Results of a 10‐year survey of workload for 10 treatment vaults at a high‐throughput comprehensive cancer center

Abstract The workload for shielding purposes of modern linear accelerators (linacs) consists of primary and scatter radiation which depends on the dose delivered to isocenter (cGy) and leakage radiation which depends on the monitor units (MUs). In this study, we report on the workload for 10 treatment vaults in terms of dose to isocenter (cGy), monitor units delivered (MUs), number of treatment sessions (Txs), as well as, use factors (U) and modulation factors (CI) for different treatment techniques. The survey was performed for the years between 2006 and 2015 and included 16 treatment machines which represent different generations of Varian linear accelerators (6EX, 600C, 2100C, 2100EX, and TrueBeam) operating at different electron and x‐ray energies (6, 9, 12, 16 and 20 MeV electrons and, 6 and 15 MV x‐rays). An institutional review board (IRB) approval was acquired to perform this study. Data regarding patient workload, dose to isocenter, number of monitor units delivered, beam energies, gantry angles, and treatment techniques were exported from an ARIA treatment management system (Varian Medical Systems, Palo Alto, Ca.) into Excel spreadsheets and data analysis was performed in Matlab. The average (± std‐dev) number of treatment sessions, dose to isocenter, and number of monitor units delivered per week per machine in 2006 was 119 ± 39 Txs, (300 ± 116) × 102 cGys, and (78 ± 28) × 103 MUs respectively. In contrast, the workload in 2015 was 112 ± 40 Txs, (337 ± 124) × 102 cGys, and (111 ± 46) × 103 MUs. 60% of the workload (cGy) was delivered using 6 MV and 30% using 15 MV while the remaining 10% was delivered using electron beams. The modulation factors (MU/cGy) for IMRT and VMAT were 5.0 (± 3.4) and 4.6 (± 1.6) respectively. Use factors using 90° gantry angle intervals were equally distributed (~0.25) but varied considerably among different treatment techniques. The workload, in terms of dose to isocenter (cGy) and subsequently monitor units (MUs), has been steadily increasing over the past decade. This increase can be attributed to increased use of high dose hypo‐fractionated regimens (SBRT, SRS) and the increase in use of IMRT and VMAT, which require higher MUs per cGy as compared to more conventional treatment (3DCRT). Meanwhile, the patient workload in terms of treatment sessions per week remained relatively constant. The findings of this report show that variables used for shielding purposes still fall within the recommendation of NCRP Report 151.


| INTRODUCTION
The National Council on Radiation Protection and Measurements (NCRP) has offered helpful recommendations and technical information related to the design and installation of structural shielding for megavoltage x-ray radiotherapy facilities in Report 151. 1 For primary barrier shielding considerations due to primary and scatter radiation, the most important parameter is the workload, defined as the time integral of the absorbed-dose rate determined as the depth of maximum absorbed dose, at 1 m from the source. 1 When designing and evaluating linac shielding, the value for workload is usually specified as the absorbed dose delivered in cGy to the isocenter in a week, and is selected for each accelerator based on its projected use. This is usually estimated from the number of patients (or fields) treated in a week and the absorbed dose delivered per patient (or field). 1 Shielding calculation for secondary barrier, due to leakage radiation in the treatment head, depends largely on the monitor units delivered (MU). 2,3 Intensity modulated radiation therapy (IMRT) has become the standard of care in the treatment of prostate, head and neck (H&N) and other sites. 4,5 IMRT delivery is inefficient due to multi-leaf collimator (MLC) modulation and requires more MU per treatment. 4 Previous reports have shown that machines treating with IMRT have higher workloads than non-IMRT machines 3,6,7 and that workloads for IMRT machines could reach approximately 100,000 MU/wk. The NCRP recognizes this difference in workload for IMRT treatments relative to three-dimensional conformal radiotherapy (3DCRT), considers the increase in leakage associated with the higher workloads when designing shielding for an IMRT room, and utilizes a modulation factor, which they term the "IMRT factor" (CI), to include the increased leakage workload in secondary barrier analyses. 1 Values of CI can range from 2 to 10 or more. 3,[7][8][9] In addition, the increased leakage due to higher MU for high energy x-ray beams (>10 MV) can lead to an increase in production of neutrons. 1,10 Since the publication of these earlier reports, volumetric modulated arc therapy (VMAT) technique was introduced into the clinic.
VMAT allows the simultaneous modulation of MLC, gantry speed, dose rate, and faster delivery time. [11][12][13][14] Similar to IMRT, VMAT requires higher MU per treatment compared to conventional 3DCRT. 15 The increase in monitor units for IMRT and VMAT does not appreciably affect the amount of radiation reaching the primary barrier on a per-plan basis, since the prescription dose, and consequently, the amount of radiation reaching the primary barrier stays the same. 1,7 If, however, more patients are treated per day, due to lower overall setup and treatment time, the total weekly dose to the primary barrier will increase. Image guided radiotherapy treatments (IGRT) has also become very common due to the availability of onboard kV imagers (kV-OBI) and MV electronic portal imager devices (EPID) on modern linacs which allows accurate patient positioning and tumor monitoring during delivery. 16 IGRT is widely used in the delivery of hypo-fractionated regimens such as stereotactic body radiotherapy (SBRT) and single fraction cranial stereotactic radiosurgery (SRS) which use higher dose per fraction. 17 However, the increase in setup and treatment time can result in lower patient throughput.
Therefore, it would be instructive to examine the impact on monitor units associated with the various modern clinical treatment techniques, as well as the overall impact on patient load, and treatment energy distribution. In this study, we wish to obtain a clearer picture of the contribution of current treatment techniques to machine workload, and to evaluate patient loads. In addition, we evaluate the use factor (U), defined as the fraction of a primary-beam workload that is directed toward a given primary barrier, which will depend significantly on the type of radiation installation and modality. In order to significantly expand on earlier studies we extracted 10 years worth of data on 16 treatment machines.

