Citations

464 Background: Based on LITESPARK-005 trail, Belzutifan was FDA-approved to treat advanced RCC following ICI and a VEGF-TKI. We conducted a retrospective cohort study to report the real-world outcomes of Belzutifan in pre-treated RCC patients. Methods: A retrospective chart review of all patients who received Belzutifan at the Mayo Clinic Comprehensive Cancer Center between Dec 2023 and Sep 2025 was conducted. Patients treated for sporadic RCC were eligible; those with von Hippel–Lindau disease, non-RCC pathologies (e.g., Pheochromocytoma, ovarian cancer) were excluded. Radiologic response was assessed by trained curators using RECIST-defined criteria, approximated through systematic review of radiology reports with clinician adjudication when needed. Progression-free survival (PFS), defined as the time from treatment initiation to radiographic progression or death, was estimated using the Kaplan–Meier method. Non-progressors were censored at the last radiology assessment. Patients who discontinued therapy for non-progression reasons were censored at the last radiology assessment prior to discontinuation. Any-cause hypoxia and anemia were reported using CTCAEv5 definitions. Results: A total of 158 patients were identified; 83 (52.5%) who received Belzutifan for sporadic RCC were included. The remaining were excluded for VHL disease (68, 43%) or other pathologies (7, 4.4%). Median follow up was 12 months, 67% were male, median age was 70 (range 39, 94) with 57 patients (69%) were ≥ 65. 28 patients (34%) had a poor IMDC score at baseline. The cohort received a median of 3 previous lines of therapy (range 1,8). Median treatment duration was 3 months (IQR: 1-9.5). Median PFS was 5.7 months (95% CI, 3.8–8.7). At 12 and 18 months, 23.1% and 13.9% of patients, respectively, remained alive and progression-free. Corresponding OS rates at these time points were 60.2% and 51.5%. A confirmed ORR occurred in 12 patients (14.4%). Hypoxia occurred in 44 (53%) patients, median onset 31 days (range 1–270) and 29 (34.9%) required hospitalizations. Median hemoglobin (Hb) levels at baseline were 12.65 g/dl (range, 7.2-15.5), and 68 (82%) patients developed Hb decline on Belzutifan, 27 (32.5%) required transfusions. Conclusions: In an older, high-risk cohort with higher number of prior lines of therapy, Belzutifan yielded PFS/OS and ORR landmarks comparable to the trial supporting its effectiveness in the real world. Outcomes compared with LITESPARK-005 trial. Variable Real world cohort (Mayo Clinic) LITESPARK-005, Belzutifan arm Median PFS (months, 95% CI) 5.7 (3.8-8.7) 5.6 (3.9-7) 12-month PFS rate (%, 95% CI) 23.1% (14.4-33.4) 33.4% (28.4-33.5) 18-month PFS rate (%, 95% CI) 13.9% (7.7-23.9) 24% (19-29.4) 12-month OS rate (%, 95% CI) 60.2% (48.9-70.8) 67.9% (62.9-72.4) 18-month OS rate (%, 95% CI) 51.5% (40.6-62.9) 55.2% (50-60.1) ORR (%, 95% CI) 14.4% (7.7-23.9) 21.9% (17.8-26.5)

405 Background: Clinical trials are critical to advancing prostate cancer treatment, yet patient enrollment remains a major bottleneck. Despite strong interest in AI for trial matching, real-world use remains limited due to reasons such as complex infrastructure requirements. To address this, we developed a lightweight, scalable AI framework leveraging the Google Healthcare Search API—requiring minimal technical expertise—to automate trial eligibility screening. Methods: We used a pre-trained GPT-5 large language model (LLM) to decompose the eligibility criteria of 13 prostate cancer clinical trials into 269 YES/NO questions, such that for an eligible patient of a given trial, the chart review would yield YES for inclusion and NO for exclusion questions. From 348 patients enrolled in one of the trials, we built a dataset linking each eligibility question to each patient, resulting in 4,947 patient–question pairs (2,034 inclusion pairs with YES labels and 2,913 exclusion pairs with NO labels). The task was framed as binary classification. During evaluation, for each patient–question pair, the LLM extracted search phrases from the question to retrieve relevant EHR information via the Google Healthcare Search API, then used the question and retrieved information to predict YES or NO responses. Performance was measured using accuracy, precision, recall, specificity, and F1 score. Results: Across 4,947 question–patient pairs, our framework achieved 0.81 accuracy, 0.85 precision, 0.65 recall, 0.92 specificity and a F1 score of 0.74. Results for different phases of trials is reported in Table 1, which demonstrates the approach’s robustness across trial phases. Predictions for exclusion criteria were highly precise, while inclusion criteria were more challenging. Most errors were observed in criteria involving complex nested logical conditions, interpretable clinical definitions and data not explicitly measurable or recorded in the EHR (e.g., planned treatments or clinician-assessed scores). Conclusions: This work demonstrates a practical AI framework to support clinical trial matching. The system can identify complex disease states (biochemical recurrence vs radiographic metastasis), castration status (sensitive, resistant) and genomics. Future work should focus on developing human-in-the-loop supervision and improving question decomposition to reduce errors in complex eligibility criteria. Screening performance of the proposed framework across different phases of trials. Trial Phase #Patient-Question Pairs #YES/NO labels Accuracy Precision Recall Specificity F1 score I, II, I/II 4,227 Yes: 1,735 No: 2,492 0.80 0.84 0.66 0.92 0.74 III 720 Yes: 299 No: 421 0.82 0.90 0.63 0.95 0.74 All 4,947 Yes: 2,034 No: 2,913 0.81 0.85 0.65 0.92 0.74 #YES/NO labels: The number of patient-question pairs that are labeled YES/NO among total patient-question pairs of the same row.

