Browse Articles

AI-assisted prediction of radiation failure patterns and survival after stereotactic radiosurgery for brain metastases: A multicenter study.

2023 Background: Brain metastases occur in approximately 20%–40% of patients with cancer and are associated with substantial morbidity and mortality. Stereotactic radiosurgery (SRS) is an effective and widely used treatment; however, a significant proportion of patients experience local, distant, or leptomeningeal failure after treatment. Early prediction of radiation failure patterns is critical for guiding individualized surveillance and treatment strategies. Current clinical tools remain inadequate for accurately predicting radiation failure in patients with brain metastases. Methods: In this multicenter retrospective study, we included 1,079 patients with brain metastases from three medical centers. We developed a novel deep learning model, the Global-to-Local Multiple-Instance Learning Mixture-of-Experts (GL-MIL MoE) framework. The model integrates a mask-guided multiscale encoder to process global MRI volumes and a multiple-instance learning (MIL) module to extract features from high-resolution local tumor patches. The primary endpoint was radiation failure pattern, defined as local failure, distant failure, or leptomeningeal failure. The secondary endpoint was overall survival (OS). Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), concordance index (C-index), and decision curve analysis (DCA). Model interpretability was assessed using SHAP (Shapley Additive Explanations) and Grad-CAM analyses. Results: The proposed AI model demonstrated strong and consistent performance, achieving AUCs ranging from 0.78 to 0.83 across validation cohorts. C-indices for OS prediction remained robust and significantly outperformed established baseline models (p < 0.05). Multivariable logistic and Cox regression analyses confirmed the model-derived risk score as an independent predictor of both radiation failure patterns and OS (p < 0.05). The model effectively stratified patients into high- and low-risk groups across all cohorts. Decision curve analysis demonstrated meaningful and consistent clinical utility. Conclusions: This AI-based deep learning model enables accurate prediction of radiation failure patterns and survival in patients with brain metastases treated with SRS. Prospective studies are warranted to evaluate its clinical utility in guiding personalized treatment and surveillance strategies in combination with standard clinicopathologic factors.

Adolescent and young adult cancer death burden in the United States from 1990 to 2023.

e22599 Background: Reports of increasing adolescent and young adult (AYA) cancer burden in the United States (US) may be influenced by detection bias. Thus, we examined AYA cancer mortality over time, a more stable indicator of cancer control progress. Methods: Using Global Burden of Disease 2023 estimates, we examined cancer death counts and rates (per 100000) among US AYAs (ages 15–49) from 1990–2023, overall and by sex. Temporal trends were quantified using estimated annual percentage change (EAPC). Results: In 1990, cancer was the leading cause of death among AYAs with 42599 deaths. By 2023, after declining to 31761 deaths (−25%), cancer ranked third, behind substance use disorders and cardiovascular diseases. Between 1990−2023, 1353140 AYAs died from cancer (53% female), while mortality rates declined from 31.7 to 20.7 (−35%). In 1990, the leading causes of AYA cancer death were lung cancer, breast cancer, colorectal cancer (CRC), leukemia, and brain and central nervous system (CNS) cancer. From 1994-1997, non-Hodgkin lymphoma (NHL) briefly supplanted leukemia among the top five causes. Lung and breast cancer ranked first and second from 1990−2012, until breast surpassed lung cancer in 2013. CRC ranked third from 1990−2014, overtook lung cancer in 2015, and surpassed breast cancer in 2021 to become the leading cause of AYA cancer death. In 2023, leading causes were CRC (4611 deaths), breast cancer (4523), lung cancer (2881), brain and CNS cancer (2642), and leukemia (1940). EAPCs for mortality rates from 1990−2023 for these leading cancers were 0.37 (95% Cl, 0.24, 0.51) for CRC, −2.18 (−2.30, −2.07) for breast cancer, −3.89 (−4.42, −3.36) for lung cancer, −0.78 (−0.90, −0.66) for brain and CNS cancer, and −1.89 (−1.98, −1.81) for leukemia. In males, NHL replaced breast cancer in the top five causes of AYA cancer death from 1990−2009. Pancreatic cancer replaced NHL from 2010−2023, except in 2011. Lung cancer was the leading cause of death from 1990−2013, until colorectal cancer became leading in 2014. Leukemia and brain and CNS tumors remained in the top five from 1990–2023. In females, breast cancer was the leading cause from 1990–2023. Lung cancer ranked second until surpassed by CRC in 2016. Cervical cancer ranked third from 1990−1998, fourth from 1999−2016, and third again from 2017−2023. The fifth ranked cancer shifted from leukemia in 1990 to ovarian cancer from 1991−2012, and to brain and CNS cancer from 2013−2023. Like overall trends, mortality rates declined for all major cancers in both sexes except CRC, for which EAPC increased: 0.43 (0.29, 0.57) in males and 0.30 (0.16, 0.43) in females. Conclusions: In the US, CRC became the leading cause of AYA cancer death in males in 2014 and overall in 2021. Mortality rates have declined in every leading AYA cancer except CRC. Despite overall progress, these findings underscore the need for age-appropriate prevention, early detection, and treatment strategies tailored to the AYA CRC population.

Safety and efficacy of perioperative use of tislelizumab in treating clinically node-positive urothelial carcinoma: A prospective clinical study.

