[This corrects the article DOI: 10.3389/fonc.2025.1695468.].

PURPOSE:To evaluate the efficacy and safety of salvage brachytherapy (SBT) as a curative treatment option for patients with locally recurrent prostate cancer after prior radiation therapy. MATERIALS AND METHODS:Between 2009 and 2023, we analyzed 108 prostate cancer patients with local recurrent after radiotherapy. SBT was performed as a sole treatment using either HDR (high dose rate) brachytherapy with a total dose of 30 Gy in three fractions (78.0 Gy EQD2 α/β=3 Gy) or PDR (pulse dose rate) brachytherapy with a total dose of 60 Gy in two sessions (71.5 Gy EQD2 α/β=3 Gy), with a three-week interval between sessions in both regimens. RESULTS:Median follow-up was 41 months (range 6-173). Median interval between initial radiation and salvage brachytherapy was 66 months (range 18-214). The 5-year cumulative incidence of local recurrence was 17.0%. Local relapse was observed in 15 patients (13.8%). The 2- and 5-year biochemical disease-free survival according to the Phoenix definition were 83.6% and 59.9%. The 2- and 5-year progression-free survival were 77.0% and 53.2%; overall survival rates were 96.9% and 87.4%, respectively. Grade ≥ 3 urinary toxicity was uncommon, with 2/93 (2.1%) patients requiring transurethral resection and 6/93 (6.5%) long-term catheterization. Most patients had grade 0-1 incontinence; grade 2 and grade ≥ 3 occurred in 13/79 (16.5%) and 7/79 (8.8%) at 24 months. Proctitis was rare; no grade ≥ 3. CONCLUSIONS:Salvage brachytherapy offers an effective and well-tolerated curative option for patients with locally recurrent prostate cancer after prior radiation therapy.

BACKGROUND:Intraoperative cone-beam computed tomography (CBCT) provides a valuable option for accurate three-dimensional applicator positioning in gynecologic brachytherapy, but is associated with radiation exposure and increased intervention time especially in case of repeated CBCT imaging being required for creating a sufficient implant arrangement. PURPOSE:To reduce the need for multiple CBCT scans for corresponding applicator verification, this work proposes two methods for needle path navigation, including corrections of potential bending in situ, by combining infrared tracking with planar x-ray imaging for enabling accurate intraoperative needle guidance. METHODS:An examined 200 mm brachytherapy needle was rigidly mounted on an infrared-reflective tracking tool to enable real time tracking. Two planar x-ray images, acquired from varying distinct angles, were used to determine the exact 3D position of the needle tip region via backprojection. A spline was fitted through the obtained coordinates to reconstruct the full needle path. Based on this, only a single initial CBCT scan was required to visualize the predicted needle path within this scan. Additionally, a second approach for needle prediction was presented focusing on only one planar x-ray image by incorporating prior needle bending information from the initial CBCT scan. Both methods were evaluated in preclinical studies and validated against a corresponding ground-truth obtained from CBCT. RESULTS:The proposed method considering two planar x-ray images successfully reconstructed the needle path with deviations of less than 1 mm from the CBCT reference scan, when using at least 20° offset between the x-ray image acquisitions. The single-scan approach, using prior bending information, yielded promising results with deviations at the tip of below 1.3 mm. CONCLUSIONS:Both described methods demonstrated their feasibility in preclinical studies, showing potential to improve and accelerate clinical implantation workflows by means of needle navigation in the future.

OBJECTIVE:To report clinical outcomes and toxicity of interstitial brachytherapy (IBT), with or without external beam radiotherapy, in a predominantly frail, multimorbid cohort of patients with locally advanced vulvar cancer, analyzed according to treatment intent and clinical setting. METHODS:We analyzed 47 patients with vulvar malignancies treated between 1998 and 2024 with pulse dose rate-IBT, either postoperatively or with definitive intent. Fifteen patients received pulse dose rate-IBT alone (median total dose of 57.2 Gy, range, 44.7-67.2), while 32 patients underwent combined external beam radiotherapy followed by an IBT boost (median total dose of 61.4 Gy; range, 50.7-68.1). Median age was 67 years (range, 38-90). Survival outcomes were estimated using Kaplan-Meier methods. RESULTS:Median follow-up was 2.2 years (range, 4-210). Two- and five-year overall survival rates were 58.2% and 45.7%, and disease-free survival rates were 62.5% and 53.1%, respectively. Concurrent chemotherapy was administered to 22/47 (47%) patients and was not completed in 7/22 (32%). In a predefined subgroup treated with curative intent for primary vulvar cancer, five-year overall survival, disease-free survival, and cumulative local recurrence rate were 61.7%, 66.7%, and 29.3%, respectively. Severe toxicity (≥grade 3), including ulceration, occurred in 14.2% of patients. CONCLUSION:In this real-world cohort, interstitial brachytherapy was feasible with a low rate of severe toxicity. Despite a frail, multimorbid patient population, acceptable local control was achieved. With appropriate patient selection, IBT may represent a viable option to improve local control while limiting mucosal toxicity.

