Dosimetric comparison of AcurosBV with AAPM TG43 dose calculation formalism in cervical intraductal high-dose-rate brachytherapy using three different applicators

Purpose: To compare the dosimetric effects of American Association of Physicists in Medicine (AAPM) TG43 dose formalism and AcurosBV (grid-based Boltzmann solver, GBBS) formalism on high-dose-rate (HDR) brachytherapy planning for cervical cancer patients irradiated using three different applicators. Methods: ATG43planandaAcurosBVplanweregeneratedforeachofthe30patients. Twenty patients who had undergone whole pelvic radiotherapy followed by cervical HDR brachytherapy and the remaining 10 patients who underwent total hysterectomy only gave HDR brachytherapy also were enrolled in this study. The patients were divided into three groups according to (T&O), (T&R), consideredwhenusingT&Oapplicatorsorothermaterialswithamuchhigherorlower density (metal or air) than soft tissue. However, If the density is close to that of soft tissue, considering AcurosBV algorithm requires more calculation time, TG43 could still be selected when using applicators in clinical.


INTRODUCTION
Cervical cancer is one of the most common malignant tumors diagnosed in China. According to cancer statistics for China, cervical cancer accounts for 6.6-26.5% of cancer-related morbidity and is the fourth leading cause of cancer-related death in women. 1 In addition to radical surgery, radiotherapy is a common treatment used for cervical cancer.
In addition, postoperative chemoradiotherapy can effectively reduce local recurrence rates and improve the survival rates of patients with cervical cancer. 2,3 Intracavitary brachytherapy (BT) is an essential component of cervical carcinoma treatment 4 and has been shown to significantly improve the radiotherapeutic outcome by improving the target dose distribution while reducing rectal and bladder toxicities. 4 The method currently in use globally for accurately determining the dose delivered in brachytherapy treatments is based on the American Association of Physicists in Medicine Task Group (TG) 43. 5 TG43 dose calculation formalism assumes the radiation is transported through an infinite homogeneous water phantom and therefore does not account for any heterogeneities within or outside the patient. This has implications for the accuracy of the dose calculation in regions close to air or bone. Since 2012, the TG186 report has described model-based dose calculation methods in brachytherapy that account for these heterogeneities. 6 According to previous studies, the TG43 formalism is known to overestimate the radiation dose in the air cavity and underestimate the dose at the high-density substance. [4][5][6] Several studies have suggested that in brachytherapy cases such as prostate and cervix, if the tissue is relatively homogeneous, the treatment technique uses unshielded plastic applicators, and there are no air pockets nearby, then the model-based dose calculation algorithms (MBDCAs) have an average dosimetric influence of smaller than 5% compared with TG43. 7,8 The Acuros BV algorithm solves the linear Boltzmann transport equation or the grid-based Boltzmann solver (GBBS) algorithm and is similar to the Monte Carlo method (MC). 9,10 The accuracy of Acuros BV dose calculation is defined by comparing it to the Monte Carlo simulation (MCNPX, converged statistics to 1%). Calculated dose distribution is generally required to be within 2% (2 mm). The accuracy limit is slightly less strict in regions close to the source (distance less than 5 mm) or in boundary areas of rapid material density change, where a 15% difference is allowed. 8,11,12 Studies have shown that Acuros BV has been reported to estimate dose deposition more accurately than TG43 in heterogeneous media. 13,14 The present study aimed to analyze the dosimetric effect and compare the dose difference of AcurosBV and TG43 plans using three types of applicators for cervical cancer to guide clinicians regarding the algorithm selection for different applicators. grid size was set at 2.5 mm for all 30 cases. The source step size was 5 mm. The total dose was equal to or greater than 28 Gy/4f, the fraction dose was no less than 7 Gy, and the prescription dose was required to surround the 90% volume of the CTV-HR was required to cover the prescription dose. We also considered the dose of point A, which was defined as a point 2 cm lateral to the central canal of the uterus and 2 cm up from the mucous membrane of the lateral fornix, in the axis of the uterus according to the ICRU Report 38. All 30 treatment plans generated using the TG43 algorithm and were also used in the patients' clinical treatment in this study. Retrospectively, the plans were recalculated using the Acuros BV algorithms. The plans were not reoptimized and therefore the structure set, dwell positions, and dwell times were identical between the two plans.

