Browse Articles
Discover research articles across all indexed journals
AI-driven multi-scale histopathologic classification of epithelial ovarian cancer to support precision oncology.
5608 Background: Epithelial ovarian cancer (EOC) comprises five histologically distinct subtypes associated with divergent molecular profiles, therapeutic responsiveness, and prognostic outcomes. Accurate histologic classification is essential for precision oncology but remains limited by morphologic overlap and interobserver variability. Deep learning architectures capable of multi-scale spatial feature extraction may enable objective and reproducible subtype discrimination. We evaluated the performance of a multi-scale factorized convolutional neural network (InceptionV3) for automated EOC subtype classification. Methods: Digitized hematoxylin–eosin–stained histopathology images were obtained from the University of British Columbia Ovarian Cancer Subtype Classification and Outlier Detection (UBC-OCEAN) dataset, including high-grade serous, low-grade serous, endometrioid, clear cell, and mucinous carcinomas. An ImageNet-pretrained InceptionV3 architecture was fine-tuned using a five-class classification head. The model leverages parallel convolutional branches with factorized asymmetric kernels to simultaneously capture fine cellular features and global tissue architecture. Images were normalized and augmented. Training was performed using 299×299 inputs, Adam optimizer (learning rate 1×10⁻⁴), cross-entropy loss, and six epochs. Performance was assessed using accuracy, precision, recall, F1-score, and macro-averaged AUROC. External clinical validation was conducted by 57 board-certified gynecologic pathologists and oncologists across 14 geographic regions. Results: The model achieved an overall accuracy of 95.0% (491/517 test images). Subtype-specific F1-scores were 0.96 for high-grade serous carcinoma, 0.94 for endometrioid, 0.93 for clear cell, 0.92 for mucinous, and 0.88 for low-grade serous carcinoma. Macro-averaged AUROC was 0.957 (95% CI, 0.948–0.966). Computational efficiency was preserved with 23.9 million parameters and 11.6 billion FLOPs. Independent clinician validation demonstrated strong concordance with AI predictions, with mean inter-rater agreement of 95.3% (κ=0.936; 95% CI, 0.923–0.949), consistent across institutional practice settings. Conclusions: Multi-scale factorized convolutional modeling enables near-expert performance for automated EOC subtype classification by integrating cellular- and tissue-level morphologic features. High inter-rater concordance with an international specialist cohort supports translational reliability. This framework represents a scalable decision-support tool with potential to improve diagnostic standardization and precision therapeutic stratification. Prospective multi-center validation is warranted.
Phase II study of adjuvant anlotinib for esophageal squamous cell carcinoma with residual disease following neoadjuvant therapy.
e16138 Background: Patients with esophageal squamous cell carcinoma (ESCC) who fail to achieve a pathological complete response (pCR) following neoadjuvant therapy face a poor prognosis. While adjuvant immunotherapy is a standard of care, alternative strategies are needed for diverse patient populations. We evaluated the efficacy and safety of adjuvant anlotinib, a multi-target tyrosine kinase inhibitor, in ESCC patients with residual pathologic disease. Methods: This prospective, single-arm, phase II trial enrolled patients with locally advanced ESCC harboring residual disease (≥ypT1 and/or ypN+) following neoadjuvant therapy (chemoradiotherapy, chemotherapy, or chemoimmunotherapy) and R0 resection. Patients received oral anlotinib (12 mg once daily; 2 weeks on/1 week off) for up to 8 cycles. The primary endpoint was the 6-month progression-free survival (PFS) rate. Secondary endpoints included safety (CTCAE v5.0), overall survival (OS), and treatment completion rate. Survival outcomes were estimated using the Kaplan-Meier method. Results: From December 2023 to January 2026, 17 patients were enrolled (study terminated early due to slow accrual; planned N = 39). All patients were male (median age: 63.0 years). Preoperative regimens included chemoradiotherapy (35.3%, 6/17), chemoimmunotherapy (23.5%, 4/17), chemotherapy (17.6%, 3/17), and chemoradiotherapy plus immunotherapy (23.5%, 4/17). All patients received anlotinib and comprised the safety/efficacy population (median follow-up: 10.2 months). The most common treatment-related adverse events were hand-foot syndrome (29.4%, 5/17) and diarrhea (29.4%, 5/17), all of which were grade 1-2. Grade 1 hypertension was reported in 1 patient (5.9%). No grade ≥3 adverse events were observed. Eleven patients (64.7%) completed 8 cycles; 3 (17.6%) discontinued due to toxicity, and 3 (17.6%) remained on treatment at data cutoff. Disease recurrence occurred in 3 patients (17.6%), involving lung, liver, and abdominal lymph nodes. Median PFS was 11.3 months. One death (5.9%) occurred; median OS was not reached. The 1-year OS rate was 94.1%. Conclusions: Adjuvant anlotinib demonstrated a highly favorable safety profile and promising preliminary efficacy in ESCC patients with residual disease. Despite the limited sample size, anlotinib represents a viable, chemotherapy-free adjuvant option that may complement current immunotherapy strategies. Further investigation is warranted. Clinical Trial Registration: MR-51-23-050661 (National Healthcare Security Information Platform Medical Research Registration and Filing System) Clinical trial information: MR-51-23-050661. Clinical trial information: MR-51-23-050661 .
