- Email: [email protected]
Treatment parameters for beta and gamma devices in peripheral endovascular brachytherapy
Int. J. Radiation Oncology Biol. Phys., Vol. 60, No. 5, pp. 1652–1659, 2004 Copyright © 2004 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/04/$–see front matter
doi:10.1016/j.ijrobp.2004.07.716
PHYSICS CONTRIBUTION
TREATMENT PARAMETERS FOR BETA AND GAMMA DEVICES IN PERIPHERAL ENDOVASCULAR BRACHYTHERAPY CHRISTIAN KIRISITS, D.SC.,* BORIS POKRAJAC, M.D.,* DANIEL BERGER, M.SC.,* ERICH MINAR, M.D.,† RICHARD PÖTTER, M.D.,* AND DIETMAR GEORG, PH.D.* *Department of Radiotherapy and Radiobiology, and †Division of Angiology, Department of Internal Medicine II, Medical University of Vienna, Vienna, Austria Purpose: To determine dosimetric parameters, such as radial and longitudinal dose profiles, for  and ␥ devices in peripheral endovascular brachytherapy. Methods and Materials: An 192Ir high-dose rate stepping source, a 90Sr source train, and a 32P-coated radiation balloon were investigated. The treatment-planning software PLATO, Monte Carlo code EGSnrc, and GafChromic film dosimetry were used to analyze the dose distribution of these devices. Results: For a 5-mm-diameter vessel, the ratio between the dose at 2 mm depth and the dose at the lumen surface was 1.8, 3.4, and 16.2 for the 192Ir, 90Sr, and 32P devices, respectively. The dose variation at the reference depth of 2 mm into the vessel wall was 7–18 Gy, for different analyzed dose prescriptions. The reference lumen dose was different by a factor >8. For all three devices, the reference isodose length was not <5 mm on the proximal and distal edge of the active source length. Conclusions: A complete set of dose parameters for  and ␥ sources has to be considered for appropriate treatment planning and performance, including reporting of reference depth dose, reference lumen dose, and reference isodose length. © 2004 Elsevier Inc. Endovascular brachytherapy, Peripheral arteries, Restenosis, Dosimetry.
INTRODUCTION
(AAPM) Task Group 60 (11), the German Society for Medical Physics (DGMP) Working Group 18 (12, 13), and the endovascular (EVA) Working Group of the Groupe Europeen Curietherapie–European Society of Therapeutic Radiation and Oncology (GEC ESTRO) (14) recommend specifying or reporting the reference depth dose (RDD) at a distance of vessel lumen radius plus 2 mm reference depth into the vessel wall. In addition, the EVA GEC ESTRO recommendations introduced several other terms and parameters for prescribing and reporting endovascular brachytherapy. The aim of this article is to provide an overview of the radial dose distributions and recommended dose and length parameters of three source types for treatment planning. Dosimetric results are based on independent calculations and measurements, as well as on existing data in the literature. In addition, different treatment protocols for peripheral endovascular brachytherapy are analyzed.
The technique of endovascular brachytherapy (EVBT) in peripheral arteries is based on traditional endoluminal radiotherapy procedures with a 192Ir high-dose rate (HDR) afterloading device (1, 2). Several clinical trials investigated the effect of radiation after percutaneous interventions in peripheral arteries: the Peripheral Artery Radiation Investigational Study (PARIS) and the Vienna, Bern, and Swiss trials (3–10). Recently, some  devices have been adapted for use in peripheral arteries. In May 2001, a pilot trial was conducted to investigate the effect of peripheral brachytherapy with a 32P-coated balloon (the Radiation After PTA In De novo lesions [RAPID] trial), and in 2002 a system based on a 90Sr source train positioned in a CO2-filled centering catheter balloon was first tested in femoropopliteal arteries (the MOre patency with Beta for In-stent restenosis in the Lower Extremity [MOBILE] trial). To date, both systems have only been used in clinical trials and are currently not available commercially. For all these studies, dose prescription was not performed in a uniform way. Hence, the doses reported for all the clinical trials and experiences are often not comparable with each other. The American Association of Physicists in Medicine
METHODS AND MATERIALS Source designs and delivery devices HDR 192Ir afterloader. The calculation of the radial dose profile presented in this work is based on the dosimetric
Reprint requests to: Christian Kirisits, D.Sc., Department of Radiotherapy and Radiobiology, Division of Medical Radiation Physics, Medical University of Vienna, Währinger Gürtel 18-20, Vienna A-1090, Austria. Tel: (⫹43) 1-40400-2695; Fax: (⫹43)
1-40400-2696; E-mail: [email protected] Received Feb 19, 2004, and in revised form Jul 22, 2004. Accepted for publication Jul 27, 2004. 1652
Treatment parameters in peripheral EVBT
parameters of the microSelectron classic HDR source (Nucletron, Veenendal, The Netherlands). This source type has an active source core 3.5 mm in length and 0.6 mm in diameter and is encapsulated in stainless steel, with a total diameter of 1.1 mm. Dosimetric characteristics are well described in the literature (15). For EVBT applied in peripheral vessels, the source is delivered with an afterloader using noncentered endoluminal catheters or specially designed centering catheters with inflatable balloons with 5-mm step size. 90 Sr source train. The radioactive source design of the Corona system (Novoste, Norcross, GA) is identical to the Beta-Cath device (Novoste) used for coronary treatments. Currently, the 5F version with an active source length (ASL) of 60 mm and containing 24 individual seeds is adapted for peripheral treatments. In each 2.5-mm-long seed, a solid/ceramic 90Sr/90Y core of 0.56-mm diameter is encapsulated in a stainless steel cylinder of 0.64-mm outer diameter. In contrast to the device used for coronary treatments, the delivery device consists of a 7F (2.34-mm outer diameter) catheter, which is at its distal end surrounded by a nylon balloon of 5– 8-mm outer diameter. After positioning at the correct position in the vessel, the balloons are inflated with CO2 at 2 atm. 32 P-coated balloon. The Peripheral RDX System (Radiance Medical Systems, Irvine, CA) is based on a balloon with 32P powder encapsulated within two layers of its outer surface. The active layer has a thickness of ⬍10 m. Until the treatment procedure, the balloon is stored in protective plastic shielding. For irradiation, the balloon is positioned with an attached catheter inside the artery and inflated with 2 atm of saline water to a nominal diameter of 4 mm or 5 mm. The ASL of this device is 40 mm. The activity distribution is not homogenous at the edges of the balloon. Within a 3-mm margin at each edge of the ASL, the activity per length is decreased to 50%. Hence, a margin of 3 mm is proposed for the overlap of two inflations in multisegmental treatment. Calculations with a treatment-planning system The radial dose profile in the case of 192Ir HDR treatments was calculated with the PLATO Brachytherapy Planning System (BPS) version 14.2.3 (Nucletron, Veenendal, The Netherlands). The ASL was 100 mm with 21 dwell positions and 5-mm step size. The dose distribution was normalized and optimized on dose points at a 4.5-mm distance from the source center. This distance corresponds to the reference depth of 2 mm into the vessel wall in the case of a 5-mm lumen diameter. Dose points were located parallel to the source axis at each dwell position, excluding the first and the last position. Dose point optimization increases the dwell times at the edges of the ASL. For further comparison, the same plan was calculated with PLATO BPS version 13.7 with the anisotropy function, the treatment-planning system used in most of the patients enrolled in the Vienna clinical trials (4 – 8). The third plan was based on the algorithm without anisotropy factor. Finally, another
● C. KIRISITS et al.
1653
plan was created to illustrate the impact of dose point optimization by normalizing to only one dose point in the central plane and equal dwell times for each source position.
Monte Carlo calculation The Monte Carlo code EGSnrc in combination with the user code DOSRZnrc was used to calculate radial and longitudinal dose profiles for both  devices (16). This user code provides a planar– cylindrical geometry, with the z axis coinciding with the central source axis and r as the radius in transverse direction. For radial dose profiles, the dose was calculated in cylindrical shells of water with a thickness of ⌬r ⫽ 0.01 mm (0.005 mm for the bin adjacent to the balloon surface) and a length of ⌬z ⫽ 10 mm surrounding the radiation delivery device. Longitudinal dose profiles were based on scoring bins of ⌬r ⫽ 0.1 mm and ⌬z ⫽ 0.2 mm. The source material for the 90Sr seed train was taken as elemental Sr ( ⫽ 2.6 g cm⫺3) encapsulated by a SS304 steel encapsulation with a density of 7.82 g cm⫺3. Encapsulations at the top and bottom of each individual seed and radiopaque marker seeds were neglected. In contrast to the construction of the 5F system, the seed source is assumed to be in the center of the delivery catheter, simulated by a cylindrical shell of water with 2.34-mm thickness. The centering balloon is modeled with its outer diameter of 5 mm and filled with CO2 at 2 atm, which is equal to a density of 3.92 10⫺3 g cm⫺3. A thickness of 0.04 mm was assumed for the balloon foil. The tapering at the ends of the balloon catheter was not modeled. The 32P-coated balloon was simulated once as a small shell of 0.1-mm thickness with an outer diameter of 5 mm and a second time with 0.01-mm thickness and a 0.01-mm outer wall thickness emitting the 32P  ray spectrum. As an approximation, it was assumed that the medium for the whole simulation geometry is homogenous and water equivalent. The calculation of longitudinal dose profiles was performed by adding two profiles, once without any activity within the 3-mm margin at the edges containing only half the linear activity. The effective nuclide of the 90Sr seed trains is the daughter nuclide 90Y, present in radioactive equilibrium. Beta particles emitted from 90Sr were assumed to have too low an energy for penetration to regions where we calculated dose values and were therefore not simulated. The energy spectrum for 90Y and 32P was taken from the LUND database (17) and verified with the spectrum diagrams given by Cross et al. (18). Global cutoff energies for electrons and photons are AE ⫽ ECUT ⫽ 0.010 MeV and AP ⫽ PCUT ⫽ 0.521 MeV, respectively. The upper cutoff energies for each medium are UE ⫽ UP ⫽ 3.0 MeV. Most improvements of the EGSnrc code as exact boundary crossing algorithm, the electron-step algorithm PRESTA-II, spin and binding effects are turned on. Simulations were performed with 107 emitted particles on a personal computer (AMD K7 processor, 500 MHz) operating under LINUX.