Patient selection and contouring
For each patient, the cumulative dose was computed, consisting of EBRT and BT contributions normalized to 2 Gy per fraction (EQD2) using the linear-quadratic model with α/β ratios of 10 and 3 Gy for of CTV-HR and OARs, respectively. 19,20 The total dose limitations of the CTV-HR and OARs are shown in Table 1. 21

Dosimetric comparison of target
The dose distributions of the plans using the three types of applicators are shown in Table 2 Figure 3 shows the DVH of CTV-HR in the TG43 and AcurosBV plans using three types of applicators. Among the three applicators, the T&O applicator still had the greatest impact on the dose distribution for the TG43 and AcurosBV plans. From Table 2 and Figures 1 and 2, we can draw the following conclusion: regardless of the type of applicator used, the dose distribution of the target in the TG43 plans was higher than that of the Acuros BV plans.

Dosimetric comparison of OARs and point A
Quantification statistics were used in the present study. The dose distributions of the OARs and point A are presented in Table 3 and Figure 3. The dose difference between the two algorithms was similar to that of the targets mentioned above. In all patients, the dose to both OARs and point A was higher in TG43 plans than those in AcurosBV plans.
In the T&O group, the dose difference was the biggest in three groups: compared to Acuros BV plans, the point A of TG43 plan had a 9.6% higher dose (6.03 vs. 5.50 Gy, p = 0.002), the D 0.1cc and D 2cc of bladder, sigmoid, rectum, and small bowel in TG43 plans had a 6-10% higher dose, and differences of the parameters were statistically significant (all p < 0.05).

F I G U R E 2
The DVH of CTV_HR in the TG43 (3D, black full line) and AcurosBV (AXB, red dashed line). Planning of three different applicators for cervical cancer, respectively: (A) the DVH of CTV_HR using cylinder applicator, (B) the DVH of CTV_HR using T&O applicator, and (C) the DVH of CTV_HR using T&R applicator. In the cylinder group, the Acuros BV plan yielded a smaller D 0.1cc and D 2cc for the bladder, rectum, and small bowel (all p < 0.05) compared to the TG43 plans. The dose differences ranged from 1% to 3%

TA B L E 3 Dosimetry of point A and
for the TG43 and AcurosBV plans.
The dose difference was largest for T&O brachytherapy planning and smallest for cylinder planning in the three types of applicators. This conclusion can be confirmed from Figure 3. Among the three applicators, the T&O applicator still had the greatest impact on the dose distribution for the TG43 and AcurosBV plans. We also found that, regardless of the type of app used, the dose distribution of the OARs in the TG43 plans was higher than that of the Acuros BV plans.

DISCUSSION
An example of a comparison of the AcurosBV dose and TG43 for a shielded cylinder applicator (GM11004380 06) in a water phantom shows that during plan optimazation, more caculation time is needed but a better accuracy dose disrtibution can be gotten by using AcurosBV algorithm than TG-43. [26][27][28] Many studies have confirmed this conclusion. 29 Table 4 shows the material composition of these applicators in detail.
From Table 4, we can see that the T&O applicators consist of stainless steel and titanium materials, and have a higher density than the other two applicators. We also found that in T&O applicators, titanium and stainless steel make up 70-80% of the total mass. These two metals have a much higher density than water, therefore the dose difference was the highest among the three types of applicators in TG43 and Acuros BV plans. On the contrary, checking the Varian product instructions, the cylinder applicator is made solely of polyphenylsulfone, which has a density of 1.3 g/cm 3 , and the composition elements are carbon, hydrogen, oxygen, and sulfur. The density is close to water, which can explain the small dose difference between the two algo-

CONCLUSION
The TG43 algorithm would overestimate the dose of the target and OARs compared with using the AcurosBV algorithm in the plans with the same dwell position and dwell time, although the plan of applying the two algorithms can meet the clinical requirements. In clinical practice, the material composition of the T&O applicator is very different from the surrounding tissue (around 10%), so AcurosBV is clinically recommended when using T&O applicators. However, in the plans based on the T&R and cylinder applicators, although the TG43 algorithm overestimated the tissue dose, the difference in dose distribution caused by the two algorithms was almost negligible because the difference in dose distribution was not much (1-5%) and both were located around the applicator. Considering the AcurosBV algorithm requires more calculation time, TG43 can still be selected clinically when using T&O or cylinder applicators.