Natural language processing (NLP) is a key technique for developing medical artificial intelligence (AI) systems that leverage electronic health record data to build diagnostic and prognostic models. NLP enables the conversion of unstructured clinical text into structured data that can be fed into AI algorithms. The emergence of transformer architecture and large language models (LLMs) has led to advances in NLP for various healthcare tasks, such as entity recognition, relation extraction, sentence similarity, text summarization, and question-answering. In this article, we review the major technical innovations that underpin modern NLP models and present state-of-the-art NLP applications that employ LLMs in radiation oncology research. However, it is crucial to recognize that LLMs are prone to hallucinations, biases, and ethical violations, which necessitate rigorous evaluation and validation prior to clinical deployment. As such, we propose a comprehensive framework for assessing the NLP models based on their purpose and clinical fit, technical performance, bias and trust, legal and ethical implications, and quality assurance prior to implementation in clinical radiation oncology. Our article aims to provide guidance and insights for researchers and clinicians who are interested in developing and using NLP models in clinical radiation oncology.
暂无摘要（点击查看详情）
Spinal stereotactic body radiotherapy (SBRT) provides durable local control for metastatic spine disease but creates new demands on post-treatment surveillance, particularly in healthcare systems where magnetic resonance imaging (MRI) access is limited. Existing guidelines recommend calendar-based imaging that is challenging to implement in resource-constrained environments. We aimed to develop practical, risk-stratified follow-up guidelines for post-SBRT surveillance, grounded in real-world feasibility across province-wide oncology centres. Seven radiation oncologists from four SBRT-providing centers (three tertiary academic, one regional) participated in a structured modified Nominal Group Technique. Panelists reviewed literature on imaging surveillance, contributed center-specific considerations, and iteratively refined recommendations. Nineteen candidate statements across four items were generated. Consensus thresholds were predefined as ≥80% ("strong"), 60-79% ("moderate"), and <60% ("no consensus"). In Round 1, 11/19 statements achieved strong consensus, 5 moderate, and 3 failed to meet threshold. Following structured adjudication, wording refinement, and one item removal, all remaining 18 statements achieved strong consensus in Round 2. Key themes included: (1) anchoring imaging intensity to actionable clinical intent; (2) prioritizing early post-SBRT MRI only when salvage therapy is feasible; (3) de-escalating surveillance in low-risk or systemically progressing patients; (4) tailoring modality selection to resource availability; and (5) integrating tumor-specific considerations, including prostate cancer-specific pathways. The resulting framework replaces rigid schedules with risk-adapted intervals, emphasizes clinical triggers and feasibility, and promotes thoughtful resource use. These consensus-derived guidelines present a pragmatic surveillance strategy for spine SBRT, calibrated to the realities of MRI access and clinical workflows. By embracing risk-adapted imaging, the framework reduces unnecessary investigations, preserves resources, and focuses surveillance where it is most likely to influence clinical outcomes. While rooted in the local context, the recommendations are broadly applicable to settings facing similar constraints. Prospective validation is warranted to evaluate oncologic and patient-reported outcomes under this de-escalated, resource-conscious model.
Isolated locoregional recurrence of pancreatic adenocarcinoma occurs in 25-50% of cases after surgical resection and has limited treatment options. Data on ablative stereotactic body radiation therapy (SBRT) with daily online adaptation for this condition are limited. Our objective was to illustrate the efficacy and safety of adaptive radiation for isolated locoregional recurrences of pancreatic adenocarcinoma. This was a retrospective cohort study of 46 patients treated with adaptive 5-fraction SBRT for isolated locoregional recurrence of pancreatic adenocarcinoma after surgical resection between 2015-2024. Kaplan-Meier curves with log-rank analysis and Cox regression were conducted to assess oncologic outcomes including locoregional recurrence-free survival, distant metastasis-free survival, and overall survival. Variables of clinical significance including biologically effective dose, planning target volume, and receipt of salvage chemotherapy in conjunction with salvage radiation were assessed. Local failure was defined by RECIST criteria. Toxicities were assessed using CTCAE v5.0 criteria. Radiation dose median was 50 Gy in 5 fractions. Seven patients had prior radiation and 27 patients received chemotherapy as part of their salvage treatment. The median duration of chemotherapy was 3.9 months. 46 patients were treated with a median age of 65, the majority were White and male. Median time to local recurrence from initial diagnosis was 20 months. Online adaptation was performed for 91% of fractions. Median follow-up from recurrent diagnosis and end of salvage radiation was 14 and 10.4 months, respectively. One-year local control was 73%, and 1- and 2-year overall survival were 69% and 42%, respectively. Chemotherapy had a statistically significant effect on progression-free survival (HR 0.45, 95% CI 0.23-0.89, p = 0.02). No acute grade 3+ toxicities occurred but three late grade 3+ toxicities were documented. Adaptive SBRT permits high biologically effective dose in 5 fractions and results in local control concordant with conventionally fractionated salvage radiation for locoregionally recurrent PDAC after definitive surgery, even after prior radiation with a reasonable toxicity profile. Future work may combine adaptive radiation with coverage of extra-pancreatic neural tract anatomy and additional chemotherapy to improve outcomes.
