The dosimetric effect of electron density overrides in 3DCRT Lung SBRT for a range of lung tumor dimensions

Abstract The combined effects of lung tumor motion and limitations of treatment planning system dose calculations in lung regions increases uncertainty in dose delivered to the tumor and surrounding normal tissues in lung stereotactic body radiotherapy (SBRT). This study investigated the effect on plan quality and accuracy when overriding treatment volume electron density values. The QUASAR phantom with modified cork cylindrical inserts, each containing a simulated spherical tumor of 15‐mm, 22‐mm, or 30‐mm diameter, was used to simulate lung tumor motion. Using Monaco 5.1 treatment planning software, two standard plans (50% central phase (50%) and average intensity projection (AIP)) were compared to eight electron density overridden plans that focused on different target volumes (internal target volume (ITV), planning target volume (PTV), and a hybrid plan (HPTV)). The target volumes were set to a variety of electron densities between lung and water equivalence. Minimal differences were seen in the 30‐mm tumor in terms of target coverage, plan conformity, and improved dosimetric accuracy. For the smaller tumors, a PTV override showed improved target coverage as well as better plan conformity compared to the baseline plans. The ITV plans showed the highest gamma pass rate agreement between treatment planning system (TPS) and measured dose (P < 0.040). However, the low electron density PTV and HPTV plans also showed improved gamma pass rates (P < 0.035, P < 0.011). Low‐density PTV overrides improved the plan quality and accuracy for tumor diameters less than 22 mm only. Although an ITV override generated the most significant increase in accuracy, the low‐density PTV plans had the additional benefit of plan quality improvement. Although this study and others agreed that density overrides improve the treatment of SBRT, the optimal density override and the conditions under which it should be applied were found to be department specific, due to variations in commissioning and calculation methods.


| INTRODUCTION
Lung Cancer is one of the world's most commonly diagnosed cancer types, as well as the most common cause of cancer death with an estimated 1.6 million deaths worldwide per year. 1 Non-small cell lung cancer (NSCLC) contributes to approximately 85% of all lung cancers. 2 For patients whom surgery is not an option, conventional or stereotactic radiotherapy is frequently used. 3,4 One of the main toxicities stemming from radiation therapy in NSCLC is Radiation Pneumonitis (RP). 5,6 The use of stereotactic body radiotherapy (SBRT) to reduce the Planned Target Volume (PTV) margin and increase PTV edge dose gradients can improve local control and reduce the chance of toxicities such as RP. 7 One of the issues associated with treating lung cancer with radiotherapy is motion of the tumor caused by patient breathing.
In SBRT, this issue becomes an even greater challenge due to the addition of the smaller expansion of the PTV around the Internal Target Volume (ITV), with a steep dose gradient beyond this target volume. 8 The most common technique for managing temporal tumor variation is four-dimensional imaging, including respirationcorrelated 4-Dimensional CT (4DCT) scanning, 9 however, this can result in motion artifacts. 10 Artifacts can be caused by irregular breathing traces, for example, coughing or patient motion during the scanning process. 11 The presence of inhomogeneous media can also affect dose calculation accuracy. Several studies have examined the impact of different dose calculation algorithms on dose delivered to inhomogeneous media, in particular lung. [12][13][14][15] It is recommended that collapsed cone convolution (CCC) algorithms be used when complex algorithms such as Monte Carlo or Acuros XB (Varian Medical Systems, Palo Alto, CA) are not readily available. 12,13 One issue with CCC algorithms is that the model is unable to accurately calculate dose at the interface between lung and tumor. 12 This is due to the assumption of transient charged particle equilibrium (TCPE) occurring at the tumor-lung interface not being true, and CCC algorithms cannot accurately model this effect. 15 These errors have been shown to increase with smaller treatment volumes, where the ratio of tumor-lung interface surface area to tumor volume increases with decreasing target volume. 12 During treatment, as the GTV moves through the ITV as defined by the 4DCT scan, the dose to the tumor will change compared to the treatment plan. As there is preferential dose buildup in higher density areas, as the GTV moves to a region of the ITV that is underdosed on the treatment plan, the GTV will receive a larger dose than expected. 14,16 One method to overcome the issues associated with inhomogeneity corrections in lung and tumor motion is to override the electron densities of the ITV/PTV. [17][18][19] A study by Fu et al. 17 devised a method for overriding the density of the PTV to 0.8 g/cm 3 to reduce the planned MU while still delivering sufficient dose to the tumor for SBRT lung planning. Wiant et al. 18 compared the use of freebreathing CT scans, time average scans, and ITV/PTV tissue-density overridden scans for lung SBRT to evaluate the accuracy of each method for predicting dose deposition in lung tissue. Accuracy was assessed from measurements using Gafchromic film in a QUASAR phantom. 18 A further study by Wiant et al. 19 looked at volumetric modulated arc therapy (VMAT) plans, and introduced a hybrid override with the ITV set to tumor density and PTV-ITV set to an intermediate density.
The studies by Wiant et al. 18,19 were performed using the Eclipse planning system, and the method has not currently been extended to any other dose calculating algorithms. Also any implications due to tumors size or overriding the density of the PTV to a variety of low densities between lung and water have not been investigated.
This study will look into quantifying the effect of density overrides to establish a trend based upon the relative sizes of the ITV and PTV.