1654
I. J. Radiation Oncology
● Biology ● Physics
Measurement setup Dosimetric measurements were performed for the 90Sr source train deployed into the gas-filled centering balloon. The balloon filled with CO2 at 2 atm was positioned on a GafChromic film HD810 (Nuclear Associates, New York, NY). The longitudinal dose profile of the 192Ir source was measured with an endoluminal catheter fixed to the film. Solid water plates below the film and water equivalent gel poured on to this setup ensured full scatter conditions with at least 1 cm of material around the emitting source and the sensitive film area. A thin (8-m) plastic sheath prevented contact of the GafChromic media with the gel. This film type consists of a 7-m-thick layer of radiation-sensitive material on a 100.5-m-thick Mylar base (19). Films were scanned with an He-Ne laser densitometer (Ultro Scan XL Laser Densitometer, Pharmacia LKB Biotechnology, Uppsala, Sweden) with a spot size of 100 m and a step size of 80 m. A linear relationship between dose and net optical density (OD) was verified for our dedicated film and densitometry system between 40 Gy and 200 Gy (20). The irradiation time for all measurements was adjusted to deliver a dose within that range to the film area used for analysis. Films were scanned at least 24 h after irradiation. Background values were obtained directly before the measurement but at least 24 h after cutting the films to their appropriate size. Net OD values were derived by subtracting the respective background values. Analysis of the scanned OD matrix was performed with MATLAB 5.1 (MathWorks, Natick, MA). The radial dose profile was taken as the average of the scanned profiles transverse to the source axis over the centered 10 mm at the central plane. The distance to the source center was obtained according to a geometric relationship based on the nominal balloon diameter, the plastic sheath thickness, half of the sensitive film layer thickness, and the position of the scanned OD on the film. For longitudinal dose profiles, rotation symmetry was assumed, using the average between the values on the right and left side parallel to the source. RESULTS HDR 192Ir afterloader In contrast to line sources, the dose distribution showed inhomogeneities in the vicinity of the central axis because it consisted of single dwell positions. Therefore, the radial dose distribution was different at the position of a source and between two dwell positions, as presented in Fig. 1. At the reference depth of 2 mm (i.e., a distance of 4.5 mm from the source axis), the difference was below 0.5% but increased to 7.3% at a 2.5-mm distance, which is at the lumen surface in the case of a centered position inside a 5-mmdiameter vessel. For further comparison, the radial dose distribution is presented as the mean value of these two extreme values. The results obtained with the older version of PLATO (version 13.7) are as follows: the dwell times normalized to
Volume 60, Number 5, 2004
Fig. 1. Radial dose profile in the case of endovascular treatment with an HDR 192Ir source. Values are normalized to 1 at a distance r ⫽ 4.5 mm from the source axis, which corresponds to the reference depth dose at 2-mm depth from a 5-mm-diameter balloon. Because a step size of 5 mm results in a gap, the dose is decreased between two subsequent dwell positions. BPS ⫽ PLATO Brachytherapy Planning System (Nucletron, Veenendal, The Netherlands).
the source strength (i.e., comparing the Total reference air kerma [TRAK]) for an optimized treatment plan with anisotropy correction were only 1.1% lower, compared with a plan created with version 14.2. The radial dose profile below 4 mm showed significant differences. The profile at the position of a source was overestimated, and the profile between dwell positions was underestimated, both up to 7%. However, the mean value was in agreement with the averaged profile of the new version 14.2 (⬍1%). If neglecting the anisotropic dose distribution of the HDR source, the dwell times were 7% lower, resulting in an overestimation of dose. Figure 2 shows an example of longitudinal dose profiles for an ASL of 10 cm. At a 2.5-mm distance from the source axis, which is at the lumen surface in the case of a centered position inside a 5.0-mm-diameter vessel, each dwell position results in a peak. At 2-mm and 4-mm depths into the vessel wall, the dose distribution becomes homogenous. The reference isodose length (RIL) is defined as the length at the reference depth covered by at least 90% of the RDD. With the dose point optimization, the calculated RIL was only 2 mm shorter on each side of the ASL. If a dwell time optimization was not used, the dose fall-off margin at each edge increased to 8 mm. The calculated dose profiles are in good agreement (within 1–2 mm) with the measurements. 90
Sr source train For this source type, Monte Carlo calculations and film measurements were performed. There was only moderate agreement between the resulting radial dose profiles, presented in Fig. 3. The maximum deviation of up to 15% can
Treatment parameters in peripheral EVBT
Fig. 2. Longitudinal dose profiles related to the 192Ir source at the lumen surface, at 2-mm depth, and at 4-mm depth into the vessel wall for a 5-mm-diameter vessel (solid lines: calculations; dashed lines: film measurements). The first dwell position of the HDR source is located at z ⫽ 0 mm and the last dwell position at z ⫽ 100 mm. The dose is normalized to the reference depth dose (RDD) at radius plus 2 mm at the central plane. Reference isodose length can be determined by measuring the length covered by at least 90% of the RDD. For this example, with an active source length (ASL) of 100 mm, the RIL is 96 mm with the dose point optimization. The increase of dwell times at both ends of the ASL results in a more pronounced peak at the outer dwell positions.
be observed at the balloon surface. At 3.0 mm from the source axis, the agreement was within 10%; at 4 mm, which is close to the normalization point, 3%. This was related to high geometric uncertainties in handling the flexible balloon. For comparison, the values published recently by Wang et al. (21) are also presented in Fig. 3. The agreement of both Monte Carlo– calculated profiles was within 5% for all distances 0 – 4 mm from the balloon surface. The deviation showed its maximum at a 2.8-mm distance from the source axis and was below 2% at 3.5–5 mm.
Fig. 3. Radial dose profiles for a 90Sr source train device. Values are normalized to 1 at a distance r ⫽ 4.5 mm from the source axis, which corresponds to the reference depth dose at 2-mm depth from a 5-mm-diameter balloon. Data for comparison from Wang et al. (21).
● C. KIRISITS et al.
1655
The RIL for this kind of source was determined from the longitudinal dose profile calculated with EGSnrc. For one position with an ASL of 60 mm, the RIL was 52.8 mm for a 5.0-mm balloon diameter. This is in excellent agreement with Wang et al., who reported 52 mm for a length covered by 90% of dose line at 2 mm from the balloon surface (21). In Fig. 4, longitudinal dose profiles are compared with measured values. The RIL could be verified within 1 mm. The calculated dose values were also used for determination of absolute dose rates related to the source strength calibration. The reference absorbed dose rate in water at 2 mm from the source axis without any catheters is given on the calibration certificate. For this case, the dose per simulated particle has been taken from a previously published report by our group about calculations and measurements for the  rail system of Novoste (20). The dose per particle at 2 mm and 0.5 mm from the CO2-filled balloon was compared with this already-available value. For a 5-mm balloon, the ratio was 0.286 at a 2-mm distance from the balloon surface and 0.755 at 0.5 mm. The second value was different from the ratio given on the calibration certificate provided by the manufacturer, with a ratio of 0.837 to get the dose at 0.5 mm from the balloon surface (no values were available for 2 mm). Wang et al. (21) reported deviations on the same order of magnitude between their Monte Carlo results and National Institute of Standards and Technology (NIST) calibrations for dose rate. However, when using the dose rate factor presented by this group, stated in Gy min-1 mCi-1, dose rate values would be less than 50% of all the other results discussed above. 32
P-coated balloon The balloon geometry was again simulated with Monte Carlo calculations. The geometry with a 0.01-mm-thick
Fig. 4. Longitudinal dose profiles related to the 90Sr source at the lumen surface, at 2-mm depth, and at 4-mm depth into the vessel wall for a 5-mm-diameter vessel (solid lines: EGSnrc calculations; dashed lines: film measurements). The 60-mm source train starts at z ⫽ 0 mm. The dose is normalized to the reference depth dose at radius plus 2 mm at the central plane. The minor difference in the dose gradient at both edges of the profile results from the simplified simulation geometry of the balloon catheter but is still within a 1-mm range.