Abstract Purpose Deep inspiration breath hold (DIBH) is widely used to reduce cardiac dose in left-sided breast radiotherapy. However, the reliability of DIBH at CT simulation, has not been systematically evaluated despite the importance of CT simulation as the basis for treatment planning. We investigated the clinical impact of surface-guided breathing training on DIBH reliability and cardiac dose sparing. Methods In this retrospective cohort study, 115 breast cancer patients undergoing DIBH CT simulation were analyzed before and after implementation of surface-guided breathing training using an SGRT system. DIBH performance was quantified by partial lung volume increase (PLVI) and breath-hold variability. Correlations between external surface motion and internal DIBH depth were assessed. Cardiac dose metrics were compared, with a subgroup analysis for patients receiving lymphatic drainage area (LDA) irradiation. Results Prior to training, insufficient DIBH CTs—including negative PLVI cases—were observed. After training implementation, all CT scans were acquired in intended breath-hold states, eliminating insufficient DIBH acquisitions. Mean PLVI increased by ~ 9%, and breath-hold variability decreased by ~ 26%. Sternum motion correlated most strongly with PLVI, with multiregion surface monitoring yielding the highest internal correlation. These improvements were reflected in significant reductions in cardiac dose, particularly in the LDA subgroup, with selected heart dose metrics reduced up to 96%, while maintaining target coverage. Conclusions DIBH at CT simulation with surface-guided breathing training proves to be reliable. The training prevents planning CTs in wrong breathing states and thus enables clinically meaningful cardiac dose reduction in breast radiotherapy.

IntroductionLarge language models (LLM) have shown great potential in clinical decision support and medical education. GPT-5 is a novel LLM system that has been specifically marketed towards oncology use. This study comprehensively benchmarks GPT-5 for the field of radiation oncology.MethodsPerformance was assessed using two complementary benchmarks: (i) the American College of Radiology Radiation Oncology In-Training Examination (TXIT, 2021), comprising 300 multiple-choice items, and (ii) a curated set of 60 authentic radiation oncologic vignettes representing diverse disease sites and treatment indications. For the vignette evaluation, GPT-5 was instructed to generate structured therapeutic plans and concise two-line summaries. Four board-certified radiation oncologists independently rated outputs for correctness, comprehensiveness, and hallucinations. Inter-rater reliability was quantified using Fleiss’ κ. GPT-5–14 results were compared to published GPT-3.5 and GPT-4 baselines.ResultsOn the TXIT benchmark, GPT-5 achieved a mean accuracy of 92.8%, outperforming GPT-4 (78.8%) and GPT-3.5 (62.1%). Domain-specific gains were most pronounced in dose specification and diagnosis. In the vignette evaluation, GPT-5’s treatment recommendations were rated highly for correctness (mean 3.24/4, 95% CI: 3.11–3.38) and comprehensiveness (3.59/4, 95% CI: 3.49–3.69). Hallucinations were rare, flagged in 10.0% of all individual reviewer assessments (24 of 240), and no patient case reached majority consensus for their presence. Inter-rater agreement was low (Fleiss’ κ 0.083 for correctness), reflecting inherent variability in clinical judgment. Errors clustered in complex scenarios requiring precise trial knowledge or detailed clinical adaptation.DiscussionGPT-5 clearly outperformed prior model variants on the radiation oncology multiple-choice benchmark. Although GPT-5 exhibited favorable performance in generating real-world radiation oncology treatment recommendations, correctness ratings indicate room for further improvement. While hallucinations were infrequent, the presence of substantive errors underscores that GPT-5-generated recommendations require rigorous expert oversight before clinical implementation. In addition, considerable inter-rater variability highlights the challenge of achieving consistent expert evaluation.

