Can Technological Improvements Reduce the Cost of Proton Radiation Therapy?

Can Technological Improvements Reduce the Cost of Proton Radiation Therapy? Jacobus Maarten Schippers, PhD,⁎ Anthony Lomax, PhD,⁎ Adriano Garonna, PhD,† and Katia Parodi, PhD‡ In recent years there has been increasing interest in the more extensive application of proton therapy in a clinical and preferably hospital-based environment. However, broader adoption of proton therapy has been hindered by the costs of treatment, which are still much higher than those in advanced photon therapy. This article presents an overview of on-going technical developments, which have a reduction of the capital investment or operational costs either as a major goal or as a potential outcome. Developments in instrumentation for proton therapy, such as gantries and accelerators, as well as facility layout and efficiency in treatment logistics will be discussed in this context. Some of these developments are indeed expected to reduce the costs. The examples will show, however, that a dramatic cost reduction of proton therapy is not expected in the near future. Although current developments will certainly contribute to a gradual decrease of the treatment costs in the coming years, many steps will still have to be made to achieve a much lower cost per treatment. Semin Radiat Oncol 28:150-159 C 2018 Published by Elsevier Inc.

Introduction

I

n the initial phase of proton therapy, accelerator laboratories were the only places where it could be performed. In 1991, 41 years after the first treatment with a proton beam, the first hospital-based particle therapy facility1 began operation. The facility was based on a synchrotron specially developed for proton therapy, together with the first rotating gantries. Since 2000, the number of clinical facilities in a hospital-based environment has increased steadily and, because of increasing interest by commercial companies during the last 10 years, more than 60 facilities are now in operation.2 In addition, a clear shift out of the accelerator laboratories has taken place: currently more than 50. dedicated clinical facilities are linked to or are located at a hospital. However, the major obstacle to broader adoption of proton therapy is the cost of treatment, which is still approximately a factor of 2-3 higher than the cost of technically advanced radiation therapy with photons. The major factor determining the cost differential is the capital ⁎

Paul Scherrer Institut, Villigen, Switzerland. TERA Foundation, CERN, 1211 Genève, Switzerland. ‡ Ludwig-Maximilians-Universität München, Munich, Germany. Address reprint requests to J.M. Schippers, Paul Scherrer Institut, WBGA/C36, 5232 Villigen, Switzerland. E-mail: [email protected] †

150

https://doi.org/10.1016/j.semradonc.2017.11.007 1053-4296/& 2018 Published by Elsevier Inc.

investment needed for the equipment and the building. Other important contributions to the costs are the operation, the more intensive quality assurance procedures required for highly accurate proton dose delivery, and the less efficient patient throughput. Therefore, apart from a reduction in the investment costs, treatment costs can also be reduced by an efficiency increase in the treatment workflow. In this overview, current technological developments in these fields will be discussed in the context of their cost reduction potential. In relation to current technological developments, the assumption is that smaller equipment will lead to smaller and thus less costly facilities. In parallel with this facility size reduction, novel technology is also aiming at further improving treatment quality.3,4 Although the latter developments do not necessarily aim for a lower equipment cost, cost reduction could be a convenient spin-off. The ultimate goal of facility size reduction is to reach a footprint similar to what is currently needed for photon treatment techniques. The research on equipment size reduction has led to 2 developments which are currently being implemented. In the first group of developments, the price reduction is achieved just by cheaper accelerators and beam delivery systems for the typical multiroom facility setups. Although the initial investment will still be quite large, the cost per treatment is aimed to be lower than achievable in older facilities. The second group of developments is aiming at

Technologies to Reduce Rroton-Therapy Costs

151

systems that are attractive to institutions seeking single-room solutions. Our overview will concentrate on the technologies that might play a role in a cost reduction of the facility as well as on methods to improve the efficiency of the patient handling and throughput. The stated quantities and dimensions in this article are just approximations to indicate the order of magnitude and should thus not be used explicitly. Developments in gantries and in accelerators will be discussed first, followed by possible improvements in patient handling and facility layout.

Beam Delivery Systems in a Gantry

SAD ~2 m

gantry radius ~5 m

Current developments in gantry design aim at a combination of size reduction, a simpler mechanical design and, in some projects, novel beam transport methods applying new technology. The application of imaging techniques in relation to the gantries is discussed in section “Technology to Improve the Efficiency of Patient Throughput.” Although the layout has developed considerably, the basic concept and specifications of the gantries have not changed much since the first gantries for proton therapy,1 in which the methods to achieve the most important characteristics can already be recognized. These first gantries were of the “corkscrew design,”5 which is characterized by a bending of the beam, which is performed in 2 perpendicular planes after each other (eg, 90º upwards followed by 270º sideways). This has resulted in a gantry radius of 6 m with a length of only 7 m. Most other gantries6 have a magnet layout that bends the beam within a single so-called bending plane (eg, 45º up followed by 135º down) toward the patient. Compared to gantries for photon treatment, proton gantries are very large. The main reason for this is the mass of the protons, which is over 1800 times the mass of the electrons, which are the particles used to produce clinical photon beams. Therefore, proton tracks will have a bending radius of ~1.5 m, which requires large magnets. This and the other most important parameters determining gantry size are shown schematically in Figure 1. The magnet layout is determined by the beam dynamics constraints to achieve the correct pencil beam characteristics at the patient (only PBS—pencil beam scanning—systems will be discussed here). One of these is the

gantry rotaon axis

coils + scan magnet ~1 m

Scan x Scan y

nozzle equipment

isocenter

gantry length ~10 m

Figure 1 The most important parameters determining the total size of a gantry for proton therapy. (Color version of figure is available online.)