e16615 Background: Recent studies suggest that the preservation of tumor-draining lymph nodes (TDLN) may enhance the sensitivity to immunotherapy. This study aims to evaluate the efficacy of neoadjuvant therapy with tislelizumab in patients with clinically node-positive urothelial carcinoma and to explore the role of preserved TDLN in subsequent immunotherapy for patients achieving a clinical complete response (cCR) in the primary tumor. Methods: In this single-arm, single-center, prospective clinical trial, we enrolled patients with no prior anti-tumor immunotherapy and imaging-confirmed cT2-T4NxM0 urothelial carcinoma (2023-SR-541). The treatment regimen included cisplatin (70 mg/m²) on Day 2-4, gemcitabine (1000 mg/m²) on Days 1 and 8, and tislelizumab (200 mg) on Day 8, administered every 21 days for three cycles. Patients who achieved clinical CR underwent radical surgery within six weeks of the last dose. Sentinel TDLN biopsy, accompanied by rapid pathological assessment, was conducted intraoperatively. If the biopsy results were negative, no further lymph node dissection was performed; if positive, standard lymph node dissection was carried out. Postoperatively, patients received adjuvant immunotherapy with tislelizumab (200 mg intravenously) every three weeks for one year or until disease relapse or trial termination. Results: In this study, ten patients have been enrolled to date, with eight patients had T2-stage tumors, while two patients had a T3-stage tumor. Imaging suggested TDLN metastasis in all cases. Following the completion of neoadjuvant therapy, all patients underwent radical cystectomy, and imaging results indicated a complete response in both primary tumors and TDLN. Postoperative pathology showed no residual pathological disease in any patient. At the most recent follow-up, with a median duration of 14.3 months, all ten patients remain free from recurrence or metastasis. Conclusions: The preservation of TDLN in patients achieving clinical CR after neoadjuvant therapy with tislelizumab is safe. This approach provides novel insights and methods for lymph node management and presents a potential new treatment option for the clinical management of this patient population.

Cholangiocarcinoma care across America: A tale of two coasts.

e16241 Background: Cholangiocarcinoma (CCA) is an aggressive biliary cancer with a historically poor prognosis and a recent rise in global incidence. Clinical trends suggest geographic location as a major social determinant, with literature indicating disproportionate disease burden in the Northeast (NE) and West (W) regions of the United States. Identifying the underlying causes of this disparity is essential to achieve health equity. This study utilizes the National Inpatient Sample (NIS) database to analyze the characteristics, clinical outcomes, and healthcare costs of CCA hospitalizations in the NE and W regions of the United States. Methods: A retrospective cohort study was conducted using the 2019–2023 NIS database. Adult patients (≥18 years) hospitalized with CCA were identified using ICD-10 codes and divided into Northeastern (NE) and West (W) regions. NE was taken as reference. Baseline characteristics and outcomes were compared among NE and W using univariate and multivariate logistic regression analysis. Results: Among 189,820 CCA hospitalized patients, 16.7% (n = 31,699) occurred in the NE and 16.8% (n = 31,889) in the W. Patients' ages were slightly similar in both regions (68 vs 69 years; coefficient: -0.50; P = 0.077), Male were the predominant gender (NE 16,800; 53% and W 17,220; 54%). Racial distribution varied significantly, White predominantly in both groups (NE 22,189; 70% and W 16,582; 52%), followed by hispanic race (NE 3,170; 10% W 7,653; 24%). Clinical outcomes were more favorable in the W, with the NE demonstrating higher in-hospital mortality (8% vs 7%; OR: 0.84; P < 0.05) and longer length of stay (LOS) (7.3 vs 6.6 days; Coefficient: 0.70; P < 0.001). However, W exhibits higher hospital charger costs ($102,559 vs $120,933; Coefficient: $18,374; P < 0.001). Most cases were managed at large hospitals (NE: 58%; W: 65%) and teaching institutions (NE: 91%; W: 82%; p < 0.001). Medicare was the primary payer (NE: 61%; W: 60%), followed by private insurance (NE: 27%; W: 24%). Conclusions: Significant regional disparities exist in CCA outcomes, with higher prevalence, mortality and longer LOS in the NE, but higher healthcare costs in the W. Though this corroborates recent trends indicating geographical disparity in CCA, the reasons remain unclear. Differential hospital courses, resource utilization, cost inflation and socioeconomic variables uniquely impact regional care. Further research is needed to identify the specific drivers and optimize equitable, cost-effective management.

Multicenter clinical study of [ ¹⁸ F]AlF-HER2-BCH PET/CT as a molecular pathology tool for linear quantification of HER2 expression levels in breast cancer.

3071 Background: Determining the human epidermal growth factor receptor 2 (HER2) status is essential in metastatic breast cancer (MBC) management. To establish the clinical validity of [ 18 F]AlF-HER2-BCH PET/CT (HER2-PET), we assessed the diagnostic accuracy of qualitative and quantitative HER2-PET assessment to predict HER2 expression by immunohistochemistry (IHC), including HER2-negative and -low breast cancer. Methods: This prospective multicenter trial enrolled treatment-naïve patients with newly diagnosed, non-rapidly progressive breast cancer across all molecular subtypes. Paired HER2 PET and [ 18 F]FDG PET/CT (FDG-PET) scans were acquired within a standardized 7-day. Tumor uptake was quantified using maximum standardized uptake value (SUVmax) and target-to-background ratios (TBR), and analyzed via general linear mixed models. Results: Between 2020 to June 2025, 240 patients were analyzed. Comparative analysis revealed superior diagnostic performance of whole-body HER2-PET versus FDG-PET for malignant lesion detection: sensitivity (97.6% vs 89.9%), specificity (83.9% vs 77.4%%), and overall accuracy (96.4% vs 88.9%). Quantitative HER2-PET analysis revealed significantly different SUVmax values across HER2 expression categories (median [IQR]: HER2-positive 13.1 [9.2-20.7] vs HER2-low 7.3 [5.1-8.9] vs HER2-negative 3.2 [2.7-4.5]; P<0.001), demonstrating strong correlation with IHC/FISH-defined HER2 status. In contrast, FDG-PET uptake showed no association with HER2 status (P=0.5). Among 10 discordant cases between HER2-PET and standard IHC/FISH: eight (80%) represented confirmed biological heterogeneity, with differing HER2 status across metastatic sites; two (20%) IHC 1+ cases exhibited PET-positive/FISH-amplified patterns, suggesting potential IHC false-negative results. Conclusions: [ 18 F]AlF-HER2-BCH PET/CT provides comprehensive HER2 profiling beyond single-biopsy IHC, offering new therapeutic insights for HER2-low/negative metastatic breast cancer through whole-body molecular visualization. Clinical trial information: NCT04547309 . Diagnostic performance of Al 18 F-HER2-BCH PET/CT and 18 F-FDG PET/CT, in the detection of primary and metastatic lesions in the evaluable imaging core population (N participant=240, N Biopsy-proven lesion=359). HER2 PET FDG PET Difference^ P Value* Participants True positive 207 199 -- -- True negative 26 22 -- -- False positive 2 4 -- -- False negative 5 15 -- -- Sensitivity 97.6% (94.7 to 99.1) 93.0% (88.6 to 95.9) 4.65% (0.32 to 8.98) 0.036* Specificity 92.8% (77.9 to 99.0) 84.6% (66.4 to 94.7) 8.24% (-6.8 to 23.3) 0.342 Positive predictive value 99.0% (96.6 to 99.9) 98.0% (95.2 to 99.4) 1.0 (-1.1 to 3.1) 0.398 Negative predictive value 83.9% (67.2 to 93.7) 59.5% (43.3 to 74.1) 24.4 (6.8 to 42) 0.009** Accuracy 97.1% (94.2 to 98.7) 92.1% (87.8 to 95.2) 5.0 (1.3 to 8.7) 0.008**