Artificial intelligence–integrated profiling of JAK-STAT, MAPK, and RTK–RAS signaling alterations in chemotherapy-treated early-onset colorectal cancer among disproportionately affected populations.
e15510 Background: Early-onset colorectal cancer (EOCRC; < 50 years) is increasing rapidly, with disproportionate burden among Hispanic/Latino (H/L) populations that remain underrepresented in molecular studies. FOLFOX (folinic acid, fluorouracil, oxaliplatin) is a standard chemotherapy backbone in CRC, yet how major oncogenic signaling pathways respond to treatment across ancestry and age remains incompletely characterized. The JAK-STAT, MAPK, and RTK–RAS pathways are central regulators of tumor growth, survival, and therapeutic response. We conducted an artificial intelligence (AI)–enabled, multi-pathway analysis. Methods: Somatic mutation and clinical data from 2,515 CRC patients were analyzed across public datasets. Patients were stratified by age at diagnosis (EOCRC vs. late-onset CRC [LOCRC]), ancestry, and FOLFOX exposure. JAK-STAT, MAPK, and RTK–RAS pathway alterations were defined using curated gene sets and analyzed categorically (altered vs. wild-type). Mutation frequencies were compared using Fisher’s exact or chi-square tests. Overall survival (OS) was assessed using Kaplan–Meier analyses. All analyses were exploratory and unadjusted for multiple testing. AI-HOPE, AI-HOPE-JAK-STAT, AI-HOPE-MAPK, and AI-HOPE-RTK–RAS conversational AI platforms enabled natural language–driven cohort construction, stratification, and multi-parameter analysis. Results: Pathway alterations demonstrated substantial heterogeneity across ancestry, age, and treatment context. JAK-STAT alterations were more frequent in FOLFOX-untreated H/L EOCRC compared with treated counterparts and enriched in untreated EO H/L relative to NHW patients. MAPK pathway alterations varied by subgroup, with FGFR3, NF1, and RPS6KA6 enriched in untreated EO H/L tumors, while PDGFRB alterations were more frequent in FOLFOX-treated EO H/L disease. RTK–RAS alterations also showed treatment-linked shifts, including reduced ERBB2 and NF1 frequencies in FOLFOX-treated EO H/L patients and distinct patterns among NHW strata. Survival analyses revealed pathway-specific OS associations in selected subgroups, particularly among NHW patients, though signals in H/L EOCRC were limited by small mutation-positive sample sizes. Conclusions: This AI-integrated analysis highlights context-dependent variation in JAK-STAT, MAPK, and RTK–RAS pathway alterations across ancestry, age, and FOLFOX treatment status in CRC. The findings underscore the biological heterogeneity of EOCRC and emphasize the importance of ancestry- and treatment-aware biomarker evaluation. Conversational AI platforms facilitated rapid, reproducible multi-pathway discovery and support hypothesis generation for future validation studies aimed at advancing equitable precision oncology.
Prognostic impact of circulating tumor DNA in pancreatic ductal adenocarcinoma: A meta-analysis.
e16430 Background: Pancreatic ductal adenocarcinoma (PDAC) is an aggressive malignancy and the third leading cause of cancer-related death in the US, with poor prognosis across stages. Circulating tumor DNA (ctDNA) is a promising non-invasive biomarker. This systematic review and meta-analysis evaluates the prognostic impact of baseline (pre-treatment) ctDNA status on progression-free survival (PFS) and overall survival (OS) in PDAC. Methods: We systematically searched major databases for studies assessing baseline ctDNA status before treatment in PDAC patients. PFS was reported in eight studies, OS in 14 (seven overlapping). Pooled hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated using random-effects models. Subgroup analyses stratified by disease stage (localized/resectable vs. advanced/metastatic). Heterogeneity was assessed via I² statistic. Results: A total of 1,050 PDAC patients (median age 69.9 years, range 34–87.3) were included: 482 with resectable/localized disease, 568 with advanced/metastatic. Undetectable baseline ctDNA was reported in 520 patients. Pooled median PFS/HRs showed detectable ctDNA associated with significantly worse PFS across eight studies (HR 2.11, 95% CI 1.72–2.60; low heterogeneity). Subgroup analysis confirmed similar impact in resectable (HR 2.11, 95% CI 1.72–2.60) and advanced disease (HR 2.11, 95% CI 1.64–2.72). Detectable baseline ctDNA was also associated with inferior OS across 14 studies (pooled HR 2.23, 95% CI 1.80–2.77; I² = 0%). Subgroup analyses demonstrated consistent poorer OS in resectable disease (HR 2.36, 95% CI 1.37–4.08; I² = 0%) and advanced disease (HR 2.23, 95% CI 1.72–2.89; I² = 20.8%). Conclusions: Baseline ctDNA positivity is strongly associated with inferior PFS and OS in PDAC, independent of disease stage, with low heterogeneity across studies. These findings underscore ctDNA as a robust, stage-agnostic prognostic biomarker and support its integration into clinical practice for improved risk stratification, treatment personalization, and potential guidance of adjuvant/neoadjuvant strategies. Prospective validation and standardization of ctDNA assays are warranted.
