To summarize the technical delivery parameters of proton plans delivered at the Mayo Clinic in Rochester, Minnesota.
The database of treated patient proton plans was queried to extract field parameters such as gantry angle, patient support angle, minimum and maximum water-equivalent depth (WED) treated, number of layers, field size, patient orientation, and monitor units. The plans were analyzed in aggregate, by disease site, and by fractionation.
There were 2963 proton plans for 2023 distinct treatment sites delivered between June 2015 and September 2018. The mean number of fields per plan was 2.8. The mean number of energy layers per field was 51.9. The mean monitor unit per field was 117.4. The median maximum field dimension was 12.4 cm; 95% of the fields had a maximum dimension < 28.7 cm, and the maximum field dimension was 39.8 cm. The median maximum field WED was 16.4 cm; 95% of the fields reached a maximum WED of ≤ 26.4 cm, and the maximum field WED was 32.4 cm.
A large variety of disease sites were treated using the maximum field size (40 cm) and WED (32.4 cm) capabilities of our half-gantry system.
Charged particle therapy is a growing field with > 30 centers open in the United States and 94 centers worldwide as of April 2019 . As facilities are designed, the specifications written in the request for proposals and further negotiated with vendors effectively lock in machine parameters that will have large effects on achievable plan quality. Once maximum field size, energy range, nozzle angles (full gantry, partial gantry, or fixed beam), and patient supports have been specified, modifications may be difficult and expensive, if even possible. However, there is difficulty in specifying the parameters for a charged-particle delivery system without having had experience treating with a similar system. We offer the analysis of proton plans delivered at Mayo Clinic in Rochester, Minnesota, since 2015, as an illustrative example of what a new practice might expect to treat. Our experience was shaped by our chosen technologies, our patient population, and our local planning standards of practice.
Materials and Methods
The Mayo Clinic is equipped with a Probeat-V system (Hitachi Ltd, Tokyo, Japan). A single synchrotron, capable of accelerating protons to 97 energies between 71.3 MeV and 228.8 MeV, provides scanned pencil-beam protons to 4 half-gantry treatment rooms and 1 fixed-beam room. The 4 identical gantries rotate 190° (5° past vertical) with a 0.1° resolution. The nozzle touch sensor is 36.7 cm from the isocenter, defining the effective radius. The patient support device is a couch top (Civco Medical Solutions, Coralville, Iowa) coupled to a robotic arm (Forte Automation Systems Inc, Machesney Park, Illinois) with 6° of freedom. Although greater rotation is possible with inferior isocenters close to the robotic arm mount, planning practice is to have couch angles within ± 90° of the nominal 270° position because of arm reach and typical isocenter location. The maximum field size is 30 cm × 40 cm. This field size is still achievable when a range shifter with a 25-mm water-equivalent depth (WED) or a 45-mm WED is inserted into the nozzle, referred to as RS25 and RS45, respectively. The downstream surface of either range shifter is 42.5 cm from the isocenter. An extended range shifter of 45 mm WED (ERS45) can be mounted off the nozzle to bring the downstream surface to 30 cm from the isocenter, but it reduces the field size to 25 cm × 30 cm.
The Department of Radiation Oncology at Mayo Clinic, Rochester, uses the ARIA oncology information system (Varian Medical System, Palo Alto, California) for comprehensive information and image management. For this technical note, the ARIA database was queried for details of plans delivered to patients since the first patient was treated in June 2015.
Treatment courses and dose point volumes (DPVs) were counted. The presence of multiple DPVs in a course indicated multiple treatment sites, and a course was counted again for every DPV. Course fractionation was determined per DPV dose recorded. For courses with x-ray and proton fractions, only the proton fractions were counted. If a course was completed early, the delivered number of fractions was tabulated, rather than the originally prescribed number of fractions. Field-based analysis treated the initial plans, the midtreatment replans, and boost plans on equal footing. Each planned field entered the analysis once, regardless of how many times it was delivered.