1656
I. J. Radiation Oncology
● Biology ● Physics
active layer showed 19% higher dose values at the surface of the balloon compared with the geometry based on a 0.1-mm-thick active layer. At distances of ⬎0.5 mm, the difference was ⬍3% for both geometries. For all the following analyses and discussions, the profile based on the 0.01-mm active layer presented in Fig. 5 is used. The radial dose profile for the 5-mm-diameter balloon was compared with the values in a figure of a 32P-impregnated balloon published by Yue et al. (22). The agreement was within 14% in the whole analyzed region and approximately 2%, on average, between 2.5 mm and 4.5 mm, because the deviations were equally distributed in positive and negative directions. However, this is only a rough comparison because the calculations of Yue et al. (22) were based on a different geometry, with the activity distributed only on the balloon surface. The RIL was calculated to be 31.6 mm (i.e., a dose fall-off margin of 4.2 mm on each side). This margin was based on the 3-mm part coated with only 50% linear activity on each side of the balloon. Neglecting this inhomogeneity, the RIL would be 35.8 mm. Figure 6 shows the calculated longitudinal dose profiles for three different depths into the vessel wall. DISCUSSION The radial dose profile for the HDR 192Ir source showed the most flat dose gradient. At distances close to the source axis, inhomogeneities occurred owing to the spacing between dwell positions. Plato BPS version 13.7 uses a point source approximation. The impact on dose distribution at distances of ⬎4.0 mm was ⬍1%. At closer distances the difference increased to 7%. According to more detailed analyses, the error in assuming point sources is compen-
Fig. 5. Radial dose profile normalized to 2-mm reference depth for a 5.0-mm balloon coated with 32P. The radial distance r is related to the central axis of the balloon and the lumen. The values given from Yue et al. (22) are only for a rough comparison because the underlying geometry is not completely identical.
Volume 60, Number 5, 2004
Fig. 6. Longitudinal dose profiles related to the 32P source at the lumen surface, at 2-mm depth, and at 4-mm depth into the vessel wall for a 5-mm-diameter vessel (EGSnrc calculations). The 40-mm active balloon starts at z ⫽ 0 mm. The dose is normalized to the reference depth dose at radius plus 2 mm at the central plane. Owing to the steep radial dose gradient for this nuclide and specific source design, the relative dose is given in logarithmic scale. The step on both sides of the profile at the balloon (⫽ lumen) surface results from the proximal and distal part of the balloon coated with 50% of the activity.
sated by subsequent dwell positions, especially at distances that are typically used for dose prescribing (23, 24). We conclude that the use of anisotropy functions is essential for treatment planning in endovascular brachytherapy. In principle, our findings should be representative for 192Ir HDR sources in general. Absolute dose rates for Ir sources are calculated with treatment-planning systems, preferably using the AAPM TG43 concept, based on the air kerma strength (Gy m2 h⫺1) of the individual source and alreadywell-established and verified dose parameters (e.g., 15). The calculation, performed by a treatment-planning system based on the TG43 formalism, is related to full water equivalence of all materials involved. In the case of 192Ir, this can be assumed for the plastic parts of the catheters used. Therefore, the given results can be used for systems with or without centering catheter devices. The radial dose distribution for the 90Sr source device showed good agreement with the values published by Wang et al. (21). Small deviations might be due to a different density for the CO2 at 2 atm, which is higher by a factor of 2 in the cited publication. The absolute dose rate for any distance was calculated by multiplying the reference absorbed dose rate at the reference distance with the respective radial dose ratio, as shown in Fig. 3. Owing to the limited range of the  particles, which is much smaller than the active lengths of the investigated sources, dose rates normalized to the reference dose rate remain constant independently from the active source length. In contrast to relative dose values, the uncertainty in determination of the reference absorbed dose rate is large and needs further attention
Treatment parameters in peripheral EVBT
● C. KIRISITS et al.
1657
by primary standard laboratories. To maintain some consistency, the NIST calibration at 0.5 mm from the balloon surface (given on the calibration certificate) can be used by now. The calculated radial dose profiles of all three source types and 5-mm reference lumen diameter are compared in Fig. 7. The steepest dose gradient can be observed for the coated balloon; the most flat profile is present in the case of 192 Ir HDR treatment. All estimations are based on the water equivalence of vessel tissue. Calcification might further decrease the dose in cases of  radiation. To date, different dose prescriptions have been used in clinical trials of peripheral EVBT. The Vienna 1 and 2 and the Bern trials are prescribing a dose of 12 Gy to a fixed distance from the source axis: 3 mm for Vienna 1 and 2 (4 – 6), 5 mm for Bern (9). The PARIS (3), Swiss (10), and Vienna 4 and 5 (7) trials have prescribed a dose of 14 Gy to the reference depth of 2 mm into the vessel wall. In the Vienna 3 trial, the dose at the same depth was increased to 18 Gy (8). In the RAPID trial, using the RDX 32P-coated balloon, a prescription of 20 Gy at 1-mm tissue depth was used, not changing the protocol from the coronary trials using the same system (25). The MOBILE trial, using the 90 Sr source train, is also based on the experiences with coronary treatments. In the STents And Radiation Therapy (START) trial for coronary EVBT, the dose was prescribed at 2 mm from the source axis (26). This would correspond to a depth of 0.5 mm into the vessel wall for a typical coronary artery diameter of 3.0 mm. The dose was based on the actual diameter of the treated vessel, which was 18.4 Gy in the case of 3.0-mm diameter. The same depth and same dose of 18.4 Gy was chosen for treatment of peripheral arteries. The recently started Limb Ischemia treatment and Monitoring post Vascular Brachytherapy to prevent Reste-
nosis (LIMBER) feasibility trial is using a prescription of 14 Gy at 2-mm reference depth. Figure 8 compares the dose distribution for all these given examples. Independent from dose prescription, the EVA GEC ESTRO working group recommends reporting the RDD at the RDD point at 2 mm into the vessel wall and the reference lumen dose (RLDo) at the RLDo point (14). The dose should be reported for the centered situation, as illustrated in Table 1. A reference distance of radius plus 2 mm for peripheral treatments is also recommended by DGMP and AAPM (12–13). This is different from coronary applications, in which DGMP and ESTRO recommend a 1-mm reference depth and AAPM a fixed distance of 2 mm from the source center. In the EVA GEC ESTRO recommendations, this is based on the assumption of clinical relevance for target dose assessment. However, the treated depth, defined as the depth enclosed by the minimum target dose, might differ from the reference depth. The adventitia and peri-adventitia supposed to be the target is located at approximately a 2-mm depth. Intravascular ultrasound analysis resulted in a distance from balloon surface to adventita of 1.5 mm (27). Adding 0.5 mm thickness for the adventita would confirm a distance of 2 mm for the reference depth. If no centering device is used, the minimum and maximum values for both RDD and RLDo should be given in addition (28). The presented dose profiles are of interest related to upper dose constraints for the RLDo. The steep dose gradient of  sources is limiting a dose escalation to an upper limit for the RLDo. To date, no clear data are available on the maximum dose tolerated by the endothelial layer of a peripheral vessel lumen. Based on experiences in coronary arteries, a maximum dose of ⬎100 Gy was feasibly safe. However, one has to distinguish between maximum dose of small volume in the
Fig. 7. Comparison of the radial dose profiles for three different devices treating a 5.0-mm vessel with the reference depth dose normalized to 1. For better readability, the comparable high dose on the surface of the 32P-coated balloon is not shown in this diagram (compare with Fig. 5). HDR ⫽ high-dose rate.
Fig. 8. Summary of the dose distribution within the vessel wall for several trials of endovascular brachytherapy. All values are based on prescriptions for a 5.0-mm-diameter vessel. Solid lines are used for trials using 192Ir, dashed lines for 90Sr, and the dot-dashed line for 32P.