Defining safe pre-chemotherapy absolute neutrophil count thresholds in early-stage breast and colon cancer: A real-world retrospective study conducted in a single hospital setting in Saudi Arabia.
e12617 Background: A pre-chemotherapy absolute neutrophil count (ANC) threshold of ≥1.5 × 10⁹/L is widely used in clinical practice, despite limited real-world evidence supporting its role in preventing febrile neutropenia (FN), particularly in patients with ethnic neutropenia. In patients with early-stage breast and colon cancer, strict adherence to this cutoff may lead to treatment delays without a clear safety advantage, consistent with previous observations of inferior clinical outcomes among patients with baseline neutropenia. Methods: We conducted a retrospective cohort study of adult patients at a single hospital in Saudi Arabia, predominantly of Arab ethnicity with early-stage breast or colon/rectal cancer who received adjuvant chemotherapy between 2018 and 2025. Patients were grouped according to febrile neutropenia (FN) occurrence and baseline pre-chemotherapy ANC (≥1.5, 1.0–1.49, and < 1.0 × 10⁹/L). FN incidence and relative risk (RR) were assessed overall and stratified by cancer type. Results: The study included 732 patients, of whom 96 (13.1%) developed FN and 636 (86.9%) did not. Among patients without FN, baseline ANC was ≥1.5 ×10⁹/L in 560 patients (88.1%), 1.0–1.49 ×10⁹/L in 60 (9.4%), and < 1.0 ×10⁹/L in 16 (2.5%).Of the 96 FN events, 76 (79.2%) occurred in patients with breast cancer and 20 (20.8%) in those with colon/rectal cancer. Among breast cancer patients who developed FN, ANC was ≥1.5 ×10⁹/L in 62 patients (81.6%), 1.0–1.49 ×10⁹/L in 7 (9.2%), and < 1.0 ×10⁹/L in 7 (9.2%). In colon/rectal cancer patients with FN, ANC was ≥1.5 ×10⁹/L in 14 patients (70.0%) and 1.0–1.49 ×10⁹/L in 6 (30.0%), with no FN events observed at ANC < 1.0 ×10⁹/L.Using ANC ≥1.5 ×10⁹/L as the reference, FN incidence increased from 12.2% to 17.8% and 30.4% at lower ANC categories, corresponding to relative risks of 1.46 and 2.49, respectively. Despite these increases, most FN events across both cancer types occurred in patients with ANC values meeting conventional treatment thresholds. Conclusions: In this real-world cohort of patients with early-stage breast and colon cancer, the majority of febrile neutropenia events occurred despite pre-chemotherapy ANC values considered acceptable for treatment initiation. The risk of febrile neutropenia varied by cancer type and was not consistently predicted by baseline ANC alone. These findings suggest that rigid ANC thresholds may have limited clinical utility and support a more individualized, risk-adapted approach to chemotherapy decision-making. Prospective trials are needed to further define safe pre-chemotherapy ANC thresholds in patients with baseline neutropenia.
Survival outcomes with immune checkpoint inhibitors among patients with cancer from minoritized racial and ethnic groups: A systematic review and meta-analysis.
1648 Background: Immune checkpoint inhibitors (ICIs) have transformed the treatment landscape across multiple malignancies; however, minoritized racial/ethnic groups remain underrepresented in pivotal trials, limiting the generalizability of efficacy estimates. Whether treatment outcomes with ICIs differ across these groups remains incompletely characterized. Thus, we conducted a systematic review and meta-analysis of published randomized controlled trials (RCTs) to evaluate differences in overall survival (OS) and progression-free survival (PFS) among these groups treated with ICIs. Methods: We systematically searched MEDLINE, Embase, Cochrane CENTRAL, and ClinicalTrials.gov for RCTs published through December 31, 2024, that evaluated ICIs in adult patients with cancer and reported OS or PFS by race/ethnicity. Risk of bias was independently assessed using the Cochrane Risk of Bias tool for randomized trials (RoB 2). Hazard ratios (HRs) and 95% confidence intervals (CIs) were extracted. Between-study heterogeneity was quantified using I². Random-effects meta-analyses were performed using Review Manager (RevMan) and stratified by: non-White and White patients; Asian, Black, and White patients; Asian and non-Asian patients; and Hispanic and non-Hispanic patients. Publication bias was evaluated using funnel plots. This review followed Cochrane methodology and PRISMA guidelines and was registered in PROSPERO (CRD420251027941). Results: A total of 97 trials comprising 59,949 patients were included. For OS, ICIs were associated with significant benefit in both non-White (HR 0.70, 95% CI 0.63–0.77; p<0.001) and White patients (HR 0.75, 95% CI 0.67–0.84; p<0.001). Significant OS benefit was also observed in analyses stratified by Asian (HR 0.81, 95% CI 0.76–0.86; p<0.001) and White patients (HR 0.86, 95% CI 0.82–0.91; p<0.001), and by Asian (HR 0.83, 95% CI 0.76–0.91; p<0.001) and non-Asian patients (HR 0.79, 95% CI 0.73–0.86; p<0.001). Similar patterns were observed for PFS. Estimates for Black and Hispanic patients were not statistically significant, reflecting limited sample sizes. Heterogeneity ranged from low to substantial (I² 0–87%), with higher heterogeneity observed in PFS analyses. Only one trial reported race-specific survival outcomes for American Indian/Alaska Native (AI/AN) patients, and none reported outcomes for Native Hawaiian/Pacific Islander (NH/PI) patients, precluding analyses for these groups. Conclusions: ICIs were associated with improved OS and PFS across racial/ethnic groups, with no consistent evidence of differential treatment effects. However, the persistent underrepresentation of minoritized populations, particularly AI/AN, Black, Hispanic, and NH/PI patients, limits definitive conclusions and highlights the need for more equitable representation in oncology trials.