Positron Emission Tomography (PET) has evolved from a purely diagnostic modality into a cornerstone of precision radiation oncology. PET now informs patient selection, target delineation, treatment personalization, and post-therapy evaluation across External Beam Radiation Therapy (EBRT) and Radiopharmaceutical Therapy (RPT). Radiotracers provide quantitative data on tumor biology, heterogeneity, receptor expression, and therapeutic response, enabling a shift from morphology-based to biology-driven oncology. PET-guided therapy is increasingly used to select patients for molecular radiotherapy, guide EBRT boost volumes, monitor receptor occupancy, and personalize activity prescription. Combined modality approaches-such as EBRT plus PSMA-RLT in prostate cancer or EBRT plus SSTR-RLT in neuroendocrine tumors-are supported by biological rationales involving synergy between external and internal radiation sources. Adaptive strategies based on mid-treatment PET show promise in improving local control while minimizing toxicity. This review summarizes the current landscape and emerging applications of PET-guided therapy, highlighting methodological synergies between EBRT and RPT, strategies for treatment sequencing, biological dose painting, and adaptive therapy. It provides practical recommendations for implementing PET-guided workflows and discusses advances in radiobiology-informed dosimetry, whole-body PET technologies, and novel imaging biomarkers, including fibroblast activation protein inhibitors (FAPI), as key drivers of innovation. As PET technology evolves toward ultra-low-dose, ultra-fast total-body systems, the role of molecular imaging in therapeutic decision-making is expected to expand, ushering in a new era of biologically guided radiation oncology.
Radiation dermatitis is among the most common adverse events associated with radiotherapy for early-stage glottic cancer. This study evaluated the effect of cutting out a portion of the thermoplastic mask overlapping with the irradiation field on surface dose and investigated its benefit in reducing radiation dermatitis in patients with early-stage glottic cancer. The surface dose on the anthropomorphic phantom with and without a thermoplastic mask was measured using Gafchromic EBT4 films (Ashland). The treatment plan was created with two opposed lateral fields, with a field size of 5.5 cm × 6.5 cm, and photon energies of 4 and 6 MV produced by TrueBeam (Varian). For clinical investigation, we retrospectively reviewed patients with early-stage glottic cancer who received definitive radiotherapy between January 2008 and April 2024 at our institution. The inclusion criteria were clinical stages I and II, total irradiation dose of 60 Gy or more, and a two-opposed lateral field technique employed for radiotherapy delivery. The incidence of radiation dermatitis in grade ≥2 and grade ≤1 for age, clinical stage, radiotherapy fractionation, total dose, and cut-out thermoplastic mask was compared using Fisher's exact test. P values less than 0.05 were considered statistically significant. The percentage increase in the surface dose by a thermoplastic mask was 47.1% for 4 MV and 59.4% for 6 MV. Fifty-nine patients were included in the clinical investigation. For the incidence of ≥grade 2 dermatitis, the only significant factor was the cut-out thermoplastic mask (with cut-out thermoplastic mask, 41.7% vs. without cut-out thermoplastic mask, 74.5%, P = 0.042). Cutting out the thermoplastic mask over the irradiation field reduced the surface dose by eliminating the bolus effect, thereby mitigating radiation dermatitis in patients with early-stage glottic cancer.
