Browse Articles

Global sex-based disparities in pancreatic cancer clinical trial enrollment: A systematic review and meta-analysis.

1654 Background: Sex-based disparities in oncology trial enrollment are well recognized, yet their magnitude and consistency in pancreatic cancer clinical trials remain poorly-defined. We conducted a global systematic review and meta-analysis to quantify women’s representation in pancreatic cancer trials relative to disease incidence. Methods: We identified phase I–III randomized and prospective interventional pancreatic cancer trials published through November 2025 in the top 25 general medicine, oncology, and gastroenterology journals. Women’s enrollment was compared with sex-specific pancreatic cancer incidence from Global Burden of Disease estimates. Enrollment disparity difference (EDD) was calculated as the proportion of women enrolled minus the proportion of women in the underlying disease population. Random-effects meta-analyses, subgroup analyses, and temporal meta-regression were performed. Results: Among 191 eligible trials including 37,688 participants, 16,393 (43.5%) were women. Overall, women were significantly under-enrolled, with a pooled EDD of −0.050 (95% CI, −0.058 to −0.042). Under-enrollment was observed across all geographic regions, with the largest disparity in European trials (EDD −0.069; 95% CI, −0.080 to −0.058). Significant disparities were present across all trial phases except combined phase I/II studies and were most pronounced in trials enrolling fewer than 50 participants (EDD −0.081; 95% CI, −0.105 to −0.057). Women remained under-enrolled across chemotherapy, immunotherapy/biologic, vaccine, and multimodal trials, while no significant disparity was observed in radiotherapy studies. Trials with women in leadership roles demonstrated slightly attenuated but persistent under-enrollment (EDD −0.046 vs −0.052). Temporal analyses showed narrowing of disparities between 2016 and 2020, followed by worsening under-enrollment from 2021 to 2025 (EDD −0.033). Sensitivity analyses limited to chemotherapy, immunotherapy/biologic, and phase III trials confirmed consistent patterns of under-representation. Conclusions: Women are persistently under-enrolled in pancreatic cancer clinical trials worldwide, with disparities evident across regions, trial phases, and treatment modalities. Recent reversals in progress highlight the need for durable, equity-focused trial design and enrollment strategies.

Adverse events and their management in patients with tenosynovial giant cell tumor (TGCT) treated with pimicotinib, after longer-term follow-up from the global phase 3 MANEUVER trial.

11571 Background: TGCT is a rare, locally aggressive, soft-tissue tumor associated with significant morbidity. Pimicotinib (pimi) is an oral, highly selective, small molecule colony-stimulating factor-1 receptor (CSF-1R) inhibitor. In the Phase 3 MANEUVER trial (NCT05804045), the primary and key secondary endpoints were met; pimi demonstrated robust tumor responses and symptomatic and functional improvements, with a tolerable safety profile. Here, we report the longer-term safety profile and adverse event management strategies with at least one year of pimi treatment. Methods: Adults patients (pts) with symptomatic unresectable TGCT were randomized (2:1) to pimi 50 mg once daily or placebo for 24 weeks (Part 1), followed by open-label pimi for an additional 24 weeks (Part 2) and an extension phase. Treatment-emergent adverse events (TEAEs) were monitored for 30 days after the last dose and graded using the Common Terminology Criteria for Adverse Events, version 5.0. Results: In total, 63 pts received pimi (median follow-up: 14.3 months). The most frequent laboratory abnormality TEAE was increased creatinine phosphokinase (CPK; 71.4%) and the most common clinical TEAE was pruritus (60.3%). Fatigue occurred in 28.6% of pts. Most TEAEs were Grade 1–2 (Table). Grade ≥3 TEAEs occurred in 46.0% of pts, primarily elevated CPK (15.9%) and rash (6.3%), with no Grade ≥3 periorbital or facial edema observed. TEAEs were managed with dose interruptions (66.7%; median 11 days) and reductions (25.4%), maintaining a high median percentage intended dose (88.2%). Discontinuations were infrequent (6.3%). Concomitant medications were commonly used for AE management; local corticosteroids and antihistamines for rash and pruritus, and diuretics for edema. Conclusions: This analysis confirmed the distinct and manageable safety profile of pimi. Most TEAEs were low grade and managed through dose modifications and standard concomitant medications, enabling sustained dosing and minimal discontinuations. These findings support the longer-term tolerability of pimi in TGCT. Clinical trial information: NCT05804045 . Grades of most common TEAEs (≥30%) in patients treated with pimi in the MANEUVER phase 3 trial. TEAEs, n (%) a Any grade Grade 1 Grade 2 Grade ≥3 Blood CPK increased b Blood LDH increased b AST increased b Amylase increased b 45 (71.4)36 (57.1)35 (55.6)24 (38.1) 20 (31.7)35 (55.6)33 (52.4)16 (25.4) 15 (23.8)1 (1.6)2 (3.2)8 (12.7) 10 (15.9) c 000 Pruritus d Face edema d Rash d Periorbital edema d 38 (60.3)31 (49.2)24 (38.1)23 (36.5) 26 (41.3)22 (34.9)18 (28.6)18 (28.6) 10 (15.9)9 (14.3)2 (3.2)5 (7.9) 2 (3.2)04 (6.3)0 a TEAEs were not mutually exclusive; patients may have had more than one TEAE. b Asymptomatic laboratory abnormalities. c Included one patient with Grade 4 event. d Clinical adverse events. LDH, lactate dehydrogenase; AST, aspartate aminotransferase.