2.A | Phantom study
To assess the impact of various density overrides on SBRT lung plans, a phantom study was conducted using the Elekta Synergy proportions. This value was consistent to the values seen clinically in our department, rounded to one significant figure. An average water ED of 1.000 was used. The overrides corresponded to 75% lung material and 25% water (ED = 0.475), 50% lung material and 50% water (ED = 0.650), and 25% lung material and 75% water (ED = 0.825).
These were applied to the PTV, with or without the ITV set to an ED of 1.000 to form a hybrid plan. In total, eight override datasets were generated for each plan, and compared to the AIP and 50% central phase plans (Table 1). No beam weighting changes were applied with each override. The dose was rescaled to cover the PTV volume with the prescription dose set to the relative isoline of 80%.

2.B | Plan quality and coverage
To assess the relative coverage of each override plan compared to the baseline plans (AIP and 50% central phase), five PTV coverage metrics were assessed. This including the Mean Dose (Gy), Maximum Dose (Gy), Minimum Dose (Gy), D90, and D95. As these plans were forward planned using the collapsed cone algorithm, the Maximum and Minimum Doses (Gy) refers to dose at a point.

Plan type ITV ED PTV ED
50% Phase n/a n /a AIP n/a n /a (TV1), the target volume (TV), and the total volume of the prescription isodose (VR1). (1) For CI(100%) and CI(50%), the CI(X%)-type metrics indicated the ratio of the percentage (x) isodose volume (V X% ) to the target volume (TV). The target volume in this case was the PTV.
The Heterogeneity Index (HI) was also directly calculated from  This results in the total dose being measured as an average across the volume that the active part of the ion chamber covers. As the amplitude of motion and the dimensions of the ion chamber cavity are both known, the active volume of the chamber was contoured as a structure in Monaco and the mean dose in Gy compared to that measured by the ion chamber.

| RESULTS
The effectiveness of each density override was assessed in two ways, including comparing the plan quality in the treatment planning system and the measurable aspects of the plan with point dose and radiochromatic film measurements.