An increase in targeting accuracy for Stereotactic Body Radiation Therapy (SBRT) to the prostate allows for a reduction in Planning Target Volume (PTV). This can reduce treatment-related side effects. In this study, we used a small gauge radiopaque catheter as a fiducial marker for prostate SBRT treatment delivery. Preparation and early engagement of radiation therapists is critical to the success of a catheter-guided prostate SBRT program. This study included forty-two patients from two cancer centers in Alberta, Canada that underwent prostate SBRT treatment. The average patient age was 72 years old (69-76). Thirty-six (86%) received radical prostate SBRT with intraprostatic dose escalation and six (14%) received prostate SBRT as salvage therapy. A small gauge, balloonless, radiopaque catheter was inserted at CT Simulation and for each fraction. The catheter served as a fiducial marker to guide urethral sparing treatment plans, and during cone beam CT-guided (CBCT) image matching. CBCT scans were taken prior to each of two Volumetric Modulated Arc Therapy (VMAT) arcs during treatment delivery. The radiopaque catheter was used as a reliable and accurate fiducial marker for image matching in prostate SBRT. Image matching guidelines were utilized as described in this paper. All patients tolerated the catheter without complications during treatment delivery. This technique is feasible, well-tolerated, and may be beneficial for radiation oncology departments that do not have the resources for surgical implantation of gold seed fiducials. Catheter-guided prostate SBRT is a feasible technique for urethral sparing in prostate SBRT with intraprostatic dose escalation. However, the benefit of intraprostatic dose escalation needs to be demonstrated more robustly to justify the additional department resources required for this technique. Radiation treatment for prostate cancer must be aimed very precisely to protect nearby parts of the body. This study tested using a thin catheter that can be seen on scans to help guide targeted radiation and checked how well people tolerated it. This study found that the catheter guided treatment accurately and was comfortable for everyone who received it. This matters because it offers a practical option for clinics that cannot use surgically placed markers and may help lower treatment side effects.
Stereotactic body radiation therapy (SBRT) is an emerging therapeutic modality in modern oncology, offering highly conformal, ablative radiation in a few fractions. However, implementing SBRT in limited-resource settings presents substantial challenges. This editorial describes the current SBRT landscape in the Philippines, a large lower-middle-income country in Southeast Asia with a rising cancer burden but constrained radiation therapy capacity. Of the 62 radiation therapy centers nationwide, only 14 (25%) have SBRT-capable technology, all privately operated and largely concentrated in the National Capital Region. SBRT remains markedly underutilized because of limitations in infrastructure, trained personnel, and patient access. As SBRT capacity continues to expand, we highlight the importance of establishing sustainable, comprehensive SBRT programs that incorporate robust clinical and technical quality assurance and outline practical strategies relevant to limited-resource settings, including streamlined workflows, hub-and-spoke care models, collaborative partnerships, strengthened data infrastructure, and equitable financing mechanisms. The commitment of dedicated practitioners and leadership from the national professional society will be essential to expanding safe, high-quality SBRT access in the Philippines.
Preclinical small animal experiments play an indispensable role in proton therapy research. However, accurate dose calculation poses a significant challenge because of the low beam energy and the requirement for submillimeter spatial resolution. Although the Monte Carlo method offers the necessary precision, its high computational cost hinders efficient implementation. This study aims to develop a GPU-accelerated radiation dose engine for proton radiotherapy (pGARDEN) based on the Monte Carlo method, specifically designed for fast and accurate dose calculation in small animal irradiation. In pGARDEN, we optimized the particle transport algorithm to better align with the GPU architecture. Moreover, various acceleration techniques were implemented to boost computational efficiency. To enhance precision, physical parameters, such as energy cutoffs for proton and electron, were tuned to better suit small animal conditions. The performance of pGARDEN was validated against Geant4 simulations and measurements across various beams and phantoms. To demonstrate its practical utility, pGARDEN was applied to calculate a multi-beam proton treatment plan for a lung tumor-bearing mouse model. Compared to Geant4, the engine achieved a > 1000-fold speedup and a 3D gamma passing rate of > 97% with a strict 1%/0.15 mm criterion in all phantom testing scenarios. The integrated depth dose curves and dose profiles showed good agreement with measurements. In the in vivo validation, the 2D gamma passing rates with a 2%/0.3 mm criterion were 95.52% ± 0.74% for the abdomen and 94.18% ± 1.08% for the thorax. Furthermore, pGARDEN calculated the treatment plan with < 1% statistical uncertainty in 4.3 s on an NVIDIA GeForce RTX 4070 Ti GPU, achieving a 100% 3D gamma passing rate with a 2%/0.3 mm criterion. pGARDEN can calculate proton dose distribution rapidly and accurately at submillimeter resolution for small animal. It provides a valuable tool for supporting small animal proton radiation experiments, such as the investigation of relative biological effectiveness (RBE) and new therapeutic strategies.