Abstract Natural killer (NK) cells represent key effectors of antitumor immunity, yet emerging evidence highlights populations with distinct roles in cancer. Despite such expanded diversity within the NK cell repertoire, we lack an understanding of how this heterogeneity affects immune responses and downstream clinical outcomes. Using single-cell RNA sequencing, we systematically profiled NK cells across cancer and uncovered a dichotomous phenotypic and functional landscape of tumor-infiltrating NK cells shaped by opposing intrinsic signaling programs that drive the expression of IFNG or TGFB1. These divergent programs are associated with distinct transcription factor circuits that integrate cues within the tumor microenvironment and skew NK cells toward proinflammatory or suppressive functions. We found that the capacity for NK cells to engage in either functional direction is intrinsically linked to their phenotypic identity. Canonical NK cells recruited from circulation predominantly directed suppressive TGFB1 signals toward effector CD8+ T cells in tumors. Of note, these subsets exhibited higher TGFB1 expression than intratumoral myeloid cells across tumor types. In contrast, a tissue-resident (TR) adaptive subset exhibited exclusively proinflammatory IFNG-driven profiles and was associated with prolonged survival in both primary and metastatic tumor settings. Moreover, these TR adaptive NK cells, but not other subsets, were linked to response to immune checkpoint blockade. Collectively, our study reveals a previously unrecognized regulatory axis in NK cells that shapes NK cell diversity and augments broader antitumor immune responses.

T cell fitness has emerged as a critical determinant of the efficacy and persistence of Chimeric Antigen Receptor (CAR)-T cell therapy. Defined by the capacity of T cells to proliferate, resist exhaustion, persist in vivo , and exert sustained effector functions, T cell fitness reflects the integration of a dynamic network of intrinsic and extrinsic regulatory mechanisms. In this review, we present a comprehensive synthesis of the molecular and cellular foundations underlying T cell fitness, emphasizing the influence of differentiation trajectories, signaling pathways, metabolic reprogramming, and epigenetic modifications. We further discuss the impact of patient-specific conditions such as age as well as disease biology, prior therapeutic exposures, and timing and quality of T cell collection, on the phenotypic and functional efficacy of CAR-T cell products. Beyond delineating these determinants, we highlight emerging strategies aimed at enhancing T cell fitness. Importantly, we propose T cell fitness as an integrated, multi-layered systems property emerging from the interaction between differentiation state, signaling architecture, metabolic–mitochondrial competence, epigenetic stability, and host-specific inflammatory and treatment-related pressures. We introduce a mechanistic framework that links these layers across the CAR-T therapeutic timeline from leukapheresis to post-infusion tumor engagement and outline how this framework can be operationalized into measurable parameters to guide patient stratification, manufacturing decisions, and rational therapeutic interventions.