1658
I. J. Radiation Oncology
● Biology ● Physics
Volume 60, Number 5, 2004
Table 1. Summary of dose parameters for clinical trials of endovascular brachytherapy Trial Frankfurt Vienna 1, 2 Vienna 3 PARIS, Swiss Vienna 4, 5 Bern MOBILE LIMBER RAPID
Prescribed dose (Gy)
At distance
RDD (Gy) (2-mm depth)
At 1-mm depth (Gy)
At 0.5-mm depth (Gy)
RLDo (Gy) (0-mm depth)
7.9
10.2
12.0
14.4
12
3 mm
18 14
RLDi ⫹ 2 mm RLDi ⫹ 2 mm
18.0 14.0
23.2 18.0
27.2 21.1
32.7 25.4
12 18.4 14 20
5 mm RLDi ⫹ 0.5 mm RLDi ⫹ 2 mm RLDi ⫹ 1 mm
13.4 7.0 14.0 7.0
17.3 13.4 26.8 20.0
20.2 18.4 36.8 35.7
24.4 24.0 48.1 ⬃ 113
Prescribed doses are related to different distances. In most trials, this distance is based on the reference lumen diameter (RLDi) of the treated artery. The reference depth dose (RDD) is reported at a reference depth of 2 mm into the vessel wall; the reference lumen dose (RLDo) gives the value at the lumen surface. In addition, dose values are given at a depth of 1 mm and 0.5 mm into the vessel wall. The comparison is based on an RLDi of 5.0 mm.
case of a noncentered device or a uniform high dose on the whole luminal surface in the case of a centered device and high RLDo. Based on the experience from recent trials, a dose of at least 14 Gy should be given at 2 mm into the vessel wall. For the 192Ir and the 90Sr device, the dose range at the luminal surface would then be ⬍50 Gy. In the case of the 32P balloon, such a dose prescription exceeds 100 Gy to the vessel lumen. The RIL for a given ASL is, in general, lower for ␥ sources compared with  sources, with their steep dose gradient (20). If sources are used with stepping source technology and increased dwell times at the edges, the RIL can be adjusted to a value close to ASL. The optimization method used in this work is one example of available techniques. Dwell time optimization can also be done by several other geometric algorithms (29). For the procedure in the Vienna trials, using the described dose point optimization, the RIL was at least ASL minus 5 mm on both edges. A rounded value for the RIL of the 90Sr source device is in total 8 mm smaller than the ASL, 4 mm on each source
edge (RIL ⫽ ASL ⫺ 8 mm). This was calculated for a 60-mm ASL, but can be applied for multisegmental treatment. The limited range of the emitted  particles does not change the longitudinal dose profile on the source edge by increasing the ASL. If following the protocol recommended by the vendor, a three-segment treatment results in an ASL of 3 ⫻ 60 mm minus 2 ⫻ 2.5 mm (overlap of one seed for each new position), which is 175 mm. The resulting RIL would then be 167 mm. For the radioactive coated 32P balloon, a good approximation for the RIL is the ASL minus 4.5 mm on each side of the ASL (RIL ⫽ ASL ⫺ 9 mm). This difference between RIL and ASL is not based on the nuclide or the balloon geometry alone but on the inhomogenous activity distribution at the edges. Adequate treatment planning has to be based on these parameters to avoid geographic miss of the interventional length (30). Safety margins have to be fulfilled on the edges, and large dose inhomogeneities have to be avoided during multisegmental treatment (31).
REFERENCES 1. Böttcher HD, Schopohl B, Liermann D, et al. Endovascular irradiation—A new method to avoid recurrent stenosis after stent implantation in peripheral arteries: Technique and preliminary results. Int J Radiat Oncol Biol Phys 1994;29:183–186. 2. Tripuraneni P, Giap H, Jani S. Endovascular brachytherapy for peripheral vascular disease. Semin Radiat Oncol 1999;9: 190 –202. 3. Waksman R, Laird JR, Jurkovitz CT, et al. Intravascular radiation therapy after balloon angioplasty of narrowed femoropopliteal arteries to prevent restenosis: Results of the PARIS feasibility clinical trial. J Vasc Interv Radiol 2001;12: 915–921. 4. Minar E, Pokrajac B, Ahmadi R, et al. Brachytherapy for prophylaxis of restenosis after long-segment femoropopliteal angioplasty: Pilot study. Radiology 1998;208:173–179. 5. Pokrajac B, Pötter R, Maca T, et al. Intraarterial 192Ir HDR brachytherapy for prophylaxis of restenosis after femoropopliteal percutaneous transluminal angioplasty: The prospective randomized Vienna-2-trial radiotherapy parame-
6.
7.
8.
9.
10.
ters and risk factors analysis. Int J Radiat Oncol Biol Phys 2000;48:923–931. Minar E, Pokrajac B, Maca T, et al. Endovascular brachytherapy for prophylaxis of restenosis after femoropopliteal angioplasty: Results of a prospective, randomized study. Circulation 2000;102:2694 –2699. Wolfram R, Pokrajac B, Ahmadi R, et al. Endovascular brachytherapy for prophylaxis against restenosis after longsegment femoropopliteal stenting: Initial results. Radiology 2001;220:724 –729. Pokrajac B, Schmid R, Pötter R. Endovascular brachytherapy prevents restenosis after femoropopliteal angioplasty: Results of the Vienna-3 multicenter study [Abstract]. Int J Radiat Oncol Biol Phys 2003;57(Suppl.):S250. Zehnder D, von Briel C, Baumgartner I, et al. Endovasculer brachytherapy after percutaneous transluminal angioplasty of recurrent femorpopliteal obstructions. J Endovasc Ther 2003; 10:304 –311. Greiner RH, Mahler F, Jäger K, et al. Swiss trials of peripheral
Treatment parameters in peripheral EVBT
11. 12. 13. 14.
15. 16.
17. 18. 19.
20.
21.
endovascular brachytherapy for restenosis prevention after PTA [abstract]. Radiother Oncol 2003;66(Suppl. 1):S6. Nath R, Amols H, Coffey C, et al. Intravascular brachytherapy physics: Report of the AAPM Radiation Therapy Committee Task Group No. 60. Med Phys 1999;26:119 –152. Quast U, Kaulich TW, Flühs D. Guideline for medical physical aspects of intravascular brachytherapy. Part I: Guideline. Z Med Phys 2002;12:47– 64. Quast U, Kaulich TW, Flühs D. Guideline for medical physical aspects of intravascular brachytherapy. Part II: Samples and examples. Z Med Phys 2002;12:133–148. Pötter R, Van Limbergen E, Dries W, et al. Recommendations of the EndoVAscular GEC ESTRO Working Group (EVA GEC ESTRO): Prescribing, recording, and reporting in endovascular brachytherapy. Quality assurance, equipment, personnel and education. Radiother Oncol 2001;59:339 –360. Williamson J, Li Z. Monte Carlo aided dosimetry of the microSelectron pulsed and high dose-rate 192Ir sources. Med Phys 1995;22:809 – 819. Kawrakow I, Rogers DWO. The EGSnrc code system: Monte Carlo simulation of electron and photon transport. Technical report, NRC report PIRS 701. Ottawa, Ontario, Canada: National Research Council of Canada, 2000. Chu SYF, Ekstroem LP, Firestone RB. WWW table of radioactive isotopes, database version 1999-02-28. Available at: http://nucleardata.nuclear.lu.se/nucleardata/toi. Cross WG, Ing H, Freedman N. A short atlas of beta-ray spectra. Phys Med Biol 1983;28:1251–1260. Niroomand-Rad A, Blackwell CR, Coursey BM, et al. Radiochromic film dosimetry: Recommendations of AAPM Radiation Therapy Committee Task Group 55. Med Phys 1998;25: 2093–2115. Kirisits C, Georg D, Wexberg P, et al. Determination and application of the reference isodose length (RIL) for commercial endovascular brachytherapy devices. Radiother Oncol 2002;64:309 –315. Wang R, Li AX, Lobdell J. Monte Carlo dose characterization
22.
23. 24. 25. 26.
27.
28.
29. 30.
31.
● C. KIRISITS et al.
1659
of a new 90Sr/90Y source with balloon for intravascular brachytherapy. Med Phys 2003;30:27–33. Yue N, Nath R, Hehrlein C. Dosimetry calculation for a novel phosphorus-32 impregnated balloon angioplasty catheter for intravascular brachytherapy. Cardiovasc Radiat Med 1999;1: 349 –357. Dries W. Monte Carlo calculated dose distribution for endovascular HDR brachytherapy with Ir-192. Radiother Oncol 1997;45:77– 82. Wong D, Wallace S, Fernando W, et al. Dose errors in the near field of an HDR brachytherapy stepping source. Phys Med Biol 1999;44:357–363. Stone GW, Mehran R, Midei M, et al. Usefulness of beta radiation for de novo and in-stent restenotic lesions in saphenous vein grafts. Am J Cardiol 2003;92:312–314. Popma JJ, Suntharalingam M, Lansky AJ, et al. Randomized trial of 90Sr/Y90 beta-radiation versus placebo control for treatment of in-stent restenosis. Circulation 2003;106:1090 – 1096. Russo R, Apostol J, Silva P, et al. Is intravascular ultrasound necessary for dose prescription during femoral artery brachytherapy? Observations from the Paris trial [Abstract]. Circulation 2000;102(Suppl. II):2065. Pokrajac B, Schmid R, Kirisits C, et al. Possible impact of iridium-192 source centering on restenosis rate after femoropopliteal angioplasty and endovascular brachytherapy in Vienna-2 study. Radiother Oncol 2002;63:97–102. Bratengeier K. A fast algorithm for homogeneous Ir-192 afterloading-irradiation of cylindrical surfaces. Z Med Phys 2001;11:102–107. Schmid R, Kirisits C, Syeda B, et al. Evaluation of geographic miss in intracoronary brachytherapy. Recommendations for determining the planning target length to avoid geographic miss. Radiother Oncol 2004;71:311–318. Coen VLMA, Marijnissen JPA, Ligthart JMR, et al. Inaccuracy in manual multisegmental irradiation in coronary arteries. Radiother Oncol 2002;63:89 –95.