To identify dosimetric advantages of biaxially rotational dynamic radiation therapy (BROAD-RT) for stereotactic body radiation therapy (SBRT) with focal boost for non-metastatic prostate cancer (PCa), compared with coplanar volumetric-modulated arc therapy (co-VMAT). BROAD-RT is a unique beam delivery technique, which facilitates sequential non-coplanar beam delivery without the need to rotate the couch or reposition the patient. For 15 patients with non-metastatic PCa, two different plans (BROAD-RT and co-VMAT) were created, and these plans and dosimetric indices were compared. The prescribed dose was 35 Gy in 5 fractions to the whole prostate gland, and that to the intra-prostatic dominant lesions (IPDLs) was increased up to 50 Gy. Quality assurance (QA) was performed for BROAD-RT and co-VMAT plans from the 15 patients, and the calculated and measured dose distributions were evaluated according to global gamma analysis using ArcCHECK. The dose coverage of the target volumes and the high-dose exposure to organs at risk (rectum, bladder, and urethra) were not significantly different between BROAD-RT and co-VMAT plans. The normal tissue dose outside of the planning target volume (PTV) (maximum dose 2 cm away from PTV [D2 cm], and 30%, 50%, or 70% isodose volume divided by the volume of PTV [R30, R50, and R70]) were significantly lower in the BROAD-RT plan (p < 0.001, p < 0.001, p < 0.001, and p < 0.001, respectively). In QA, average (± standard deviation) passing rates were 98.2 ± 0.7% for BROAD-RT and 99.0 ± 1.0% for co-VMAT. Non-coplanar VMAT via BROAD-RT improved the dose distribution, mainly outside of the PTV and for some of the organs at risks, in prostate SBRT with focal boost compared with coplanar VMAT. Therefore, as BROAD-RT enables practical implementation of non-coplanar VMAT, it is considered a promising radiotherapy method of SBRT with focal boost for non-metastatic PCa.
To provide a case-based, practical example outlining step-by-step calculations and documentation for reirradiation (reRT) evaluation. This representative example demonstrates the application of the Reirradiation Collaborative Group (ReCOG) consensus for dosimetric assessment and reporting in reRT and illustrates how these standards can be implemented in clinical practice. Two spine reRT cases were selected. Cumulative organ-at-risk (OAR) doses were assessed using three approaches: direct point dose summation, point dose summation within overlap regions, and image registration-based 3D dose summation. Step-by-step calculations were performed for each approach, and new dose objectives for subsequent planning iterations were derived and compared. The example reRT cases detail the special medical physics consult (reRT-SMPC) process, highlighting different cumulative dose estimation strategies depending on the availability of voxel-based equieffective dose calculation and visualization tools. Selection of an assessment strategy should balance clinical resources with patient-specific considerations. Calculation and visualization of equieffective dose distributions facilitate cumulative dose assessment and strengthen image registration-based methods. If 3D equieffective dose calculation is not available, anatomically corresponding volumes and doses in each plan contributing to the cumulative dose in the areas of overlap can be determined manually to improve the accuracy of point dose-based summation. The case examples demonstrate how to apply the ReCOG consensus for dosimetric assessment and reporting in reRT and illustrate multiple strategies for cumulative dose evaluation. This work supports patient care and contributes to enhancing the field's knowledge and delivery of safer reRT.
A MASCC Clinical Practice Statement (CPS) serves as a brief, practical, clinician-oriented tool that highlights essential information for managing complications from anti-cancer treatments in cancer patients. This CPS provides guidance on the use of topical corticosteroids (TC) in the prevention and management of acute radiation dermatitis (ARD). This CPS was developed based on a critical evaluation of the literature identified through a search in MEDLINE till November 23, 2025, followed by a structured discussion of a group of experts of the MASCC Oncodermatology Study Group. The information is summarised in concise bullet points to create a brief manual on optimal standard care practices. Recommendations were provided regarding patient selection criteria for TC in the prevention and management of ARD. Guidance was also presented on the optimal choice of TC, its timing, schedule, application details, and monitoring of clinical effects to achieve the best outcomes based on existing literature. TCs are recommended for the prevention of ARD in selected patients with head and neck and breast cancers at a high risk of ARD. TC is also a treatment option for the management of patients with early ARD without evidence of moist desquamation. Medium potency TCs, such as mometasone furoate 0.1% cream, are preferred. Healthcare professionals should monitor patients' skin weekly during radiation and guide steroid use, discontinuing it if moist desquamation or other complications arise.
