Acute toxicity outcomes from salvage high-dose-rate brachytherapy for locally recurrent prostate cancer after prior radiotherapy.

Purpose: Isolated intra-prostatic recurrence of prostate adenocarcinoma after definitive radiotherapy presents a challenging clinical scenario. Salvage options require specialized expertise and pose risks of harm. This study aimed to present the acute toxicity results from using salvage high-dose-rate brachytherapy (sHDR-BT) as treatment in locally recurrent prostate cancer cases. Material and methods: Seventeen consecutive patients treated with sHDR-BT between 2019 and 2022 were evaluated retrospectively. Eligible patients had to have received curative intent prostate radiotherapy previously, and showed evidence of new biochemical failure. Evaluation with American Urological Association (AUA) and Common Terminology Criteria for Adverse Events (CTCAE) symptom assessments were performed for each case. Results: The median (inter-quartile range) age prior to salvage treatment was 68 (66-74) years. The median post-sHDR-BT follow-up time was 20 (13-24) months. At baseline prior to sHDR-BT, 8 (47%) patients had significant lower urinary tract symptoms. The median AUA score prior to sHDR-BT was 7 (3-18). Three (18%) patients reported irregular bowel function and 2 (12%) reported hematochezia prior to sHDR-BT. One-month post-treatment, the median AUA score was 13 (8-21, p = 0.21). Using CTCAE scoring, there were no cases of grade 2+ bowel or rectal toxicity, and no cases of grade 3+ urinary toxicity. Reported grade 2 urinary toxicities included 10 (59%) cases of bladder spasms, 2 (12%) cases of incontinence, 1 (6%) urinary obstruction, and 4 (24%) reports of urinary urgency. All these adverse events were temporary. Conclusions: This study adds to the existing literature by demonstrating that the acute toxicity profile of sHDR-BT is acceptable even without intra-operative magnetic resonance (MR) guidance or image registration. Further study is ongoing to determine long-term efficacy and toxicity of treatment.

Prospective validation of a machine learning model for applicator and hybrid interstitial needle selection in high-dose-rate (HDR) cervical brachytherapy.

PURPOSE: To Demonstrate the clinical validation of a machine learning (ML) model for applicator and interstitial needle prediction in gynecologic brachytherapy through a prospective clinical study in a single institution. METHODS: The study included cervical cancer patients receiving high-dose-rate brachytherapy using intracavitary (IC) or hybrid interstitial (IC/IS) applicators. For each patient, the primary radiation oncologist contoured the high-risk clinical target volume on a pre-brachytherapy MRI, indicated the approximate applicator location, and made a clinical determination of the first fraction applicator. A pre-trained ML model predicted the applicator and IC/IS needle arrangement using tumor geometry. Following the first fraction, ML and radiation oncologist predictions were compared and a replanning study determined the applicator providing optimal organ-at-risk (OAR) dosimetry. The ML-predicted applicator and needle arrangement and the clinical determination were compared to this dosimetric ground truth. RESULTS: Ten patients were accrued from December 2020 to October 2022. Compared to the dosimetrically optimal applicator, both the radiation oncologist and ML had an accuracy of 70%. ML demonstrated better identification of patients requiring IC/IS applicators and provided balanced IC and IC/IS predictions. The needle selection model achieved an average accuracy of 82.5%. ML-predicted needle arrangements matched or improved plan quality when compared to clinically selected arrangements. Overall, ML predictions led to an average total improvement of 2.0 Gy to OAR doses over three treatment fractions when compared to clinical predictions. CONCLUSION: In the context of a single institution study, the presented ML model demonstrates valuable decision-support for the applicator and needle selection process with the potential to provide improved dosimetry. Future work will include a multi-center study to assess generalizability.

Estimated absolute percentage of biopsied tissue positive for Gleason pattern 4 (eAPP4) in low dose rate prostate brachytherapy: Evaluation of prognostic utility in a large cohort.