so-called achromaticity. In an achromatic gantry, protons within a range of 1%-2% energy spread (typical for most cyclotron facilities), will be guided through the gantry without losses and without affecting the pencil beam size and position at isocenter. Usually the magnet layout is such that the beam is initially bent by 45°-60°, followed by bending 135°-120° in the opposite direction toward the isocenter. The latter bigger bending apparatus can consist of a set of magnets. The gantries can be made achromatic by using appropriate distances and the addition of several quadrupole magnets. These and additional beam dynamics requirements strongly limit the possibilities to make a more compact gantry design. Present proton gantry designs typically have a length of 8-10 m and a radius of 4-5 m. As shown in Figure 1, in most gantries this radius is the sum of: • 1.5 m for the bending radius of the proton beam, • 0.5-1 m for the magnet widths of the bending magnets, • 1 m for the space required for magnet coils and the scanning magnets, and • 1.5-2.5 m for the source-to-axis-distance (SAD), the virtual source (location of scanning magnets) to axis distance, which is an important parameter to determine the maximum field size. The typical nozzle equipment such as monitors etc., is located downstream of the scanning magnets, and thus within the SAD. As can be seen in Figure 1, the most important contribution to the gantry radius is determined by the required SAD. A reduction of the SAD will lead to either complications due to the strong inclination of the pencil beams near the lateral edges of the field, or to a smaller field size to prevent this effect. However, one can also locate the scanning magnets before the last bending magnet (eg, at the Paul Scherrer Institute [PSI], Heidelberg Ion Therapy [HIT] facility, and the National Institute of Radiological Science [NIRS]),7-11 or in between the last bending magnets, like in, for example, IBA’s ProteusONE gantry.12 In these “upstream scanning” layouts (Fig. 2), the space between the last bending magnet and the patient can be reduced significantly, since it is only required to accommodate nozzle equipment. In the PSI, HIT, and NIRS gantries, the last bending magnet has been designed such that the scanning is performed as a parallel displacement of the pencil beam in the transverse plane at the isocenter. In this way, a very large or infinite SAD has been achieved with a gantry radius in the order of only 3.5 m.8 A gantry radius of only 2 m has been achieved by also including an off-centered table, which is counter rotating with the gantry.7 Due to the required beam optics for the parallel scanning, these upstream scanning gantries are longer than the gantries with down-stream scanning. It is important to realize, however, that parallel scanning has some advantages in field patching, treatment planning, and dosimetry-related quality assurance procedures. But, given that most of these types of gantries have been developed to test new scanning concepts, and that there is no experience yet with commercial versions of this gantry type, it is not yet clear if these advantages will have a very big effect on treatment cost reduction.

J.M. Schippers et al.

152 Down-stream scanning: + small aperture, - large radius because of SAD - small spots difficult X Y

Rgantry X Y SAD ~2 m

Up-stream scanning: + parallel + small gantry radius + small spots possible - wide aperture SC magnet

Rgantry

Figure 2 Conceptual difference between an “upstream” and a “downstream” scanning gantry. (Color version of figure is available online.)

The recently introduced use of superconducting (SC) magnets in gantries (Fig. 3), is expected to enable several types of improvements or cost reductions.11,13,14 First of all, the mass of the bending magnets will be strongly reduced. This will also allow a much lighter and simpler mechanical construction, so that the total weight of the gantry can be reduced by a factor 2-4.15,16 In general, the acquisition costs of SC magnets are at least as high as the normal-conducting room temperature magnets, but due to the lack of resistivity, their operating costs will be much lower. The main technical advantage of SC magnets is their possibility to make stronger magnetic fields, which will reduce the bending radius. However, since this parameter does not contribute very much to the radius of a proton gantry, the gantry radius will only be reduced by 1 m, or so. However, since the bending magnets will also be shorter, the gantry length can be reduced to 5-6 m. So just a simple replacement of warm magnets by SC magnets will not dramatically affect the proton-therapy gantry dimensions, but of course, its mass will be reduced by around a factor 4. In the case of gantries for carbon-ion therapy, both the reductions in size and in mass can be more significant.11 In the carbon-ion gantry at NIRS the SC magnet technology has enabled magnets of relatively low weight with a large aperture, so that “upstream scanning” can also be performed with SC magnets. This first SC gantry in particle therapy started treatments in May 2017.

Other interesting recent studies have investigated the possibility of using SC magnets in which different magnetic field types (for bending and for focusing) have been combined within a single SC magnet. Several of such gantry designs are being proposed and they can be designed such that they can accept a very broad spectrum (up to ± 25%) of proton energies without changing magnet settings.14,17,18 This extreme achromaticity will help to shorten the treatment time and would enable ultra-fast 3D scanning, since all proton energies necessary in a certain field can be sent through the gantry without an adjustment of the magnetic field. The more relaxed ramping schedule of such a magnet will also require less cooling capacity, which limits the costs. In addition, when mounting a degrader on such a gantry (Fig. 4), one could refrain from the usual degrader and energy selection system.17,19 This would enable a dramatic reduction in size of the whole system. Currently ongoing studies are still at an early stage, and it is expected that it will take several years until these gantries will be ready for introduction into the clinic. In summary, most studies on gantries with SC magnets are aiming at a reduction of the gantry size and especially the weight. In the coming years, the most likely developments in gantries with SC as well as with non-SC magnets will be focused on a footprint reduction by a combination of sections in the beam transport or by limiting the gantry rotation to 180°. It is expected that these footprint and weight reductions will reduce the costs. Other conceptual gantry designs have been proposed, which are based on peculiar characteristics of the beam production, such as for example those gantries on which a proton linear accelerator has been mounted, as will be discussed in section “Linear Accelerator.” As will also be discussed later are systems based on high-power lasers. Such a design can be based on a light isocentric (360° rotation) compact gantry of only 2 m radius and 3.5 m length.20 In parallel, several developments in beam delivery technique are in progress. These are dealing with optimized or variable pencil beam size, scanning speed, speed of energy modulation, continuous scanning, adjustable dose rate (ie, beam intensity), and mitigation of organ motion. Typical technologies in these fields are focused on beam dynamics, beam monitoring and control, (SC) magnet and power supply design, and organ position monitoring or imaging. These developments will not dramatically reduce the investment costs, but may reduce the operational costs due to an increase of efficiency.

Proton Accelerators

Figure 3 The first 2 SC gantries. Left: the gantry for carbon-ion therapy.11 Right: the gantry for proton therapy.15 (Color version of figure is available online.)

Proton accelerators have their origin in physics laboratories. Their first commercial application was to provide beams for isotope production. This stimulated the development and production of dedicated accelerators, which were mostly cyclotrons. In order to keep the cost low and to minimize the need for service, the machines have been made as simple as possible, but optimized for their application. When proton therapy became of commercial interest for companies after

Technologies to Reduce Rroton-Therapy Costs

153

Superconducng achromac bending secon Collimators

Scanning magnets

Degrader

Rotaon axis

Isocenter

Figure 4 PSI design of a gantry with SC magnets in the last bending section, which is preceded by a degrader and a collimator system.17,19 This bending section has been designed to have an energy acceptance that is large enough to accept all beam energies (ie, degrader settings) that are applied in a given gantry orientation (field) with only 1 or 2 settings of the magnetic field. (Color version of figure is available online.)