Small-field dosimetry plays a pivotal role in stereotactic body radiation therapy (SBRT), where high-precision dose delivery is essential. In multi-institutional clinical trials, quality assurance (QA) becomes increasingly critical due to the variability in linear accelerator (linac) output calibration and treatment planning system (RTPS) configurations. Conventional relative dose factor, which assess relative dose outputs, lack absolute dose information and thus cannot verify the consistency of linac calibration or monitor unit (MU)-to-dose conversion across institutions. These limitations may compromise treatment quality and undermine the integrity of clinical trials. Therefore, an enhanced QA method that incorporates absolute dosimetry is required. This study introduces and implements a novel credentialing procedure for small-field dosimetry in SBRT clinical trials using the absorbed dose per MU, expressed in cGy/MU. By integrating absolute dose information into the verification process, this method aims to evaluate both linac output calibration and small-field modeling accuracy across institutions. A nationwide dosimetric survey was conducted involving 545 beams from 120 institutions participating in Japan Clinical Oncology Group (JCOG) trials. Treatment plans were created in each institution's RTPS by prescribing 10 Gy to the isocenter for square field sizes of 2 × 2 to 10 × 10 cm2. Doses were calculated at depths of 5 and 10 cm in a virtual water phantom under a fixed source-to-surface distance of 90 cm. The absorbed dose per MU values were calculated and compared against multicenter averages stratified by linac type and beam energy. A ±3% tolerance criterion was applied for credentialing. Institutions with deviations were contacted and corrective actions were taken as needed. Of the 545 beams analyzed, 97.8% met the ±3% tolerance criteria. Discrepancies in the ten beams were attributed to small-field beam modeling errors or incorrect RTPS MU-to-dose settings, which were subsequently corrected through feedback. Although variations were observed between the dose calculation algorithms and RTPSs, the differences were within 0.6%, 0.4%, and 1.6%, respectively. The credentialing process proved robust in standardizing dose calculations for clinical trial participation. The proposed absorbed dose per MU-based method enables comprehensive small-field dosimetry evaluation by incorporating absolute dose verification into the credentialing process. This approach effectively identifies calibration inconsistencies that are not detectable using traditional relative dose factor methods. Its application across a large cohort of institutions demonstrated high accuracy and inter-institutional consistency, reinforcing its utility as a practical QA tool in SBRT clinical trials. By enhancing standardization in RTPS performance and linac output verification, this method strengthens the foundation of safe and reliable SBRT implementation in multi-center clinical research.
The conventional computed tomography (CT)-based consultation to simulation process for hippocampal-sparing whole-brain radiation therapy (HS-WBRT) typically requires several days, delaying treatment initiation for patients who would benefit from expedited care. We developed a simulation-free, magnetic resonance imaging (MRI)-only workflow for HS-WBRT to accelerate treatment start and maintain dosimetric quality through online adaptive radiation therapy. An in-house deep learning algorithm was developed to generate synthetic CT (sCT) images from diagnostic MRI scans. To address the limited field-of-view of diagnostic MRI, a hybrid sCT-reference CT stitching method was developed to create a full planning sCT suitable for HS-WBRT, through preserving patient-specific brain anatomy while supplementing inferior regions with matched reference anatomy. Using the resulting sCT, a patient-specific reference plan was created and then adapted to the setup position and patient anatomy during the first fraction. Subsequent fractions were delivered using the first-fraction adapted plan or readapted as needed. Commissioning included 5 retrospective emulator cases and a phantom end-to-end test to validate workflow feasibility, plan quality, and interoperability. Clinical feasibility was prospectively assessed in 5 patients. All retrospective and phantom tests met predefined endpoints, including protocol compliance, secondary quality assurance ≥95% (3%/2 mm), and error-free data transfer. In the prospective cohort, hippocampal contours were finalized in all cases during the online session without resimulation. The first-fraction adaptive plan met all HS-WBRT protocol constraints. For patients prescribed 30 Gy in 10 fractions (n = 4), median hippocampus Dmax and D100% were 14.6 Gy (range, 13.0-15.7 Gy) and 7.2 Gy (range, 7.0-7.5 Gy), respectively. For the patient prescribed 20 Gy in 10 fractions, hippocampus Dmax and D100% were 14.6 Gy and 5.2 Gy. All other organs-at-risk met institutional limits. The workflow eliminated CT simulation, enabling a clinical approved plan within 2 days of diagnostic imaging. Median on-couch adaptation time was 39 minutes. This MRI-only, simulation-free workflow is clinically feasible, shortens time to treatment, and enhances patient experience without compromising dosimetric quality.