Real-world pregnancy outcomes following temporary interruption of adjuvant endocrine therapy in early-stage hormone receptor–positive breast cancer.

e12702 Background: The POSITIVE trial established the feasibility of temporary interruption of adjuvant endocrine therapy (ET) to allow pregnancy in selected patients with hormone receptor–positive early breast cancer. However, trial populations may not fully reflect the demographic, racial, and treatment heterogeneity encountered in routine clinical practice. Real-world benchmarking of pregnancy outcomes following ET interruption remains limited. We evaluated pregnancy outcomes in a contemporary institutional cohort. Methods: We conducted a retrospective cohort study of patients with stage I–III hormone receptor–positive breast cancer who temporarily interrupted adjuvant ET for pregnancy intent between 1999–2023 at a tertiary academic cancer center. Patient demographics, tumor characteristics, treatment history, duration of ET prior to interruption, and pregnancy outcomes were abstracted. Descriptive statistics were used. Results: Among 94 eligible patients, the median age at diagnosis was 32 years (IQR 29–35). Most patients received prior chemotherapy (80%) and underwent mastectomy (66%). At ET interruption, 66% were receiving tamoxifen and 34% were receiving an aromatase inhibitor with ovarian suppression (AI/OFS). Median duration of ET prior to interruption was 28.7 months (IQR 14.7–42.8). Overall, 65.9% of patients achieved a successful pregnancy following ET interruption. Pregnancy success was higher in patients <35 compared with ≥35 (75.8% vs 46.9%, p = 0.01), and in White compared with Black patients (80% vs 47%, p = 0.03). Success rates were similar for patients receiving AI/OFS versus tamoxifen (71.9% vs 62.9%, p = 0.49). After delivery, 20.3% resumed ET. Prior live-birth rates at the time of ET interruption were identical in both the successful and failed pregnancy attempt groups (66%). The overall median time to follow up was 72 months. Median DFS was 73.6 months in the failed-pregnancy cohort and 68.6 months in the successful-pregnancy cohort, while overall survival did not differ between groups. Conclusions: In this diverse real-world cohort, pregnancy success following temporary endocrine therapy interruption was high and consistent with trial-reported outcomes. Younger age, White race, and higher parity were associated with higher pregnancy success, while overall survival remained comparable between groups. These findings support the feasibility of ET interruption for pregnancy attempts across a broader patient population. Larger cohort study is planned to compare outcomes to external controls who received uninterrupted endocrine therapy.

The impact of potentially inappropriate medications on performance status, functional status, and overall survival in older adults with prostate cancer.

