Limiting treatment plan complexity by applying a novel commercial tool

Abstract Purpose A recently introduced commercial tool is tested to assess whether it is able to reduce the complexity of a treatment plan and improve deliverability without compromising overall quality. Methods Ten prostate and ten oropharynx plans of previously treated patients were reoptimized using the aperture shape controller (ASC) tool recently introduced in Eclipse TPS (Varian Medical Systems, Palo Alto, CA). The performance of ASC was assessed in terms of the overall plan quality using a plan quality metric, the reduction in plan complexity through the analysis of 14 of the most common plan complexity metrics, and the change in plan deliverability through 3D dosimetric measurements. Similarly, plans optimized limiting the total number of delivered monitor units was assessed and compared. The two strategies were also combined to assess their potential combination. Results The plans optimized by exploiting the ASC generally show a reduced number of total Monitor Units, a more constant gantry rotation and a MLC modulation characterized by larger and less complicated shapes with leaves traveling shorter overall lengths. Conclusions This first experience suggests that the ASC is an effective tool to reduce the unnecessary complexity of a plan. This turns into an increased plan deliverability with no loss of plan quality.

However, the potential downsides of VMAT include the augmented low dose radiation to surrounding normal tissue 2 and the large complexity of delivery characterized by intensive MLC modulation, that is, a large amount of small MLC subfields. Such complexity might undermine dose calculation correctness and increase the uncertainty of the dose verification process. [6][7][8][9] These drawbacks of VMAT have led to the proliferation of studies dedicated to the measurement of VMAT plan complexity and related plan quality and deliverability. It has been shown that unnecessary amount of complexity negatively affects the accuracy of both dose calculations and treatment delivery, in particular for TPS based on simple MLC physics models. [10][11][12][13][14][15][16][17][18][19] Part of these studies were dedicated to the exploration of strategies to limit this complexity. In this respect, many strategies have been proposed to limit plan complexity, primarily reducing the delivered MU or directly simplifying the shape of the MLC during optimization. 9,11,[20][21][22][23][24] To limit VMAT complexity, a similar strategy has recently been introduced in v. 15.5 of the Eclipse TPS (Varian Medical Systems, Palo Alto, CA). The aperture shape controller (ASC) is a new component in the leaf sequencer for VMAT in the photon optimizer (PO) algorithm that tends to increase the size and decrease the complexity of the MLC aperture. 25 In this study, a first experience with ASC is reported. The strength of ASC in limiting VMAT plan complexity has been tested and compared to the MU limiting strategy implemented in Eclipse since v.8.6. Ten prostate and ten oropharynx clinical plans were reoptimized following different strategies to limit plan complexity.
Every new plan was compared to the clinical plan to assess whether it is possible to limit plan complexity without compromising the overall clinical plan quality and if this reduction in complexity turns into an improved plan deliverability.

2.A | Aperture shape controller
The aperture shape controller (ASC) is a new component in the leaf sequencer for VMAT in the PO algorithm. It was introduced in version 15.5 of the Eclipse TPS. The ASC favors apertures of minimal local curvature, which is measured as the positions of the tips of adjacent leaves that modulate the same spatially continuous target projection. ASC acts through a penalization term in the total cost function, the weight of which can be partially controlled by the user via a calculation option that can be set to five different values, ranging from very low to very high. 25

2.B | VMAT plans
Ten patients treated for low-risk prostate cancer and ten patients treated for oropharyngeal cancer at our institution were retrospectively selected. The prostate patients were treated with 70 Gy in 28 fractions (2.5 Gy/fx) to the prostate gland only. The oropharynx patients were treated with 69.96 Gy to the tumor bed and 59.4 Gy to the nodal chains in 33 fractions (2.12 and 1.8 Gy/fx). All patients were treated using a TrueBeamSTx with high-definition MLC (Varian Medical Systems, Palo Alto, CA, USA) and 6 MV energy.
According to manufacturer's specifications, ASC attempts to join disconnected apertures defined by adjacent leaf pairs, which, in turn, may reduce the number of MUs delivered. On the other hand, it is also well known that limiting the delivered MU is an indirect way to limit MLC movement and modulation. 22,26 For this reason, three different strategies to limit MLC modulation were tested and compared: three plans were optimized by the same expert planner for each of the selected patients. For each patient the optimization was started with a set of constraints that was maintained practically unaltered for all the relative plans. The human interaction was minimal and devoted only to avoid possible major deviations. No postoptimization normalization was applied. A limit on the maximum Monitor Units was imposed, setting the MU objective to a total ratio of MUs/cGy equal to 3 with a fixed objective weight; this plan will be referred to as MU limit. 16 Another plan was optimized setting the ASC to very high penalty; this plan will be referred to as ASC. A further plan was optimized coupling the ASC penalty and the MU limit; it will be indicated as ASC + MU limit. The ASC was tested using the maximum available penalty to probe the limits of its capability. A detailed evaluation of which penalty level to use in a clinical environment is outside the scope of this study.
These plans were compared with the original clinical plans, which were optimized with no strategies to limit plan complexity. All plans

