Introduction: Vision loss remains a significant concern for patients undergoing radiation treatment for uveal melanoma. Herein, we identify predictors of vision loss after plaque brachytherapy at our academic medical center and, using this independent data set, evaluate a published vision prognostication tool. Methods: A retrospective chart review identified 252 posterior uveal melanoma patients treated with iodine-125 plaque brachytherapy from 2013 to 2018. Interval-censored nonparametric survival analysis estimated risk for moderate (<20/50) and severe (<20/200) vision loss at 3 and 5 years post-brachytherapy. Factors affecting vision loss were evaluated using interval-censored proportional hazards regression models. The prognostication tool was evaluated using a receiver operating curve and comparing predicted risk to observed risk. Results: Increased age, tumor height, and radiation dose to optic disc were associated with vision loss for patients with baseline VA >= 20/50. For those with baseline VA >= 20/200, risk factors included dose to optic disc, female sex, and baseline vision worse than 20/50. The vision prognostication tool demonstrated a moderate ability to differentiate patients with 3-year vision <20/200, with an area under the curve of 0.75. Conclusion: The vision prognostication tool was predictive for vision loss but generally overestimated risk, possibly reflecting evolving practice patterns including anti-VEGF use for radiation retinopathy.

Commercial autosegmentation has entered clinical use, however real-world performance may suffer in certain cases. We aimed to assess the influence of anatomic variants on performance. We identified 112 prostate cancer patients with anatomic variations (edge cases). Pelvic anatomy was autosegmented using three commercial tools. To evaluate performance, Dice similarity coefficients, and mean surface and 95% Hausdorff distances were calculated versus clinician-delineated references. Deep learning autosegmentation outperformed atlas-based and model-based methods. However, edge case performance was lower versus the normal cohort (0.12 mean DSC reduction). Anatomic variation presents challenges to commercial autosegmentation.

Purpose:Outside of randomized clinical trials, it is difficult to develop clinically relevant evidence-based recommendations for radiation therapy (RT) practice guidelines owing to lack of comprehensive real-world data. To address this knowledge gap, we formed the Learning from Analysis of Multicenter Big Data Aggregation consortium to cooperatively implement RT data standardization, develop software solutions for data analysis, and recommend clinical practice change based on real-world data analyzed. The first phase of this "Big Data" study aimed at characterizing variability in clinical practice patterns of dosimetric data for organs at risk (OARs) that would undermine subsequent use of large-scale, electronically aggregated data to characterize associations with outcomes. Evidence from this study was used as the basis for practical recommendations to improve data quality. Methods and Materials:Dosimetric details of patients with head and neck cancer treated with radiation therapy between 2014 and 2019 were analyzed. Institutional patterns of practice were characterized, including structure nomenclature, volumes, and frequency of contouring. Dose volume histogram (DVH) distributions were characterized and compared with institutional constraints and literature values. Results:Plans for 4664 patients treated to a mean plan dose of 64.4 ± 13.2 Gy in 32 ± 4 fractions were aggregated. Before implementation of TG-263 guidelines in each institution, there was variability in OAR nomenclature across institutions and structures. With evidence from this study, we identified a targeted and practical set of recommendations aimed at improving the quality of real-world data. Conclusions:Quantifying similarities and differences among institutions for OAR structures and DVH metrics is the launching point for next steps to investigate potential relationships between DVH parameters and patient outcomes.

Purpose/Objective(s)To assess the performance of commercially available autosegmentation solutions of pelvic anatomy in edge cases of anatomic variation among patients receiving definitive radiotherapy (RT) for prostate cancer.Materials/MethodsIn a single institution retrospective cohort of 830 patients with prostate cancer receiving definitive RT between 2011 and 2019, we identified 112 patients with anatomic variations seen on simulation CT imaging which may present challenging cases for precise delineation of anatomy, including reasons such as arthroplasty metal artifacts, extensive median lobe hypertrophy, “droopy” seminal vesicles, presence of urinary catheter, and history of transurethral resection of prostatic tissue. Three commercially available solutions for pelvic anatomy autosegmentation (AS) employing deep learning and model-based segmentation were applied to treatment planning CT scans to generate segmented volumes of the prostate, rectum, bladder and femoral heads in DICOM-RT format. To quantify the accuracy of software-generated contours, Dice similarity coefficients (DSC) and Hausdorff distances (HD) were calculated for each AS contour with a comparator manually segmented reference contour. Calculations were performed in R using the RadOnc package and manually segmented contours were approved by an expert radiation oncologist with 19 years of experience.ResultsA total of 112 patients were identified for inclusion after manual review of treatment planning CTs. Contours were generated using two atlas/model-based segmentation products and a deep learning segmentation method. The mean DSC for the prostate, bladder, rectum was calculated for each segmentation method (Table 1). Deep learning segmentation outperformed model-based methods for all structures with the highest mean DSC but still had significant disagreement with manually segmented structures. Hip arthroplasty in particular reduced overall performance more than other anatomical edge cases, followed by prostatic hypertrophy.ConclusionAnatomic edge cases present a challenging and relevant consideration in clinical implementation of autosegmentation software.

