Patient's specific integration of OAR doses (D2 cc) from EBRT and 3D image‐guided brachytherapy for cervical cancer

Abstract The objective of this study was to assess the recommended DVH parameter (e.g., D2 cc) addition method used for combining EBRT and HDR plans, against a reference dataset generated from an EQD2‐based DVH addition method. A revised DVH parameter addition method using EBRT DVH parameters derived from each patient's plan was proposed and also compared with the reference dataset. Thirty‐one biopsy‐proven cervical cancer patients who received EBRT and HDR brachytherapy were retrospectively analyzed. A parametrial and/or paraaortic EBRT boost were clinically performed on 13 patients. Ten IMRT and 21 3DCRT plans were determined. Two different HDR techniques for each HDR plan were analyzed. Overall D2 cc and D0.1 cc OAR doses in EQD2 were statistically analyzed for three different DVH parameter addition methods: a currently recommended method, a proposed revised method, and a reference DVH addition method. The overall D2 ccEQD 2 values for all rectum, bladder, and sigmoid for a conformal, volume optimization HDR plan generated using the current DVH parameter addition method were significantly underestimated on average −5 to −8% when compared to the values obtained from the reference DVH addition technique (P < 0.01). The revised DVH parameter addition method did not present statistical differences with the reference technique (P > 0.099). When PM boosts were considered, there was an even greater average underestimation of −8~−10% for overall OAR doses of conformal HDR plans when using the current DVH parameter addition technique as compared to the revised DVH parameter addition. No statistically significant differences were found between the 3DCRT and IMRT techniques (P > 0.3148). It is recommended that the overall D2 cc EBRT doses are obtained from each patient's EBRT plan.


| INTRODUCTION
Integration of concomitant chemotherapy, external beam radiotherapy (EBRT), and intracavitary brachytherapy (BT) is the standard of care in the curative management of locally advanced cervical cancer. 1 Using a BT boost is linked with improved pelvic control 2 and overall survival. 2,3 The first use of BT for the treatment of cervical cancer dates back to 1903. 4 The use of three-dimensional (3D) imaging techniques, such as computerized tomography (CT) and magnetic resonance imaging (MRI), have been rapidly replacing planar x-ray imaging in BT treatment planning. This follows the recommendations of the Groupe Europ een de Curieth erapie-European Society for Therapeutic Radiology and Oncology (GEC-ESTRO), [5][6][7] the American Brachytherapy Society (ABS), 8,9 EMBRACE (An intErnational study on MRI-guided BRachytherapy in locally Advanced CErvical cancer) protocol, 10 and a recent International Commission on Radiation Units and Measurements (ICRU) Report #89. 11 Volumetric dose parameters for targets and organs-at-risk (OARs) were introduced and used, allowing clinicians to customize isodose lines with the goal of achieving maximal coverage of the high-risk clinical target volume (HR-CTV) while irradiating OARs as little as possible. These adaptive, conformal BT approaches have resulted in significantly improved clinical outcomes. 12 Volumetric OAR dose constraints, such as the minimal dose of the 2 cc of normal tissue with the highest dose (D2cc) or D0.1 cc, have been investigated [13][14][15] as an alternative to conventional rectum and bladder point doses. These alternatives originated from the ICRU Report #38, 16 and are mainly applicable to Point A-based BT planning techniques. In order to integrate overall volumetric OAR doses (D2 cc and D0.1 cc) from EBRT and BT, it was recommended that the EBRT and BT doses be added even though the location of given hot-spots (D2 cc or D0.1 cc regions) may not be identical for each of the plans. This was initially called a "worst case assumption" in the GEC-ESTRO recommendation, 7 but a worse case would occur due to intra-fraction organ or applicator motion. The adopted EMBRACE protocol phrase for this is "DVH parameter addition". 10 In this DVH parameter addition technique, the EBRT component dose distributions (at least for the volumetric OAR parameters (D2 cc and D0.1 cc)), are assumed to be completely uniform EBRT prescription doses following the recommendations of the EMBRACE protocol. 5,7,10 There have been efforts to accurately estimate overall doses from EBRT and HDR BT plans 17,18,19 but he previous studies were performed using either a phantom study 17 or a dosimetric planning study with no statistical analysis for six or fewer patients, 18,19 and they did not present a practical approach on how to estimate the overall OAR doses (e.g., D2 cc EQD2 ) without exporting and processing dose DICOM files (dose distribution (DVH EQD2 ) addition) or using DIR-based DVH analysis. In this study, we present a practical revised DVH parameter addition method where the volumetric OAR parameters (e.g., D2 cc) are simply obtained from each patient's EBRT plan, instead of assuming a completely uniform EBRT prescription dose.
The proposed, revised DVH parameter addition method was compared with the current DVH parameter addition method that has been used in the overall dose integration framework of GEC-ESTRO guidelines 7 and the EMBRACE protocol, 10 and it assumes the com-  8,9 The prescription dose was 33-36 Gy in 5-7 fractions, typically 5.5 Gy 9 5 fractions or 7 Gy 9 4 fractions). The clinical Point A plans were generated on 3 Tesla T2-and T1-weighted MRI data sets 20 (MAGNETOM Trio TM , Siemens Medical System Inc., Erlangen, Germany). A staff physician contoured the bladder, rectum, and sigmoid structures using T2weighted MR images. 11 A T&O applicator was reconstructed (digitized) on T1-weighted MR images. 6 The details of the HDR workflow, imaging, and planning have been previously described. [21][22][23][24] An adaptive/conformal volume optimization HDR plan was retrospectively created for each clinical Point A plan through a hybrid-inverse optimization process that includes a combination of an inverse optimization and manual forward planning. The hybridinverse optimization process 21,22,25,26 includes three main steps: (a) generate a conventional Point A plan, (b) set dose-volume objective constraints for inverse optimization based upon the resulting DVH parameters, and (c) perform final dose shaping using graphical optimization based upon DVH parameters and isodose lines. As a last step, a physician reviews the isodose lines for each slice on coronal, sagittal, and axial views. A patient's initial EBRT plan isodose lines, PA boost, and PM boost are depicted in Fig. 1 Two different approaches to assess overall OAR DVH parameters (e.g., D2 cc and D0.1 cc) from EBRT and each HDR plan were tested. The first is the recommended GEC-ESTRO 7 and EMBRACE protocol 10 DVH parameter addition technique where a completely uniform EBRT prescription dose is assumed. EBRT DVH parameters (e.g., D2 cc) were assumed to receive the full EBRT prescription dose from the initial EBRT plan and PA boost but receive no additional doses from the PM boost due to its central block. A 4 cm central block was used for all EBRT PM boost plans. The second method is a revised DVH parameter addition technique where EBRT DVH parameters (e.g., D2 cc) are obtained from each patient's EBRT plan. In both approaches, their physical D2 cc and D0.1 cc parameters were converted into EQD2-based values (D2cc EQD2 and D0.1cc EQD2 ) according to a linear-quadratic cell survival model following GEC-ESTRO guidelines 5,7 and the ICRU Report #89. 11 The a/ b ratio of 3 and repair halftime ðT 1=2 Þ of 1.5 hr were used. 11 Here N, d, and g represent a fraction number, a dose per fraction, and an incomplete repair function that is 1 for HDR. Afterward, the DVH parameters (e.g., D2 cc) in EQD2 were added for each EBRT and HDR plan. Both approaches can be simply done using an Excel spreadsheet (Microsoft Corporation, Redmond, WA, USA) available as a template on the American Brachytherapy Society website (www.americanbrachytherapy.org).