Background and purpose:Managing respiratory motion is crucial for computed tomography (CT) in radiotherapy. Clinical surrogate systems and visual coaching can exhibit inaccuracies due to table motion or sag, limiting effectiveness in deep inspiration breath-hold (DIBH) CT and four-dimensional CT (4DCT). This study evaluated a motion-corrected novel surrogate and feedback system to improve DIBH and 4DCT breathing quality. Materials and methods:The feedback system was evaluated in phantom tests for table sag compensation under incremental loads (up to 104 kg). For patients, DIBH plateau stability metrics (n = 2) and baseline shifts in 4DCT (n = 3) were assessed. In a volunteer study (n = 10), audio and visual feedback were compared in DIBH and 4DCT scenarios, assessing breath-hold stability and breathing regularity, respectively. Results:In phantom measurements, the impact of table sag on the breathing signal was effectively reduced at maximum load, with baseline shifts limited to -0.2 mm, compared to up to -5 mm in clinical systems. In patients, the system improved DIBH signal stability, as reflected in the plateau drift (-0.2 mm/s vs. -0.8 mm/s), and substantially reduced baseline shifts in 4DCT (-0.2 ± 0.2 mm) when compared to clinical systems (-1.7 ± 0.3 mm). Volunteer tests demonstrated improved DIBH reproducibility with visual feedback (standard deviation: 0.5 mm vs. 1.1 mm with audio feedback). In 4DCT scenarios, visual feedback unified irregular breathers but offered no consistent improvement over audio guidance. Conclusion:The novel system compensated for table motion in phantom, patient, and volunteer measurements. In feedback scenarios, it performed well in DIBH, while its performance in 4DCT requires further optimization.

Purpose:Model-based dose calculation algorithms (MBDCAs) are increasingly applied in brachytherapy, but their considerations in dose-response analyses is still lacking. This study aimed to assess correlations between both TG-43 and MBDCA dosimetry and reports on clinical outcomes for oral cavity brachytherapy. Methods:We considered 158 patients with oral cavity cancer treated in our institution between 2012 and 2021. Survival outcomes and toxicity (soft tissue necrosis, osteoradionecrosis, mucositis, xerostomia) were reported for a median follow-up of 80 months (2-152 months). All clinical, TG-43 based treatment plans were re-calculated using a MBDCA integrated into our planning system. Differences considering several target volume, tissue, and bone dose parameters were evaluated. Parameter correlations with clinical outcomes and thresholds associated with increased toxicity were investigated. Results:Cumulative local recurrence, soft-tissue necrosis, osteoradionecrosis, mucositis, and xerostomia rate were 21 %, 22 %, 28 % and 79 % considering all patients. Substantial differences between MBDCA and TG-43 were observed, especially regarding high-dose areas with changes up to 19 %. A number of dose-toxicity correlations were observed, as for osteoradionecrosis (1.6 % vs. 10.3 % for bone D2ccm ≥ 59.3 Gy), soft tissue necrosis (16 % vs. 32 % for tissue D5ccm ≥ 87.7 Gy), and local recurrence (13 % vs. 25 % for dose non-uniformity ratio < 0.29) (using the MBDCA). Target volumes ≥ 10.2-11.8ccm were associated with increased rates of soft tissue necrosis and mucositis. Dosimetric thresholds and correlations differed between TG-43 and MBDCA. Conclusion:For oral cavity brachytherapy, several important dosimetric thresholds associated with increased toxicity were determined. MBDCAs for corresponding dose calculations should be advanced and used in clinical practice.

Introduction: Postoperative radiotherapy is generally recommended for pT1/2pN0 squamous cell carcinoma of the oral cavity (OSCC) if risk factors are present. Prospective studies are missing. Interventional radiotherapy offers a precise dose application. In this paper we analyze long-term efficacy and toxicity in a large single-center cohort of patients with early OSCC receiving sole postoperative brachytherapy (BT). Material and methods: From 1998 to 2023, 217 patients were postoperatively treated with sole BT in our institute. The median follow-up was 110 months (range: 2-316). The primary objective was local control. Secondary outcomes were overall survival, cancer specific survival, and toxicity. Results: The local recurrence rates for 12, 24, and 60 months were 7.1 %, 9.1 %, and 12.6 %. The disease-free survival was 89.7 %, 86.1 %, and 79.3 %. The overall survival rates at 12, 24, and 60 months were 94.4 %, 89.6 %, and 77.9 %. The cancer-specific survival was 97.1 %, 96.6 %, and 92.9 %, respectively. At two years, the rate of regional recurrence was 8.3 %. Patients without neck dissection had a significantly increased risk for lymph node recurrence (p = 0.025). Side effects >= grade 3 were seen in 14 % (30/217). 17 % (37/217) of patients developed a soft tissue necrosis (STN). Osteoradionecrosis (ORN) was seen in 7 % (15/217) of patients. A target volume > 15 cm(3) was significantly associated with the occurrence of STN (p = 0.011) and ORN (p = 0.004). Conclusions: Postoperative interventional radiotherapy for previously not irradiated patients with early-stage OSCC is a safe and efficient treatment. Randomized trials are needed to compare these results to omission of postoperative radiotherapy as well as external beam radiotherapy.