Assessment of CAR-T clinical trial availability and accessibility in the United States.

1558 Background: Chimeric antigen receptor T-cell (CAR-T) therapy has transformed outcomes in hematologic malignancies and is increasingly evaluated in solid tumors. CAR-T trials requires specialized infrastructure, potentially limiting access by geography and institutional concentration. We assessed the national distribution of CAR-T clinical trials, population distant from trial sites, and institutional clustering. Methods: All active U.S. interventional Phase I-III CAR-T trials (adult/pediatric) were identified via ClinicalTrials.gov (accessed January 2026). Trial locations were geocoded by ZIP code and linked to 2020 Census ZIP Code Tabulation Areas. Population access was defined as residence within 30 or 60 miles of a trial site using Haversine distance. Rurality (RUCA codes), income quartiles, and race/ethnicity (ACS 2023) were evaluated to assess disparities. Institutional concentration was assessed by the proportion of trial sites contributed by individual institutions. Results: Out of 264 identified trials, 189 were selected, conducted across 356 unique institutions. More than 85% of trials were early phase(Phase I–II), spanning lymphoma, myeloma, leukemia, and solid tumor indications. Access was limited. At a 30-mile radius, 43.3% of the U.S. population (~145 million individuals) lacked proximity to an active CAR-T trial site; 27.1% (~91 million) remained without access at 60 miles. Within 30 miles, lack of access affected 94.1% of rural residents, compared with substantially lower coverage in urban areas, and 63.2% of individuals in the lowest income quartile. By race and ethnicity, 37.4% of Hispanic and 32.0% of Black populations lacked access within 30 miles. These patterns were consistent when access was defined at 60 miles. Trial availability was highly centralized. The top 10 of institutions accounted for over 60% of all CAR-T trials, led by large academic centers such as City of Hope, Memorial Sloan Kettering, MD Anderson, Texas Children’s Hospital, and Stanford University. These centers are located in major metropolitan regions, resulting in dense trial availability for populations living nearby while leaving large geographic regions without access. In contrast, approximately 60% of participating institutions hosted only a single CAR-T trial, indicating limited diffusion of trial infrastructure beyond major academic hubs. Conclusions: CAR-T clinical trial access in the United States is restricted by both geographic exclusion and marked institutional concentration. As CAR-T indications expand, addressing these structural bottlenecks will be critical to improving equitable access and ensuring broader participation in cellular therapy development. Access to CAR-T clinical trials. Population Group 30 mi 60 mi Overall US Pop 43.3% 27.1% Rural Pop 94.1% 65.6% Lowest income quartile 63.2% 45.8% Hispanic 37.4% 25.9% Black 32.0% 20.1%

Global lung cancer burden attributable to environmental and occupational exposures.

e22677 Background: Lung cancer without tobacco exposure is a growing and preventable source of global morbidity driven by environmental and occupational exposures, with marked regional variation and unevenly addressed prevention strategies. Methods: Global Burden of Disease (GBD) 2023 estimates were used to quantify lung cancer disability-adjusted life years (DALYs) and age-standardized mortality rates (ASMRs) attributable to ambient particulate matter (PM2.5), household air pollution (HAP), residential radon, and occupational carcinogens across seven GBD super-regions from 1990–2023. Population attributable fractions (PAFs) and temporal trends were assessed, with sex-stratified analyses for radon and occupational exposures. Results: Air pollution accounted for the largest attributable lung cancer DALYs in low and middle-income regions, particularly South Asia, where PM2.5 (PAF 15.9%) and HAP (PAF 16.8%) together accounted for 32.7% of DALYs. In Sub-Saharan Africa, HAP alone accounted for 21.4%. Occupational carcinogens contributed the largest proportion in high-income regions (PAF 21.7%) and were associated with the highest DALY rate (230/100,000) and ASMR (13/100,000). From 1990–2023, HAP-attributable lung cancer burden declined across all regions (40–70%), consistent with transitions away from solid household fuels. In contrast, PM2.5-attributable DALY rates declined by 30–40% in high-income regions but increased by 40–50% in South Asia and the Middle East and North Africa. Radon- and occupational carcinogen–attributable burdens remained largely stable across most regions, although high-income regions experienced declines in radon (18.4%) and occupational carcinogen (32.1%) attributable DALY rates. Radon showed a modest male excess (male-to-female ratios 1.2–1.5), whereas occupational carcinogens demonstrated marked male predominance (often > 4). Conclusions: Despite substantial decrease in HAP related lung cancer burden, other modifiable factors including PM2.5, occupational carcinogen and radon exposure remain major contributors, specially in high-income regions where occupational exposures account for the largest PAF. These patterns suggest that reducing household energy exposure alone is insufficient. We need regional exposure targeted prevention strategies that extend beyond tobacco control and are relevant to oncology practice, survivorship and policy advocacy. Population-attributable fractions of lung cancer burden from environmental and occupational exposures across gbd super-regions. GBD Super-Region Air Pollution (PM2.5 + HAP) Residential Radon Occupational Carcinogens High-income 6.2 4.6 21.7 Central Europe, Eastern Europe & Central Asia 11.3 6.3 10.9 Latin America & Caribbean 12.5 4.1 11.9 North Africa & Middle East 21 3.7 12.7 South Asia 32.7 4.3 10 Sub-Saharan Africa 30.7 4.3 9.4 Southeast Asia, East Asia & Oceania 23.6 3.4 9.8

Gut microbiome characteristics in patients with hormone receptor–positive (HR+) HER2− breast cancer treated with adjuvant abemaciclib: Biomarker analysis of the TRADE trial.