2.C | Integrated, single EQD2-based DVH as a reference dataset
In order to test these two different approaches, an integrated single EQD2-based, differential DVH was generated as a reference dataset.
This was done through three steps: (a) each physical dose map (i.e., dose DICOM file) was converted into an EQD2 dose map to account for the different fractionation schemes between the EBRT and HDR BT plans, (b) a differential DVH was generated from each EQD2 dose map that is EQD2-based, differential DVH (DVH EQD2 ), and (c) all differential DVH EQD2 were combined to create a single, integrated differential DVH EQD2 . The radiobiological plan evaluation tool, RadioBioEval, was developed in-house as a stand-alone software application in order to convert physical dose maps in DICOM format from EBRT treatment planning system (TPS) (Pinnacle, Philips Healthcare, Inc.,) and HDR TPS (BrachyVision, Varian Medical System, Inc.) into EQD2 dose maps, to generate a single, differential DVH EQD2 from EQD2 dose maps of EBRT and HDR plans and to evaluate an overall D2 cc EQD2 , D0.1 cc EQD2 , and gEUD EQD2 (see Otherwise, an a/b ratio of 3 is used. In this study, the a/b ratio of 3 was used for all OARs. In this demonstration case, a composite EBRT

2.D | EQD2-based, generalized EUD (gEUD EQD2 ) as an additional plan evaluation metric
Currently, the GEC-ESTRO working group, 5,7 EMBRACE protocol, 10 and the ICRU report #89 11 all recommend assuming maximal dosevolume parameters. D2cc values sufficiently represent each OAR's dose distribution including whole DVH. As an additional plan evaluation metric, the use of an EQD2-based, generalized equivalent uniform dose (gEUD EQD2 ) was proposed. The gEUD EQD2 was obtained from the integrated, differential DVH EQD2 that was described in the previous section. The gEUD EQD2 is determined by solving the following eq. (2): where n is a volume effect parameter, and EQD2 i is the differential dose bin obtained from a single integrated, differential DVH EQD2 .
The n values of the rectum, bladder, and sigmoid were 0.23, 29 0.5, 30 and 0.17, 30 respectively, and were based upon the available literature. The values of gEUD EQD2 were compared with the values of current OAR plan evaluation metric, D2cc EQD2 .

2.E | Statistical correlation analysis
The percent differences of the two different DVH parameter addition techniques were statistically analyzed in comparison with the reference dataset. The statistical differences between the two were also measured. The impact of EBRT techniques such as 3D In-house radiobiological evaluation tool (RadioBioEval) for integrated EBRT and each HDR brachytherapy plans through which physical EBRT and HDR DICOM dose map plans (solid lines on DVH Graph) are converted into EQD2-dose maps and DVHs (dashed lines). In this demonstration case, a composite EBRT plan with EBRT boost and three HDR plans of fraction #1-#3 were imported.

| RESULTS AND DISCUSSION
The overall D2cc EQD2 parameters for the rectum, bladder, and sigmoid that were obtained from a revised DVH parameter addition technique presented no statistical differences (P > 0.0981) with the reference dataset values regardless of conformal, volume optimization, and Point A HDR plans (see Table 1). Gy EQD2 ), respectively (see Table 2). One way to eliminate the complexity of dose integration of EBRT and BT is using EBRT boosts as an alternative to BT boost.
Pioneering studies have investigated the feasibility of using IMRT or intensity-modulated proton therapy (IMPT) boosts. 36

CONF LICT OF I NTEREST
The authors declare that there is no conflict of interest.