Results from a pilot feasibility study of a novel community health worker–led intervention delivering stepped palliative care and care navigation for patients with pancreatic ductal adenocarcinoma.
12081 Background: Pancreatic Ductal Adenocarcinoma (PDAC) is associated with substantial morbidity, poor quality of life, and poor overall survival. Early palliative care (PC) has been shown to improve quality of life and reduce symptom burden in PDAC, but specialty PC services are often unavailable for this fast-progressing illness, and even then are only utilized by a minority of PDAC patients. To address this challenge, we developed a digitally enhanced CHW-led intervention to deliver stepped PC to patients with PDAC and to report its feasibility and preliminary efficacy. Methods: Adults (aged ≥18 years) with newly diagnosed PDAC (within 2 weeks of diagnosis) were enrolled in a single-arm pilot feasibility study from January to August 2025 (N = 48). The intervention involved early palliative care engagement with targeted escalation (stepping) for patients with untreated symptoms, leading to expedited PC visits, patient education, care navigation, and providing targeted financial resources for patients. The digital component involves using short educational videos and the Twilio SMS to administer the Edmonton Symptom Assessment Scale weekly, with patients with high symptom scores being stepped up to PC visits. The CHW-led intervention was delivered over 12 weeks. Feasibility outcomes were patient engagement with the CHW and completion of the intervention. The acceptability outcome was defined using the AIM-IAM-FIM instrument. Secondary outcomes include symptom burden (NCCN/FACT Hepatobiliary Cancer Symptom Index-18), Advanced Care Planning (ACP) compliance rates, and financial toxicity (measured by the COST instrument). Results: The median age of pts (N = 48) was 69 (IQR 62-76), and 29 (60%) were Hispanic. 25 (52%) had metastatic PDAC, 14 (29.1%) had borderline resectable PDAC, and 9 (18.8%) had resectable PDAC at diagnosis. Four patients died before the intervention was completed, while the remaining patients completed it. Regarding acceptability, 94% strongly agreed that the intervention was acceptable. Themes identified from exit-interviews included the critically important role of the CHW in the process, including appointment scheduling, especially as symptoms worsened. Patient’s symptom burden (median 68 vs. 55, P = 0.02) and rates of financial toxicity (39% vs. 25%, p = 0.01) decreased from baseline to 12 weeks, while 83% of patients completed ACP by 12 weeks. Conclusions: Our digitally-enhanced CHW-led intervention delivering stepped PC was feasible and acceptable to patients with PDAC. The intervention was associated with improvement in symptom burden, reduction in financial toxicity, and higher ACP planning rates. Future prospective comparative studies are needed to understand the intervention’s ability to help manage the morbidity associated with PDAC.
Adaptive radiotherapy as a strategy for universal dosimetric accuracy across tumor sizes and locations in lung SBRT for NSCLC.
e20102 Background: Inter-fractional anatomical changes in lung SBRT raise concerns about dosimetric accuracy and potential target under-dosage. Adaptive radiotherapy (ART) is a proposed strategy to correct these variations. This study aims to quantify the dosimetric degradation from these changes and assess the necessity of ART to restore plan quality. Methods: This retrospective planning study retrospectively included 33 lung SBRT patients, each prescribed with 50Gy/5Fx radiation therapy on an emulator. For each patient, three plans were generated: 1) the original approved Reference plan (REF); 2) the non-adaptive Scheduled plan (SCH), created by recalculating the REF plan on the daily anatomy; and 3) the re-optimized Adaptive plan (ADP) on the same daily anatomy. Dosimetric parameters for planning target volume (PTV) coverage (PTV D99%, PTV D95%) and organs at risk (OARs) were compared using the Wilcoxon signed-rank test. The ART benefit (ADP-SCH improvement) was correlated with tumor diameter and lobe location using the Wilcoxon rank-sum test. Results: Compared to the REF plan, the SCH plan showed significant target coverage degradation. The median PTV D99 dropped from 48.0 Gy (REF) to 43.1 Gy (SCH) (p < 0.001) . The ADP plan significantly recovered this loss, restoring the median PTV D99 to 47.5 Gy (p < 0.001 vs SCH), a value nearly identical to the original REF plan. In terms of PTV D95, nine patients showed dose degradation over 5.0 Gy in SCH plans with a maximum of 14.5 Gy. Crucially, ADP plans corrected a “false sparing” effect seen in SCH plans for serial OARs; for example, median spinal cord Dmax increased from 8.3 Gy (SCH) to 9.15 Gy (ADP) (p = 0.022), aligning with the REF plan's intent. This dosimetric benefit was necessary regardless of tumor characteristics. There was no significant correlation between the magnitude of PTV D99 improvement and tumor diameter (p = 0.72) or lobe location (e.g., Upper vs. Mid/Lower, p = 0.13, Upper/Mid vs. Lower, p = 0.82). Conclusions: Inter-fractional anatomical variations in lung SBRT lead to clinically significant target under-dosage when a standard IGRT workflow (SCH) is used. ART (ADP) effectively and accurately restores target coverage to the intended (REF) level. The need for ART is ubiquitous and cannot be predicted by baseline tumor size or location. These findings suggest that a verification-based ART strategy is dosimetrically necessary to ensure treatment quality and safety in lung SBRT.
Expression patterns of circulating microRNAs in classification models: New opportunities in brain tumor diagnosis.