1990, their knowledge and experience in simplifying a cyclotron was successfully applied for proton therapy. For now and the near future, the cyclotron, synchrotron, and synchrocyclotron will be the most commonly used accelerators in proton therapy. In the coming years, these machines will continue to be improved. The commercial suppliers of these machines put a strong emphasis on optimization in their production and installation processes as well on standardization and optimization for their service. As will be described here, all are aiming at price reduction by decreasing the size of the accelerator. In parallel in various laboratories, some accelerator developments are being extended to possible applications in proton therapy. In this context, we will discuss the proton linac, of which the first clinical device is being built, as well as more recent proposals such as the Fixed‐Field Alternating Gradient (FFAG) and laser accelerators, and how these new ideas and technologies could contribute to a cost reduction of proton therapy. More detailed descriptions of the accelerators discussed in this contribution are available.21,22

Synchrotron Proton-synchrotrons are composed of a preaccelerator, which injects the protons from a proton source into a ring of 4-8 bending magnets with quadrupole and sextupole magnets in between them. In the ring, an radiofrequency (RF) cavity is also mounted, in which an oscillating electric field is used for further acceleration of the protons. The RF-frequency is varied synchronously to the increasing revolution frequency of the protons, and the magnets increase their strengths synchronous to the acceleration process. The ring has a typical diameter of 6-8 m. The preaccelerator is a separate 6-10 m long chain of accelerators, usually an RFQ (Radio Frequency Quadrupole) accelerator, followed by a linac. The maximum number of protons that can be injected into the ring is limited, but can increase with the injection energy. For the application of 1 field at the patient, one typically needs 1-3 filling and acceleration sequences (“spills”). See example in Ref.23

Smaller and thus cheaper synchrotrons have been developed with diameters down to 5 m, see example in Ref.24 Also, the footprint of several machines has been reduced by mounting the injector within or on top of the synchrotron ring. Ring diameter reductions have primarily been achieved by using improved magnets, modification of the beam optics, and optimization of interfaces to the injector. Improvements in the injection chain can also reduce the power consumption of synchrotron systems. A recent prototype uses a new RFQ25 directly injecting into the synchrotron, thus reducing the length of the injection line and power consumption. This was also the strategy adopted in the compact (diameter of 5.5 m) synchrotron design “Radiance 330,”26 which has been installed at McLaren, Flint (MI), and at MGH, Boston (MA), where the number of synchrotron elements has been reduced to a minimum. Additional improvements in the power consumption and treatment time can be achieved via special field-regulation of synchrotron magnets. First implementations of this technique at the Heidelberg facility (HIT) showed a reduction of 30% in cycle time.27 Although this has the most significant effect in carbon-ion systems, it will also be relevant for proton machines. Another approach to decrease the “dead time” between the spills is to increase the ramping speed of the magnets, as was studied by Best Medical and Brookhaven National Laboratories.28 Despite the large progress in SC magnets for synchrotrons in high-energy physics, the application of SC magnets for medical synchrotrons has only recently started to be investigated. This is due to their typically smaller scale and necessarily faster ramping speeds. NIRS researchers are working on a new carbon-ion synchrotron dubbed “SuperMinimac,” with strong SC magnets and a diameter of only 7 m,29 instead of the typical 25 m diameter. It is expected, however, that for protonsynchrotrons such a significant reduction in size will still require much more time. In Europe and especially in Japan, important other developments are in progress in synchrotrons for therapy. In NIRS (Chiba, Japan), one has recently introduced the possibility to change the energy of the extracted beam within 0.1 second during extraction, so within a spill.30 Although it would reduce treatment times considerably, changing energy during extraction has not yet been achieved in a proton synchrotron. At HIT several improvements have focused on the stability of the extracted beam intensity,31 which is a crucial specification to perform continuous scanning. Recently, Danfysik has proposed a synchrotron mounted on a gantry. This would be interesting for a single-room facility using a synchrotron as an accelerator. At the moment, this idea is still in a very early stage of R&D. So, although several important improvements in facility operation are expected soon, no substantial facility size reductions are expected in the near future in facilities for proton therapy driven by a synchrotron. Cost reduction can certainly be achieved by developments in magnets, RF systems, injection and extraction schemes, a higher filling of the ring, and a faster and more accurate control of the beam parameters. However, this is a slow process of many small steps.

154 Fortunately, however, it is also ongoing continuous development, yielding a gradual cost reduction.

J.M. Schippers et al.

Ironless k250 Magnet Design (MIT) the US patent number: US8975836 B2

Cyclotron and Synchrocyclotron Cyclotrons for proton therapy are single magnet machines, with a typical diameter of 5 m and a weight of 200 tons. In between the magnet poles, an oscillating (RF) electric field is constructed between Dee-shaped electrodes. The protons are accelerated in a plane between the magnet poles. Due to the magnetic field and their increasing energy from the RF field, they follow a spirally shaped orbit of a few hundred turns to the outer radius of the magnet, from where they are extracted and sent into the beam line. Cyclotrons for proton therapy accelerate protons to a fixed energy between 230 and 250 MeV. Using a degrader (adjustable amount of material in the proton beam to slow down the protons) and an energy selection system (a set of two ~45° bending magnets and an energy selecting aperture in the beam), all necessary lower energies can be obtained. This system can even be made fast enough to enable sufficiently fast energy modulation. However, quite a lot of shielding is necessary around this system. To reduce the size of a cyclotron, a stronger magnetic field is needed. This is only possible by using a SC magnet. The first SC proton cyclotron for therapy32,33 has a diameter 3.5 m and a weight of 100 tons. Several of these cyclotrons are in operation or in construction, and quite a few facilities have indicated serious interest in this SC cyclotron. The next step in SC cyclotrons has also been driven by commercial companies. By increasing the magnetic field by a factor 2-4, cyclotrons of only 50-30 tons and a diameter 1-2 m have been created.34,35 As expected, this has led to a significant reduction in the price of a cyclotron. In one of the commercially available systems the cyclotron is mounted directly on a rotating gantry.35 In such a proton therapy facility, a very small footprint and volume have been achieved. It should be noted, however, that cyclotrons with such a strong magnetic field can only operate in a pulsed proton beam intensity mode (500 Hz-1 kHz) and are called “synchrocyclotrons.” Such a pulsed beam imposes only a minor limitation in the spot-scanning technique and no problem at all for the passive scattering technique. However, in contrast to the continuous beam intensity from the “isochronous cyclotrons” discussed before, the pulsed beam from the synchrocyclotron does not allow advanced dose application techniques such as continuous line scanning. In addition, limitations on dose rate and accuracy of the dose per pulse could result in longer treatment times. To prevent these limitations and to keep the higher flexibility of their beam handling system, several companies also offer a single-room facility with a compact arrangement of a gantry with the existing isochronous cyclotron, to provide a continuous beam intensity. In the field of SC cyclotrons, studies have also been started to design a cyclotron with a magnet that has no iron yoke.36 Normally the magnet of a cyclotron is made of a coil embedded in a heavy iron yoke. The role of the yoke is to shape the magnetic field in the cyclotron and to shield the environment of the cyclotron from its magnetic field. As