PURPOSE: To determine whether a system to estimate Absolute Percentage of Biopsied Tissue Positive for Gleason Pattern 4 (eAPP4) is useful as a prognostication tool for patients with intermediate risk prostate cancer (IR-PCa) undergoing low dose rate prostate brachytherapy. METHODS: 497 patients with IR-PCa and known grade group 2 or 3 disease treated with low dose rate seed brachytherapy (LDR-BT) at a quaternary cancer centre were retrospectively reviewed. Prostate biopsies for each patient included Gleason grading with synoptic reporting that did not include percentage of pattern 4 disease found within the sample. Each core was assigned a grade grouping, however, and that was used with optimized estimates of percentage of pattern four disease to estimate eAPP4. Outcomes including cumulative incidence of recurrence (CIR), treatment of recurrent disease (RRX), and metastasis-free survival (MFS) were then reviewed and the prognostic value of eAPP4 evaluated. RESULTS: 428 (86%) patients had Gleason grade group 2 and 69 (14%) patients had Gleason grade group 3 disease. 230 (46%) patients had National Comprehensive Cancer Network (NCCN) favourable intermediate at baseline, while 267 (54%) of patients had NCCN unfavourable intermediate at baseline. Median follow-up was 7.3 (5.5-9.6) years. eAPP4 was predictive of CIR (p = 0.003), RRX (p = 0.003), or MFS (p = 0.001) events, while Gleason grade grouping alone was not. eAPP4 was strongest as a predictor for MFS when estimates of 30% (grade group 2) and 80% (grade group 3) were used [HR 1.07 (1.03-1.12); p = 0.001]. CONCLUSIONS: eAPP4 was strongly predictive of recurrence and metastasis-free survival in a large cohort of patients receiving LDR-BT treatment for IR-PCa. Treatment of future patients with IR-PCa could include the use of eAPP4 prognostication.

Assessment of tissue toxicity risk in breast radiotherapy using Bayesian networks.

PURPOSE: The purpose of this analysis is to predict worsening post-treatment normal tissue toxicity in patients undergoing accelerated partial breast irradiation (APBI) therapy and to quantitatively identify which diagnostic, anatomical, and dosimetric features are contributing to these outcomes. METHODS: A retrospective study of APBI treatments was performed using 32 features pertaining to various stages of the patient's treatment journey. These features were used to inform and construct a Bayesian network (BN) based on both statistical analysis of feature distributions and relative clinical importance. The target feature for prediction was defined as a measurable worsening of telangiectasia, subcutaneous tissue induration, or fibrosis when compared against the observed baseline. Parameter learning for the network was performed using data from the 299 patients included in the ACCEL trial and predictive performance was measured. Feature importance for the BN was quantified using a novel information-theoretic approach. RESULTS: Cross-validated performance of the BN for predicting toxicity was consistently higher when compared against conventional machine learning (ML) techniques. The measured BN receiver operating characteristic area under the curve was 0.960 ± $\,{\pm}\,$ 0.013 against the best ML result of 0.942 ± $\,{\pm}\,$ 0.021 using five-fold cross-validation with separate test data across 100 trials. The volume of the clinical target volume, gross target volume, and baseline toxicity measurements were found to have the highest feature importance and mutual dependence with normal tissue toxicity in the network, representing the strongest contribution to patient outcomes. CONCLUSIONS: The BN outperformed conventional ML techniques in predicting tissue toxicity outcomes and provided deeper insight into which features are contributing to these outcomes.

Corrigendum: Development of a Machine Learning Model for Optimal Applicator Selection in High-Dose-Rate Cervical Brachytherapy.

[This corrects the article DOI: 10.3389/fonc.2021.611437.].

MRI-TRUS registration methodology for TRUS-guided HDR prostate brachytherapy.