Abstract In a mixed‐vendor radiation oncology environment, it is advantageous if the department's treatment planning system (TPS) supports the linear accelerators of different vendors. In this publication beam data collection and modeling for the Versa HD linear accelerator in Monaco, Pinnacle, and Eclipse are discussed. In each TPS static field, Intensity‐Modulated Radiation Therapy (IMRT) step and shoot, and Volumetric‐Modulated Arc Therapy (VMAT) plans for flattened and flattening‐filter free photon beams of all available energies were evaluated for field sizes >3 × 3. To compare passing rates, identical beam model validation plans were calculated in each TPS. Eclipse, Monaco, and Pinnacle beam models passed validation measurements in homogeneous materials for a variety of treatment fields, including static, IMRT, and VMAT. In the case of Eclipse, the “dosimetric leaf gap” parameter was found to be critical for passing rates of VMAT plans. The source size parameter plays an important role as well for small fields. In the case of Pinnacle the multileaf collimator offset table needed to be optimized for better VMAT QA results. Each of the investigated treatment planning systems met the criteria to be used clinically in conjunction with Elekta Versa HD linear accelerators. It can be of great advantage to have the option to operate a TPS and linear accelerator from different vendors, as decisions surrounding linear accelerator or TPS purchases are very complicated and not just limited to technical considerations.

To assess the performance of computer algorithm(s) for autosegmentation of pelvic anatomy in anatomical edge cases among patients receiving radiation therapy for prostate cancer. In a single institution retrospective cohort of prostate adenocarcinoma patients receiving definitive radiation therapy between 2011 and 2018, we identified 110 patients with anatomical variation(s) identified on treatment planning CT imaging which may present challenging cases for precise delineation of anatomy, including reasons such as arthroplasty metal artifacts, extensive prostate median lobe hypertrophy, prior brachytherapy seeds, extensive prostatic calcifications, "droopy" seminal vesicles, presence of urinary catheter, and history of transurethral resection of prostatic tissue. Autosegmentation algorithms for male pelvic anatomy trained using commercial atlas-based autosegmentation and deep learning platforms were applied to treatment planning CT scans to generate autosegmented volumes of the prostate, seminal vesicles, penile bulb, rectum, bladder, and bilateral femoral heads in DICOM-RT format. To quantify the accuracy of software-generated contours, median Hausdorff distance (HD) and Dice similarity coefficients (DSC) were calculated based on manually segmented reference contours. Preliminary analysis of 6 patients with median lobe hypertrophy, "droopy" seminal vesicles, prostatic surface irregularity including extracapsular extension, presence of a Foley catheter, and prostatic median lobe protrusion was performed. Among structures, autosegmentation of the left and right femoral heads performed best (DSC of left femoral head = 0.92; interquartile range (IQR) = 0.03; DSC of right femoral head = 0.90; IQR = 0.05), whereas autosegmentation of the penile bulb performed worst (Dice similarity coefficient = 0.36; IQR = 0.12). Among different anatomical variant classes, autosegmentation of structures from cases with prostatic surface irregularity had better performance (DSC = 0.77; IQR = 0.47), while autosegmentation of structures from patients with "droopy" seminal vesicles performed the worst (DSC = 0.61; IQR = 0.24). Autosegmentation performance varies between structures and cases with anatomic variations. Better understanding the limitations of autosegmentation platforms with regards to performance on anatomical edge cases can lead us to building more robust segmentation algorithms for trustworthy implementation.