As the use of range shifting devices in the nozzle or next to the patient altered the energies needed to treat a target, we report the WED treated, instead of the energies used. For the fields without any range shifter, the minimum energy of 71.3 MeV has a range in water (R80 = depth to the distal 80% dose) of 4.0 cm, and the maximum energy of 228.8 MeV has R80 = 32.4 cm. For fields that use a 25-mm or 45-mm range shifter in the nozzle, the WED treated is straightforward to calculate. For fields that use a custom bolus device close to the patient, the WED shift assumes a normal beam incidence for all fields.
The field size is derived from the equation (xmax − xmin) × (ymax − ymin), where the position of every spot in a field is considered. This mimics isocenter placement in the center of each beam's eye view, a situation that is not always achievable (or desirable) with more than 2 beams. We often shift treatment isocenter to maintain clearance, so the reported field sizes are an underestimate of the actual rectangular field sizes 2|x|max × 2|y|max used for treatment. Because the x and y of each spot are in the nozzle coordinate system, their values depend on the couch and gantry angles. For example, an anterior field that is 5 × 10 cm2 would be 10 × 5 cm2 with a 90° couch rotation. As a simplification, we report the maximum and minimum dimensions of each field.
The gantry angle reported is between −5° and 185°. In this coordinate system, gantry 0° would provide an anterior-posterior (AP) field, and gantry 180° provides a posterior-anterior (PA) field. To collapse the patient support angle and the gantry angle into a single variable with a parallel to x-ray treatments, we introduce the angle from coplanar. This angle is akin to the couch “kick” that would be needed for an x-ray linear accelerator system to deliver an x-ray beam superior (+) or inferior (−) to an axial slice. For example, an AP or PA field would be 0° from coplanar, whereas a vertex field would be +90° from coplanar.
Because the fields treated are influenced by the referral patterns of our practice, plans were sorted by physician-entered diagnosis codes.
Monitor unit (MU) is defined so that 100 MU delivered to a 10 × 10 × 10 cm3 cube of water centered at a depth of 15 cm will result in a uniform dose of 1 Gy.
There were 2963 patient proton plans for 2023 distinct treatment sites delivered between June 2015 and September 2018 at the Mayo Clinic, Rochester, Minnesota. Midcourse replans and sequential boost plans were included in this total. There were 8300 fields, yielding an average 2.8 fields/plan. The average number of energy layers per field was 51.9. The average MU per field was 117.4. The median maximum field dimension was 12.4 cm; 95% of the fields had a maximum dimension < 28.7 cm, and the maximum field dimension was 39.8 cm. The median maximum field WED was 16.4 cm; 95% of the fields reached a maximum WED of ≤ 26.4 cm, and the maximum field WED was 32.4 cm. The Table shows the course and plan statistics separated by anatomic site treated and course fractionation.
Figure 1 shows the differential (left subfigure) and cumulative (right subfigure) distributions of gantry angles by anatomic site treated. The cumulative curve for breast fields reveals that most fields were anterior beams, with 80% of fields having a gantry angle ≤ 55°. Abdomen/thorax sites were mostly treated with posterior beams because 60% of fields had a gantry angle ≥ 130°. The steps in the pelvis curve reflect the heavy use of anterior obliques and laterals in prostate cancer plans.
Of the 2963 plans treated, 714 (24%) were delivered at a single couch angle. The remaining plans had ≥ 2 couch angles treated, from which, the difference between the extremes was calculated. There were 2189 plans (74%) with ≥ 90° couch rotation and 1563 plans (53%) with ≥ 180° couch rotation.
Figure 2 shows the WED of targets by displaying the maximum WED (corresponding to the highest energy) treated on the x-axis and the minimum WED treated on the y-axis. The high density of points with min(WED) = 0 cm in the breast figure indicates that most fields treated the skin surface. The pelvis figure shows a cluster of deep targets with max(WED) > 20 cm that are relatively small—max(WED) − min(WED) ≈ 10 cm—as would be expected for prostate cancer plans.
