Background: Large language models (LLMs) hold promise for supporting clinical tasks, particularly in data-driven and technical disciplines such as radiation oncology. While prior evaluation studies have focused on examination-style settings for evaluating LLMs, their performance in real-life clinical scenarios remains unclear. In the future, LLMs might be used as general AI assistants to answer questions arising in clinical practice. It is unclear how well a modern LLM, locally executed within the infrastructure of a hospital, would answer such questions compared with clinical experts. Objective: This study aimed to assess the performance of a locally deployed, state-of-the-art medical LLM in answering real-world clinical questions in radiation oncology compared with clinical experts. The aim was to evaluate the overall quality of answers, as well as the potential harmfulness of the answers if used for clinical decision-making. Methods: Physicians from 10 departments of European hospitals collected questions arising in the clinical practice of radiation oncology. Fifty of these questions were answered by 3 senior radiation oncology experts with at least 10 years of work experience, as well as the LLM Llama3-OpenBioLLM-70B (Ankit Pal and Malaikannan Sankarasubbu). In a blinded review, physicians rated the overall answer quality on a 5-point Likert scale (quality), assessed whether an answer might be potentially harmful if used for clinical decision-making (harmfulness), and determined if responses were from an expert or the LLM (recognizability). Comparisons between clinical experts and LLMs were then made for quality, harmfulness, and recognizability. Results: There were no significant differences between the quality of the answers between LLM and clinical experts (mean scores of 3.38 vs 3.63; median 4.00, IQR 3.00-4.00 vs median 3.67, IQR 3.33-4.00; P=.26; Wilcoxon signed rank test). The answers were deemed potentially harmful in 13% of cases for the clinical experts compared with 16% of cases for the LLM (P=.63; Fisher exact test). Physicians correctly identified whether an answer was given by a clinical expert or an LLM in 78% and 72% of cases, respectively. Conclusions: A state-of-the-art medical LLM can answer real-life questions from the clinical practice of radiation oncology similarly well as clinical experts regarding overall quality and potential harmfulness. Such LLMs can already be deployed within the local hospital environment at an affordable cost. While LLMs may not yet be ready for clinical implementation as general AI assistants, the technology continues to improve at a rapid pace. Evaluation studies based on real-life situations are important to better understand the weaknesses and limitations of LLMs in clinical practice. Such studies are also crucial to define when the technology is ready for clinical implementation. Furthermore, education for health care professionals on generative AI is needed to ensure responsible clinical implementation of this transforming technology.

Glioblastoma is a highly heterogeneous primary malignant brain tumor with marked inter-/intratumoral diversity and a poor prognosis. It may contain a population of neural stem cells (NSC) and glioblastoma stem cells that have the capacity for migration, self-renewal and differentiation. While both may contribute to resistance to therapy, NSCs may also play a role in brain tissue repair. The subventricular zone (SVZ) is the main reservoir of NSCs. This study investigated the impact of bilateral SVZ radiation doses on patient outcomes. We included 147 patients. SVZs were delineated and the dose administered was extracted from dose–volume histograms. Tumors were classified based on their spatial relationship to the SVZ. The dose and outcome correlations were analyzed using the Kaplan–Meier and Cox proportional hazards regression methods. Median progression-free survival (PFS) was 7 months (range: 4–11 months) and median overall survival (OS) was 14 months (range: 9–23 months). Patients with an ipsilateral SVZ who received ≥50 Gy showed significantly better PFS (8 versus 6 months; p < 0.001) and OS (16 versus 11 months; p < 0.001). Furthermore, lower doses (<32 Gy) to the contralateral SVZ were associated with improved PFS (8 versus 6 months; p = 0.030) and OS (15 versus 11 months; p = 0.001). Targeting the potential tumorigenic cells in the ipsilateral SVZ while sparing contralateral NSCs correlated with an improved outcome. Further studies should address the optimization of dose distribution with modern radiotherapy techniques for the areas surrounding infiltrated and healthy SVZs.

Background: Surgery and radiotherapy are well-established standards of care for unilateral stage 0 and I early-stage glottic cancer (ESGC). Based on comparative studies and meta-analyses, functional and oncological outcomes after both treatment modalities are similar. Historically, radiotherapy (RT) has been performed by irradiation of the whole larynx. However, only the involved vocal cord is being treated with recently introduced hypofractionated concepts that result in 8 to 10-fold smaller target volumes. Retrospective data argues for an improvement in voice quality with non-inferior local control. Based on these findings, single vocal cord irradiation (SVCI) has been implemented as a routine approach in some institutions for ESGC in recent years. However, prospective data directly comparing SVCI with surgery is lacking. The aim of VoiceS is to fill this gap. Methods: In this prospective randomized multi-center open-label phase III study with a superiority design, 34 patients with histopathologically confirmed, untreated, unilateral stage 0-I ESGC (unilateral cTis or cT1a) will be randomized to SVCI or transoral CO₂-laser microsurgical cordectomy (TLM). Average difference in voice quality, measured by using the voice handicap index (VHI) will be modeled over four time points (6, 12, 18, and 24 months). Primary endpoint of this study will be the patient-reported subjective voice quality between 6 to 24 months after randomization. Secondary endpoints will include perceptual impression of the voice via roughness – breathiness – hoarseness (RBH) assessment at the above-mentioned time points. Additionally, quantitative characteristics of voice, loco-regional tumor control at 2 and 5 years, and treatment toxicity at 2 and 5 years based on CTCAE v.5.0 will be reported. Discussion: To our knowledge, VoiceS is the first randomized phase III trial comparing SVCI with TLM. Results of this study may lead to improved decision-making in the treatment of ESGC. Trial registration: ClinicalTrials.gov NCT04057209. Registered on 15 August 2019. Cantonal Ethics Committee KEK-BE 2019-01506

Background: Intraoperative radiation therapy (IORT) is the delivery of a single large dose of radiation to a limited volume of tissue at the time of surgery. The aim of this study is to assess outcomes in patients who underwent IORT for gynecologic malignancies and to identify parameters that can predict a good outcome. Methods: This is a retrospective single-center cohort study including all women treated with surgery and IORT for a primary or recurrent gynecologic cancer between 2014 and 2022 at the Bern University Hospital, Switzerland. Results: A total of 30 patients with gynecologic malignancies were treated with surgery and IORT. Of these patients, 63.3% presented with cervical cancer, 23.3% with sarcoma, 10% with endometrial cancer, and 3.3% with carcinosarcoma of the ovary. Seventy percent (21/30) of women had an ECOG performance status of 0 at time of IORT. There was no difference in survival among women with incomplete surgical resection (R1/2 vs. R0) at time of IORT. Fifty percent of patients suffered postoperative complication of Clavien-Dindo grade >III, but there was no significant correlation of these complications to overall survival (p = 0.58). Three-year disease-free survival was 53.3%, and five-year overall survival was 53.3%. ECOG status was a significant parameter in DSS (p = 0.002) and OS (p = 0.02). Conclusion: Surgery with IORT is potentially a good treatment option in selected patients with recurrent or locally advanced cervical or endometrial cancer. An ECOG status of 0 is a significant parameter for good outcomes and should be taken into consideration for treatment decisions.

Riggenbach E, et al. J Gynecol Oncol. 2025 Dec;37:e41. https://doi.org/10.3802/jgo.2026.37.e41