563 Background: Adjuvant abemaciclib reduces recurrence risk and improves overall survival in patients with high-risk node-positive early HR+/HER2- breast cancer. However, treatment-emergent diarrhea may lead to dose modifications, treatment discontinuation, and impaired quality of life. The gut microbiome influences treatment response and toxicity across multiple cancer therapies; its role in abemaciclib-associated diarrhea remains undefined. We evaluated features of the gut microbiome during the initiation of adjuvant abemaciclib and any association with diarrhea. Methods: 90 patients with HR+/HER2− early breast cancer enrolled in the TRADE trial (NCT06001762) and were treated with dose-escalated adjuvant abemaciclib (50 mg BID x 14 days, then 100 mg BID x 14 days, then 150 mg BID onwards) plus endocrine therapy. Stool samples were collected at baseline, cycle 2 day 1 (C2D1), and end of treatment (EOT). Whole-genome shotgun metagenomic sequencing was performed, and species-level profiles were generated using MetaPhlAn4. Diarrhea was analyzed as a binary outcome (ever vs none). Associations between microbiome features and clinical variables were evaluated using linear mixed modeling with a random effect for repeated samples from the same individual (alpha diversity) or per-timepoint using PERMANOVA (beta diversity) and MaAsLin3 (species and genes, with false discovery rate correction). Models were adjusted for age, cancer stage, and type of endocrine therapy. Results: A total of 142 samples were analyzed (77 baseline, 58 C2D1, and 7 EOT; 77 total subjects, 28 with diarrhea, 49 without). Treatment was associated with a consistent microbiome shift from baseline to C2D1, characterized by a decline in alpha diversity (-0.26, 95% CI -0.39 to -0.14, p < 0.001) and a phylum-level shift with increased Bacteroidota and decreased Firmicutes abundance (p < 0.002 and p = 0.001, respectively, paired T-test). When assessing by timepoint, having any-grade diarrhea was not associated with differences in beta diversity or alpha diversity, nor with individual microbial species; pathway-level enrichment suggested diarrhea-associated functional trends that did not meet FDR significance. Limited EOT samples suggested partial recovery following treatment. Conclusions: Adjuvant abemaciclib induced consistent, treatment-related shifts in gut microbiome composition, characterized by reduced diversity and a phylum-level shift from Firmicutes to Bacteroidota, suggesting an early broad ecological treatment effect from exposure to a CDK4/6 inhibitor. This pattern has been reported in patients with inflammatory bowel disease. A species-specific diarrhea signature or predictive relationship between microbiome signature and the development of diarrhea was not observed. Further evaluation of the microbiome's evolution at EOT is ongoing. Clinical trial information: NCT06001762 .

Safranin-O dye removal by synthesized copper oxide nanoparticles: Isotherms, kinetics, and thermodynamics

Decoupling Chemo‐Mechanical Degradation for Scalable Silicon‐Based Solid‐State Batteries

ABSTRACT Silicon‐based solid‐state batteries (Si‐SSBs) have emerged as a pivotal next‐generation energy storage technology to surpass the energy density ceiling of conventional lithium‐ion batteries. However, their practical deployment is impeded by severe chemo‐mechanical degradation at the silicon anode–solid electrolyte interface due to the substantial volumetric expansion and interfacial contact loss. This review systematically reviews the underlying chemo‐mechanical failure mechanisms through advanced operando characterization, connecting atomic‐scale dynamics to macroscopic performance decay. This review then evaluates interfacial stabilization strategies for solid electrolytes through utilizing the viscoelastic buffers via in situ polymerization, and surface wettability and passivation of inorganic solid electrolytes, and the design of the mechanically reinforced polymer composites, with the objective of harmonizing ionic conductivity with mechanical compliance. Furthermore, the coupling between mechanical stress and electrochemical stability is elucidated through integrated material design and multiscale modeling. Finally, critical engineering considerations for scalability and manufacturing are discussed to bridge the gap between laboratory and practical Si‐SSBs.

Improving multidisciplinary care in the treatment of muscle-invasive bladder cancer: A quality improvement initiative.

e23315 Background: The rapidly evolving treatment landscape for bladder cancer has introduced several new therapy classes, including immunotherapies, antibody-drug conjugates, and kinase inhibitors. Integrating these therapies into routine clinical practice requires coordinated multidisciplinary decision-making including through regular tumor boards; however, real-world implementation remains variable. To address these gaps, PRIME implemented a data-driven quality improvement (QI) initiative aimed at optimizing patient care and enhancing multidisciplinary collaboration. Methods: Between July – October 2025, healthcare professionals (HCPs) who treat bladder cancer (n = 60; 82% physician, 3% NP/PA, 7% nurse, 8% pharmacist) from US academic and community oncology centers were surveyed to evaluate current practices and gaps in multidisciplinary treatment decision making and care in muscle invasive bladder cancer (MIBC). HCPs (n = 81) from participating sites then engaged in audit-feedback (AF) sessions to assess site-specific gaps identified in surveys and develop action plans for improvement. Results: Providers reported several key barriers to multidisciplinary care in MIBC, including lack of or infrequent multidisciplinary tumor boards (61%), inconsistent referrals from other specialists (51%), and coordination and communication challenges (methods, frequency or differing philosophies) (31%) . Although 85% of HCPs reported at least monthly participation in multidisciplinary tumor boards (MTBs) and 98% reported that MTBs included medical oncology, less than 70% reported inclusion of urology, pathology or radiation oncology, and less than 50% reported surgery or radiology participating. Additionally, while 78% of HCPs surveyed reported they were likely or extremely likely (4/5 on 5-point Likert scale) to recommend systemic treatment and 77% to recommend cystectomy for a patient with MIBC prior to discussing the case at an MTB, multidisciplinary case review was not uniformly embedded into clinical workflows, with only 24% of HCPs reporting that every MIBC case is discussed , 13% reporting only complex cases are discussed, and 3% reporting no MIBC patients are ever discussed in an MTB. To mitigate these barriers, HCPs reported that standardized referral workflows (63%), dedicated care coordinators or navigators (53%), and improved communication tools/platforms (37%) would most improve multidisciplinary collaboration and treatment planning in MIBC. Finally, HCP teams developed action plans in AF sessions to increase usage of electronic health record templates, increase MTB utilization for coordination of care, and document progress in implementing multidisciplinary care. Conclusions: Findings from this QI initiative suggest key opportunities to improve implementation of MTBs in patient testing and evaluation for patients with bladder cancer.