PURPOSE: High-dose-rate (HDR) prostate brachytherapy is an established technique for whole-gland treatment. For transrectal ultrasound (TRUS)-guided HDR prostate brachytherapy, image fusion with a magnetic resonance image (MRI) can be performed to make use of its soft-tissue contrast. The MIM treatment planning system has recently introduced image registration specifically for HDR prostate brachytherapy and has incorporated a Predictive Fusion workflow, which allows clinicians to attempt to compensate for differences in patient positioning between imaging modalities. In this study, we investigate the accuracy of the MIM algorithms for MRI-TRUS fusion, including the Predictive Fusion workflow. MATERIALS AND METHODS: A radiation oncologist contoured the prostate gland on both TRUS and MRI. Four registration methodologies to fuse the MRI and the TRUS images were considered: rigid registration (RR), contour-based (CB) deformable registration, Predictive Fusion followed by RR (pfRR), and Predictive Fusion followed by CB deformable registration (pfCB). Registrations were compared using the mean distance to agreement and the Dice similarity coefficient for the prostate as contoured on TRUS and the registered MRI prostate contour. RESULTS: Twenty patients treated with HDR prostate brachytherapy at our center were included in this retrospective evaluation. For the cohort, mean distance to agreement was 2.1 ± 0.8 mm, 0.60 ± 0.08 mm, 2.0 ± 0.5 mm, and 0.59 ± 0.06 mm for RR, CB, pfRR, and pfCB, respectively. Dice similarity coefficients were 0.80 ± 0.05, 0.93 ± 0.02, 0.81 ± 0.03, and 0.93 ± 0.01 for RR, CB, pfRR, and pfCB, respectively. The inclusion of the Predictive Fusion workflow did not significantly improve the quality of the registration. CONCLUSIONS: The CB deformable registration algorithm in the MIM treatment planning system yielded the best geometric registration indices. MIM offers a commercial platform allowing for easier access and integration into clinical departments with the potential to play an integral role in future focal therapy applications for prostate cancer.

Excessive waitlists and delays to treatment with low-dose-rate brachytherapy predict an increased risk of recurrence and metastases in intermediate-risk prostatic carcinoma.

PURPOSE: It has previously been shown that increased wait times for prostatectomy are associated with poorer outcomes in intermediate-risk prostatic carcinoma (PCa). However, the impact of wait times on PCa outcomes following low-dose-rate brachytherapy (LDR-BT) are unknown. METHODS AND MATERIALS: We retrospectively reviewed 466 intermediate-risk PCa patients that underwent LDR-BT at a single comprehensive cancer center between 2003 and 2016. Wait times were defined as the time from biopsy to LDR-BT. The association of wait times with outcomes was evaluated using Cox and Fine-Gray regression in both univariate and multivariate models. RESULTS: Median (interquartile range) follow-up and wait time for all patients were 8.1 (6.3-10.4) years and 5.1 (3.9-6.9) months, respectively. Among NCCN unfavourable intermediate-risk (UIR) patients (n = 170; 36%), increased wait times predicted both a greater cumulative incidence of recurrence [MHR = 1.01/month of wait time (95% CI: 1.00-1.03); P = 0.044] and metastases [MHR = 1.04/month of wait time (95% CI: 1.02-1.06); P < 0.001] in multivariate modeling. In NCCN favourable intermediate-risk (FIR) patients, there was no significant association between wait time and recurrence or metastases risk. Among all intermediate-risk patients, wait time was associated with an increase in the incidence of metastases [MHR = 1.03/month of wait time (95% CI: 1.02-1.05); P < 0.001], but not recurrence in multivariate models. There was no association between wait time and overall survival in the UIR, FIR, or all intermediate-risk cohorts. CONCLUSIONS: Resource constraints within this center's public healthcare system have contributed to waitlists exceeding 5-months in length. This study finds that patients with UIR PCa experience a 1% increase in the risk of recurrence and 4% increase in the risk of metastases with each additional month of delay in definitive disease management. Preventing such extended management delays in LDR-BT may improve disease-related outcomes in patients with PCa.