Large scale, real-world data has potential not only to augment clinical trial design and validation but also to improve radiotherapy (RT) plan quality and patient outcomes by characterizing and reducing practice variability. Assessment of RT clinical practice patterns is challenged by the complexity of non-standardized dosimetric data. This "Big Data" study from 5 collaborating institutions electronically aggregated organ at risk (OAR) dosimetric data of 4660 head and neck (HandN) cancer RT plans from 2014 to 2019. Statistical and machine learning methods were used to detail distributions of values and categorize evidence for sources and impact of inter-institutional variability that could undermine future efforts to pool dose volume histogram (DVH) data. Results were reviewed in the context of institutions' optimization constraints and literature recommendations in order to develop the following recommendations for LAMBDA members: · Adopt TG-263 nomenclature. · 13 structures were contoured on ≥ 50% of patients in ≥ 3/5 of institutions and we recommend this as a contour set to ensure complete datasets: brain, brainstem, spinal cord, eyes, cochleas, optic nerves, mandible, parotid/submandibular glands, oral cavity, esophagus, larynx, and constrictor muscles. At a minimum, contour those within 3 cm of PTVs. As per ASTRO's recent consensus paper, other OAR structures should be included based on disease site treated. If contouring submandibular and/or parotid glands, include both left and right structures if present. Given reported relationships between dysphagia and dose to individual components, separation of Musc_Constrict_S and Musc_Constrict_I is recommended versus Pharynx. · If using OAR-PTV volumes, contour the corresponding OAR volume. < = 2 institutions agreed on high dose constraints for SpinalCord, Brainstem, or Bone_Mandible; D0.03cc [Gy] is recommended (versus Max [Gy] or D0.1cc [Gy]), to ensure interoperability. Consider reducing constraint values for DVH metrics where real-world values are below literature set limits (e.g. esophagus LAMBDA consortium Mean 21 Gy Median [15 Q1,28 Q3] significantly lower than QUANTEC Mean <34 Gy p = 0.006 [0.003,0.17]) Kmeans clustering of inter vs intra-institutional volume variability coupled with quantile analysis of DVH values identified larynx, oral cavity, esophagus and constrictors as structures posing data quality risks to multi-institutional studies; contour practice assessment will be an essential next step. This multi-institutional Big Data study has identified patterns of HandN RT practice variation. Resultant clinical practice recommendations for LAMBDA members pave way for next steps to improve plan quality through standardization and facilitate future studies of dosimetric data and patient outcomes.

Purpose:To investigate the variation in I‐125 seed activities ordered by various clinics for their plaque brachytherapy cases under a standardized set of assumptions.Methods:A majority of the plaque programs in North America were contacted and a survey was designed to give a few standardized cases to allow inter‐comparison of seed activities ordered. Tumor dose, treatment duration, number of seeds, plaque, and tumor apex were held constant in order to reveal differences in prescription point, seed type, and seed activity.Results:While the survey is presently underway, preliminary results show alarmingly wide variations between centers. Differences up to 45% have been found with 15% differences being common.Conclusion:Though knowledge of the TG‐43 dose calculation formalism is common, a number of factors in the field of plaque brachytherapy lead to alarming differences in activity of I‐125 seeds being ordered for a given tumor. Knowledge of the present reality of widely varying treatment activities, and thus doses to tumor and normal structures, should serve as motivation for centers involved in this modality to review their programs with others in the community and share their experiences.

To compare pacemaker maximum doses calculated by TPS versus measurements by in vivo dosimeters in head and neck cancer patients with pacemakers located outside of the treatment volume. From July 2011 to July 2013, 6 patients with squamous cell carcinoma of the head and neck were treated with IMRT with a pacemaker device located in their left superior chest. All pacemakers were located near but outside of the planned radiation treatment volume. Each patient was treated to 70 Gy in 35 fractions to the primary tumor with bilateral cervical neck nodal volumes, including left level four. The pacemaker volume was contoured in the TPS covering the device but not including the wires from the device into the atrium or ventricles. Five out of the six patients' treatment plans were developed for both Tomotherapy and linear accelerator based IMRT; one patient had a linear accelerator IMRT plan only. Eclipse, Tomotherapy, and Pinnacle TPS were used. Dose constraints, for treatment planning purposes, on pacemakers were set to attempt to keep the total pacemaker maximum dose less than 5 Gy. All patients with a Tomotherapy plan were treated with that plan. On the first day of treatment, optically stimulated luminescence dosimeters (OSLDs) were placed on the skin over the pacemaker with or without a lead shield with 5mm bolus underneath, as a means of in vivo dosimetry. Analysis of the maximum dose measured by in vivo dosimetry was compared to the maximum dose calculated on the TPS. OSLDs measurements included dose from image guidance: cone-beam computed tomography or mega-voltage computed tomography. Six patients' plans and OSLD results were reviewed. The maximum pacemaker dose in our tomotherapy plans ranged from 1.54 - 3.58 Gy(mean: 2.28 Gy), the linear accelerator plans ranged from 1.76 - 7.1 Gy (mean: 3.89 Gy), and the OSLD doses ranged from 2.2-10.4 Gy (mean: 4.67 Gy). OSLD maximum doses were consistently measured higher than those calculated by the TPS with doses ranging from 127 - 468% higher. Using in vivo OSLD measurements as a surrogate for pacemaker dose showed TPS calculations consistently underestimate the maximum pacemaker dose. Based on these preliminary results, it is recommended that in vivo measurements be performed and compared to the TPS calculated maximum pacemaker dose in order to ensure manufacturer recommendations are not exceeded.