Background In postmastectomy radiation therapy (PMRT), respiratory motion compromises dose delivery. Quantitative comparisons between virtual bolus (VB) and robust optimization (RO) remained limited, and there is no definitive consensus regarding optimal skin-flash strategies. This study evaluated the relative motion robustness of VB and RO in PMRT and explored the influence of VB thickness and density on dosimetric robustness. Methodology This study included 20 patients with left-sided PMRT planned in RayStation using standard optimization (SO), VB (27 combinations of three planning target volume (PTV) margins, three VB thicknesses, and three densities), and RO (three uncertainty settings). Respiratory motion was simulated by shifting the isocenter by 3-15 mm. Robustness was quantified using changes in chest wall D98% and D2% relative to the nominal plan. Results VB configurations showed substantial variability. The most robust plans consistently involved a VB thickness equal to the PTV margin plus 8 mm at 0.4 g/cm³. Under a 5-mm shift, median D2% changes for SO, VB, and RO were 2.6%, -0.8%, and 0.7%, while D98% changes were -8.7%, -0.3%, and -2.6%, respectively. Top-ranked VB settings maintained dose stability even when motion exceeded the applied PTV margin. Conversely, thin, low-density VB configurations caused marked D2% escalation (up to 3%) and were considered suboptimal. For representative configurations, changes in heart Dmean and lung V20Gy under isocenter shifts were small and comparable among SO, VB, and RO. Conclusions Under the tested conditions, appropriately configured VB demonstrated greater dosimetric robustness of target coverage than RO and SO. Robustness relied on the combined effects of PTV margin, VB thickness, and density. A configuration with VB thickness equal to the PTV margin plus 8 mm at 0.4 g/cm³ achieved the most stable performance, providing practical dosimetric insight for robustness-oriented PMRT planning.
Growing evidence supports stereotactic body radiation therapy (SBRT) over conventional radiation therapy for spine bone metastases, with an expanding role in non-spine bone metastases (NSBM). Our case-based review aims to inform radiation oncologists in the appropriate utilization of SBRT for representative cases of NSBM. Three cases were selected for discussion: (1) rib, (2) skull base, and (3) femur. Relevant literature was reviewed, and areas for future investigation were discussed. SBRT can be effectively delivered in NSBM with appropriate patient selection, target volume delineation, prescription dose, organs at risk dose constraints, and treatment planning. The Radiosurgery Society's case-based review offers guidance on the appropriate use of SBRT in NSBM with discussions and consensus recommendations from experts. SBRT can be considered for an oligometastatic patient with favorable prognosis in whom the goal is durable local control and/or symptom relief. It can be considered for radioresistant histologies and improved OAR sparing. Available MRI or PET/CT should be fused to improve target volume delineation. A CTV margin, generally of 5 mm, should be considered to cover microscopic disease. As bones are easily visualized on daily images acquired for accurate and precise set-up of patients, a PTV margin should be kept less than or equal to 3 mm. While SBRT can be delivered in 1 fraction, fractionated SBRT may be preferred to meet dose constraints when the CTV is adjacent to the OARs. NSBM of long bones that are weight bearing, lytic, or have a high MIRELS score should be evaluated by an orthopedic surgeon.