CheckMate-901: Association of tumor and peripheral biomarker profiles with durable disease control in metastatic urothelial carcinoma treated with nivolumab plus ipilimumab.

4525 Background: Patients (pts) with untreated, unresectable, or metastatic urothelial carcinoma (mUC) have limited therapeutic options and poor prognosis. The phase 3, global, open-label, randomized CheckMate 901 trial compared nivolumab plus ipilimumab (NIVO+IPI) vs gemcitabine–carboplatin/gemcitabine–cisplatin (gem–carbo/gem–cis) with previously untreated unresectable or mUC. A subset of pts treated with NIVO+IPI experienced durable disease control. This exploratory analysis seeks to identify biomarkers that might characterize pts deriving particular benefit from NIVO+IPI. Methods: Serum samples from CheckMate 901 (NIVO+IPI, n = 288–300; gem–carbo/gem–cis, n = 222–236) underwent multiplex analysis at Myriad RBM using Luminex and Simoa technology. Pts were stratified by progression-free survival (<8 months [early progressor; EP] vs ≥8 months [durable benefit; DB]), and clinical laboratory variables (granulocyte count [GRANC], hemoglobin [HB], thyroid-stimulating hormone [TSH], neutrophil-lymphocyte ratio [NLR]) and cytokine/protein biomarkers (alpha-1 antitrypsin [AAT], beta2-microglobulin [B2M], C-reactive protein [CRP], ferritin [FRTN], interleukin [IL]-2 receptor alpha [IL-2ra], IL-8, vascular cell adhesion molecule-1 [VCAM-1], and von Willebrand factor [vWF]) were compared in EP vs DB pts. In addition, tumor RNA-seq was used for gene signature analysis (NIVO+IPI, n = 134; gem–carbo/gem–cis, n = 117), including 8-gene epithelial–mesenchymal transition (EMT)/stroma core, 9-gene transforming growth factor beta (TGFβ) activation, 13-gene high-risk TGFβ, and cluster of differentiation 4 (CD4). Results: Distinct biomarker patterns emerged between EP and DB pts. Pts achieving DB vs EP with NIVO+IPI exhibited significantly lower levels of GRANC, TSH, and NLR, and higher HB (all P < 0.05). Similarly, levels of proteins AAT, B2M, CRP, FRTN, IL-2ra, IL-8, VCAM-1, and vWF were significantly lower in NIVO+IPI-treated pts achieving DB vs EP (all P < 0.05). RNA-seq analysis showed significant differences between NIVO+IPI-treated pts achieving DB vs EP for 8 EMT/stroma core and 9 TGFβ activation ( P < 0.05). Conclusions: In untreated, unresectable, or mUC, distinct clinical and cytokine/protein biomarker profiles were associated with DB vs EP with NIVO+IPI therapy. Pts with EP exhibit features in the tumor microenvironment and circulation suggestive of tumor-promoting inflammation, several of which have been linked to resistance to immune checkpoint blockade in prior analyses. These findings underscore the heterogeneity of mUC and the potential of routinely measured circulating analytes for risk stratification and treatment decisions. Prospective studies should confirm their clinical applicability. Clinical trial information: NCT03036098 .

A machine learning model for predicting recurrence and survival in stage I-III colorectal cancer.

e15518 Background: While 5-year overall survival for stage I-III colorectal cancer (CRC) is relatively high, the risk of recurrence remains substantial, ranging from approximately 7% in stage I to 29% in stage III disease. Accurate identification of high-risk patients is therefore critical for personalizing postoperative surveillance and adjuvant therapy. This study aimed to develop and validate a machine learning (ML) model using routinely available clinical data to precisely predict 5-year recurrence and mortality, enabling improved patient risk stratification. Methods: In this retrospective study, we analyzed data from 7,253 patients diagnosed with non-metastatic CRC who underwent curative-intent surgery at three clinical centers. From an initial set of 143 clinical, laboratory, and treatment-related variables extracted from electronic medical records, a parsimonious predictor set was selected through iterative CatBoost-based feature elimination. This process identified the most informative and clinically accessible features, including: NRAS mutation status, number of positive lymph nodes, preoperative CEA level, absolute lymphocyte count, hemoglobin, patient weight, age at treatment initiation, platelet count, leukocyte count, surgical approach, and pathological T-stage (pT). Model development and comparison against logistic regression, Random Forest, XGBoost, and LightGBM were conducted using stratified 5-fold cross-validation. Hyperparameters were optimized using the Optuna framework with AUC as the objective. The final model's performance was subsequently validated on an independent hold-out test cohort (20% of the data) to assess generalizability. Results: The optimized CatBoost model demonstrated superior predictive performance. Final cross-validated AUCs reached 0.84 for recurrence and 0.81 for mortality, representing a substantial improvement over the initial comparative analysis (AUCs of 0.749 and 0.770, respectively). The model maintained strong and robust discrimination on the independent validation set. Conclusions: We developed a validated ML model for accurate recurrence and mortality risk stratification in CRC. Crucially, all selected predictors are simple, routinely documented clinical parameters that can be automatically extracted from existing medical information systems. This facilitates seamless clinical integration, enabling data-driven identification of high-risk patients for personalized surveillance and adjuvant therapy, potentially improving long-term outcomes.