doi:10.1016/j.ijrobp.2004.07.716
PHYSICS CONTRIBUTION
TREATMENT PARAMETERS FOR BETA AND GAMMA DEVICES IN PERIPHERAL ENDOVASCULAR BRACHYTHERAPY CHRISTIAN KIRISITS, D.SC.,* BORIS POKRAJAC, M.D.,* DANIEL BERGER, M.SC.,* ERICH MINAR, M.D.,† RICHARD PÖTTER, M.D.,* AND DIETMAR GEORG, PH.D.* *Department of Radiotherapy and Radiobiology, and †Division of Angiology, Department of Internal Medicine II, Medical University of Vienna, Vienna, Austria Purpose: To determine dosimetric parameters, such as radial and longitudinal dose profiles, for  and ␥ devices in peripheral endovascular brachytherapy. Methods and Materials: An 192Ir high-dose rate stepping source, a 90Sr source train, and a 32P-coated radiation balloon were investigated. The treatment-planning software PLATO, Monte Carlo code EGSnrc, and GafChromic film dosimetry were used to analyze the dose distribution of these devices. Results: For a 5-mm-diameter vessel, the ratio between the dose at 2 mm depth and the dose at the lumen surface was 1.8, 3.4, and 16.2 for the 192Ir, 90Sr, and 32P devices, respectively. The dose variation at the reference depth of 2 mm into the vessel wall was 7–18 Gy, for different analyzed dose prescriptions. The reference lumen dose was different by a factor >8. For all three devices, the reference isodose length was not <5 mm on the proximal and distal edge of the active source length. Conclusions: A complete set of dose parameters for  and ␥ sources has to be considered for appropriate treatment planning and performance, including reporting of reference depth dose, reference lumen dose, and reference isodose length. © 2004 Elsevier Inc. Endovascular brachytherapy, Peripheral arteries, Restenosis, Dosimetry.
INTRODUCTION
(AAPM) Task Group 60 (11), the German Society for Medical Physics (DGMP) Working Group 18 (12, 13), and the endovascular (EVA) Working Group of the Groupe Europeen Curietherapie–European Society of Therapeutic Radiation and Oncology (GEC ESTRO) (14) recommend specifying or reporting the reference depth dose (RDD) at a distance of vessel lumen radius plus 2 mm reference depth into the vessel wall. In addition, the EVA GEC ESTRO recommendations introduced several other terms and parameters for prescribing and reporting endovascular brachytherapy. The aim of this article is to provide an overview of the radial dose distributions and recommended dose and length parameters of three source types for treatment planning. Dosimetric results are based on independent calculations and measurements, as well as on existing data in the literature. In addition, different treatment protocols for peripheral endovascular brachytherapy are analyzed.
The technique of endovascular brachytherapy (EVBT) in peripheral arteries is based on traditional endoluminal radiotherapy procedures with a 192Ir high-dose rate (HDR) afterloading device (1, 2). Several clinical trials investigated the effect of radiation after percutaneous interventions in peripheral arteries: the Peripheral Artery Radiation Investigational Study (PARIS) and the Vienna, Bern, and Swiss trials (3–10). Recently, some  devices have been adapted for use in peripheral arteries. In May 2001, a pilot trial was conducted to investigate the effect of peripheral brachytherapy with a 32P-coated balloon (the Radiation After PTA In De novo lesions [RAPID] trial), and in 2002 a system based on a 90Sr source train positioned in a CO2-filled centering catheter balloon was first tested in femoropopliteal arteries (the MOre patency with Beta for In-stent restenosis in the Lower Extremity [MOBILE] trial). To date, both systems have only been used in clinical trials and are currently not available commercially. For all these studies, dose prescription was not performed in a uniform way. Hence, the doses reported for all the clinical trials and experiences are often not comparable with each other. The American Association of Physicists in Medicine
METHODS AND MATERIALS Source designs and delivery devices HDR 192Ir afterloader. The calculation of the radial dose profile presented in this work is based on the dosimetric
Reprint requests to: Christian Kirisits, D.Sc., Department of Radiotherapy and Radiobiology, Division of Medical Radiation Physics, Medical University of Vienna, Währinger Gürtel 18-20, Vienna A-1090, Austria. Tel: (⫹43) 1-40400-2695; Fax: (⫹43)
1-40400-2696; E-mail: [email protected] Received Feb 19, 2004, and in revised form Jul 22, 2004. Accepted for publication Jul 27, 2004. 1652
Treatment parameters in peripheral EVBT
parameters of the microSelectron classic HDR source (Nucletron, Veenendal, The Netherlands). This source type has an active source core 3.5 mm in length and 0.6 mm in diameter and is encapsulated in stainless steel, with a total diameter of 1.1 mm. Dosimetric characteristics are well described in the literature (15). For EVBT applied in peripheral vessels, the source is delivered with an afterloader using noncentered endoluminal catheters or specially designed centering catheters with inflatable balloons with 5-mm step size. 90 Sr source train. The radioactive source design of the Corona system (Novoste, Norcross, GA) is identical to the Beta-Cath device (Novoste) used for coronary treatments. Currently, the 5F version with an active source length (ASL) of 60 mm and containing 24 individual seeds is adapted for peripheral treatments. In each 2.5-mm-long seed, a solid/ceramic 90Sr/90Y core of 0.56-mm diameter is encapsulated in a stainless steel cylinder of 0.64-mm outer diameter. In contrast to the device used for coronary treatments, the delivery device consists of a 7F (2.34-mm outer diameter) catheter, which is at its distal end surrounded by a nylon balloon of 5– 8-mm outer diameter. After positioning at the correct position in the vessel, the balloons are inflated with CO2 at 2 atm. 32 P-coated balloon. The Peripheral RDX System (Radiance Medical Systems, Irvine, CA) is based on a balloon with 32P powder encapsulated within two layers of its outer surface. The active layer has a thickness of ⬍10 m. Until the treatment procedure, the balloon is stored in protective plastic shielding. For irradiation, the balloon is positioned with an attached catheter inside the artery and inflated with 2 atm of saline water to a nominal diameter of 4 mm or 5 mm. The ASL of this device is 40 mm. The activity distribution is not homogenous at the edges of the balloon. Within a 3-mm margin at each edge of the ASL, the activity per length is decreased to 50%. Hence, a margin of 3 mm is proposed for the overlap of two inflations in multisegmental treatment. Calculations with a treatment-planning system The radial dose profile in the case of 192Ir HDR treatments was calculated with the PLATO Brachytherapy Planning System (BPS) version 14.2.3 (Nucletron, Veenendal, The Netherlands). The ASL was 100 mm with 21 dwell positions and 5-mm step size. The dose distribution was normalized and optimized on dose points at a 4.5-mm distance from the source center. This distance corresponds to the reference depth of 2 mm into the vessel wall in the case of a 5-mm lumen diameter. Dose points were located parallel to the source axis at each dwell position, excluding the first and the last position. Dose point optimization increases the dwell times at the edges of the ASL. For further comparison, the same plan was calculated with PLATO BPS version 13.7 with the anisotropy function, the treatment-planning system used in most of the patients enrolled in the Vienna clinical trials (4 – 8). The third plan was based on the algorithm without anisotropy factor. Finally, another
● C. KIRISITS et al.
1653
plan was created to illustrate the impact of dose point optimization by normalizing to only one dose point in the central plane and equal dwell times for each source position.
Monte Carlo calculation The Monte Carlo code EGSnrc in combination with the user code DOSRZnrc was used to calculate radial and longitudinal dose profiles for both  devices (16). This user code provides a planar– cylindrical geometry, with the z axis coinciding with the central source axis and r as the radius in transverse direction. For radial dose profiles, the dose was calculated in cylindrical shells of water with a thickness of ⌬r ⫽ 0.01 mm (0.005 mm for the bin adjacent to the balloon surface) and a length of ⌬z ⫽ 10 mm surrounding the radiation delivery device. Longitudinal dose profiles were based on scoring bins of ⌬r ⫽ 0.1 mm and ⌬z ⫽ 0.2 mm. The source material for the 90Sr seed train was taken as elemental Sr ( ⫽ 2.6 g cm⫺3) encapsulated by a SS304 steel encapsulation with a density of 7.82 g cm⫺3. Encapsulations at the top and bottom of each individual seed and radiopaque marker seeds were neglected. In contrast to the construction of the 5F system, the seed source is assumed to be in the center of the delivery catheter, simulated by a cylindrical shell of water with 2.34-mm thickness. The centering balloon is modeled with its outer diameter of 5 mm and filled with CO2 at 2 atm, which is equal to a density of 3.92 10⫺3 g cm⫺3. A thickness of 0.04 mm was assumed for the balloon foil. The tapering at the ends of the balloon catheter was not modeled. The 32P-coated balloon was simulated once as a small shell of 0.1-mm thickness with an outer diameter of 5 mm and a second time with 0.01-mm thickness and a 0.01-mm outer wall thickness emitting the 32P  ray spectrum. As an approximation, it was assumed that the medium for the whole simulation geometry is homogenous and water equivalent. The calculation of longitudinal dose profiles was performed by adding two profiles, once without any activity within the 3-mm margin at the edges containing only half the linear activity. The effective nuclide of the 90Sr seed trains is the daughter nuclide 90Y, present in radioactive equilibrium. Beta particles emitted from 90Sr were assumed to have too low an energy for penetration to regions where we calculated dose values and were therefore not simulated. The energy spectrum for 90Y and 32P was taken from the LUND database (17) and verified with the spectrum diagrams given by Cross et al. (18). Global cutoff energies for electrons and photons are AE ⫽ ECUT ⫽ 0.010 MeV and AP ⫽ PCUT ⫽ 0.521 MeV, respectively. The upper cutoff energies for each medium are UE ⫽ UP ⫽ 3.0 MeV. Most improvements of the EGSnrc code as exact boundary crossing algorithm, the electron-step algorithm PRESTA-II, spin and binding effects are turned on. Simulations were performed with 107 emitted particles on a personal computer (AMD K7 processor, 500 MHz) operating under LINUX.