Headaches in patients with brain tumors can be multifaceted and difficult to treat, significantly affecting quality of life. We conducted a narrative synthesis of the literature to discuss the most common causes of and treatments for headache associated with brain tumors. Although increased intracranial pressure due to space-occupying lesions may be the pain driver in some instances, many patients experience secondary headaches that should be treated based on their primary characteristics, such as migraine or tension-like headache. Similarly, the location of tumors can affect pain type, inciting nerve, meningeal, or skull pain, and secondary headaches may also be induced by treatment effects from entities such as posterior reversible encephalopathy syndrome PRES or radiation necrosis. All these headache phenotypes should be treated based on symptoms and pain driver, with a focus on a wholistic approach incorporating both pharmacologic and non-pharmacologic options to improve the quality of life of patients with brain tumors.
Primary cutaneous lymphomas are a heterogeneous group of lymphoid neoplasms with predominant skin involvement, among which mycosis fungoides (MF) and Sézary syndrome (SS) are the most frequent and clinically challenging entities. This narrative review provides a multidisciplinary, compartment-oriented approach to the diagnosis, staging and management of MF/SS. We summarize the therapeutic principles across the disease continuum, from early-stage, skin-directed management to advanced-stage systemic therapy, including oral retinoids, methotrexate, extracorporeal photopheresis, mogamulizumab, brentuximab vedotin, chemotherapy and allogeneic stem-cell transplantation. We also address large-cell transformation, response assessment, follow-up strategies, and practical issues surrounding treatment maintenance, spacing and discontinuation in a chronic, relapsing disease. Finally, we highlight the importance of supportive care, particularly pruritus control and palliative integration when no further effective options remain, and provide pragmatic algorithms to facilitate real-world clinical decision-making.
Major salivary gland carcinoma (SGC) comprises diverse histologies with multiple interrelated risk factors, making it difficult to determine which patients truly benefit from post-operative radiotherapy (PORT). We developed a nomogram to predict the risk of locoregional failure (LRF) and model postoperative risk estimates under different PORT scenarios. Major SGC patients treated with curative-intent surgery between 2000 and 2021 across five tertiary cancer centers were identified. Prognostic factors (p < 0.05) from multivariable analysis and clinicopathologically relevant characteristics were utilized to construct a nomogram estimating five-year LRF risk. Subsequently, we modeled the five-year LRF and overall survival (OS) risks for each nomogram point-value, with and without PORT. A total of 1175 patients were included in the analysis. The median follow-up was 5.3 years. The nomogram for prediction of five-year LRF risk comprised six statistically significant negative prognostic factors (lymphovascular invasion [100 points], WHO high-risk pathology [84], involved resection margins [61], parotid primary tumor [69], pathologic T3-4 category [44], and the non-utilization of PORT [57]), and two relevant factors (pathologic nodal involvement [34] and perineural invasion [7]). The corrected C-index was 0.77. We estimated the values of five-year LRF and OS depending on the PORT. For example, a patient with 271 points had a five-year LRF risk of 20% without PORT versus 11% with PORT, and a five-year OS of 65% versus 79%, respectively. This nomogram provides individualized estimates of LRF after surgery for major SGC and offers model-based risk estimates under different PORT scenarios. It may serve as a practical tool to guide personalized adjuvant treatment decisions.
To explore the effectiveness of utilizing ChatGPT 4.0 to assist human physicians in assessing dysphagia in patients undergoing radiotherapy for head and neck cancer. This prospective study included 100 head and neck cancer (HNC) patients who visited our hospital between January 2025 and October 2025. All participants first underwent an independent dysphagia assessment in the control group conducted by a human physician (Physician A). Subsequently, they were evaluated in the experimental group by a similarly qualified physician (Physician B) with the assistance of ChatGPT 4.0. The comprehensive assessment results from an expert group consisting of two senior head and neck surgeons with ten years of experience served as the "gold standard." Consistency comparisons of the evaluation results among the three groups were conducted to validate the effectiveness of the language model-assisted assessment. The consistency Kappa index between the experimental group and the expert group was 0.87, indicating a "good" level of consistency, significantly superior to the control group's 0.70. Subgroup analysis of different EAT-10 and MDADI score ranges showed that in 85 patients with EAT-10 scores ≥ 3: the control group accurately identified 72 cases, achieving an accuracy of 84.7%; the experimental group accurately identified 80 cases, with an accuracy of 94.1%. Among 78 patients with MDADI scores ≤ 69, the control group accurately identified 65 cases (accuracy of 83.3%), while the experimental group identified 73 cases accurately (93.6%). The assessment model combining large language models with human physicians effectively improves the accuracy and consistency of dysphagia assessment in patients undergoing radiotherapy for head and neck cancer.