CAPTURE: A phase II trial evaluating alpelisib + fulvestrant vs capecitabine in advanced breast cancer patients with <i>PIK3CA</i> -mutant circulating DNA following CDK4/6 inhibition.

1064 Background: PIK3CA mutations are a key therapeutic target in ER+ advanced breast cancer (ABC) readily detectable through circulating tumor DNA (ctDNA). The PI3Kα inhibitor alpelisib is approved in ER+ HER2- ABC but it remains unclear when to incorporate it into the current treatment algorithm. To address this, we conducted a phase II study assessing superiority of alpelisib + fulvestrant vs capecitabine in ABC pts with detectable PIK3CA mutant ctDNA following CDK4/6i plus aromatase inhibition (AI). Methods: Eligible pts had ER+ HER2- ABC, prior progression on CDK4/6i+AI and PIK3CA mutations detected via ctDNA ddPCR. Pts were randomized 1:1 to receive alpelisib (300mg daily) + fulvestrant (Arm A), or capecitabine (1000-1250mg/m 2 BD D1-14 q21d) (Arm B); strata were prior chemo for ABC and visceral disease. Primary endpoint was progression free survival (PFS) defined from randomization until progressive disease per RECIST 1.1 or death. Secondary endpoints included objective response rate (ORR), clinical benefit rate (CBR) and adverse events (AE). Imaging was q8 weeks. Exploratory objectives included efficacy according to baseline ctDNA levels. Sample size was reduced from 140 to 66 pts randomized for 53 PFS events to detect median PFS of 9 vs 5 mo (HR=0.56;1-sided α=0.1,80% power). Interim inefficacy analysis (60% events) estimated HR&gt;1.0 and IDMC recommended early termination. Follow-up continued until all pts completed alpelisib. Reported are stratified Cox model HR (2-sided 80% CI) and 2-sided log rank test. Results: From July 2020-Oct 2024, 396 pts were screened, 113 (29%) had PIK3CA mutant ctDNA, 58 were randomized and 55 initiated treatment (Arm A: 29/30; Arm B: 26/28). Median age was 56 (32-82 yrs), 75% had visceral disease; 87% had no prior chemo for ABC. At final analysis, 43 PFS events occurred. Median PFS was 7.4 mo for alpelisib + fulvestrant vs 9.4 mo for capecitabine (HR 1.28, 80% CI: 0.84-1.96, p=0.45). For pts with visceral disease, median PFS was 5.4 mo (80% CI:3.7-5.1) for alpelisib + fulvestrant and 12.1 mo (80% CI:5.4-15) for capecitabine. ORR was 24.1% vs 50.0% and CBR 55.2% vs 61.5% for alpelisib + fulvestrant vs capecitabine. Overall, 79.3% pts on alpelisib + fulvestrant had grade 3/4 AEs with 24% grade 3 hyperglycemia vs 46.2% on capecitabine with 12% grade 3 palmar-plantar erythrodysesthesia. Six of 30 pts discontinued alpelisib due to AEs; 5 of the 6 continued fulvestrant. There was no evidence of association between baseline ctDNA levels (PIK3CA mutant copies/ml; median 90; range 2-13678) and PFS (HR 1.03; 90%CI:0.90-1.19). Conclusions: In our trial of pts with PIK3CA mutant ABC after progression on CDK4/6i plus AI, targeted therapy with alpelisib + fulvestrant was not superior to standard chemotherapy with capecitabine. Compared with the targeted therapy, capecitabine was better tolerated and resulted in longer PFS. Clinical trial information: ACTRN12619001117101.

Clinical impacts of the 2025 Southern California wildfires among childhood cancer patients and survivors: A pilot study.

e13515 Background: Climate change-related natural hazards threaten healthcare delivery, and wildfires are one such hazard. Wildfires may impede childhood cancer patients and survivors (survivors) from accessing specialized cancer care. Survivors may also be highly sensitive to wildfire smoke due to their young age and medical vulnerability. The 2025 Southern California wildfires, 14 fires that burned 57,000 acres in 1/2025, were an opportunity to evaluate how a major wildfire affected cancer care access and health symptoms among childhood cancer survivors. Methods: Healthcare providers at Children’s Hospital Los Angeles (CHLA) recruited caregivers (e.g. parents, grandparents) of children with a cancer history (on or off therapy) for a 1:1 interview from 11/2025-1/2026. Eligibility included: 1) have a child age &lt; 18 years with a cancer history (on or off therapy), 2) resided in Los Angeles County during 1/2025. Interviews asked about cancer care and general healthcare access, smoke exposure, and respiratory symptoms during 1/2025 as well as evacuation, information seeking, and current impacts on health and finances. We identified common themes and concerns. Results: Eleven interviews were completed. All caregivers expressed high distress about their child’s cancer history but moderate-low wildfire-related distress, which was largely in relation to how evacuation or road closures would impact CHLA access as well as cancer recurrence. Only three caregivers had appointments at CHLA in 1/25, and they adapted to road closures by adding 1-2 hours of travel time. One intentionally missed cancer care to prevent the child’s emotional distress related to wildfire damage. All caregivers noted visible smoke, ash, and odor; one noted a lingering odor and accompanying respiratory discomfort in their household until 4/2025. All were concerned about the child’s wildfire smoke exposure, but not their own. Caregivers of children with concomitant respiratory conditions expressed their greatest concern and took actions including masking and using air filters. Three caregivers noted respiratory symptoms (e.g. coughing) in their children during 1/2025, but none required urgent care. Three caregivers received evacuation alerts, and the two that evacuated noted the lack of shelters appropriate for medically vulnerable children. Caregivers obtained health information and evacuation information from television, friends, family, and phone-based apps. Conclusions: Wildfires pose challenges to deliver pediatric cancer care due to road closures and evacuation notices, and respiratory symptoms are another concern. Healthcare providers should communicate information about wildfire smoke avoidance to patients. Our findings underscore the need for disaster preparedness and communication strategies that ensure continuity of pediatric cancer care and minimize wildfire smoke exposure.