Development of a Machine Learning Model for Optimal Applicator Selection in High-Dose-Rate Cervical Brachytherapy.

Purpose: To develop and validate a preliminary machine learning (ML) model aiding in the selection of intracavitary (IC) versus hybrid interstitial (IS) applicators for high-dose-rate (HDR) cervical brachytherapy. Methods: From a dataset of 233 treatments using IC or IS applicators, a set of geometric features of the structure set were extracted, including the volumes of OARs (bladder, rectum, sigmoid colon) and HR-CTV, proximity of OARs to the HR-CTV, mean and maximum lateral and vertical HR-CTV extent, and offset of the HR-CTV centre-of-mass from the applicator tandem axis. Feature selection using an ANOVA F-test and mutual information removed uninformative features from this set. Twelve classification algorithms were trained and tested over 100 iterations to determine the highest performing individual models through nested 5-fold cross-validation. Three models with the highest accuracy were combined using soft voting to form the final model. This model was trained and tested over 1,000 iterations, during which the relative importance of each feature in the applicator selection process was determined. Results: Feature selection indicated that the mean and maximum lateral and vertical extent, volume, and axis offset of the HR-CTV were the most informative features and were thus provided to the ML models. Relative feature importances indicated that the HR-CTV volume and mean lateral extent were most important for applicator selection. From the comparison of the individual classification algorithms, it was found that the highest performing algorithms were tree-based ensemble methods - AdaBoost Classifier (ABC), Gradient Boosting Classifier (GBC), and Random Forest Classifier (RFC). The accuracy of the individual models was compared to the voting model for 100 iterations (ABC = 91.6 ± 3.1%, GBC = 90.4 ± 4.1%, RFC = 89.5 ± 4.0%, Voting Model = 92.2 ± 1.8%) and the voting model was found to have superior accuracy. Over the final 1,000 evaluation iterations, the final voting model demonstrated a high predictive accuracy (91.5 ± 0.9%) and F1 Score (90.6 ± 1.1%). Conclusion: The presented model demonstrates high discriminative performance, highlighting the potential for utilization in informing applicator selection prospectively following further clinical validation.

A modern mold room: Meshing 3D surface scanning, digital design, and 3D printing with bolus fabrication.

PURPOSE: This case series represents an initial experience with implementing 3-dimensional (3D) surface scanning, digital design, and 3D printing for bolus fabrication for patients with complex surface anatomy where traditional approaches are challenging. METHODS AND MATERIALS: For 10 patients requiring bolus in regions with complex contours, bolus was designed digitally from 3D surface scanning data or computed tomography (CT) images using either a treatment planning system or mesh editing software. Boluses were printed using a fused deposition modeling printer with polylactic acid. Quality assurance tests were performed for each printed bolus to verify density and shape. RESULTS: For 9 of 10 patients, digitally designed boluses were used for treatment with no issues. In 1 case, the bolus was not used because dosimetric requirements were met without the bolus. QA tests revealed that the bulk density was within 3% of the reference value for 9 of 12 prints, and with more judicious selection of print settings this could be increased. For these 9 prints, density uniformity was as good as or better than our traditional sheet bolus material. The average shape error of the pieces was less than 0.5 mm, and no issues with fit or comfort were encountered during use. CONCLUSIONS: This study demonstrates that new technologies such as 3D surface scanning, digital design and 3D printing can be safely and effectively used to modernize bolus fabrication.

Verification of Acuros XB dose algorithm using 3D printed low-density phantoms for clinical photon beams.