1654
I. J. Radiation Oncology
● Biology ● Physics
Measurement setup Dosimetric measurements were performed for the 90Sr source train deployed into the gas-filled centering balloon. The balloon filled with CO2 at 2 atm was positioned on a GafChromic film HD810 (Nuclear Associates, New York, NY). The longitudinal dose profile of the 192Ir source was measured with an endoluminal catheter fixed to the film. Solid water plates below the film and water equivalent gel poured on to this setup ensured full scatter conditions with at least 1 cm of material around the emitting source and the sensitive film area. A thin (8-m) plastic sheath prevented contact of the GafChromic media with the gel. This film type consists of a 7-m-thick layer of radiation-sensitive material on a 100.5-m-thick Mylar base (19). Films were scanned with an He-Ne laser densitometer (Ultro Scan XL Laser Densitometer, Pharmacia LKB Biotechnology, Uppsala, Sweden) with a spot size of 100 m and a step size of 80 m. A linear relationship between dose and net optical density (OD) was verified for our dedicated film and densitometry system between 40 Gy and 200 Gy (20). The irradiation time for all measurements was adjusted to deliver a dose within that range to the film area used for analysis. Films were scanned at least 24 h after irradiation. Background values were obtained directly before the measurement but at least 24 h after cutting the films to their appropriate size. Net OD values were derived by subtracting the respective background values. Analysis of the scanned OD matrix was performed with MATLAB 5.1 (MathWorks, Natick, MA). The radial dose profile was taken as the average of the scanned profiles transverse to the source axis over the centered 10 mm at the central plane. The distance to the source center was obtained according to a geometric relationship based on the nominal balloon diameter, the plastic sheath thickness, half of the sensitive film layer thickness, and the position of the scanned OD on the film. For longitudinal dose profiles, rotation symmetry was assumed, using the average between the values on the right and left side parallel to the source. RESULTS HDR 192Ir afterloader In contrast to line sources, the dose distribution showed inhomogeneities in the vicinity of the central axis because it consisted of single dwell positions. Therefore, the radial dose distribution was different at the position of a source and between two dwell positions, as presented in Fig. 1. At the reference depth of 2 mm (i.e., a distance of 4.5 mm from the source axis), the difference was below 0.5% but increased to 7.3% at a 2.5-mm distance, which is at the lumen surface in the case of a centered position inside a 5-mmdiameter vessel. For further comparison, the radial dose distribution is presented as the mean value of these two extreme values. The results obtained with the older version of PLATO (version 13.7) are as follows: the dwell times normalized to
Volume 60, Number 5, 2004
Fig. 1. Radial dose profile in the case of endovascular treatment with an HDR 192Ir source. Values are normalized to 1 at a distance r ⫽ 4.5 mm from the source axis, which corresponds to the reference depth dose at 2-mm depth from a 5-mm-diameter balloon. Because a step size of 5 mm results in a gap, the dose is decreased between two subsequent dwell positions. BPS ⫽ PLATO Brachytherapy Planning System (Nucletron, Veenendal, The Netherlands).
the source strength (i.e., comparing the Total reference air kerma [TRAK]) for an optimized treatment plan with anisotropy correction were only 1.1% lower, compared with a plan created with version 14.2. The radial dose profile below 4 mm showed significant differences. The profile at the position of a source was overestimated, and the profile between dwell positions was underestimated, both up to 7%. However, the mean value was in agreement with the averaged profile of the new version 14.2 (⬍1%). If neglecting the anisotropic dose distribution of the HDR source, the dwell times were 7% lower, resulting in an overestimation of dose. Figure 2 shows an example of longitudinal dose profiles for an ASL of 10 cm. At a 2.5-mm distance from the source axis, which is at the lumen surface in the case of a centered position inside a 5.0-mm-diameter vessel, each dwell position results in a peak. At 2-mm and 4-mm depths into the vessel wall, the dose distribution becomes homogenous. The reference isodose length (RIL) is defined as the length at the reference depth covered by at least 90% of the RDD. With the dose point optimization, the calculated RIL was only 2 mm shorter on each side of the ASL. If a dwell time optimization was not used, the dose fall-off margin at each edge increased to 8 mm. The calculated dose profiles are in good agreement (within 1–2 mm) with the measurements. 90
Sr source train For this source type, Monte Carlo calculations and film measurements were performed. There was only moderate agreement between the resulting radial dose profiles, presented in Fig. 3. The maximum deviation of up to 15% can
Treatment parameters in peripheral EVBT
Fig. 2. Longitudinal dose profiles related to the 192Ir source at the lumen surface, at 2-mm depth, and at 4-mm depth into the vessel wall for a 5-mm-diameter vessel (solid lines: calculations; dashed lines: film measurements). The first dwell position of the HDR source is located at z ⫽ 0 mm and the last dwell position at z ⫽ 100 mm. The dose is normalized to the reference depth dose (RDD) at radius plus 2 mm at the central plane. Reference isodose length can be determined by measuring the length covered by at least 90% of the RDD. For this example, with an active source length (ASL) of 100 mm, the RIL is 96 mm with the dose point optimization. The increase of dwell times at both ends of the ASL results in a more pronounced peak at the outer dwell positions.
be observed at the balloon surface. At 3.0 mm from the source axis, the agreement was within 10%; at 4 mm, which is close to the normalization point, 3%. This was related to high geometric uncertainties in handling the flexible balloon. For comparison, the values published recently by Wang et al. (21) are also presented in Fig. 3. The agreement of both Monte Carlo– calculated profiles was within 5% for all distances 0 – 4 mm from the balloon surface. The deviation showed its maximum at a 2.8-mm distance from the source axis and was below 2% at 3.5–5 mm.
Fig. 3. Radial dose profiles for a 90Sr source train device. Values are normalized to 1 at a distance r ⫽ 4.5 mm from the source axis, which corresponds to the reference depth dose at 2-mm depth from a 5-mm-diameter balloon. Data for comparison from Wang et al. (21).
● C. KIRISITS et al.
1655
The RIL for this kind of source was determined from the longitudinal dose profile calculated with EGSnrc. For one position with an ASL of 60 mm, the RIL was 52.8 mm for a 5.0-mm balloon diameter. This is in excellent agreement with Wang et al., who reported 52 mm for a length covered by 90% of dose line at 2 mm from the balloon surface (21). In Fig. 4, longitudinal dose profiles are compared with measured values. The RIL could be verified within 1 mm. The calculated dose values were also used for determination of absolute dose rates related to the source strength calibration. The reference absorbed dose rate in water at 2 mm from the source axis without any catheters is given on the calibration certificate. For this case, the dose per simulated particle has been taken from a previously published report by our group about calculations and measurements for the  rail system of Novoste (20). The dose per particle at 2 mm and 0.5 mm from the CO2-filled balloon was compared with this already-available value. For a 5-mm balloon, the ratio was 0.286 at a 2-mm distance from the balloon surface and 0.755 at 0.5 mm. The second value was different from the ratio given on the calibration certificate provided by the manufacturer, with a ratio of 0.837 to get the dose at 0.5 mm from the balloon surface (no values were available for 2 mm). Wang et al. (21) reported deviations on the same order of magnitude between their Monte Carlo results and National Institute of Standards and Technology (NIST) calibrations for dose rate. However, when using the dose rate factor presented by this group, stated in Gy min-1 mCi-1, dose rate values would be less than 50% of all the other results discussed above. 32
P-coated balloon The balloon geometry was again simulated with Monte Carlo calculations. The geometry with a 0.01-mm-thick
Fig. 4. Longitudinal dose profiles related to the 90Sr source at the lumen surface, at 2-mm depth, and at 4-mm depth into the vessel wall for a 5-mm-diameter vessel (solid lines: EGSnrc calculations; dashed lines: film measurements). The 60-mm source train starts at z ⫽ 0 mm. The dose is normalized to the reference depth dose at radius plus 2 mm at the central plane. The minor difference in the dose gradient at both edges of the profile results from the simplified simulation geometry of the balloon catheter but is still within a 1-mm range.