2.C | Plan quality assessment
To assess the overall plan quality and limit the subjectivity of judgment, the Plan Quality Metric (PQM) was adopted as a global measure of quality. PQM was first introduced by Nelms 27  (PQM%), that is, the ratio of the achieved score to the maximum achievable thus represents a relative measure of plan soundness. In this study, two dedicated PQM algorithms were used to judge prostate and oropharynx plans. The algorithms were delineated in accordance with the clinicians and were inspired by published standards (RTOG0126, RTOG0522, RTOG0615, RTOG0619, RTOG0625, RTOG1016) and prior experience. 28,29 The detailed description can be found in the supplementary materials.
To pose further attention on dose conformity and dose spread, we also evaluated the variation of Conformity Index and the volume of healthy tissue receiving low doses. In particular, the following quantities were collected and compared: Conformation Number of 95% prescription dose to prostate PTV and to low dose oropharynx PTV 30 ; the healthy tissue volume at 95%, 50%, and 30% of the prescription doses. The plans were considered deliverable if at least one of the following action levels of the gamma passing rate (GPR%) was met: 90% for 2%L/2 mm, 95% for 2%G/2 mm, 93% for 3%L/3 mm, and 97% for 3%G/3 mm. These cutoffs are based on the center's experience and were set following the recommendation of AAPM report 119. 32

2.F | Data analysis
The clinical, complexity, and dosimetric outcome of the different optimization strategies were compared through two-tailed paired Student's t-test, with a significance level of 0.05, once their normality was ensured using a Shapiro-Wilk test.

3.A | Plan quality
The PQM% of the entire pool of plans is shown in Fig. 1. In general, the overall quality of the clinical plan was maintained by all the other optimization strategies. In Fig. 2 an example of a prostate and an oropharynx plan obtained for the different strategies is given. Table I

3.B | Modulation parameters
In the case of prostate treatments, all of the adopted strategies induced significant changes in approximately all the considered modulation parameters, as can be seen in Table II  The case of oropharynx treatments, reported in Table III, is similar to what is described above. The only differences are that the dose rate variation is not significantly changed and the reduction in delivery time, even if significant, is not clinically relevant. In this case, the ASC seems to be more incisive than the MU limitation, while the coupling of the two is again the more effective strategy to reduce plan complexity, independent of the considered metric. When the ASC is coupled with the limit on the total MU, the best deliverability is reached for prostate plans while no clear advantage can be seen in the case of oropharynx plans.

| DISCUSSION
The reported results show that prostate and oropharynx plans optimized with PO algorithm v15.5 contain a discrete amount of unnecessary plan complexity that might be mitigated by limiting the total MU or making use of the newly introduced ASC tool during optimization.
Concerning the MU limitation, the results presented here match those already reported in the literature for prostate and hypopharynx, despite the different optimizers. 20,22 No clinically significant differences in the plan quality was measured for any of the optimization strategies. Not even low dose T A B L E I Change in dose conformity. A partial motivation of this can be found by the positive linear relationship between the uncertainty of 2D pretreatment verification and the proportion of dose from small fields reported in, 9 but at the moment no clear explanation for such a result can be given because no conclusive relationship has been reported between a single complexity metric and the result of dosimetric verification. A more indepth evaluation of the matter is outside the scope of the present work but it is interesting and will be addressed in a subsequent investigation.

Metric
This work reports the first experience and the positive impact of ASC on plan complexity and deliverability. The results reported herein suggest the need for further investigation on a larger patient cohort to tweak the ASC setting to the specific treatment site, eventually coupling it to the MU limit.

| CONCLUSION S
The use of ASC in Eclipse optimization leads to a reduction in plan complexity comparable or superior to a 15% reduction in the total delivered monitor unit. On average, the reduced complexity does not compromise the overall clinical plan quality but ensures an ameliorated plan deliverability. Detailed results might vary among different treatment sites. ASC deserves additional studies for a thorough evaluation of its capability and methods of use.

CONFLI CT OF INTEREST
None.
The author(s) received no specific funding for this work. A. Scaggion and M. Fusella were responsible for statistical analyses of the entire work.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article.