2051 Background: Diffuse gliomas are the most common primary brain tumors and, according to the 2021 WHO classification, are divided into three main types: IDH-wild-type glioblastomas, IDH-mutant astrocytomas, and IDH-mutant oligodendrogliomas with 1p/19q codeletion. Oligodendrogliomas are neuroepithelial tumors of the central nervous system that have histological and molecular characteristics distinct from other subtypes of diffuse gliomas. The problem of early diagnosis of oligodendrogliomas, due to their location, requires the search for accessible and informative markers. Circulating microRNAs in blood plasma appear to be the most promising in this aspect. The purpose of the study was to develop a model for the diagnosis of oligodendrogliomas based on the expression profiles of microRNAs circulating in blood plasma. Methods: The study included 251 participants: 84 patients with glioblastomas, 51 patients with astrocytomas, 12 patients with oligodendrogliomas, 47 patients with meningiomas, 14 patients with lung cancer metastases, 25 patients with breast cancer metastases, and 18 healthy donors. The miRNeasy Serum/Plasma Kit (Qiagen, Germany) was used to isolate the microRNA fraction from blood plasma. The expression levels of hsa-miR-30c-5p, hsa-miR-186-5p, hsa-miR-194-5p, hsa-miR-484, hsa-miR-19b-3p, hsa-miR-431-5p, hsa-miR-363-3p, hsa-miR-128-3p, hsa-miR-3168, hsa-miR-192-5p, hsa-let-7c-5p, hsa-miR-340-5p, hsa-miR-182-5p and hsa-miR-144-5p were measured by stem loop RT-qPCR. Results: The model for diagnosing cases with oligodendrogliomas was as follows: K o =-1.59×hsa-miR-128-3p-1.97×hsa-miR-194-5p+1.78×hsa-miR-19b-3p-2.47×hsa-miR-3168+3.41×hsa-let-7c-5p+0.91×hsa-miR-192-5p+1.39×hsa-miR-340-5p-1.8, where K o is the coefficient used to differentiate cases with oligodendrogliomas from other cases. The optimal cut-off value for K o was 0.38 with a Youden index of 0.86. Model validation on the test dataset revealed that cases with oligodendrogliomas typically have calculated coefficient values ≥ 0.38. ROC analysis yielded the following model metrics: sensitivity – 100%, specificity – 85.7%, accuracy – 92.9%, and AUC – 0.93. Conclusions: The developed model opens up new opportunities for the early diagnosis of oligodendrogliomas and may eventually become part of a test system for the differential diagnosis of brain tumors.
RNDO-564-001: A first-in-human, phase 1/1b study of RNDO-564, a costimulatory bispecific antibody for the treatment of advanced bladder cancer and other Nectin-4–positive solid tumors.
TPS4640 Background: RNDO-564 is a novel CD28 x Nectin-4 bispecific antibody (bsAb) with high affinity Nectin-4 binding and optimized CD28 costimulatory arms designed to maximize the therapeutic window. RNDO-564-001 (NCT07218003) is an ongoing, open label, multicenter, phase 1/1b dose escalation and dose optimization study of RNDO-564 as monotherapy, and in combination with pembrolizumab (anti-programmed death PD-1 monoclonal antibody), in adults with relapsed/refractory (R/R), locally advanced or metastatic urothelial cancer (la/mUC) and other Nectin-4 positive solid tumors. Methods: The objectives of the study are to assess the safety, tolerability, and to determine the recommended phase 2 dose (RP2D) of intravenous RNDO-564 monotherapy given on Days 1, 8, and 15 of a 21-day cycle. Monotherapy dose escalation will include an accelerated titration component using four single participant (pt) cohorts, followed by more gradual dose increments using a Bayesian Optimal Interval (BOIN) with backfill design. Pts with R/R la/mUC, cervical, head and neck, esophageal, gastric, gastroesophageal, non-small cell lung, or triple negative breast cancer, who have exhausted, are ineligible for, or declined available standard treatment options, are eligible for the dose escalation phase of the study. Pts must have an ECOG performance status of 0-1. Pts with prior Nectin-4 directed therapies and Monomethyl Auristatin E (MMAE) exposure and those with ongoing Grade ≤ 2 peripheral neuropathy are eligible. Pts with a history of, or with active, inflammatory skin conditions are ineligible. Enrollment into Cohorts 1 and 2 have been completed without dose limiting toxicity (DLT). Enrollment to cohort 3 opened in January 2026. Safety endpoints include incidence of adverse events per CTCAE v6.0. Secondary endpoints include clinical activity, pharmacokinetics, and incidence of anti-drug antibody development. Retrospective assessment of Nectin-4 expression on tumor tissue by immunohistochemistry (IHC), and measurement of CD28 receptor occupancy, T cell activation, and serum cytokines in peripheral blood will be performed as exploratory endpoints. To determine the optimal RP2D in R/R la/mUC, up to two 20 pt dose optimization cohorts at two different dose levels will be enrolled. Separate dose escalation and optimization stages in combination with pembrolizumab will be conducted to assess safety, clinical activity, and RP2D for further development of combination treatment. Based on emerging clinical data, additional monotherapy and/or combination therapy expansion cohorts may be opened for pts with non-UC Nectin-4 positive solid tumors. Conclusions: RNDO-564-001 is an ongoing phase 1/1b study to determine the safety, tolerability, clinical activity, and RP2D of RNDO-564 in R/R la/mUC and other Nectin-4 positive solid tumors. Clinical trial information: NCT#07218003 .