| MATERIALS AND METHODS
To perform this survey, we utilized data from 16 Varian Linear Accelerators (LINACs) that were operational at different time periods between 2006 and 2015 in 10 treatment vaults at the main center in our institution as illustrated in Table 1. The linacs represent different generations of Varian machines (6EX, 600C, 2100C, 2100EX, and TrueBeam) operating at different electron and photon energies (6, 9, 12, 16 and 20 MeV electrons and, 6 and 15 MV photons). We note that some machines in certain rooms were de-commissioned during this 10 yr period and were replaced primarily with state-of-the-art TrueBeam machines with one exception where the vault (#6) could not accept the TrueBeam accelerator and a Varian 6EX was installed (Table 1).
At our institution, some machines are designated for specific treatment sites based on machine energy and the availability of ancillary devices (ExacTrac, Calypso, Align RT, 6DOF couch, etc.).
As an example, the machine in vault #4 is used primarily for total body irradiation (TBI) while the machine in vault #9 serves as a backup and the machine in vault #6 is a single energy machine were backward compatible. We would like to mention that because of machine compatibility issues with R&V system prior to 2005 causing some missing information for some machines, we opted to report on the workload in the past 10 yr.
An IRB approval was acquired to perform this study. Patient treatment records were exported from ARIA using custom reports and stored in Excel spreadsheets. Data were processed in MATLAB (R2011a, Ver 7.12) and anonymized to mask all protected health information (PHI) for further analysis.
All linacs were calibrated per TG-51 to deliver 1 cGy/MU at 100 SAD for x-ray beams and 1 cGy/MU at D max (100 SSD) for electron beams using 600 MU/MIN dose rate 18 The workload per machine was calculated based on the number of treatment sessions per week, dose in cGys delivered to isocenter, and the corresponding MUs. An inverse square correction was applied to calculate the dose at isocenter for TBI treatments.
For use factor (U) calculations, the beam angles were binned into 12 bins with 30°gantry angle intervals, and the dose (cGy) delivered by beams in each bin was divided by the total dose (cGy). To estimate use factor for VMAT, knowing the starting and stopping angle along with directionality, each arc was subdivided into 30°intervals. It was assumed that the dose delivered per arc is divided equally among sub-   Fig. S2.  The increased modulation varied based on the treatment delivery technique as shown in Fig 2. The average modulation factor for IMRT was 5.0 which is slightly higher than that for VMAT (4.7) but similar to previously published results; 3,8,19 The workload (cGy) delivered using 6 MV (60%) was twice the workload for 15 MV (30%).
The 6 MV beam was predominantly used for cranial SRS, lung, and This study also showed that the average number of beams/fraction has been increasing due to the adoption of complex IMRT treatments such as H&N and hypo-fractionated regimens such as para-spinal SBRT, and single fraction Cranial SRS which require multiple fields (7-10) compared with the more traditional 3DCRT (4-5 fields) and palliative treatments with two opposed beams (Fig. 5). We note that we upgraded our in-house treatment plan- This study did not take into account the extra dose (cGy) and resulting MU from daily and monthly QA routines which are usually done off-hours. It also ignored the imaging dose generated during patient setup. Moreover, flattening filter free beam (6FFF) has not been heavily utilized at our center but is expected to play a bigger role in SRS treatment because of faster delivery due to its high dose rate. 20,21 In addition, the use factor for IMRT and VMAT beams can be determined more accurately using EPID dosimetry.

| CONCLUSION
This study showed that the workload, in terms of dose (cGy) and    Table S1. Weekly workload per machine in cGy Table S2. Weekly workload per machine in MU   Table S3. Use-factors (U) in 2015 at 90°gantry angle intervals.