The percentage of fields using a nozzle range shifter was 9.9% RS25, 9.8% RS45, and 37.8% ERS45. The remaining 42.5% did not use a nozzle-based range shifter. Custom bolus was used for 581 fields (7.0%); 46 fields (0.5%) used both custom bolus and a nozzle range shifter.
Figure 3 displays the differential (left subfigure) and cumulative (right subfigure) distributions of the maximum WED by anatomic location treated. The rightmost position of the cumulative pelvis curve shows fields treating the pelvis must reach the largest WED, with 60% of fields treating deeper than 20 cm WED. In contrast, > 95% of fields treating breast locations had a max(WED) ≤ 20 cm.
Figure 4 shows the width and length of each field, with the largest dimension always on the x-axis of the figure. The clusters of points in the lower left of the central nervous system (CNS), breast, and pelvis figures indicate small fields that were almost square in the beam's eye view (max[dimension] = min[dimension], but conforming to the target). The breast figure has a second grouping of points in its upper right, indicating large, nearly square fields. The long, narrow, spinal craniospinal irradiation fields are grouped in the CNS figure in the lower right with max(dimension) ≈ 25 cm and min(dimension) < 10 cm.
Figure 5 contains the differential (left subfigure) and cumulative (right subfigure) distributions for the largest field dimension by anatomic location treated. Every group contains a field with max(dimension) > 30cm. If the field size had been limited to 20 cm, 60% of the breast fields and 10% of the prostate fields would have required multiple isocenters.
Of the 2963 plans treated, 2917 (98.4%) oriented the patient head-first supine, 39 (1.3%) were head-first prone, and 7 (0.2%) were feet-first supine.
Of the 8300 fields treated, 7095 (85.5%) were coplanar. For fields within 15° of coplanar, that fraction increases to 91.6% (n = 7605).
Less than 20% of treatment courses in aggregate had 5 or fewer fractions. However, when considered by year, that percentage increased from 3% in 2015 to 26% in 2018. The anatomic sites with the largest percentage of hypofractionated courses in 2018 were the pelvis (47%), breast (38%), and bone/miscellaneous (31%).
One of the first decisions a new proton or carbon ion particle facility makes is to choose full-gantry, half-gantry, and/or fixed-beam delivery systems. There are arguments to be made based on facility footprint, building costs, and technology costs. The facility footprint will be influenced by site constraints and predicted patient volumes. Simulations can be run to calculate the number of treatment rooms needed based on expected disease mix, workflow, and treatment times, with assumptions for maximum charge, dose rate, room switching time, and layer switching time. A proton center will also have to choose whether to use superconducting technology and whether the accelerator will be gantry mounted. Plans for integrated image guidance (in-room computed tomography [CT], cone beam CT, or magnetic resonance) will influence gantry design. Because the technology and vendor landscape is rapidly evolving and each center has its own goals and constraints, we cannot advocate a single solution for all. We can present considerations and our treatment data to inform institutions that are about to make a multimillion-dollar decision.
With our half-gantry and robotic couch, we have limited ability to tip beams inferiorly on head-first patients. The cumulative gantry angle distribution in Figure 1 shows that all gantry angles available in the half-gantry system were used. For a horizontally positioned patient, a few fixed-beam angles might be sufficient to treat a single site, such as the prostate, where our primary planning technique uses laterals or laterals plus anterior obliques. For most sites, our experience has been that changing the gantry angle by a few degrees can improve the quality of a treatment plan, perhaps by avoiding metal, changing the angle with respect to a material interface, or moving the region of biologic enhancement. A full-gantry system would have the same available gantry angles for coplanar treatments as our half-gantry system without requiring any couch rotation. The radius of a full gantry system will determine whether the same noncoplanar beams could be achieved. This data set had 8.4% (n = 695) of fields greater than 15° from coplanar, with 95% of those superiorly tipped fields treating targets in the head (groups CNS and head and neck). A small-radius gantry could limit the ability to deliver vertex fields by restricting couch rotation. However, a large-radius full-gantry or half-gantry system with different couches might be able to access more inferiorly tipped fields. Our center does not use a seated position for treatment, but the feasibility of treating with a fixed beam for that patient orientation has been discussed in another publication by Yan et al .