Spatial analysis of CD8+ T cell neighbors using multiplexed immunofluorescence (MxIF) in esophageal squamous cell carcinoma (ESCC).
e16100 Background: Little is known of how the spatial orientation of the tumor immune microenvironment (TIME) contributes to ESCC progression. Higher CD8+ T cell number correlates with improved response to immunotherapy and better survival. However, which cell types are close to CD8+ T cells at different stages or differentiation status of ESCC is unknown. We used MxIF to identify the location of CD8+ T cell neighbors (fibroblasts, epithelial cells, macrophages, eosinophils, CD4+ T cells, Tregs) and cell states (proliferation, apoptosis, DNA damage) to determine whether there are specific cell types neighboring CD8+ T cells associated with stage or differentiation. Methods: We created a tissue microarray of 27 ESCC resection specimens from Japanese patients who had no prior therapy and performed MxIF for the aforementioned cell types/states. We generated probability maps of positive markers in a substructure specific manner using machine learning to segment cells. Each cell was summed for marker combinations related to cell type/state for accurate distances between cells. Demographic characteristics were summarized using descriptive statistics. We calculated effective neighbor count by multiplying neighbor proportion of a cell type by cell type intensity to define the expected number of neighboring cells within 50 µm of each CD8+ T cell. Proportion is the number of CD8+ T cells with that cell type as a neighbor divided by the total number of CD8+ T cells (e.g. 55% of CD8+ T cells neighbor a macrophage). Intensity is the average number of neighbors per CD8+ T cell with 1 neighbor of that cell type (e.g. each CD8+ T cell with 1 macrophage neighbor had 20.5 macrophage neighbors, for an effective macrophage neighbor count of 11.3). Effective neighbor counts were compared across differentiation status, T- and N-stage using the Kruskal–Wallis or Wilcoxon rank-sum test, as appropriate. Results: Demographics are shown in the Table. The most effective neighbor cell types are macrophages (11.3), CD4+ T cells (9.3), and fibroblasts (6.0). In well versus poorly differentiated tumors, significantly more fibroblasts neighbor CD8+ T cells (p = 0.012), but fewer CD4+ T cells neighbor CD8+ T cells (p < 0.01). Stage T2/3 specimens exhibit more neighboring fibroblasts (p = 0.017) and fewer Treg neighbors (p = 0.064) to CD8+ T cells. There were no significant differences when comparing N0 vs N+. Conclusions: In these 27 patients with ESCC, CD8+ T cells more commonly neighbor fibroblasts in poorly differentiated and higher T-stage tumors. Future studies should focus on understanding the interaction of CD8+ T cell/fibroblast crosstalk during ESCC progression. Characteristic N = 27 Age (mean) 67 years Male Gender 67% Stage T1: 59%T2: 11%T3: 30% N-stage N0: 52%N+: 48% Differentiation Well: 22%Moderate: 59%Poor: 19% Tobacco use (mean) 33.7 pack years Alcohol use Social: 7%Habitual: 56%Heavy: 37%
Association of secondary hematologic malignancies with radioactive iodine therapy for differentiated thyroid cancer: A systematic review and meta-analysis of retrospective cohort studies.
e18076 Background: Radioactive iodine (RAI) ablation therapy is a minimally invasive yet effective intervention for treating differentiated thyroid cancer (DTC), however, its association with second primary malignant neoplasms continues to warrant concern. Moreover, current evidence for association with hematologic malignancies is limited. Aims and Objectives: This systematic review and meta-analysis aimed to evaluate the odds of secondary hematologic malignancies in DTC patients following RAI exposure. Methods: Following PRISMA guidelines, a systematic literature search was conducted on the following databases: PubMed and Web of Science to identify relevant studies that met the inclusion criteria from inception to January 2026. All statistical analyses were performed using R 4.5.2 using a random-effects model, and heterogeneity was assessed using the I 2 statistic. Results: Nineteen eligible studies comprising 642,218 DTC patients (RAI treatment group, n = 305,225; non-RAI treatment group, n = 336,993) were included in the final meta-analysis. The pooled odds ratio for secondary hematologic malignancies among patients treated with RAI was higher (OR = 1.25, 95%CI 1.12 to 1.39, P < 0.0001) compared to non-RAI group with high statistical significance and low heterogeneity ( I 2 = 8.4). Further subgroup analyses were performed based on different subtypes of haematologic malignancies. The odds ratio estimates for each subtype except for leukemia (OR = 1.48, 95%CI 1.11 to 1.98, P = 0.0082) were non-significant: Hodgkin’s lymphoma (OR = 1.27, 95%CI 0.68 to 2.38, P = 0.4557), myeloma (OR = 0.89, 95%CI 0.64 to 1.25, P = 0.4994), non-Hodgkin’s lymphoma (OR = 1.04, 95%CI 0.82 to 1.32, P = 0.7523). Conclusions: This meta-analysis suggests higher odds of secondary hematologic malignancies, especially leukemia, in DTC patients treated with RAI ablation therapy. However, further large-scale, multicenter prospective studies are needed to validate these findings.