Figure 5 Design of an iron-free superconducting cyclotron , using a set of main coils and 6 field shaping coils above and below the acceleration plane for the magnetic field in which the protons are accelerated. The 2 sets of 3 shielding coils will compensate the magnetic field outside the cyclotron (US patent number: US8975836 B2).36 (Color version of figure is available online.)

shown in Fig. 5, in the design of a cyclotron without iron yoke, the shaping of the magnetic field in the cyclotron is done by SC coils around the acceleration volume in the cyclotron. The shielding is achieved with a system of SC coils surrounding the cyclotron. These coils provide a so-called active shielding and thus replace the shielding role of the iron yoke. Since an iron yoke contributes to at least 90% of the mass of a cyclotron, this will reduce the weight enormously. Another very interesting advantage of an “iron-free” cyclotron may be the possibility to change the energy of the protons directly coming out of the cyclotron within a minute or so. The associated fast changes of the magnetic field and the RF-frequency are not possible in a cyclotron with an iron yoke. It would of course be extremely interesting if a small isochronous cyclotron could be made in an iron-free version, providing a continuous beam intensity at the maximum energy needed in a field. Together with an appropriate gantry (such as the one designed at PSI17,19) this would enable many improvements in speed and dose accuracy, as well as in facility size. As mentioned before, such a gantry would not require the usual degrader and energy selection system. Therefore, the building costs are expected to be much lower due to a smaller layout and a reduced need for shielding. However, since these ideas are still at an early design stage, estimates on price and availability cannot be made yet. Currently, the recent big steps in cyclotron related developments, such as facility size reduction, SC cyclotrons, and single-room layouts, are entering into clinical facilities. These technical developments are enabling lower initial investments, but are not yet expected to decrease the cost per treatment significantly and the consequences of the associated disadvantages that have been mentioned are not yet clear. However, it is expected that this industry driven achievement in cyclotron-based facilities will continue to be promoted in the coming 5-10 years. In parallel, dramatic changes in cyclotrons are under investigation, which may lead to cost reductions in equipment, operation, and in building layout, although their potential is still uncertain both in technical outcome and in time and costs.

Technologies to Reduce Rroton-Therapy Costs

155

Linear Accelerator

Fixed-Field Alternating Gradient

The study of linacs for particle therapy has naturally come from its extensive use in photon therapy, where 3 GHz electron linacs are employed. In 1989, a first design of a 66 MeV proton linac for hospital-based therapy was proposed.37 With an acceleration gradient of 2.5 MeV/m, the linac would have been 24 m long. The length of the linac has been a major obstacle for its introduction into particle therapy. However, potential advantages such as the possibility to vary the proton energy within a few milliseconds by reducing the electric field amplitude or by changing the phase in the last acceleration unit in the linac, the reduced transverse emittances, which imply lighter and cheaper beamline magnets, and the low losses, which imply reduced shielding requirements and operational problems, have motivated further developments.38 Since then, optimization of the design of the acceleration cavities and the use of 3 GHz as RF-frequency have enabled stronger electric acceleration fields. This has increased the energy gain of the linacs to 15 MeV/m, and they operate at a repetition rate of 100-300 Hz.39-41 The shorter length has been exploited in the design of TUrning LInac for Proton (TULIP)42,43 a gantry on which a linac has been mounted (Fig. 6). Recent developments in RF technology for highenergy physics have allowed the investigation of higher RF frequencies in the C- and X- bands (even to 6-12 GHz).45 In several designs, preacceleration to 24-70 MeV is necessary and this could be performed by a cyclotron (the combination being called “cyclinac”). Moreover, the recently achieved high gradients allow the extension of the same concept to carbon ions, where linac systems feature a significant power consumption reduction compared to synchrotron systems.46 Currently, several clinical facilities have indicated their interest in such a system. Following extensive simulations and successful tests of critical components, a 230 MeV linac as a first working complete system, is presently being tested by a company spun off from CERN. Hence, this very interesting linac technology has left the laboratory phase and entered the commercial space. For proton therapy, some cost reductions could come from the lighter magnets (lower transverse beam emittance), the reduced shielding, and the active energy modulation (no degrader). It may, however, not result in a significant overall cost reduction.

In the last decade, the accelerator community has investigated whether a FFAG accelerator would be suitable for proton (or helium-ion or carbon-ion) therapy.47-51 The FFAG concept is based on a clever combination of a cyclotron and a synchrotron. As in a synchrotron ring of separated sector magnets, the RF-frequency in the acceleration cavities is varied to match the revolution frequency. Similar to a cyclotron, the magnetic fields in the ring do not change their strength in time, but they feature alternating focusing and defocusing gradients. The 2 versions of this accelerator, the so-called scaling FFAG and nonscaling FFAG, use a different type of beam optics. The nonscaling FFAGs are the smaller ones, but the magnet design and large gradients of opposing signs require a very complicated design process. In addition, the RF-generator and the cavities are complex, since very strong electric fields at varying frequencies are needed. This requires a relatively large amount of electric power. For its application in proton therapy, the large energy acceptance of FFAGs may be of advantage to obtain fast energy changes,52,53 and possibly achieve higher beam intensities than in synchrotrons and synchrocyclotrons. However, typically FFAGs are larger than a synchrotron and of considerable weight. The nonscaling FFAGs are smaller, but very complex to design, to build and to operate. Both types of FFAGs need a ~10 MeV injection system, example, a cyclotron.54 Therefore, due to its still relatively experimental character and the considerable associated costs, it is expected that the FFAG will not be applicable for a hospital-based proton therapy setting in the near future. However, further research and development on this interesting type of accelerator are encouraged, especially for helium- or carbon-ion acceleration. It is important to note, however, that even if the FFAG system itself does not become small or “cheap” in the future, a spin-off of these concepts may lead to interesting developments such as a gantry design for carbon-ion therapy.55

Figure 6 A proton linac mounted on a gantry. Preliminary design of the TULIP gantry based on the linac from Benedetti et al.44 (courtesy of P. Carrio Perez, TERA Foundation). TULIP, TUrning LInac for Proton therapy. (Color version of figure is available online.)