Abstract Background The ability of large language models (LLMs) to work collaboratively and screen studies in a systematic review (SR) is under-explored. Hence, we aimed to evaluate the effectiveness of LLMs in automating the process of screening in systematic reviews. Methods This is an observational study which included labeled data (title and abstracts) for five SRs. Originally, two reviewers screened the citations independently for eligibility. A third reviewer cross-checked each citation for quality assurance. GPT-4, Claude-3-Sonnet, and Gemini-Pro-1.0 were used using zero-shot chain-of-thought prompting. Collaborative approaches included (i): conflict resolution using benefit of the doubt, (ii) majority voting using an independent third LLM and (iii) conflict resolution using an informed third LLM. Performance was assessed using accuracy, precision for exclusion, and recall for inclusion. Work saved over samples (WSS) was computed to estimate the reduction in manual human effort. Results A total of 11300 articles were included in this study. The individual models, GPT-4, Claude-3-Sonnet, and Gemini-Pro-1.0 exhibited a high precision for exclusion, achieving 99.7%, 99.7%, and 99.2% and high recall for inclusion achieving 95.5%, 96.6% and 85.7%, respectively. However, the collaborative approach utilizing the two best-performing models (GPT-4 and Claude-3S) achieved an average precision of 99.9% and a recall of 98.5% (across all collaborative approaches). Furthermore, the proposed collaborative approach resulted in an average WSS of 63.5%, compared to the average WSS of 45.2% for individual models. Conversational LLM interactions showed a consistent pattern of results. Limitations This study was limited due to reliance on proprietary models, and evaluation on oncology datasets. Conclusion Evidence shows that collaborative LLMs enable efficient, high-performing screening in systematic reviews, supporting continuous evidence updates. Primary funding source NIH (U24CA265879-01-1) and Carolyn-Ann-Kennedy-Bacon Fund.

Abstract Enhancer is a critical epigenetic feature in head and neck squamous cell carcinoma (HNSCC), yet the functional roles of individual enhancers remain poorly understood. Here, we conducted an integrative multi-omics analysis based on publicly available ATAC-seq, H3K27ac ChIP-seq, transcriptomic profiling, and genetic association datasets to systematically map HNSCC-associated enhancers and their target genes. Integrating ATAC-seq and H3K27ac ChIP-seq, we identified 20,362 enhancers and 18,040 enhancer-associated genes, highlighting widespread regulatory complexity. Functional characterization of the TERT -associated enhancer GH05J001312 (GRCh38/hg38: chr5:1312099-1317743) revealed strong transcriptional activity in HNSCC. CRISPR-mediated deletion of a core sequence significantly reduced TERT expression, impaired cellular proliferation in vitro , and suppressed tumor growth in vivo , confirming its role as a key cis-regulatory element. RNA-seq analysis of enhancer-edited cells uncovered 742 differentially expressed genes enriched in cancer-related pathways, including MAPK and IL-17 signaling, indicating a broad transcriptional impact. Collectively, our findings establish GH05J001312 as a functional enhancer driving oncogenic programs in HNSCC and suggest enhancer-targeted strategies as a potential therapeutic avenue.

Abstract INTRODUCTION The vascular endothelial growth factor (VEGF) signaling family plays a role in neurodegenerative diseases, including Alzheimer's disease (AD). Previous work has shown widespread effects of the members FLT1 , FLT4 , and VEGFB on AD outcomes. However, these analyses have focused within the non‐Hispanic White (NHW) population. METHODS The goal of this study was to analyze the effects of the VEGF family in underrepresented populations, leveraging large and diverse bulk RNA sequencing and tandem mass tag–mass spectrometry (TMT‐MS) proteomic data. Outcomes included measures of AD pathology and diagnosis. RESULTS Within underrepresented populations, we replicated previously reported effects of FLT1 and FLT4 , whereby higher protein abundance was observed in the AD brain and was associated with higher neuropathology burden. In stratified analyses, these associations were largely consistent across race and ethnicity. DISCUSSION This multi‐omic study on the role of the VEGF family in AD emphasizes the need for more representative studies focused on therapeutic targets for AD. Highlights Vascular endothelial growth factor (VEGF) genes and proteins were quantified in four different brain regions. Samples included participants from four different populations. Previously observed effects were replicated in diverse populations. This study is the largest multi‐omic study of the vascular endothelial growth factor (VEGF) genes among Alzheimer's disease (AD) participants from diverse populations.

Abstract Background Diabetes mellitus is a prevalent global health issue, affecting approximately 14% of individuals as latest reported. Among its complications, diabetic peripheral polyneuropathy (DPN) is a major cause of morbidity, impacting up to half of patients with long-standing diabetes. Given the limitations of current diagnostic methods, there is an urgent need for reliable biomarkers for early detection and management of DPN. Neurofilament light chain (NFL), a structural neuronal protein, has been proposed as a potential biomarker due to its release into the blood and cerebrospinal fluid following axonal damage. This systematic review and meta-analysis aim to evaluate serum NFL levels in diabetic patients with and without polyneuropathy compared to healthy controls. Methods A systematic search was conducted in four databases: Web of Science, PubMed, Scopus, and Embase. Studies were included if they measured serum NFL levels in diabetic patients, with or without polyneuropathy, and compared these levels to healthy controls. The Newcastle-Ottawa Scale (NOS) and JBI Critical Appraisal tools were used for quality assessment. We performed a meta-analysis using Review Manager 5.4 version, and assessed heterogeneity through chi-square tests and I² statistics. Results Eight studies (n = 2980 participants, including 950 diabetic patients and 2030 healthy controls) met the inclusion criteria. Three studies specifically evaluated serum NFL levels in diabetic patients with polyneuropathy. Meta-analysis revealed significantly elevated serum NFL levels in diabetic patients compared to healthy controls. Similarly, diabetic patients with polyneuropathy exhibited higher NFL levels than healthy controls. However, due to the limited number of studies comparing diabetic patients with and without neuropathy, a definitive conclusion on NFL's specificity to DPN remains uncertain. Heterogeneity analysis showed a moderate to high variation among studies, with a slight reduction after sensitivity analysis. Conclusion This study highlights elevated serum NFL levels in diabetic patients, irrespective of polyneuropathy status, suggesting that diabetes itself may contribute to increased NFL release. However, due to limited studies assessing NFL in DPN, further research is required to clarify its role as a specific biomarker for DPN.