Background and purpose:In the brachytherapy of cervical cancer, creating a suitable implant based on ultrasound guidance may be impacted by imaging limitations. To validate the implant if ultrasound is not sufficient, we implemented a new workflow utilizing additional intraoperative cone-beam computed tomography (CBCT). The aims of this work were to describe the newly established workflow, reflect associated (dis)advantages, and assess geometric and dosimetric benefits compared to the previous solely ultrasound-guided workflow. Materials and methods:We report the establishment of our new workflow utilizing mobile CBCT during interventions and corresponding experiences for 26 consecutive patients. Image quality was assessed by considering the applicator visualization and contrast-noise ratio (CNR) between tissues. Implant changes based on CBCT scans were analyzed with respect to the enhanced insertion depths (EIDs) of needles and their tip distances to target volume borders. Dosimetric effects were evaluated by calculating common dose-volume parameters for target volume and organs at risk (OARs) and comparing them in both a previous patient cohort and scenarios simulating sole ultrasound guidance. Implant uncertainties between intra- and postoperative imaging were analyzed using a corresponding registration as well. Results:Implementing intraoperative CBCT was associated with clinical challenges but increased safety feeling during interventions and resulted in geometric as well as dosimetric benefits. Needles could be shifted deeper into the pelvis by an EID of 14 ± 11 mm based on CBCT, associated with corresponding significant dose improvements for target volume and OARs with a mean tradeoff increase of up to 4.8 Gy. With a reasonable CNR between tissues up to 8.5 ± 3.6 and clear detectability of applicators, image quality was sufficient to fulfill intraoperative intentions. Furthermore, the CBCT scans were suitable for treatment planning purposes from a geometric uncertainty perspective. Conclusion:The implementation of intraoperative CBCT can substantially improve the quality and safety of image-guided gynecologic brachytherapy.

Computed tomography (CT) scans are vital for radiotherapy planning, providing essential data for dose calculations. This study retrospectively evaluated imaging doses, scan lengths, and protocol adherence to support imaging optimization and reduce patient radiation exposure. CT data from patients undergoing external beam radiotherapy and brachytherapy in the period 04/2021 to 12/2024 were retrieved from the institutional picture archiving and communication system (PACS). Imaging doses (volumetric CT dose index [CTDIvol] and dose length product [DLP]) were extracted from dose reports. Automated organ segmentation was used to assess standard operating procedures (SOPs) adherence by estimating anatomical scan length differences. Additional quality assurance checks assessed protocol and imaging consistency. Brain protocols exhibited the highest CTDIvol (73 ± 12 mGy), while head and neck protocols had higher DLP values (3212 ± 757 mGy·cm). The lung 4D protocol showed a higher effective dose (23 ± 9 mSv) compared to the standard lung protocol. Notable anatomical scan length differences were observed at the lower boundary in the upper abdomen (120 ± 75 mm) and spine (155 ± 159 mm), indicating opportunities for workflow improvement. Enhancing CT workflows for radiotherapy patients is important and feasible. Dose and scan length analyses suggest that revising institutional SOPs, optimizing X‑ray tube modulation, and refining scan length boundaries should be considered to achieve this goal.

Needle tracking using external prediction techniques such as optical tracking is a modern approach aimed at improving implantation accuracy in gynecologic brachytherapy. This study aims to investigate the corresponding impact of needle bending in situ and to analyze needle path deviations from the intended locations occurring in our current clinical workflow that only considers ultrasound imaging without tracking. We developed a semi-automated approach for reconstructing brachytherapy needles based on planning CTs and compared the respective accuracy to the also determined intra-observer variability of manual reconstructions. Based on this, we measured needle bending in situ for 89 patients and calculated the Euclidean distances between actual needle tips and needle tip predictions both longitudinally and laterally to the insertion direction. Furthermore, we compared actual and intended spacings between inserted needles to estimate implantation uncertainties with respect to our current clinical workflow. Our developed reconstruction featured an accuracy of 0.17 ± 0.08 mm, which was improved compared to the intra-observer variability of 0.21 ± 0.11 mm. Needle bending depended strongly on needle length and ranged from 3.6 ± 2.1 mm for 100–120 mm needles up to 7.9 ± 3.0 mm for 200–220 mm needles. Deflections in the transverse direction were substantially higher than tip deviations in the longitudinal direction. Furthermore, we found deviations from an equidistant spacing between needle paths of 1.4 ± 1.2 mm in the transverse direction. Inserting brachytherapy needles can be substantially affected by transverse needle bending in situ, which should therefore be corrected for in prediction approaches such as optical tracking.