1656
I. J. Radiation Oncology
● Biology ● Physics
active layer showed 19% higher dose values at the surface of the balloon compared with the geometry based on a 0.1-mm-thick active layer. At distances of ⬎0.5 mm, the difference was ⬍3% for both geometries. For all the following analyses and discussions, the profile based on the 0.01-mm active layer presented in Fig. 5 is used. The radial dose profile for the 5-mm-diameter balloon was compared with the values in a figure of a 32P-impregnated balloon published by Yue et al. (22). The agreement was within 14% in the whole analyzed region and approximately 2%, on average, between 2.5 mm and 4.5 mm, because the deviations were equally distributed in positive and negative directions. However, this is only a rough comparison because the calculations of Yue et al. (22) were based on a different geometry, with the activity distributed only on the balloon surface. The RIL was calculated to be 31.6 mm (i.e., a dose fall-off margin of 4.2 mm on each side). This margin was based on the 3-mm part coated with only 50% linear activity on each side of the balloon. Neglecting this inhomogeneity, the RIL would be 35.8 mm. Figure 6 shows the calculated longitudinal dose profiles for three different depths into the vessel wall. DISCUSSION The radial dose profile for the HDR 192Ir source showed the most flat dose gradient. At distances close to the source axis, inhomogeneities occurred owing to the spacing between dwell positions. Plato BPS version 13.7 uses a point source approximation. The impact on dose distribution at distances of ⬎4.0 mm was ⬍1%. At closer distances the difference increased to 7%. According to more detailed analyses, the error in assuming point sources is compen-
Fig. 5. Radial dose profile normalized to 2-mm reference depth for a 5.0-mm balloon coated with 32P. The radial distance r is related to the central axis of the balloon and the lumen. The values given from Yue et al. (22) are only for a rough comparison because the underlying geometry is not completely identical.
Volume 60, Number 5, 2004
Fig. 6. Longitudinal dose profiles related to the 32P source at the lumen surface, at 2-mm depth, and at 4-mm depth into the vessel wall for a 5-mm-diameter vessel (EGSnrc calculations). The 40-mm active balloon starts at z ⫽ 0 mm. The dose is normalized to the reference depth dose at radius plus 2 mm at the central plane. Owing to the steep radial dose gradient for this nuclide and specific source design, the relative dose is given in logarithmic scale. The step on both sides of the profile at the balloon (⫽ lumen) surface results from the proximal and distal part of the balloon coated with 50% of the activity.
sated by subsequent dwell positions, especially at distances that are typically used for dose prescribing (23, 24). We conclude that the use of anisotropy functions is essential for treatment planning in endovascular brachytherapy. In principle, our findings should be representative for 192Ir HDR sources in general. Absolute dose rates for Ir sources are calculated with treatment-planning systems, preferably using the AAPM TG43 concept, based on the air kerma strength (Gy m2 h⫺1) of the individual source and alreadywell-established and verified dose parameters (e.g., 15). The calculation, performed by a treatment-planning system based on the TG43 formalism, is related to full water equivalence of all materials involved. In the case of 192Ir, this can be assumed for the plastic parts of the catheters used. Therefore, the given results can be used for systems with or without centering catheter devices. The radial dose distribution for the 90Sr source device showed good agreement with the values published by Wang et al. (21). Small deviations might be due to a different density for the CO2 at 2 atm, which is higher by a factor of 2 in the cited publication. The absolute dose rate for any distance was calculated by multiplying the reference absorbed dose rate at the reference distance with the respective radial dose ratio, as shown in Fig. 3. Owing to the limited range of the  particles, which is much smaller than the active lengths of the investigated sources, dose rates normalized to the reference dose rate remain constant independently from the active source length. In contrast to relative dose values, the uncertainty in determination of the reference absorbed dose rate is large and needs further attention
Treatment parameters in peripheral EVBT
● C. KIRISITS et al.
1657
by primary standard laboratories. To maintain some consistency, the NIST calibration at 0.5 mm from the balloon surface (given on the calibration certificate) can be used by now. The calculated radial dose profiles of all three source types and 5-mm reference lumen diameter are compared in Fig. 7. The steepest dose gradient can be observed for the coated balloon; the most flat profile is present in the case of 192 Ir HDR treatment. All estimations are based on the water equivalence of vessel tissue. Calcification might further decrease the dose in cases of  radiation. To date, different dose prescriptions have been used in clinical trials of peripheral EVBT. The Vienna 1 and 2 and the Bern trials are prescribing a dose of 12 Gy to a fixed distance from the source axis: 3 mm for Vienna 1 and 2 (4 – 6), 5 mm for Bern (9). The PARIS (3), Swiss (10), and Vienna 4 and 5 (7) trials have prescribed a dose of 14 Gy to the reference depth of 2 mm into the vessel wall. In the Vienna 3 trial, the dose at the same depth was increased to 18 Gy (8). In the RAPID trial, using the RDX 32P-coated balloon, a prescription of 20 Gy at 1-mm tissue depth was used, not changing the protocol from the coronary trials using the same system (25). The MOBILE trial, using the 90 Sr source train, is also based on the experiences with coronary treatments. In the STents And Radiation Therapy (START) trial for coronary EVBT, the dose was prescribed at 2 mm from the source axis (26). This would correspond to a depth of 0.5 mm into the vessel wall for a typical coronary artery diameter of 3.0 mm. The dose was based on the actual diameter of the treated vessel, which was 18.4 Gy in the case of 3.0-mm diameter. The same depth and same dose of 18.4 Gy was chosen for treatment of peripheral arteries. The recently started Limb Ischemia treatment and Monitoring post Vascular Brachytherapy to prevent Reste-
nosis (LIMBER) feasibility trial is using a prescription of 14 Gy at 2-mm reference depth. Figure 8 compares the dose distribution for all these given examples. Independent from dose prescription, the EVA GEC ESTRO working group recommends reporting the RDD at the RDD point at 2 mm into the vessel wall and the reference lumen dose (RLDo) at the RLDo point (14). The dose should be reported for the centered situation, as illustrated in Table 1. A reference distance of radius plus 2 mm for peripheral treatments is also recommended by DGMP and AAPM (12–13). This is different from coronary applications, in which DGMP and ESTRO recommend a 1-mm reference depth and AAPM a fixed distance of 2 mm from the source center. In the EVA GEC ESTRO recommendations, this is based on the assumption of clinical relevance for target dose assessment. However, the treated depth, defined as the depth enclosed by the minimum target dose, might differ from the reference depth. The adventitia and peri-adventitia supposed to be the target is located at approximately a 2-mm depth. Intravascular ultrasound analysis resulted in a distance from balloon surface to adventita of 1.5 mm (27). Adding 0.5 mm thickness for the adventita would confirm a distance of 2 mm for the reference depth. If no centering device is used, the minimum and maximum values for both RDD and RLDo should be given in addition (28). The presented dose profiles are of interest related to upper dose constraints for the RLDo. The steep dose gradient of  sources is limiting a dose escalation to an upper limit for the RLDo. To date, no clear data are available on the maximum dose tolerated by the endothelial layer of a peripheral vessel lumen. Based on experiences in coronary arteries, a maximum dose of ⬎100 Gy was feasibly safe. However, one has to distinguish between maximum dose of small volume in the
Fig. 7. Comparison of the radial dose profiles for three different devices treating a 5.0-mm vessel with the reference depth dose normalized to 1. For better readability, the comparable high dose on the surface of the 32P-coated balloon is not shown in this diagram (compare with Fig. 5). HDR ⫽ high-dose rate.
Fig. 8. Summary of the dose distribution within the vessel wall for several trials of endovascular brachytherapy. All values are based on prescriptions for a 5.0-mm-diameter vessel. Solid lines are used for trials using 192Ir, dashed lines for 90Sr, and the dot-dashed line for 32P.