OBJECTIVE Brain metastases significantly impact neurocognitive function and overall survival. Stereotactic radiosurgery (SRS) is a cornerstone of treatment for patients with limited metastases and expected survival beyond 3 months. Despite current guidelines, up to 20% of patients with brain metastases undergoing SRS have been reported to die within 90 days. This study retrospectively evaluated prognostic factors associated with 90-day survival after SRS, aiming to improve patient selection. METHODS The authors retrospectively analyzed a cohort of 1546 patients who underwent Gamma Knife SRS for brain metastases at their institution between 2015 and 2023. One hundred seventy patients who survived less than 90 days after SRS were identified and case matched to 170 patients who survived over 90 days. Measured variables included patient demographic characteristics, tumor characteristics, treatment history, functional status, and control of the primary cancer. The authors modeled post-SRS 90-day survival using binomial and multivariate logistic regression. RESULTS Multivariate analysis highlighted Karnofsky Performance Score (KPS) < 70 (OR 17.4, p < 0.001), prior whole-brain radiation (OR 6, p = 0.004), and focal neurological deficits (OR 3.02, p = 0.003) as significant predictors of poor survival, whereas CNS progression before SRS (OR 0.22, p < 0.001) and control of systemic cancer (OR 0.556, p = 0.002) were associated with survival < 90 days. The predictive model demonstrated acceptable performance with an area under the curve (AUC) of 0.85, accuracy of 80%, sensitivity of 87%, and specificity of 71%. CONCLUSIONS Key predictors of 90-day survival after SRS for brain metastases include functional status (KPS), control of systemic cancer, CNS progression status, and focal neurological deficits. These findings are complementary factors that can assist in making decisions and SRS patient selection.

Early allograft failure (EAF) after living donor liver transplantation (LDLT) remains a clinical challenge. Existing prediction models developed for deceased donor transplantation poorly apply to LDLT due to distinct surgical and physiological factors. This study identifies clinical determinants of EAF and develops an LDLT-specific prediction model. We conducted a multicenter retrospective cohort study from 17 high-volume LDLT centers (January 2016–December 2020) with external validation at a tertiary center in Saudi Arabia (January 2015–December 2022). The primary outcome was EAF (graft loss or patient death ≤90 d). Multivariable mixed-effects logistic regression identified preoperative/intraoperative risk factors. The EAGLE-LDLT model was constructed using postoperative laboratory values. Performance was compared against established models (EAD, MEAF, A2ALL) using ROC analysis and decision curve analysis. The development cohort included 2944 adult LDLT recipients (67.7% male; median age 55 y; median MELD 14) with a 5.5% EAF rate. External validation included 1020 recipients (median MELD 21, 6.7% EAF). Independent risk factors for EAF were MELD (OR 1.06, 95% CI 1.04–1.08), donor BMI (OR 1.05, 95% CI 1.00–1.10), portal vein thrombosis (OR 1.73, 95% CI 1.13–2.63), and hepaticojejunostomy (OR 1.58, 95% CI 1.06–2.36). The EAGLE-LDLT model incorporating peak ALT (>468 U/L), peak INR (>1.9), and bilirubin (>3.5 mg/dL) and INR (>1.3) at POD7, demonstrated superior discrimination (AUC=0.81) compared with MEAF (AUC=0.77, p =0.004), EAD (AUC=0.67, p <0.001), and A2ALL (AUC=0.65, p <0.001). EAGLE-LDLT achieved balanced sensitivity (75.0%) and specificity (73.7%), effectively stratifying patients into high-risk (15% of patients; 40.4% EAF incidence) and low-risk groups. Preoperative and intraoperative clinical factors predict EAF in LDLT. The EAGLE-LDLT model accurately identifies LDLT recipients at the highest risk for EAF postoperatively.