Laser-Plasma Accelerator Although all aforementioned accelerator concepts exploit traditional radio-frequency sources, laser-driven ion acceleration relies on ultra-short (few tens of fs) laser pulses of ultra-high (41018 W/cm2) intensity focused onto (ultra)thin (nm-μm) plastic or metal foils. The resulting laser-plasma interaction enables generation and acceleration of protons and heavier ions such as carbon (depending on the target choice) with unprecedented acceleration gradients in the order of MV/μm, that is, 6 orders of magnitude above the RF-field strengths in state-of-the-art accelerators.56,57 In the last years, different acceleration mechanisms including target normal sheath acceleration, radiation pressure acceleration, and breakout afterburner acceleration have been investigated and proton energies close to 100 MeV have been obtained predominantly in single shot experiments. More recently, controlled in-vitro radiobiological experiments have been performed at several infrastructures using lasers of lower power—hence proton energies of up to a few tens of MeV—but at enhanced

J.M. Schippers et al.

156 repetition rates, to support the transition towards biomedical application.58,59 Currently new generation Petawatt laser installations, combining both the required high intensity for reaching the ion energies in the range of therapeutic interest and repetition rates close to 1 Hz are being installed in a few places around the world, such as the ELIMAIA facility in Prague, Czech Republic, and the CALA center in Munich, Germany. Moreover, extensive research is ongoing to optimize the interaction targets for enhanced laser absorption and, by acceleration, energy transfer to ion kinetic energy. This has brought very promising results for specially designed microstructures and nanostructures as well as for multiple thin layers added to foil targets.60-62 Laser-accelerated protons feature beam properties that are unique and significantly different from those obtained with accelerators discussed before. In particular, one should mention the ultra-short (≤ns) bunch duration, high intensity (≥107/cm2/pulse), large angular divergence (≥2°), and a poly-energetic spectrum (up to 100% ΔE/E). Therefore, not only the acceleration process, but also the beam diagnostics, transport, and delivery, as well as treatment planning and monitoring require innovative approaches currently being addressed by many groups to permit eventual successful clinical translation.63 The potential reduction of high-power laser dimensions may bring a reduction of size and probably costs over the long term. Although cost and size considerations were among the initial rationale for proposing laser-based acceleration, many groups have recognized that the major potential of laser-driven acceleration lies in the unique properties of laser-driven ion bunches and unique laser-driven capability. This uniqueness can open new opportunities such as for example, ultra-fast delivery and enabling mixed-ion production from the same or multiple targets. Although clinical application is still remote, progress enabled in this new Petawatt era of laser sources will facilitate exploration of new frontiers of laser-plasma acceleration and identify the role of the laser-driver for biomedical applications relevant to ion beam therapy.

Technology to Improve the Efficiency of Patient Throughput An additional way to reduce overall costs of proton therapy is to improve patient throughput efficiency, such that any facility or treatment room can treat more patients per day. Given the relative immaturity of commercial proton therapy solutions, it is perhaps not surprising that most commercial developments have been in optimizing delivery quality and reliability rather than in workflow efficiency. However, considerable work is now being invested into improving proton therapy workflow, as will be outlined in the following sections.

The Move to Pencil Beam Scanning In the second decade of this millennium, there has been an inexorable move in particle therapy from passive scattering to PBS.64 PBS is an inherently automated delivery method which,

in contrast to passive scattering, does not necessarily require patient or field specific hardware such as compensators or collimators. Potentially, this means that a full treatment can be delivered without the need to enter the room between fields.

In-Room Imaging Daily imaging for patient setup and positioning is also an area that has made huge advances in conventional therapy in the last years. Indeed, image guidance has been a central element of particle therapy for almost its whole history, in the form of 2D, portal X-ray images for each delivered field. In order to improve workflow, however, this has to change, and more efficient imaging systems and methods need to be introduced for particle therapy. Developments in this area are 2-fold. A number of centers are now installing diagnostic computed tomography (CT) scanners directly in the treatment-room in order to provide daily, high-quality 3D imaging for patient setup.8 Alternatively, most proton therapy manufacturers are now providing conebeam-CT solutions on their gantries, moving the patient setup procedures much closer to those used in conventional therapy. A particularly interesting solution for this is provided by MedPhoton,65 who have developed a “ring” CT system that can be directly mounted on the treatment couch. Its compatibility with particle therapy is demonstrated by such a system that has already been installed at the Med-Austron heavy ion facility in Wiener-Neustadt (Austria).

Reduced Treatment Delivery Times PBS is, in many ways, a somewhat inefficient delivery technique. Firstly, treatment delivery is inherently sequential, with individual, and narrow, pencil beams being delivered one after the other. Secondly, it is a 3-dimensional process, which means that each field can consist of many thousands of such sequentially delivered pencil beams. Although delivery times of a minute or so for volumes of a liter or so are perfectly acceptable, when combined with motion mitigation techniques such as rescanning or gating, the total delivery time for a single fraction can reach tens of minutes per field.66 Delivery times are dependent on a number of factors, but mainly on the available beam intensity, scanning dead times lateral to the field direction, and the time to change between energy levels (the third dimension of scanning). All these parameters have room for improvement, and thus potential for reducing delivery times. One approach being developed at PSI67,68 is, within 1 scan line, to move away from the discrete spot-scanning mode to a continuous scanning, with the desired proton fluence being modulated through a combination of modulated beam intensity and scanning speed. In preliminary tests at PSI, this has been shown to reduce the delivery time for a field by up to a factor of 2 in comparison to conventional discrete scanning. In the future, it is hoped that delivery times can be further reduced by also moving to higher intensity levels, however, this will involve significant improvements in the monitoring systems and associated real-time processing of the control

Technologies to Reduce Rroton-Therapy Costs systems. Further reductions in delivery times could also be achieved through optimization of energy layer switching times. For a change of Bragg peak depth of 5 mm in water, typical energy change times are currently in the range of a second. However, some systems have demonstrated much reduced energy change times, if only with relatively restricted field sizes,8,69 achieving energy layer changes of 50-100 ms. For a field with 20 such changes, this alone corresponds to a reduction in treatment time of about 18 seconds. When combined with continuous delivery and higher beam intensities, it would seem therefore possible that considerable reductions in delivery times will be possible in the coming years.

