this“cold spot”; (2) the planning system algorithm is not designed to represent a large metallic
object in the beam. While we corrected for CT artifact using manual contouring techniques, we recognize the difficulty that the Eclipse treatment planning system algorithm may have in modeling high Z materials. An attempt to address this deficiency was made by Chen et al They have described modifications of the electron density of the metallic port (in a traditional saline-filled tissue expander). This recalculated electron density was used with heterogeneity corrections in Eclipse treatment planning software. We believe it would be more challenging to consistently replicate this technique using a large metallic reservoir of varying density.
The average OSLD measurements on the superficial aspect of the expander (positions 7-12) appeared to be within the ± 5% range when compared with the Eclipse treatment plan with or without heterogeneity corrections.In an actual patient, this would represent the expander-skin interface. This is similar to the findings of the in vivo study by Kuo et al They conducted in vivo thermoluminescent dosimetry measurements on a single patient with the gasbased expander in place. These were compared with the treatment plan using heterogeneity corrections, with careful contouring of the gas cavity and the metallic reservoir. In comparison with similar thermoluminescent dosimetry readings performed on 30 nonmastectomy patients treated to the intact breast without bolus, thereadings on the patient with the gas-based expander differed by ± 12%. The authors concluded that the presence of a large gas-filled volume had a limited impact
on the skin dose in this patient. However, in this in vivo study dosimeters could not be placed at the tissueexpander interface. The use of a phantom also permitted measurements deep to the expander, representing the chest wall-expander interface. There was only 1 OSLD location in the current study that revealed the OSLD readings to be consistently higher than the calculated dose, either with or without the use of heterogeneity corrections. The measured OSLD dose directly adjacent to the metallic reservoir (position 1) was consistentlyN5% higher than the Eclipse calculated dose. The OSLDs at this position reported higher than expected measurements on both the heterogeneity corrected plan and the nonheterogeneity corrected plan. This raises concern regarding the possibility that the presence of the expander may result in a small area of increased dose at the expander-chest wall interface that may not be predicted by a treatment planning system. As we were unable to replicate this increased dose at the peripheral aspects of the expander-chest wall interface, we believe that the area consistently receiving the increased dose is small. The current study did not employ energies other than 6-MV photons, nor did we attempt to design a homogenous dose distribution with 3-dimensional or intensity modulated
radiation therapy planning. The discrepancy between the calculated and observed doses deep to the reservoir suggest using caution when attempting to employ advanced treatment planning techniques to homogenize the dose distribution. The current study demonstrates similar OSLD readings, with and without heterogeneity correction. Thus, at most OSLD locations the presence of gas rather than tissueequivalent material does not appear to have had a significant impact on the measured dose as demonstrated by the OSLD readings.
Conclusion
The presence of CO₂ and a metallic reservoir did not preclude the delivery of acceptable amounts of radiation in our phantom model. While film dosimetry demonstrated dose
perturbation in the shadow of the metallic reservoir, this effect did not lead to clinically relevant dose decrements at the AeroForm tissue-expander interface in our phantom studies. Experiments for the current study were conducted using a 3-dimensional rendition of a 2-dimensional plan using standard tangent fields with 6-MV photons. These experiments are not designed to comment on the use of intensity modulated radiation therapy. These experiments reflect only results based on using the Varian Eclipse treatment planning system, AAA-8615 calculation algorithm. Measurements Dosimetry of the AeroForm tissue expander 7 Practical Radiation Oncology: Month 2014 mayneedtoberepeatedifotherplanningsystemsare employed for patient management. Additional evaluation may assist in developing strategies to minimize alterations in
dose distribution by this novel expander.