3.A | Plan quality and coverage
Comparison of the baseline plans (AIP and 50% central phase) to the density overridden plans for the 15-mm insert shows that in every case the target coverage was at worst unchanged and mostly The target coverage DVH metrics and PTV conformity and heterogeneity metrics for a range of standard and electron density overridden treatment plans in a 15-mm tumor object in a lung phantom. For the 15-mm insert, all ED overridden plans showed a significant (P < 0.05) difference to the 50% central phase plan (Fig. 2). The largest percentage dose differences were for the 50% central phase The target coverage DVH metrics and PTV conformity and heterogeneity metrics for a range of standard and electron density overridden treatment plans in a 22-mm tumor object in a lung phantom. For the 22-mm insert, the best-performing ED overrides were the ITV = 1.00 plan and PTV = 1.00 plan (P < 0.003, P < 0.002). The positive trend that was previously seen in the 15-mm result remained between the percentage difference and total MU (R 2 = 0.9532), however, the correlation was weaker.
For the 30-mm insert, there was no density overridden plan that showed a statistically significant improvement over the original plans (average P < 0.232). The positive trend seen between the percentage difference and total MU was not significant (R 2 = 0.3051). As there was limited benefit seen in the TPS and point dose results, film work was not completed for the 30-mm tumor insert.
For the 22-mm insert, the best-performing plan was the ITV override, which showed a 3%, 3-mm pass rate of 90.2% for the 10% threshold, and 94.7% for the 40% threshold gamma criteria. However, for the 10% TH gamma criteria, no plans showed a consistent significant difference to the baseline plan.

| DISCUSSION
The key motivation for exploring the effect of ED overrides was to improve the treatment of lung cancer with SBRT. This can be achieved in two different ways: improving the quality of the treatment plan, in terms of target coverage and conformity, and improving the accuracy of the treatment delivery.
Overriding the ED of a target volume to a water-equivalent ED is associated with better target coverage due to the fact that the CCC algorithm accounts for inhomogenieties by scaling dose kernels by the relative electron densities. 22 There is a limited application for applying density overrides to large tumors. When this plan is delivered, however, regardless of the position of the GTV inside the ITV, a higher proportion of the energy fluence will be deposited inside the higher electron density tumor as compared to the surrounding lung material. As the tumor moves within the ITV, the result is a higher dose overall to the ITV. As the baseline 50% central phase plan is unable to predict this smearing out effect, the measured dose in the center of the ITV will be higher than expected, a result that was consistently seen in both the ion chamber and film measurements.
The impact of the dose smearing effect is not only seen in the ITV plan, but the lower density PTV and HPTV plans as well.
Previously published work most comparable to this study is the study by Wiant et al. 19 In their study, a large diameter tumor object was used, and the CI and mean dose values were all approximately the same, with the only significant difference occurring with the maximum dose to the PTV. The ITV plan showed the greatest increase in maximum dose compared to the 50% central phase plan, whereas the PTV and hybrid plans showed minimal differences. This is not consistent with the plan quality results seen in this study. A reason for the variation between results may stem from the differences between the two planning systems and the operation of the heterogeneity accounting algorithms.
In both studies, density overrides were shown to improve dosimetric accuracy, but in this study density overrides were shown to be clinically beneficial for tumors less than 22 mm. An additional study into the mid-sized range around 22 mm would be beneficial. Applying an established Lung SBRT planning protocol to a small selection of override options is necessary to determine the clinic-specific best fit.

| CONCLUSION
No significant statistical difference was seen between the 50% central phase and AIP plans. A trend was demonstrated where, for smaller tumors (<22-mm diameter), the geometric and dosimetric dose coverage and conformity improved when a PTV override was applied. For larger tumors (>22-mm diameter), minimal differences were seen in terms of plan quality and accuracy, suggesting the results are equipment specific. Improvements to dosimetric accuracy were seen as the tumor size decreased. The results established in this study suggest a valid method for improving outcomes to patients with NSCLC treated with SBRT, particularly for small tumors where dosimetric as well as geometric accuracy is a greater concern.

This work was supported by the Medical Physics and Bioengineering
Department at Christchurch Hospital. The authors wish to thank the members of the department who provided resources and support toward this project.

CONFLI CT OF INTEREST
The authors declare no conflicts of interest.