Development and Feasibility of a Patient-Derived Training Simulator for Image-Guided Adaptive Brachytherapy of Locally Advanced Cervical Cancers

Image-guided adaptive brachytherapy (IGABT) plays a pivotal role in denitive radiotherapy of cervical cancer. Although the combination of a tandem and ovoid applicator with interstitial needles (IC/IS brachytherapy) is an ecient IGABT technique for bulky irregular-shaped tumors, training opportunities for IC/IS brachytherapy remain limited. Thus, we developed a training simulator for IC/IS brachytherapy for locally advanced cervical cancer and tested its feasibility. simulator a soft silicone tumor with an acrylic tube mimicking the vagina. The tumor phantom was modeled on a cancer through detailed inspection of their from comparing

Cervical cancer causes more than 300 000 deaths worldwide annually, and its mortality ranks fourth among all cancers [1]. Radiotherapy plays a pivotal role in the de nitive treatment of locally advanced cervical cancers [2], with de nitive radiotherapy approaches including external beam radiotherapy (EBRT) and brachytherapy [3]. For the latter, image-guided adaptive brachytherapy (IGABT) using magnetic resonance imaging (MRI) or computed tomography (CT) has become widespread in recent years [4,5]. The introduction of IGABT technologies into de nitive radiotherapy practice has dramatically improved the oncologic outcomes of cervical cancers [6,7].
Despite the technological advancements, IGABT using a tandem and ovoid applicator alone results in suboptimal dose coverage for bulky or irregular-shaped tumors, leading to insu cient tumor control [8,9]. To deliver a su cient dose to such tumors, the tandem and ovoid applicator can be supplemented with a few interstitial needles. In 2011, our group introduced a combination of a few interstitial needles with a Fletcher-Suit Asian Paci c applicator (Elekta, Stockholm, Sweden) under in-room CT guidance [10].
Multiple studies have shown that this combined intracavitary and interstitial (IC/IS) brachytherapy has excellent outcomes for locally advanced disease [7,11]. Additionally, brachytherapy applicators equipped with a template for needle insertion have been developed by several groups. In 2006, Dimopoulos et al. introduced the Vienna applicator consisting of a tandem and a ring that enables needle insertion from prede ned directions guided through the holes in the ring [12]. In 2009, Jürgenliemk-Schulz et al.
introduced the Utrecht applicator consisting of a tandem and two oval ovoids that enables needle insertion from more-varied directions than the Vienna applicator, with the needles guided through the holes in the ovoids [13]. In 2018, Walter et al. introduced the Venezia applicator consisting of a tandem and two lunar-shaped ovoids (that form a ring in combination) functioning as a needle template [14].
The key to the success of IC/IS brachytherapy is to insert the interstitial needles in locations appropriate to achieving a highly conformal dose to the targeted tumor volume. However, the training opportunities for needle application are limited, thereby preventing this technique from becoming widely used by radiation oncologists. To address this issue, this study aimed to develop a training simulator for IC/IS brachytherapy for locally advanced cervical cancer, and to report on its feasibility with different applicator types and physicians with different levels of experience of IGABT.

Foundation of the simulator
First, the foundation of the simulator was developed using transparent acryl to provide visibility; no metal was used to ensure that the simulator was CT and MRI compatible. To enable a physician to approach through the vagina, an acrylic tube mimicking a vagina (50 mm in diameter and 90 mm in length) was installed on an acrylic plate (200 × 100 mm).

Patient-derived Tumor Phantom
Next, a tumor phantom representing a typical bulky and irregular-shaped tumor was developed for IC/IS brachytherapy training. The case on which the tumor phantom was modeled was selected through a review of in-house IGABT records. Three-dimensional (3D)-printing techniques were used to create a realscale tumor phantom based on the high-risk clinical target volume (HR-CTV) data. The feasibility of the IC/IS training simulator was investigated by comparing treatment plans derived under various application settings.
A review was made of 495 patients with newly-diagnosed and pathologically-con rmed cervical cancer who were treated with de nitive radiotherapy using IGABT at Gunma University Hospital (Maebashi, Gunma, Japan) between 2012 and 2020, and 70 FIGO stage IIIB patients were identi ed. The details of the de nitive radiotherapy are described previously [7]. The pre-treatment MRI obtained at the rst brachytherapy session of these 70 FIGO IIIB patients was inspected by three radiation oncologists (KT, KA, and DI), and ten patients harboring a bulky tumor with extensive bilateral parametrial involvement were selected. For these ten cases, the HR-CTV was delineated on the T2-weighted MRI according to the recommendations of The Groupe Européen de Curiethérapie and the European Society for Radiotherapy and Oncology [15,16]. MIM Maestro software was used for this purpose (version 6.8.7., MIM Software Inc., Cleveland, OH, USA). The HR-CTV contour data were then transferred to OsiriX 64-bit (Pixmeo, Bernex, Switzerland) and subjected to 3D-volume rendering. The three radiation oncologists inspected the 3Dreconstructed HR-CTVs from various angles and identi ed the best case for use as the phantom for the simulator. The 3D-HR-CTV data of the chosen patient were then transferred to computer-aided design software (Creo Parametric, PTC, Boston, MA, USA) and a 3D-model was printed in acrylonitrile butadiene styrene (ABS) plastic using a 3D printer (Value 3D Magix MF-1100, Mutoh Industries Ltd., Tokyo, Japan).
An aluminum mold of the ABS plastic tumor phantom was then created using a CNC (computerized numerical control) milling center (ACCUMILL 4000, DMG Mori Co., Ltd., Nagoya, Aichi). Finally, a semitranslucent soft silicon tumor phantom was created using the mold. Soft silicon was selected as the material for the tumor phantom because it allowed visualization of the needle locations.