ResNets in the diagnosis of gliomas: Knowledge distillation from ResNet101 to ResNet18 for accurate and deployable WHO grade stratification.
e14093 Background: Gliomas are the most common primary malignancies of the central nervous system and remain a leading cause of cancer-related mortality in young adults. The WHO 2021 classification stratifies gliomas into grades with markedly different prognoses, underscoring the importance of accurate, timely grading on magnetic resonance imaging (MRI). Despite advances in deep learning, many high-performing models are computationally intensive, limiting translation into routine clinical practice, particularly in resource-limited settings. Knowledge distillation offers a strategy to transfer diagnostic capability from high-capacity teacher networks to lightweight student models while preserving performance. We evaluated a ResNet-based distillation framework for glioma diagnosis and WHO grade stratification. Methods: We conducted a retrospective multicontinental analysis of 2,840 brain MRI studies, including gliomas (grades 2–3 and grade 4), meningiomas, and pituitary adenomas. Ground truth was established using WHO 2021 criteria with expert neuroradiologist consensus. A ResNet101 teacher model, leveraging deep residual connections for hierarchical feature learning across MRI sequences, was trained for glioma grading and differential diagnosis. A ResNet18 student model was trained using temperature-scaled knowledge distillation with regularized cross-entropy loss to inherit the teacher’s probabilistic decision structure while reducing computational complexity. Model performance was evaluated using accuracy, sensitivity, specificity, F1 score, and AUROC. Results: The distilled ResNet18 achieved high diagnostic performance for glioma grade stratification, with accuracy exceeding 96% and AUROC greater than 0.99 for distinguishing grade 2–3 from grade 4 disease. Differential diagnosis accuracy exceeded 95% for gliomas versus meningiomas and pituitary adenomas. Knowledge distillation reduced model parameters by over 80% and substantially decreased inference time, enabling real-time analysis on standard hardware. Neuroradiologist reviewers reported high clinical utility for diagnostic support. Conclusions: Knowledge distillation enables a lightweight ResNet18 model to achieve near–ResNet101 diagnostic performance for glioma diagnosis and WHO grade stratification while markedly reducing computational burden. This approach addresses key deployment barriers and supports scalable, real-time AI assistance in neuro-oncology imaging. Prospective validation is warranted to assess impact on diagnostic accuracy and time-to-treatment.
A randomized prospective trial evaluating duration of PD-1 and PD-L1 inhibitor therapy in advanced solid tumors.
2514 Background: Immune checkpoint inhibitors (IO) have revolutionized the treatment of advanced solid tumors. Despite their widespread adoption, the optimal duration of IO has not been clearly established, and this remains an unmet clinical need. Methods: In our prospective trial, patients (pts) with advanced solid tumors who achieved at least stable disease (SD) after 1 year of standard-of-care PD-1/PD-L1 inhibitors were randomized 1:1 to either stop or continue IO beyond one year. Pts who stopped IO and then progressed could be retreated with IO at physician discretion. Primary endpoint was cure rate, defined as the proportion of pts without progressive disease, death, or initiation of new treatment (including re-treatment with IO) at five years. Cure rate was calculated using Kaplan-Meier (KM) estimates with 95% confidence intervals (CI) and compared between arms using a multivariable logistic regression model to adjust for pre-specified stratification factors of tumor type and expected efficacy by IO indication, and reported as an odds ratio (OR) with CI and p-value. Secondary endpoints included PFS, OS, and incidence of iRAEs. Median PFS and OS were estimated from KM curves, with hazard ratios (HR) and p-values obtained from stratified Cox proportional hazards models. All analyses were conducted using R version 4.4.2. Results: From 2019 to 2025, leveraging our central hub and community network, a total of 161 pts (Stop = 81, Continue = 80) were enrolled. The trial was closed early due to slow accrual (goal 578 pts). Median age was 71 and the most common tumor types were: NSCLC(39.8%), mismatch repair-deficient tumors(18.6), RCC(11.8%), HNSCC(8.1%), and melanoma (7.5%). There was no difference in baseline characteristics between arms. With a median follow up of 43.2 months (mo), cure rate was 45% (32%-63%)in the stop arm vs 38% (27-54%) in the continue IO arm (OR 2.26, 95% CI [0.62-8.22], p = 0.22). Median PFS was 55.4(27.5-NR) vs. 31.2 mo(20-NR) HR 0.77, 95% CI (0.48-1.24), p = 0.28 and median OS was NR (55.4 vs. NR) vs. 53.5 mo(40.8-NR) HR 0.54, 95%CI (0.3-0.99), p = 0.04 for the stop vs continue IO arms, respectively. In the continue IO arm, the median number of IO doses was 11 (range 0-36) and 52 of these pts discontinued IO with the most common reason being disease progression (42.3%), and 7.7% discontinued for toxicity. In the stop IO arm, 17 pts who progressed were retreated with IO; 20% achieved subsequent partial response and the median PFS was 10.5 mo(6.5-NR) and median OS was 26.7 mo(10.5-NR) after retreatment. Conclusions: Our trial is unique as a prospective randomized evaluation of length of IO treatment in advanced solid tumors. Cure rates and PFS were similar across arms and OS was improved in pts that discontinued IO at 1 year. These findings suggest that stopping IO at 1 year may be as effective as indefinite IO therapy, though validation in a larger, adequately powered trial is needed. Clinical trial information: NCT04157985 .
Real-world comparison of ipilimumab–nivolumab dosing strategies in melanoma brain metastases.