The transport-based dose calculation algorithm Acuros XB (AXB) has been shown to accurately account for heterogeneities primarily through comparisons with Monte Carlo simulations. This study aims to provide additional experimental verification of AXB for clinically relevant flattened and unflattened beam energies in low density phantoms of the same material. Polystyrene slabs were created using a bench-top 3D printer. Six slabs were printed at varying densities from 0.23 to 0.68 g/cm3 , corresponding to different density humanoid tissues. The slabs were used to form different single and multilayer geometries. Dose was calculated with Eclipse™ AXB 11.0.31 for 6MV, 15MV flattened and 6FFF (flattening filter free) energies for field sizes of 2 × 2 and 5 × 5 cm2 . EBT3 film was inserted into the phantoms, which were irradiated. Absolute dose profiles and 2D Gamma analyses were performed for 96 dose planes. For all single slab configurations and energies, absolute dose differences between the AXB calculation and film measurements remained <3% for both fields in the high-dose region, however, larger disagreement was seen within the penumbra. For the multilayered phantom, percentage depth dose with AXB was within 5% of discrete film measurements. The Gamma index at 2%/2 mm averaged 98% in all combinations of fields, phantoms and photon energies. The transport-based dose algorithm AXB is in good agreement with the experimental measurements for small field sizes using 6MV, 6FFF and 15MV beams adjacent to various low-density heterogeneous media. This work provides preliminary experimental grounds to support the use of AXB for heterogeneous dose calculation purposes.

Modulated photon radiotherapy (XMRT): an algorithm for the simultaneous optimization of photon beamlet energy and intensity in external beam radiotherapy (EBRT) planning.

This is a proof of principle study on an algorithm for optimizing external beam radiotherapy in terms of both photon beamlet energy and fluence. This simultaneous beamlet energy and fluence optimization is denoted modulated photon radiotherapy (XMRT). XMRT is compared with single-energy intensity modulated radiotherapy (IMRT) for five clinically relevant test geometries to determine whether treating beamlet energy as a decision variable improves the dose distributions. All test geometries were modelled in a cylindrical water phantom. XMRT optimized the fluence for 6 and 18 MV beamlets while IMRT optimized with only 6 MV and only 18 MV. CERR (computational environment for radiotherapy research) was used to calculate the dose deposition matrices and the resulting dose for XMRT and IMRT solutions. Solutions were compared via their dose volume histograms and dose metrics, such as the mean, maximum, and minimum doses for each structure. The homogeneity index (HI) and conformity number (CN) were calculated to assess the quality of the target dose coverage. Complexity of the resulting fluence maps was minimized using the sum of positive gradients technique. The results showed XMRT's ability to improve healthy-organ dose reduction while yielding comparable coverage of the target relative to IMRT for all geometries. All three energy-optimization approaches yielded similar HI and CNs for all geometries, as well as a similar degree of fluence map complexity. The dose reduction provided by XMRT was demonstrated by the relative decrease in the dose metrics for the majority of the organs at risk (OARs) in all geometries. Largest reductions ranged between 5% to 10% in the mean dose to OARs for two of the geometries when compared with both single-energy IMRT schemes. XMRT has shown potential dosimetric benefits through improved OAR sparing by allowing beam energy to act as a degree of freedom in the EBRT optimization process.

Characterizing 3D printing in the fabrication of variable density phantoms for quality assurance of radiotherapy.

PURPOSE: To present characterization, process flow, and applications of 3D fabricated low density phantoms for radiotherapy quality assurance (QA). MATERIAL AND METHODS: A Rostock 3D printer using polystyrene was employed to print slabs of varying relative electron densities (0.18-0.75). A CT scan was used to calibrate infill-to-density and characterize uniformity of the print. Two printed low relative density rods (0.18, 0.52) were benchmarked against a commercial CT-electron-density phantom. Density scaling of Anisotropic Analytical Algorithm (AAA) was tested with EBT3 film for a 0.57 slab. Gamma criterion of 3% and 3 mm was used for analysis. RESULTS: 3D printed slabs demonstrated uniformity for densities 0.4-0.75. The printed 0.52 rod had close agreement with the commercial phantom. Dosimetric comparison for 0.57 density slab showed >95% agreement between calculation and measurements. CONCLUSION: 3D printing allows fabrication of variable density phantoms for QA needs of a small clinic.