Out-of-Room Patient Preparation and Imaging The modern paradigm for image-guided radiotherapy is the use of linac mounted imaging devices, a trend that began with mega-voltage portal imaging in the 1980s and 1990s and has culminated in cone-beam-CT and, more recently, the development of combined MRI-Linac machines.70 There are many advantages to in-room imaging. However, from the workflow point of view, it may not necessarily be the most efficient, as every imaging intervention, and the associated cost of assessing the images, applying a correction and reimaging, all cost time, and potentially block the use of a treatment room for its prime role—the therapeutic irradiation of patients. As such, some interest has been shown in the particle therapy community in the concept of out-of-room patient preparation and imaging. Such an approach for proton therapy has been described by Bolsi et al.71 In this concept, all patient preparation (positioning of the patient on the table, placement of fixation devices, etc.) and, in-some cases, pretreatment imaging are all performed in an area outside of the treatment room. With the potentially time-consuming preparation or imaging process taking place outside, more efficient use of the treatment room can be achieved. While 1 patient is being treated, the next patient is already being prepared (and imaged) outside, and can be brought in and irradiated immediately after the first patient has finished treatment. As described by Bolsi et al, such an approach has been used for many years on gantry 1 at PSI, and the idea has been considered (but not always adopted) in a number of other particle facilities. However, workflow simulation studies have shown that the advantage of such an approach is rather limited, and best for single, rather than multiple room systems.71,72

157 switch between treatment rooms is (at best) 10 seconds, and can be considerably longer. Although such times are small, the ability to quickly switch between rooms opens up new possibilities for beam sharing. For instance, when beam switching time is long, this inevitably leads to the scenario where beam is assigned to a particular treatment room for the whole fraction, blocking treatment in all other rooms of the facility for the duration of that treatment. On the other hand, with switching times of a few seconds or less, it is possible to switch the beam between rooms between fields of a treatment. This provides much more flexibility in beam distribution, and allows for a much more efficient use of beam. Taken to its logical conclusion, this leads also to the concept of “beam splitting,” whereby, if there is enough intensity available from the accelerator, the primary beam could be split into the different beam lines, thus, potentially providing beam at any time in any room, and completely removing any time losses owing to beam distribution.3

Facility Layout Single-room facilities are promoted by many companies, since their investment costs are within reach of many interested clinics and in some cases the small footprint allows the addition of proton therapy to an existing radiotherapy center or additional similar rooms can be added sequentially. Several types are being proposed at this moment. As mentioned before, in parallel to the small synchrocyclotron mounted on a gantry,34 several other companies are producing a (synchro) cyclotron and a separate gantry for single-room facilities. For instance in the ProteusONE system of IBA73 and in the ProBeam system of Varian.74 A footprint (but not a volume) reduction of a factor 2-3 can also be achieved by mounting the accelerator below or above the gantry with a vertical beam line in between.75 Of course, whether this makes sense will depend on the price of the land, the building environment, and the method of patient preparation. It is clear that the total investment for a single-room facility will be much lower than that for a multiroom facility and it will lower the financial threshold to start proton therapy at all. However, although the absolute investment to start proton therapy may be lower, the cost per treatment (room) will be higher in the case of single-room facilities. Therefore, singleroom facilities will offer opportunities in certain cases, but it is not clear whether single-room facilities will make proton therapy cheaper in general.

Fast Beam Switching The final workflow improvement strategy discussed here is fast switching of the particle beam between treatment rooms. By definition, this is an approach that only has a meaning for multiple treatment room facilities driven by a single accelerator. For such facilities, however, efficient beam sharing (or beam scheduling) is a nontrivial task, and becomes more and more of a limiting factor, the more treatment rooms there are attached to the single accelerator. For current facilities, the time to

Conclusions and Outlook: Moderate Steps The quest to reduce the price of proton therapy will likely continue until the goal of a price similar to photon treatment will have been reached. In this overview, the most well-known developments in technology have been discussed in the context of their potential cost reduction. Several options seem

158 to be possible. However, the history of developments in proton therapy clearly shows that all major steps in proton therapy, such as for example, gantries, dedicated accelerators, and the scanning technique, took approximately 10 years from first trials to introduction into the clinic. Therefore, since significant technical developments, such as those discussed here, are needed for a major cost reduction, it is expected that a dramatic cost reduction of proton therapy will not be reached in the near future. However, thanks to the many on-going developments, the costs of proton therapy will go down, but it will be a long process of many relatively small steps.

Acknowledgments We acknowledge the information we have received from several companies and institutes and Paul Bolton for the fruitful discussions and valuable suggestions to the manuscript.