Feasibility Test
The feasibility of the IC/IS training simulator was investigated by comparing six different treatment plans derived by one of two physicians (an expert or a resident) using either a Fletcher-Suit Asian Paci c applicator (sessions #1-#3) or a Venezia applicator (sessions #4-#6), as outlined in Table 1. In sessions #1 and #4, the application of tandem and ovoids was performed by the expert; in sessions #2 and #5, the application of tandem, ovoids, and interstitial needles was performed by the resident; and in sessions #3 and #6, the application of tandem, ovoids, and interstitial needles was performed by the expert. The expert was a Japanese Society for Radiation Oncology-board certi ed radiation oncologist who had experience of more than 800 sessions of IC/IS brachytherapy. Four interstitial needles were used and reinsertion was not allowed, although no constraints were set on the insertion locations of the needles. With the Fletcher-Suit Asian Paci c applicator, the interstitial needles were inserted in a freehand manner according to a previous report by our group [10]. HR-CTV, high-risk clinical target volume; D90, the minimum dose delivered to 90% of the volume; D2cc, the maximum dose at which any 2 cc of the volume is irradiated; T&O, tandem and ovoids.
At each session, CT images were obtained and the HR-CTV (i.e., the tumor phantom) was delineated using MIM Maestro software following the actual clinical practice used for IGABT. The CT images used to plan the rst brachytherapy for the model case were superimposed on those obtained at each test session, and the contours of the rectum and bladder were created on the test session CT images on the basis of the anatomy of the original patient. The contour data were transferred to an Oncentra (Elekta, Stockholm, Sweden) and treatment plans were created based on a high-dose rate 192 Ir remote-afterloading system (microSelectron, Elekta, Stockholm, Sweden) [7]. The highest possible dose was prescribed to the HR-CTV while keeping the dose constraints for the rectum and bladder set at D 2cc (i.e., the maximum dose at which any 2 cc of the volume is irradiated) below 6 Gy and 7.6 Gy, respectively [17,18]. D 90 (i.e., the minimum dose at which any 90% of the volume is irradiated) was utilized as the endpoint of dose coverage to the HR-CTV [17,18]. For all sessions, the target delineation and dose prescription were performed by the expert.

Results
The HR-CTV of the model patient at the rst brachytherapy session was 68 (left-right) ⋅ 49 (anteriorposterior) ⋅ 45 (cranial-caudal) mm in size (Fig. 1a, b, Supplementary Data 1), and the life-size tumor phantom was successfully created using these HR-CTV data (Fig. 1c-f).
The feasibility test of the IC/IS training simulator using the Fletcher-Suit Asian Paci c applicator without interstitial needles (session #1) resulted in an unacceptably low HR-CTV D 90 of 4.23 Gy, clearly demonstrating that the use of a tandem and ovoids alone was inadequate for this tumor, even when positioned by the expert (Table 1, Fig. 3a) [17,18]. In session #2, the resident used four interstitial needles in addition to the tandem and ovoids. Although an evident improvement over session #1 was observed, the HR-CTV D 90 was still low at 5.69 Gy because of insu cient dose coverage in the bilateral and anterior portions of the target (Table 1, Fig. 3b). By contrast, in session #3, the four needles inserted by the expert were located at peripheral sites in the bilateral and anterior portions of the HR-CTV, which led to greater dose conformity and a satisfactorily high HR-CTV D 90 of 6.70 Gy (