9558 Background: Ipilimumab 3 mg/kg plus nivolumab 1 mg/kg (Ipi3+Nivo1) is the standard immunotherapy regimen for metastatic melanoma with brain metastases (MBM) but is associated with significant immune-related toxicity, particularly neurologic toxicity. Ipilimumab 1 mg/kg plus nivolumab 3 mg/kg (Ipi1+Nivo3) has demonstrated improved tolerability in extracranial melanoma, though comparative data in MBM are limited. We compared efficacy and toxicity of these dosing strategies in a multisite real-world cohort. Methods: We retrospectively identified patients with MBM treated with ipilimumab plus nivolumab across three Mayo Clinic sites between 2016 and 2024. Patients were classified by induction dosing as Ipi1+Nivo3 or Ipi3+Nivo1. Overall survival (OS) and progression-free survival (PFS) were estimated using the Kaplan–Meier method and compared using log-rank tests. Best overall response (BOR) and toxicity outcomes were summarized descriptively. Results: Among 147 patients (Ipi1+Nivo3 n=32; Ipi3+Nivo1 n=115), 144 were evaluable for survival. Median age was higher in the Ipi1+Nivo3 group (69.5 vs 66.0 years); males predominated in both groups (71.9% vs 61.7%), and the cohort was predominantly White (>96%). Most patients had ECOG 0–1 in both groups (84.4% in Ipi1+ Nivo3 and 90.4% in Ipi3+Nivo1). BRAF mutation status was balanced between regimens, with approximately half of patients harboring BRAF-mutant disease (51.6% vs 53.2%). Median OS was 18.5 months (95% CI, 5.7–54.7) with Ipi1+Nivo3 versus 10.6 months (95% CI, 6.5–19.1) with Ipi3+Nivo1 (log-rank p=0.25). Two-year OS was 45.9% and 35.0%, and five-year OS was 24.5% and 19.2%, respectively. Median PFS was 7.3 months (95% CI, 1.8–10.9) versus 2.8 months (95% CI, 2.1–4.4) (log-rank p=0.16). BOR was available for 138 patients; objective response rates were 35.5% and 31.8%, respectively. 3–5 adverse events occurred in 24.1% of Ipi1+Nivo3 versus 30.5% of Ipi3+Nivo1 patients. Hospitalizations were less frequent (16.7% vs 36.1%), and neurologic adverse events occurred only with Ipi3+Nivo1 (11.9% vs 0%). Conclusions: In this multisite real-world MBM cohort, Ipi1+Nivo3 demonstrated comparable efficacy, numerically longer survival, and a more favorable toxicity profile compared with standard Ipi3+Nivo1. Notably, no neurologic toxicity was observed with the Ipi1+Nivo3 regimen. These findings support further prospective evaluation of lower dose ipilimumab strategies. Response and survival outcomes. Outcome Ipi1 + Nivo3 Ipi3 + Nivo1 Complete response (CR) 2 / 31 (6.5%) 8 / 107 (7.5%) Partial response (PR) 9 / 31 (29.0%) 26 / 107 (24.3%) Objective response rate (ORR = CR+PR) 11 / 31 (35.5%) 34 / 107 (31.8%) Disease control rate (DCR = CR+PR+SD) 18 / 31 (58.1%) 51 / 107 (47.7%) Median PFS, months (95% CI) 7.3 (1.8–10.9) 2.8 (2.1–4.4) 2-year PFS 27.6% 22.3% Median OS, months (95% CI) 18.5 (5.7–54.7) 10.6 (6.5–19.1) 2-year OS 45.9% 35.0% 5-year OS 24.5% 19.2%
The politics of processed foods: Consumer perceptions of policies targeting ultra-processed foods
This paper examined U.S. public support for six potential government policies designed to regulate ultra-processed foods (UPFs). Specifically, we consider support for two information-based policies (defining UPFs and providing dietary recommendations around UPFs); three restriction-based policies (UPF restrictions in grocery stores, schools, and food assistance programs); and one price-based policy (taxing UPFs). Survey results from 990 U.S. consumers showed the highest support for information-based policies, with the least support for UPF taxation. Additionally, there was considerable heterogeneity in policy support based on perceptions of UPFs, subjective knowledge, and demographics. These results may help policymakers gauge public sentiment around and appetite for regulating UPFs in the U.S.
Passive diffusion in Stratum corneum membrane of hybrid nanoemulsions based on Carapa guianensis oil and FE₃O₄ nanoparticles
Freestanding Flexible Vortex Tubes Integrated With Ferroelectric Transistor
ABSTRACT Polar topologies hold great potential in information storage. So far, thin films with polar topological structure are predominantly constrained by the substrate on which they are grown. Such substrate‐stabilized polar topologies limit their integration into memory devices, and consequently, their physical behaviors in data memory remain unclear. Here, we show a ferroelectric field‐effect transistor (Fe‐FET) memory device, which is based on large‐scale freestanding pure polar vortex tube arrays. In situ heating demonstrates that vortex tube arrays in freestanding superlattices undergo a more directly reversible phase transition, evolving from vortex tubes to single domains and ultimately to the disappearance of ferroelectricity. The flexible vortex tubes, which are obtained by removing the sacrificial layers, could bend at a 90° without breaking, demonstrating the nature of robustness. The Fe‐FET exhibits a wide and stable clockwise hysteresis loop, spanning a voltage range of −60 to 60 V and temperatures from room temperature to 450 K. Due to the stable and robust ferroelectric vortex dipole moment arrays, the Fe‐FET achieves a memory retention time of 3600 s and an endurance of 10 4 cycles.