Pulmonary metastasis in chondroblastoma: An institutional cohort and meta-analysis.

e23506 Background: Chondroblastoma is a benign bone tumor with rare potential for pulmonary metastasis. The true incidence of lung metastasis remains uncertain due to inconsistent surveillance practices. We sought to determine the rate of pulmonary metastasis in chondroblastoma and characterize current chest imaging practices using an institutional cohort and literature meta-analysis. Methods: We conducted a retrospective review (2010-2025) of 19 chondroblastoma patients at our institution and stratified them by whether chest imaging was obtained. A meta-analysis of 23 published studies comprising 411 chondroblastoma patients was also conducted. Primary outcomes were pulmonary metastasis and mortality. Imaging utilization and clinical characteristics were compared between cohorts. Results: Sex distribution was nearly identical between both groups (institutional 62.3% male vs literature 65.9% male, p = 0.994). The mean age of the institutional cohort was 18.6 ± 5.5 years. Pulmonary metastasis was identified in 1 of 19 institutional patients (5.3%) and in 23 of 411 patients from the literature (5.6%), with no difference between cohorts (p = 1.000). Chest imaging was performed in 42.1% of institutional patients and reported in 47.8% of literature cohorts (p = 0.82). Imaging was performed for routine staging in 87.5% of institutional cases and 63.6% of literature cases (p = 0.338). Among imaged patients at our institution, median tumor size was larger compared with non-imaged patients (3.66 vs 2.20 cm), approaching statistical significance (p = 0.077). The median time from diagnosis to chest imaging was 145 days at our institution. No institutional patients had lung metastasis at initial presentation compared with 3.41% reported in the literature (p = 1.000). All institutional patients were alive and disease-free at last follow-up, and only 3 cases of disease-specific mortality were reported in the literature (0.73%). Conclusions: Although exceedingly rare, lung metastasis in chondroblastoma does occur. Current surveillance practices for chondroblastoma metastasis are inconsistent and lack standardization. Despite metastasis detection, prognosis remains excellent, with only three reported cases in the literature resulting in mortality. These findings suggest that implementing image surveillance may be needed to better define true metastasis rates and guide optimal management strategies.

Intracranial Rosai–Dorfman disease: Systematic review of published case reports and small series (2000–2025).

e14108 Background: Intracranial Rosai-Dorfman disease (RDD) is a rare histiocytic disorder that often mimics meningioma and lacks standardized management. We synthesized recent evidence on presentation, imaging/pathology, treatment, outcomes, and predictors of recurrence. Methods: This systematic review was conducted in accordance with PRISMA guidelines and a registered protocol (PROSPERO: CRD420251107147). PubMed, EMBASE, Scopus, and Web of Science were searched for case reports and case series (≤10 patients) describing histopathologically confirmed intracranial RDD. Two reviewers independently screened studies and extracted data. Descriptive statistics summarized patient, imaging, pathologic, treatment, and outcome variables. Group comparisons were performed using chi-square testing. Cox proportional hazards models were used to identify independent predictors of recurrence. Results: From 1,897 records, 176 studies including 261 patients were identified (226 adults and 35 children; 65.5% male). Magnetic resonance imaging was available in nearly all cases; 88.8% of lesions were dural-based, and 32.2% of patients had multiple intracranial lesions. Imaging typically demonstrated T1 isointensity, T2 iso- or hypointensity, and homogeneous enhancement. Lesions were described as meningioma-like in 82.0% of cases, and lytic bone erosion was reported in 41.8%. Histopathologic examination showed emperipolesis in 97.0% of cases, with S-100 and CD68 positivity and CD1a negativity in approximately 96%. Surgical resection was the primary treatment approach, with gross total resection (GTR) achieved in 58.0% of patients. Adjuvant corticosteroids, radiotherapy, and chemotherapy were administered in 33.6%, 16.7%, and 10.5%, respectively. Over a mean follow-up of 30.7 months, 84.2% of patients experienced functional improvement, recurrence occurred in 16.5%, and mortality was 5.7%. On multivariable analysis, the presence of multiple intracranial lesions was independently associated with higher recurrence risk (adjusted hazard ratio [HR] 5.08; 95% CI, 2.43–10.61; P &lt; 0.001), while GTR was associated with a lower risk of recurrence compared with non-GTR surgery (adjusted HR 0.44; 95% CI, 0.21–0.89; P = 0.022). Conclusions: Intracranial RDD most often presents as a dural-based, meningioma-like mass and requires histopathologic confirmation for diagnosis. Surgical resection remains the mainstay of treatment; however, lesion multiplicity is the strongest predictor of recurrence, supporting the need for long-term surveillance. The role of adjuvant therapies remains uncertain, emphasizing the importance of multicenter registries, standardized reporting, and molecular profiling to guide future management.

HER2 expression heterogeneity (IHC/FISH) in breast cancer: A systematic review and meta-analysis of prognostic implications.