References 1. Slater JM, Archambeau JO, Miller DW, et al: The proton treatment center at Loma Linda University Medical Center: Rationale for and description of its development. Int J Radiat Oncol Biol Phys 22:383-389, 1992 2. Website of the Particle Therapy Cooperative Group PTCOG. Available at: http://www.ptcog.ch/index.php/facilities-in-operation, April 2017 3. Schippers JM, Lomax A: Emerging technologies in protontherapy. Acta Oncol 50:838-850, 2011 4. Schippers JM: Beam delivery systems for particle radiation therapy: Current status and recent developments. Reviews of Accelerator Science and Technology 2:179-200, 2009 5. Koehler AM, Proceedings of 5th PTCOG Meeting: International Workshop on Biomedical Accelerators (Lawrence Berkeley Lab. Berkeley, CA), pp 147–158, 1987 6. Flanz JB, Large medical gantries, Proc. Part. Acc. Conf 2015, IEEE Xplore (1996) 2004–2008 https://doi.org/10.1109/PAC.1995.505434 7. Pedroni E, Bacher R, Blattmann H, et al: The 200 MeV proton therapy project at the Paul Scherrer Institute: Conceptual design and practical realization. Med Phys 22:37-53, 1995 8. Pedroni E, Bearpark R, Böhringer T, et al: The PSI gantry 2: A second generation proton scanning gantry. Z Med Phys 14:25-34, 2004 9. Becher W, Schardt D, et al: Magnetic scanning system for heavy ion therapy. Nucl Instr Meth A 330:296-305, 1993 10. Pavlovic M: Beam-optics study of the gantry beam delivery system for light-ion cancer therapy. Nucl Instr Meth A 399:439-454, 1997 11. Iwata Y, Noda T, Shirai T, et.al: Design of superconducting rotating-gantry for heavy-ion therapy, in Proceedings of IPAC2012, THPPR047, 2012 12. Pearson, de Wilde O, Doyen R, et al: Magnet developments and commissioning for the IBA compact gantry. iEEE Trans Appl Supercond 24(3):4401004, 2014. https://doi.org/10.1109/TASC.2013.2284719 13. Gerbershagen A, Calzolaio C, Meer D, et al: Advantages and challenges of superconducting magnets in particle therapy particle therapy. Supercond Sci Technol 29(15pp):083001, 2016. https://doi.org/10.1088/0953-2048/ 29/8/083001 14. Wan W, Brouwer L, Caspi S, et al: Alternating-gradient canted cosine theta superconducting magnets for future compact proton gantries. Phys Rev ST Accel Beams 18:103501, 2015 15. Derenchuk V, Particle beam technology and delivery, in 55th Annual AAPM Meeting, Proton Symposium, Indianapolis, IN, 2013 16. Holder DJ, Green AF, Owen HL, A compact superconducting 330 MeV proton gantry for radiotherapy and computed tomography, Proceedings of IPAC2014, WEPRO101 pp. 2201–2204, 2014 17. Alexander Gerbershagen, Jacobus M Schippers, David Meer, et al: Novel beam optics concepts in particle therapy gantries utilizing the advantages of superconducting magnets. Z Med Phys 26:224-237, 2016. https://doi. org/10.1016/j.zemedi.2016.03.006

J.M. Schippers et al. 18. Trbojevic D, Parker B, Keil E, et al: Carbon/proton therapy: A novel gantry design. Phys Rev ST Accel Beams 10:053503, 2007 19. Sanfilippo S, Calzolaio C, Anghel A, et al: Conceptual design of superconducting combined function magnets for the next generation of beam cancer therapy gantry, Proceedings of RuPAC2016, THCDMH0, pp. 138–140, 2016 20. Masood U, Bussmann M, Cowan TE, et al: A compact solution for ion beam therapy with laser accelerated protons. Appl Phys B 117(1):41-52, 2014 21. Humphries S, Principles of Charged Particle Acceleration, John Wiley and Sons (ISBN 0-471-87878-2, QC787.P3H86) in 1986 22. Schippers JM: Chapter 3, Beam delivery systems for particle therapy. In: Ma C, Lomax A (eds): Proton and Carbon Ion Therapy. Boca Raton, FL: CRC Press, Taylor & Francis Group, 2013 23. Hiramoto K, Hiramoto K, Umezawa M, et al: The synchrotron and its related technology for ion beam therapy. Nucl Instr Meth B 261:786-790, 2007 24. Umezawa, Ebina F, Fujii Y, et al: Development of Compact Proton Beam Therapy System for Moving Organs. Hitachi Review 64(8), 2015 25. Vretenar M, Dallocchio A, Dimov VA, et al: A Compact High-Frequency RFQ For Medical Applications, THPP040, p. 935-938, LINAC14 26. Wang F, Flanz J, Hamm RW, Injection study of the protom-radiance 330 synchrotron with a 1.6 MeV RFQ linac , The 19th Particles and Nuclei International Conference July 24th–29th, Cambridge, MA, 2011 27. Cee R, Feldmeier E, Galonska M, et al: The First Magnetic Field Control (B-train) to Optimize the Duty Cycle of a Synchrotron in Clinical Operation, THPPD002, pp. 3503–3505, IPAC2012 28. Trbojevic, Alessi J, Blaskiewicz M, et al: Lattice design of a rapid cycling medical synchrotron for carbon/proton therapy, Proceedings of IPAC2011, San Sebastián, Spain WEPS028 29. Iwata Y, et al: Nov 2016 Workshop at CIEMAT 30. Iwata Y, Furukawa T, Mizushima K et al: Multiple-energy operation with quasi-DC extension of flattops at HIMAC, IPAC2010, MOPEA008, 79-81 31. Schoemers C, Feldmeier E, Naumann J, et al: The intensity feedback system at Heidelberg Ion-Beam Therapy Centre,. Nucl Instr Meth A 795:92-99, 2015 32. M Schillo et al: Compact superconducting 250 MeV proton cyclotron for the PSI PROSCAN proton therapy project 33. Marti F (ed), Cyclotrons and their applications, 16th International Conference, pp. 37–39 22, 2001 34. Website of Mevion, Available at : www.mevion.com 35. Website of IBA: Available at: https://iba-worldwide.com/proton-therapy/ proton-therapy-solutions/proteus-one 36. Radovinsky A, Minervini JV, Miller CE, et al: Superconducting magnets for ultra light and magnetically shielded, compact cyclotrons for medical, scientific, and security applications, IEEE Trans Appl Supercond 24 (3):4402505, 2014 37. Lennox J, Hendrickson R, Swenson DA, et al: Proton linac for hospitalbased fast neutron therapy and radioisotope production Fermi National Accelerator Laboratory, TM-1622 (1989) 38. Amaldi U: High frequency linacs for hadron therapy. Reviews of Accelerator Science and Technology 2:111-131, 2009 39. Amaldi U, Berra P, Crandall K, et al: LIBO—a linac-booster for protontherapy: Construction and tests of a prototype. Nucl Instr Meth A 521:512-529, 2004 40. De Martinis C, Giove D, Amaldi U, et al: Acceleration tests of a 3 GHz proton linear accelerator (LIBO) for hadrontherapy. Nucl Instrum Meth A 681:10-15, 2012 41. Benedetti S, Amaldi U, Grudiev A, et al: RF Design of a Novel S-Band Backward Travelling Wave Linac For Proton Therapy, THPP061, pp. 992–994, LINAC2014 42. Amaldi U, Braccini S, Magrin G, et al: Ion acceleration system for medical and/or other applications, Patent number: US8405056, 2006 43. Degiovanni A, Amaldi U, Bergesio D, et al: Design of a fast-cycling highgradient rotating linac for protontherapy, Proceedings of IPAC2013, THPWA008, 3642–3644 44. Benedetti, et al: Phys Rev Accel Beams. (courtesy of P. Carrio Perez, TERA Foudation) 2017