To determine whether rectal distention during IMRT treatment for prostate cancer is dosimetrically significant. From July 2011 to December 2012, 86 post-prostatectomy prostate cancer patients with localized disease were treated with adjuvant/salvage radiation to 70 Gy in 35 fx. Prior to simulation an enema was done to minimize rectal distention for treatment planning. Daily CBCT was utilized and the patient was lined up using the most apical surgical clips. If the visualized rectal volume overlapped the PTV by 1 cm or more in any axial plane, treatment was held and the patient was asked to empty their bowel. The patient was then re-imaged with CBCT to verify appropriate rectal positioning relative to the PTV and treatment was delivered. Occasionally patients were required to undergo repeated rounds of bowel emptying and CBCT. For our study, each CBCT with a distended rectum was contoured over the same cranio-caudal dimensions as the primary treatment plan and the contours from the CBCT were fused to the original treatment planning CT allowing us to compare the DVH as if the patient was treated with the distended rectum vs the actual treated rectal contours. Twenty-nine pts (33.7%) of 86 pts were asked to empty their rectum at least once during the course of treatment. For each patient, this occurred in a mean of 2.9 (8.3%) treatments during their course. DVH analysis of the rectal distension CBCTs was done on five of the patients with multiple rectal distension delays (mean of 7 delays; range, 4-12) with results indicated in the attached Table. Rectal emptying when rectal distension is noted on daily CBCT prior to radiation treatment is a clinically significant intervention that is easily implemented and more feasible than adaptive radiation treatment planning. This practice is associated with reduced rectal doses and likely decreased rectal toxicity.Digital Poster Abstract 1069; TableDosimetric results of rectal evacuationPre-evacuation CBCTPost-evacuation CBCTTreatment planV7024.1% (10.5 - 33.8)4.0% (0 - 22.6)0.2% (0-0.9)V6034.4% (20.6 - 46.0)12.4% (1.6 - 29.3)7.2% (2.3 - 14.1)V5042.0% (25.1 - 56.1)21.0% (5.1 - 44.0)16.2% (7.8 - 29.0)Rectal volume76.9 cc (24.1 - 170.7)46.3 cc (17.9 - 92.4)36.2 cc (18.6 - 58.0) Open table in a new tab

We report a series of patients treated with I-125 plaque brachytherapy for choroidal melanoma. The specific aims of the analysis were to evaluate local tumor control, overall and metastasis-free survival, loss of visual acuity versus loss of subjective vision, and describe the need for secondary enucleation. We identified 317 patients with choroidal melanoma managed by I-125 plaques from 1991 to 2011. The selected tumors had been prescribed 85 Gy to the prescription point per COMS protocol. We recorded tumor and treatment characteristics, mortality, rate of local recurrence (LR), development of metastatic disease, need and reason for secondary enucleation, changes in visual acuity, and subjective vision loss (new patient-reported loss between previous time point and current time point/total patients at current time point). Kaplan-Meier survival curves were used to calculate overall and metastasis-free survival. Log-rank analysis was used to determine factors associated with overall survival. Mean tumor thickness was 3.57 mm (range, 1-12 mm) and mean maximum basal tumor diameter was 11.00 mm (range, 4-20 mm). Ciliary body invasion was seen in 28/317 patients. Using a mean plaque size of 16.91 mm (range, 4-24 mm) the mean calculated dose of 87.64 Gy was given over a mean of 144.9 hours. One and 5-year overall survival were 98.7% and 77.4%, respectively. One and 5-year metastasis-free survival were 99.7% and 93.4%, respectively. Factors associated with overall survival (p < 0.05) included age at treatment, maximum basal tumor diameter, and retinal detachment. Five-year LR free survival was 97.8%. There were 20 secondary enucleations total. LR was cause for secondary enucleation in 10/317 cases (all LR were enucleated); 3/10 of these LR had pre-treatment invasion into the ciliary body. Pain was the main reason for secondary enucleation in 10/317 cases. Visual acuity in the treated eye of equal or better than 20/200 was seen in 87.9%, 77.3%, 54.1%, 46.7%, and 40% of patients pre-treatment, and 1, 5, 10, and 15 years post-treatment, respectively. Subjective vision loss (within treated eye) was reported in 45.0%, 31.0%, 22.3%, 18.2%, and 16.7% of patients at pre-treatment, 1, 5, 10, and 15 years follow-up. Curative intent I-125 plaque brachytherapy provides a high 5-year disease-free survival and low risk of LR. Enucleation was required in less than 6.5% of cases, with pain and local recurrence each accounting for half of the cases. Nearly one third of tumors requiring secondary enucleation due to LR had pre-treatment ciliary body invasion. Visual acuity showed a gradual decline up to 15 years post-treatment, with subjective vision loss noted predominantly in the first 5 years of follow-up.