e13023 Background: Human epidermal growth factor receptor 2 (HER2/ERBB2) is a validated predictive and prognostic biomarker in breast cancer that guides the use of HER2-targeted therapies. While current guidelines use a binary IHC/FISH classification, increasing evidence suggests clinically relevant intratumoral HER2 expression heterogeneity. This study reviewed the association between intratumoral HER2 expression heterogeneity and clinical outcomes in patients with HER2-positive breast cancer. Methods: PubMed, Scopus, Embase, Cochrane, and ClinicalTrials.gov were searched from 2018 to December 2026 in accordance with PRISMA guidelines for studies reporting the prognostic implications of HER2 expression heterogeneity (IHC/FISH) in breast cancer. Dichotomous outcomes were pooled using risk ratios (RRs) with 95% confidence intervals (CIs), while continuous outcomes were analyzed using mean differences (MDs) or standardized mean differences (SMDs) with 95% CIs. Analyses were conducted using RevMan 5.4.0. Heterogeneity was assessed using I² statistics (I² &gt; 50%). A p-value &lt; 0.05 was considered significant. Results: High intratumor HER2 heterogeneity emerged as a robust negative prognostic indicator across seven studies (n = 2847)of HER2+ breast cancer patients primarily aged 50+. High heterogeneity significantly correlated with diminished clinical outcomes, notably reduced pathological complete response rates (49.6%) and increased treatment resistance (P = 0.007). Meta-analysis confirmed 4.3 fold risk of poorer distant recurrence-free survival (HR: 4.26; 95% CI: 1.69–6.94; P = 0.00062), worse overall survival (HR: 2.02; 95% CI: 1.09–3.74; P = 0.025), all comparing heterogeneous vs. homogeneous HER2 IHC/FISH expression. Conclusions: This systematic review and meta-analysis suggested that intratumoral HER2 IHC/FISH expression heterogeneity is not just a pathological nuance but a clinically important feature that meaningfully influences outcomes in HER2-positive breast cancer. Across seven studies, largely involving patients aged 50 years and older, tumors with greater HER2 IHC/FISH heterogeneity consistently showed poorer responses to HER2-targeted therapy and worse long-term outcomes. Overall, a binary HER2 classification oversimplifies disease biology, and routine IHC/FISH heterogeneity assessment should be integrated into prognostic models and trial design.

Post-residency career intentions among oncology and related specialty residents in Armenia.

e21012 Background: In Armenia, postgraduate training for Medical, Radiation, and Surgical oncology is delivered through a single centralized department. Ongoing reforms transition the system towards a fellowship-style model consisting of one year of internal medicine followed by three years of oncology-focused specialty training. However, residency training remains fee-based, with no financial remuneration. This study aimed to assess post-residency career intentions among Armenian residents in oncology and related specialties. Methods: A cross-sectional study was conducted among Yerevan State Medical University residents training in Oncology and related specialties (i.e., Pathology, Gynecology Oncology, Thoracic and Abdominal Surgery, and Radiology). Participants were recruited using convenience sampling. Data were collected from December 16, 2025, to January 26, 2026, via an e-survey and administered through Google forms, adapted from Tomas Ferreira et al. (2023). Data were analyzed using SPSS. Results: In total, 51 residents were enrolled (M:F=1:2.4). Median age was 25 years (range 24–31). Most respondents were from the capital (76%). At the time of the survey, 52.9% were employed: 29.4% in clinical work and 5.9% in research; 17.6% were involved in non-medical activities. Residents in oncology, pathology, and surgery accounted for 62.7%, 11.8%, and 25.5%, respectively, and 44.4% of residents from other specialties reported plans to pursue this specialty. Overall, 90.2% saw their future in clinical practice, while smaller proportions planned careers in healthcare management (7.8%) or research (2.0%); all planned to obtain a second specialty qualification. Low salary and work–life balance were rated as important or very important reasons for leaving clinical practice. The desire to leave clinical practice differed significantly by specialty (p=0.013), with pathologists more often preferring a clinical setting. Most respondents were confident in their specialty choice (66.6%: very certain 51.0%, somewhat certain 15.7%), and 70.6% planned to practice in Armenia. Among those planning to leave, competitive salary (100%) and higher quality training (66.6%) were the main factors. Work–life balance (66.7%), family abroad (100%), and shorter training duration (66.6%) were reported as less important. Overall, 66.7% of those planning to leave reported no intention to return to Armenia after completing training. Conclusions: According to the study results, the majority of the respondents were planning to practice Oncology in Armenia. However, one of the most important factors contributing to the termination of medical practice or leaving Armenia was low or absence of revenue. Hence, interventions are required to ensure long-term specialty retention among young specialists.

Effect of an EHR-facilitated symptom surveillance intervention on rehabilitation utilization in cancer patients.

1587 Background: Rehabilitation services (physical medicine and rehabilitation [PM&amp;R], physical therapy [PT], occupational therapy [OT], and speech-language pathology [SLP]) improve physical function, symptom burden, and quality of life for cancer survivors, yet utilization remains low due to persistent barriers. The collaborative care model–based Enhanced Electronic Health Record–Facilitated Cancer Symptom Control (E2C2) pragmatic trial improved symptom burden and palliative care utilization. Using post hoc exploratory analyses of population-level trial data, we evaluated whether this EHR-facilitated approach also increased rehabilitation service use. Methods: E2C2 was conducted across Mayo Clinic Midwest medical oncology practices using an unblinded, stepped-wedge, cluster-randomized cohort design. The EHR-embedded intervention targeted SPPADE symptoms (sleep interference, pain, impaired physical function, anxiety, depression, and fatigue). Symptoms and physical function were assessed using 11-point numerical rating scales linked to oncology encounters. Patients received self-management materials for moderate symptoms (4–6/10) and care manager support for severe symptoms (7–10/10). All outpatient rehabilitation encounters across the Mayo Clinic Enterprise were abstracted. Mixed-effects Poisson models estimated intervention effects on rehabilitation visits during the first 365 days of exposure. Results: Between March 28, 2019, and January 31, 2023, 50,207 patients were enrolled. Intervention and control conditions included 31,719 and 28,938 patients, respectively, with 10,450 contributing data to both periods. Cross-sectionally, 25% of participants reported moderate and 9% severe physical function impairment. Of 91,220 rehabilitation encounters scheduled, 57,669 (63.3%) were completed. Completed encounters were predominantly PT (68.3%), followed by OT (16.4%), physician visits (9.4%), and SLP (4.2%). The intervention was associated with a more than fourfold increase in completed rehabilitation encounters during the six weeks following index symptom assessment (adjusted mean 2.22 [95% CI, 0.14–4.29] vs 0.54 [0.40–0.68]; incidence rate ratio 4.11 [1.43–11.82]; P =0.009). Scheduled encounters also favored the intervention but did not reach statistical significance. Conclusions: An EHR-embedded symptom surveillance intervention increased short-term rehabilitation service completion, but overall utilization remained low relative to the prevalence of physical function impairment, highlighting the need for system-level strategies to translate symptom identification into sustained rehabilitation engagement.