1658
I. J. Radiation Oncology
● Biology ● Physics
Volume 60, Number 5, 2004
Table 1. Summary of dose parameters for clinical trials of endovascular brachytherapy Trial Frankfurt Vienna 1, 2 Vienna 3 PARIS, Swiss Vienna 4, 5 Bern MOBILE LIMBER RAPID
Prescribed dose (Gy)
At distance
RDD (Gy) (2-mm depth)
At 1-mm depth (Gy)
At 0.5-mm depth (Gy)
RLDo (Gy) (0-mm depth)
7.9
10.2
12.0
14.4
12
3 mm
18 14
RLDi ⫹ 2 mm RLDi ⫹ 2 mm
18.0 14.0
23.2 18.0
27.2 21.1
32.7 25.4
12 18.4 14 20
5 mm RLDi ⫹ 0.5 mm RLDi ⫹ 2 mm RLDi ⫹ 1 mm
13.4 7.0 14.0 7.0
17.3 13.4 26.8 20.0
20.2 18.4 36.8 35.7
24.4 24.0 48.1 ⬃ 113
Prescribed doses are related to different distances. In most trials, this distance is based on the reference lumen diameter (RLDi) of the treated artery. The reference depth dose (RDD) is reported at a reference depth of 2 mm into the vessel wall; the reference lumen dose (RLDo) gives the value at the lumen surface. In addition, dose values are given at a depth of 1 mm and 0.5 mm into the vessel wall. The comparison is based on an RLDi of 5.0 mm.
case of a noncentered device or a uniform high dose on the whole luminal surface in the case of a centered device and high RLDo. Based on the experience from recent trials, a dose of at least 14 Gy should be given at 2 mm into the vessel wall. For the 192Ir and the 90Sr device, the dose range at the luminal surface would then be ⬍50 Gy. In the case of the 32P balloon, such a dose prescription exceeds 100 Gy to the vessel lumen. The RIL for a given ASL is, in general, lower for ␥ sources compared with  sources, with their steep dose gradient (20). If sources are used with stepping source technology and increased dwell times at the edges, the RIL can be adjusted to a value close to ASL. The optimization method used in this work is one example of available techniques. Dwell time optimization can also be done by several other geometric algorithms (29). For the procedure in the Vienna trials, using the described dose point optimization, the RIL was at least ASL minus 5 mm on both edges. A rounded value for the RIL of the 90Sr source device is in total 8 mm smaller than the ASL, 4 mm on each source
edge (RIL ⫽ ASL ⫺ 8 mm). This was calculated for a 60-mm ASL, but can be applied for multisegmental treatment. The limited range of the emitted  particles does not change the longitudinal dose profile on the source edge by increasing the ASL. If following the protocol recommended by the vendor, a three-segment treatment results in an ASL of 3 ⫻ 60 mm minus 2 ⫻ 2.5 mm (overlap of one seed for each new position), which is 175 mm. The resulting RIL would then be 167 mm. For the radioactive coated 32P balloon, a good approximation for the RIL is the ASL minus 4.5 mm on each side of the ASL (RIL ⫽ ASL ⫺ 9 mm). This difference between RIL and ASL is not based on the nuclide or the balloon geometry alone but on the inhomogenous activity distribution at the edges. Adequate treatment planning has to be based on these parameters to avoid geographic miss of the interventional length (30). Safety margins have to be fulfilled on the edges, and large dose inhomogeneities have to be avoided during multisegmental treatment (31).
REFERENCES 1. Böttcher HD, Schopohl B, Liermann D, et al. Endovascular irradiation—A new method to avoid recurrent stenosis after stent implantation in peripheral arteries: Technique and preliminary results. Int J Radiat Oncol Biol Phys 1994;29:183–186. 2. Tripuraneni P, Giap H, Jani S. Endovascular brachytherapy for peripheral vascular disease. Semin Radiat Oncol 1999;9: 190 –202. 3. Waksman R, Laird JR, Jurkovitz CT, et al. Intravascular radiation therapy after balloon angioplasty of narrowed femoropopliteal arteries to prevent restenosis: Results of the PARIS feasibility clinical trial. J Vasc Interv Radiol 2001;12: 915–921. 4. Minar E, Pokrajac B, Ahmadi R, et al. Brachytherapy for prophylaxis of restenosis after long-segment femoropopliteal angioplasty: Pilot study. Radiology 1998;208:173–179. 5. Pokrajac B, Pötter R, Maca T, et al. Intraarterial 192Ir HDR brachytherapy for prophylaxis of restenosis after femoropopliteal percutaneous transluminal angioplasty: The prospective randomized Vienna-2-trial radiotherapy parame-
6.
7.
8.
9.
10.
ters and risk factors analysis. Int J Radiat Oncol Biol Phys 2000;48:923–931. Minar E, Pokrajac B, Maca T, et al. Endovascular brachytherapy for prophylaxis of restenosis after femoropopliteal angioplasty: Results of a prospective, randomized study. Circulation 2000;102:2694 –2699. Wolfram R, Pokrajac B, Ahmadi R, et al. Endovascular brachytherapy for prophylaxis against restenosis after longsegment femoropopliteal stenting: Initial results. Radiology 2001;220:724 –729. Pokrajac B, Schmid R, Pötter R. Endovascular brachytherapy prevents restenosis after femoropopliteal angioplasty: Results of the Vienna-3 multicenter study [Abstract]. Int J Radiat Oncol Biol Phys 2003;57(Suppl.):S250. Zehnder D, von Briel C, Baumgartner I, et al. Endovasculer brachytherapy after percutaneous transluminal angioplasty of recurrent femorpopliteal obstructions. J Endovasc Ther 2003; 10:304 –311. Greiner RH, Mahler F, Jäger K, et al. Swiss trials of peripheral
Treatment parameters in peripheral EVBT
11. 12. 13. 14.
15. 16.
17. 18. 19.
20.
21.
endovascular brachytherapy for restenosis prevention after PTA [abstract]. Radiother Oncol 2003;66(Suppl. 1):S6. Nath R, Amols H, Coffey C, et al. Intravascular brachytherapy physics: Report of the AAPM Radiation Therapy Committee Task Group No. 60. Med Phys 1999;26:119 –152. Quast U, Kaulich TW, Flühs D. Guideline for medical physical aspects of intravascular brachytherapy. Part I: Guideline. Z Med Phys 2002;12:47– 64. Quast U, Kaulich TW, Flühs D. Guideline for medical physical aspects of intravascular brachytherapy. Part II: Samples and examples. Z Med Phys 2002;12:133–148. Pötter R, Van Limbergen E, Dries W, et al. Recommendations of the EndoVAscular GEC ESTRO Working Group (EVA GEC ESTRO): Prescribing, recording, and reporting in endovascular brachytherapy. Quality assurance, equipment, personnel and education. Radiother Oncol 2001;59:339 –360. Williamson J, Li Z. Monte Carlo aided dosimetry of the microSelectron pulsed and high dose-rate 192Ir sources. Med Phys 1995;22:809 – 819. Kawrakow I, Rogers DWO. The EGSnrc code system: Monte Carlo simulation of electron and photon transport. Technical report, NRC report PIRS 701. Ottawa, Ontario, Canada: National Research Council of Canada, 2000. Chu SYF, Ekstroem LP, Firestone RB. WWW table of radioactive isotopes, database version 1999-02-28. Available at: http://nucleardata.nuclear.lu.se/nucleardata/toi. Cross WG, Ing H, Freedman N. A short atlas of beta-ray spectra. Phys Med Biol 1983;28:1251–1260. Niroomand-Rad A, Blackwell CR, Coursey BM, et al. Radiochromic film dosimetry: Recommendations of AAPM Radiation Therapy Committee Task Group 55. Med Phys 1998;25: 2093–2115. Kirisits C, Georg D, Wexberg P, et al. Determination and application of the reference isodose length (RIL) for commercial endovascular brachytherapy devices. Radiother Oncol 2002;64:309 –315. Wang R, Li AX, Lobdell J. Monte Carlo dose characterization
22.
23. 24. 25. 26.
27.
28.
29. 30.
31.
● C. KIRISITS et al.
1659
of a new 90Sr/90Y source with balloon for intravascular brachytherapy. Med Phys 2003;30:27–33. Yue N, Nath R, Hehrlein C. Dosimetry calculation for a novel phosphorus-32 impregnated balloon angioplasty catheter for intravascular brachytherapy. Cardiovasc Radiat Med 1999;1: 349 –357. Dries W. Monte Carlo calculated dose distribution for endovascular HDR brachytherapy with Ir-192. Radiother Oncol 1997;45:77– 82. Wong D, Wallace S, Fernando W, et al. Dose errors in the near field of an HDR brachytherapy stepping source. Phys Med Biol 1999;44:357–363. Stone GW, Mehran R, Midei M, et al. Usefulness of beta radiation for de novo and in-stent restenotic lesions in saphenous vein grafts. Am J Cardiol 2003;92:312–314. Popma JJ, Suntharalingam M, Lansky AJ, et al. Randomized trial of 90Sr/Y90 beta-radiation versus placebo control for treatment of in-stent restenosis. Circulation 2003;106:1090 – 1096. Russo R, Apostol J, Silva P, et al. Is intravascular ultrasound necessary for dose prescription during femoral artery brachytherapy? Observations from the Paris trial [Abstract]. Circulation 2000;102(Suppl. II):2065. Pokrajac B, Schmid R, Kirisits C, et al. Possible impact of iridium-192 source centering on restenosis rate after femoropopliteal angioplasty and endovascular brachytherapy in Vienna-2 study. Radiother Oncol 2002;63:97–102. Bratengeier K. A fast algorithm for homogeneous Ir-192 afterloading-irradiation of cylindrical surfaces. Z Med Phys 2001;11:102–107. Schmid R, Kirisits C, Syeda B, et al. Evaluation of geographic miss in intracoronary brachytherapy. Recommendations for determining the planning target length to avoid geographic miss. Radiother Oncol 2004;71:311–318. Coen VLMA, Marijnissen JPA, Ligthart JMR, et al. Inaccuracy in manual multisegmental irradiation in coronary arteries. Radiother Oncol 2002;63:89 –95.