The most commonly used method for specifying prescription dose for intracavitary cervical brachytherapy is Point A. Slight variants of the definition of Point A have evolved over time, and the historical definition of Point A is underspecified for 3D planning, leading to wide variation in placement of Point A between brachytherapy centers. The specific aim of this study was to determine the effect that this variation has on dose to the tumor and adjacent organs at risk (OAR). We examined 13 brachytherapy plans from 3 cervical cancer patients treated at our institution who each underwent 4 - 5 intracavitary high-dose rate tandem and ovoid placements. We recalculated each plan using equal dwell time loading at a standard dose of 600cGy to several alternate dose points: the contemporary definition of Point A (left, right, average), the original Manchester definition of Point A (also known as Point H) (left, right, average). We also simulated the effect of a cranio-caudal placement error of 2mm of these points. We measured the effect that dosing to these alternate points would have on actual Point A, GTV D90, and to normal tissues (D2cc of rectosigmoid and bladder). Actual dose delivered to Point A varied widely depending upon the prescription point used for a given implant and averaged between 559 - 680cGy. Dose to tumor, GTV D90, varied on average between 651 - 784cGy. The average distance from Point A to Point H was 7.3mm cephalad (range 2.3 - 14mm). Dosing to Point H averaged 12% higher than dosing to Point A. Compared with Point H, placement errors of Point A resulted in a larger variation in tumor dose (3.96% Pt A vs. 2.88% Pt H, p = 0.003), dose to rectum (3.96% Pt A vs 2.88% Pt H, p = 0.006), and dose to bladder (3.95% Pt A vs 2.89% Pt H, p = 0.007). Actual tumor dose delivered in cervical brachytherapy can vary substantially depending upon the choice and placement of prescription points. Prescribing to Point H appears to be more stable and shows less variation due to placement error, since it is a fixed distance from and usually further away from the intravaginal sources. Until MRI-based volumetric brachytherapy dosing becomes widely available, more rigorous definitions of historical prescription points are needed for more reliable 3D brachytherapy planning.

Purpose: The purpose of this study was to design and characterize a custom HDR applicator for the treatment of cancers in the region 0–5 mm, comparing it to the dosimetric properties of the Maxima R‐100 superficial x‐ray unit. Accelerator produced electrons of energy 4–6 MeV with the addition of a buildup bolus are in competition with superficial X‐rays and are very well suited for the treatment of shallow lesions as the maximum dose, Dmax, is brought to the surface. With this applicator, HDR may be a modality suited for treatment of Basal and squamous cell carcinomas, Kaposi's sarcoma, and Merkel cell carcinomas.Method and Materials: The superficial machine was characterized in terms of percent depth dose, beam profiles, and dose rate. Similar measurements were taken from a Varian Varisource HDR unit applying a newly designed copper HDR applicator. Both data sets were compared in terms of dosimetric properties and clinical efficacy. Results: The applicator alone had an unacceptably large dose inhomogeneity across the the HDR beam so a primary filter, modified filter, and secondary leadcollimator were designed and manufactured to help shape the dose into a profile similar to the superficial applicator. Beam flatness and width were nearly replicated but beam divergence was more pronounced for the HDR applicators. Conclusions: Percent depth dose measurements for the HDR applicator showed a dose fall‐off nearly identical to superficial x‐rays rather than the assumed inverse square dose fall‐off. Flatenning the HDR beam through this filtered applicator provides means to deliver uniform dose to surface lesions with superior treatment delivery while implementing the HDR billing codes.