Discussion
In this study, we developed a training simulator for IC/IS brachytherapy for locally advanced cervical cancer and evaluated its feasibility with different applicator types and physicians with different levels of experience of IGABT. The patient-derived tumor phantom was shown to be feasible with procedures using Fletcher-Suit Asian Paci c applicator and Venezia applicator with insertion of interstitial needles. Furthermore, the phantom re ected the difference in experience between an expert and a resident, resulting in various HR-CTV D 90 doses. The results of the present study indicate that the phantom is a good representation of typical bulky and irregular-shaped cervical cancers that are treated with IGABT, and therefore the training simulator appears to be useful for the training of IC/IS brachytherapy for locally advanced cervical cancer.
Despite the e cacy of brachytherapy for cervical cancer, multiple reports indicate that its utilization rate has decreased. In 2013, Han et al. reported a decrease in the brachytherapy utilization rate from 83% in 1988 to 58% in 2009, using data from the National Cancer Institute's Surveillance, Epidemiology, and End Results program covering 7359 patients with stage IB2 to IVA cervical cancer who were treated with EBRT.
In their study, the use of brachytherapy was independently associated with better cause-speci c and overall survival [19]. More recently, Robin et al. reported that according to the National Cancer Database, less than 50% of cervical cancer patients diagnosed between 2004 and 2012 received a standard of care including EBRT and brachytherapy [20]. Therefore, to increase the utilization rate of brachytherapy, education and training on this modality have become increasingly important. However, the number of brachytherapy procedures performed by residents decreased over the period 2006-2007 to 2010-2011 according to the Accreditation Council of Graduate Medical Education nal resident case logs [21]. Moreover, the number of interstitial procedures decreased by 25% over the period. Thus, education and training on the interstitial procedure can be considered essential for residents.
Accumulating evidence suggests the e cacy of IC/IS brachytherapy for locally advanced cervical cancers. In a large-scale retrospective study (retroEMBRACE), Fokdal et al. reported that IC/IS brachytherapy resulted in signi cantly greater HR-CTV D 90 compared with intracavitary brachytherapy alone, with no difference in doses to organs at risk [22]. In their study, the 3-year local control rate for patients with an HR-CTV ≥ 30 cm 3 was signi cantly greater when they were treated with IC/IS brachytherapy. More recently, in a multicenter prospective observational study (EMBRACE-I), IC/IS brachytherapy was used in 43.0% of 1416 cervical cancers, and the overall 5-year local control was 92% [6]. In addition, a sub-group analysis of this cohort reported that IC/IS brachytherapy was used for 69.8% of the cases with bladder wall in ltration [23]. Consistent with these studies, we found that treatment plans with a tandem and ovoids without interstitial needles resulted in an insu cient dose to a bulky HR-CTV model (approximately 70 cm 3 ), whereas the addition of interstitial needles contributed to a HR-CTV D 90 greater than 6 Gy, regardless of applicator type (Table 1). It is worthy to note that an ongoing multicenter prospective interventional study (EMBRACE-II) employs strict dose constraints for the organs at risk, with the planning aim for the total equivalent dose in 2 Gy-fractions with an α/β ratio of 3 for D 2cc being set below 65 Gy and 80 Gy for the rectum and bladder, respectively [24]. Taken together, these data suggest high demand for the IC/IS technique in the near future. From this standpoint, the training simulator developed in the present study should contribute to the radiation oncology community by supporting education and dissemination of the IC/IS technique.
Optimization of the intratumoral needle location is important for broadening the therapeutic window of IC/IS brachytherapy [25]. Freehand needle application has the advantage of a high degree of freedom in needle position and direction, but the disadvantages of poor reproducibility and di culty in educating beginners. By contrast, the use of a template for needle application (e.g., Vienna, Utrecht, and Venezia applicator) provides high reproducibility at the cost of reduced degrees of freedom in needle direction [12][13][14]. To date, no clear indication has been established as to the case types most suited to either the freehand method or templates. In this study, the HR-CTV D 90 of the plan with freehand needles exceeded that with the Venezia template when the application was performed by the expert (Table 1, Fig. 3). By contrast, the HR-CTV D 90 of the Venezia-based plan outperformed the freehand-based plan when application was performed by the resident (Table 1, Fig. 3). These ndings indicate that the use of a template may be bene cial for residents, whereas freehand needle application by experts may be superior to template-guided needle application in speci c clinical situations. Future studies are needed to standardize IC/IS techniques, as well as to establish an educational program for this technique.
The training simulator developed in the present study can be easily replicated from the one-off aluminum mold, which will enable inter-practitioner, -modality, and -institute comparisons of IGABT simulation under standardized settings in the future. We can also use this simulator to examine improvements in the performance of the same practitioner or institute. Furthermore, the 3D-tumor printing pipeline reported in this study is applicable to the production of tumor phantoms tailored to individual patients, which could be used in pre-treatment or post-treatment evaluation of IGABT plans in routine clinical practice.

Conclusions
To support IC/IS brachytherapy training for bulky and irregular-shaped cervical cancers, we developed a training simulator consisting of a patient-derived soft silicon tumor phantom and an acrylic tube mimicking the vagina. The results of our feasibility test indicate that the tumor phantom is a good representation of a typical bulky and irregular-shaped cervical cancer to be treated with IGABT, and further evaluation of the training simulator is therefore warranted to con rm its educational value in IC/IS brachytherapy of locally advanced cervical cancer.

Consent for publication
The requirement for written consent for publication from participants was waived by the institutional ethical review committee of Gunma University Hospital because of the retrospective observational nature of the study.