Technologies to Reduce Rroton-Therapy Costs 45. Degiovanni A, Amaldi U: Proton and carbon linacs for hadron therapy, Proceedings of LINAC2014, ISBN 978-3-95450-142-7, Geneva, Switzerland FRIOB02, (2014) 1207–1212, 2014 46. Verdú-Andrés S, Amaldi U, Faus-Golfe Á: CABOTO, a high-gradient linac for hadrontherapy. J Radiat Res 54(suppl 1):i155-i161, 2013 47. Mori Y, Development of FFAG accelerators, Cyclotron Conference, 2004, Tokyo, Symon KR, et al. fixed field alternating gradient particle accelerators, Phys. Rev 103:1837–1859, 1956 48. Craddock MK, Symon KR: Cyclotrons and fixed-field alternating-gradient accelerators. Rev Acc Sci Technol 1:65-97, 2008 49. Trbojevic D: FFAGs as accelerators and beam delivery devices for ion cancer therapy. Rev Acc Sci Technol 2:229-251, 2009 50. Sheehy SL, High intensity and other world wide developments in FFAG accelerators, Proceedings of Cyclotrons 2016, THD01 (2016) 374–379 51. Ishi Y, Kuriyama Y, Mori Y, et al: KURRI FFAG′s future project as ADSR proton driver, THC01 Proceedings of Cyclotrons 2016, Zurich, Switzerland, THC01 366–370, 2016 52. Antoine S, Autin B, Beeckman W, et al: Principle design of a protontherapy, rapid-cycling, variable energy spiral FFAG. Nucl Instr Meth A 602:293-305, 2009 53. Verdú-Andrés S, Amaldi U, Faus-Golfe Á: Literature search on linacs and FFAGs for hadron therapy. Int J Mod Phys A 26:1659-1689, 2011 54. Keil E, Trbojevic D, Sessler AM, et al: Fixed Field Alternating Gradient Accelerators (FFAG) for Fast Hadron Cancer Therapy, PAC —Proceedings Knoxville, Tennessee, pp. 1667–1669, 2005 55. Trbojevic D, Gupta R, Parker B, et al: Superconducting non-scaling FFAG gantry for carbon/proton cancer therapy, Proceedings PAC07, 3199–3201 56. Borghesi M: Laser-driven ion acceleration: State of the art and emerging mechanisms. Nucl Instr Meth Phys Res A 740:6-9, 2014 57. Hegelich BM, Albright BJ, Cobble J, et al: Laser acceleration of quasimonoenergetic MeV ion beams. Nature 439:441-444, 2006 58. Bin J, Allinger K, Assmann W, et al: A laser-driven nanosecond proton source for radiobiological studies. Appl Phys Lett 101:243701, 2012 59. Zeil K, Baumann M, Beyreuther E, et al: Dose-controlled irradiation of cancer cells with laser-accelerated proton pulses. Appl Phys B 110: 437-444, 2013 60. Schwoerer H, Pfotenhauer S, Jäckel O, et al: Laser-plasma acceleration of quasi-monoenergetic protons from microstructured targets. Nature 439:445-448, 2006

159 61. Lübcke Andrea, Andreev Alexander A, Höhm Sandra, et al: Prospects of target nanostructuring for laser proton acceleration Article number: 44030. Sci Rep 7. https://doi.org/10.1038/srep44030 62. Bin, Ma WJ, Wang HY, et al: Ion acceleration using relativistic pulse shaping in near-critical-density plasmas. Phys Rev Lett. 115:064801, 2015 63. Hofmann KM, Masood U, Pawelke J, et al: A treatment planning study to assess the feasibility of laser-driven proton therapy using a compact gantry design. Med Phys 42(9):5120-5129, 2015 64. Jermann M, PTCOG 53, Shanghai, 2014 65. Website of Med-Photon, Available at: https://www.medphoton.at 66. Zhang Y, Huth I, Wegner M, et al: An evaluation of rescanning technique for liver tumor treatments using a commercial PBS proton therapy system. Radiother Oncol pii: S0167-8140(16)34333-X. https://doi.org/10.1016/j. radonc.2016.09.011 67. Klimpki G, Psoroulas S, Bula C, et al: A beam monitoring and validation system for continuous line scanning in proton therapy. Phys Med Biol. https://doi.org/10.1088/1361-6560/aa772e. Epub ahead of print 68. Inoue J: Line scanning apparatus utilizing irradiation position and stationary time for scanning 520 control United States Patent no. 8890090. SUMITOMO Heavy Industries, Ltd., Tokyo, Japan, 2014 69. http://www.mevion.com/products/mevion-s250i-hyperscan-adaptiveaperturetm 70. Lagendijk JJ, Raaymakers BW, van Vulpen M: The magnetic resonance imaging-linac system. Semin Radiat Oncol 24(3):207-209, 2014. https: //doi.org/10.1016/j.semradonc.2014.02.009. Review 71. Bolsi A, Lomax AJ, Pedroni E, et al: Experiences at the Paul Scherrer Institute with a remote patient positioning procedure for high-throughput proton radiation therapy. Int J Radiat Oncol Biol Phys 71:1581-1590, 2008 72. Fava G, Widesott L, Fellin F, et al: In gantry or remote patient positioning? Monte Carlo simulations for proton therapy centres of different sizes. Radiother Oncol 103:18-24, 2012 73. Website of IBA, Available at: https://iba-worldwide.com/proton-therapy/ proton-therapy-solutions/proteus-one 74. Website of Varian, Available at: https://www.varian.com/oncology/prod ucts/treatment-delivery/probeam-compact-proton-therapy-solution 75. Website of Sumitomo, Available at: http://www.shi.co.jp/quantum/eng/ product/proton/proton.html