An additional consideration between full- and half-gantry systems is the need for large couch rotations and the speed at which those can be accomplished. A full-gantry system would not need a couch rotation to treat from the left and right. In our practice, we do not find ourselves overly limited by having to rotate the couch, and 53% of our plans (n = 1563) include couch positions separated by ≥ 180°. During the time a therapist is in the room rotating the couch and gantry, the beam is available for another treatment room to use.
Without beam modifiers, the Mayo Clinic proton beam can treat WED between 4.0 cm and 32.4 cm. With additions of 25 mm or 45 mm range shifters, the reachable WED is between 1.5 cm and 29.9 cm or between 0.0 cm and 27.9 cm, respectively. Figure 2 shows that there are fields in every category, except breast, that required 28 cm of dynamic WED selection. Considering maximum WED only, 95% of fields reached ≥ 26.4 cm WED. The Table and Figure 3 shows that breast, CNS, and head and neck targets have a higher fraction of shallow targets.
In the 57.5% of fields (n = 4775) that required a range shifter to treat targets shallower than 4.5 cm WED, the ERS45 was used most frequently. The extended range shifter being closer to the isocenter allows the spot size to remain smaller than it is for the in-nozzle RS45. The RS45 is used when either clearance or large field sizes prevent the use of the ERS45. The RS25 provides a spot size smaller than the RS45 does, but it can be used only when the min WED is ≥ 2.0 cm. Our custom bolus is positioned within 10 cm of the patient surface and has a minimal effect on spot size.
The number of discrete energies in a system is chosen to balance cost and dose homogeneity. Although a narrow Bragg peak with a small distal falloff may be desirable, the creation of a spread-out Bragg peak with a constant flat-top dose will then require closely spaced energy layers. Because the Bragg peaks of higher energies will be broader due to range straggling, higher energies can be spaced further apart. For systems that need to cycle a synchrotron after one or a few energy layers, even if sufficient charge remains to treat the next layer, the number of layers can significantly affect treatment times. The use of a miniridge filter  is another way to reduce the number of layers per plan.
The maximum field size of the Mayo Clinic nozzle is 30 cm × 40 cm, except with the use of the extended range shifter, which limits the field to 25 cm × 30 cm. Figure 5 and the Table illustrate that, although some anatomic locations have smaller average field sizes, every anatomic site used fields ≥ 30 cm in their largest dimension in the beam's-eye view. A practice could treat the same patient cohort with a smaller nozzle, but multiple isocenters would be needed, which would increase imaging and delivery times. Tied to treatment field size is imaging field size. Because all relevant bony anatomy should be visualized during setup, the imaging field of view should be approximately the same as, if not larger than, the treatment volume.
The growing percentage of hypofractionated courses in our practice has been driven by 2 factors. First, the opening of 3-fraction breast cancer and 5-fraction prostate cancer protocols has increased the percentage of hypofractionated courses in the breast and pelvis grouping. Secondly, the CT on rails was commissioned in the treatment rooms in 2017. The availability of volumetric matching and pretreatment visual confirmation of target WED have increased physician comfort with stereotactic body radiation therapy procedures with protons, and the procedure's frequency has increased, especially to bony sites.
ADDITIONAL INFORMATION AND DECLARATIONS
Conflicts of Interest: Anita Mahajan, MD, is an Associate Editor of the International Journal of Particle Therapy. The authors have no other relevant conflicts of interest to disclose.
Acknowledgments: An abstract based on a subset of these findings was accepted for the Particle Therapy Co-Operative Group's (PTCOG) annual meeting (PTCOG58) in Manchester, UK, June 2019.
Ethical Approval: All patient data were collected under internal review board–approved protocol.