e17018 Background: The 2023 American Geriatric Society's Beers Criteria lists around 100 medications with potential side effects that outweigh their benefits for adults age 65 and older. These medications are collectively referred to as potentially inappropriate medications (PIM). Among the PIMs listed, particular attention is paid to benzodiazepines and medications with strong anticholinergic activity as geriatric patients are more susceptible to their adverse effects. Prostate cancer (PCa) is the fourth most common cancer by incidence worldwide. Roughly 60% of men with PCa are 65 or older. Our study aims to assess the impact of these two classes of PIMs on performance status, functional status, and overall survival in geriatric patients with PCa undergoing treatment. Methods: We utilized data from Global Collaborative Network-TriNetX to assess the impact of benzodiazepines and anticholinergics (first generation antihistamines, antidepressants, antipsychotics, antiparkinsonian agents, barbiturates, Z-drugs, antispasmodics) on performance status, functional status, and overall survival. Patients aged 65 to 90 were divided into two cohorts: those being treated for PCa and receiving PIMs (PIM cohort), and those being treated for PCa and not receiving PIMs (control cohort). Cohorts were propensity score matched based on age, sex, race, baseline ECOG performance status, frailty, falls, bed-confinement status, ambulatory dysfunction, AJCC stage of disease, PSA levels, and underlying renal, cardiovascular, and pulmonary comorbidities. Using ICD-10 codes, we evaluated the following outcomes: ECOG performance status, age-related physical disability, need for assistance with ADLs, bed confinement, ambulatory dysfunction, falls, and overall survival at 1 year. Generalized linear models were used to measure associations and estimates were presented as mean values and odds ratios with 95% confidence intervals. Results: After matching, each cohort consisted of 49,713 patients. The PIM cohort had a mean age of 74.2 +/- 8 years. Caucasians accounted for 54.6% of patients. Over a 1-year period, older adults in the PIM cohort had a statistically worse ECOG performance status (mean 0.67 vs. 0.47, p = 0.012) and a higher risk of death (OR: 2.007, CI 1.92 – 2.10, p < 0.0001), age-related physical disability (OR: 2.22, CI 1.78 – 2.77, p < 0.0001), need for assistance with ADLs (OR: 2.35, CI 1.77–3.11, p < 0.0001), bed confinement (OR: 5.01, 2.88 – 8.72, p < 0.0001), ambulatory dysfunction (OR: 2.53, CI 2.07–3.09, p < 0.0001), and falls (OR: 2.55, CI 2.22–2.93, p < 0.0001). Conclusions: Our study demonstrated that PIM use in geriatric patients undergoing treatment for PCa is associated with worse performance status, functional status, and overall survival, raising the question of whether PIM use reduces patient's eligibility to receive PCa-directed treatment.

Integrated genomic and transcriptomic profiling of a breast cancer cohort in the UAE compared with TCGA.

e13015 Background: Breast cancer (BC) in the United Arab Emirates (UAE) remains under-characterized relative to large reference cohorts such as The Cancer Genome Atlas (TCGA). We compared UAE BC tumors, including a UAE National subset, with stage-matched TCGA BC to determine whether observed differences reflect TCGA demographic composition or distinct molecular programs with potential therapeutic relevance. Methods: Clinical, genomic, and transcriptomic features were analysed in a UAE BC cohort (n = 83; mutation calls available for 79), comprising UAE Nationals (n = 26) and UAE Non-Nationals (n = 57), against the TCGA BC cohort (n = 99). Sub-analyses compared UAE Nationals with TCGA race-stratified groups: TCGA-White (n = 80) and TCGA-Asian (n = 9). All comparisons were matched by late-stage disease. Tumor mutational burden (TMB) was compared across cohorts. Recurrent protein-altering variants in canonical BC genes were identified. Transcriptomic differences were assessed using gene set variation analysis (GSVA), and tumor microenvironment composition was inferred using xCell 2.0 (adjusted p < 0.05). Results: UAE cohort was younger than TCGA overall (median age 51 vs 55 years; p < 0.05), but this difference was not observed after stratification: UAE Nationals had an intermediate median age (56 years) between TCGA-White (59 years) and TCGA-Asian (52 years). UAE cohort demonstrated higher TMB compared with TCGA (p < 0.0001). Evaluation of recurrent mutations revealed canonical BC driver events predominantly associated with luminal and estrogen receptor–driven biology, including alterations in MAP3K1, PIK3CA, and ARID1A, indicating alignment with established luminal oncogenic pathways. GSVA identified enrichment of extracellular matrix remodeling and innate-immune programs in UAE tumors, including complement, JAK/STAT, and inflammatory signaling gene sets. In contrast, TCGA tumors showed relatively higher estrogen and HER2 response signatures. Immune deconvolution suggested that UAE tumors were relatively enriched for innate and myeloid-associated programs, with increased representation of myeloid cells and Th17- and Th2-associated signals, whereas TCGA tumors showed higher enrichment of adaptive immune components, particularly CD8⁺ T cells. These pathway and immune cell differences persisted when comparing UAE nationals with TCGA-White and TCGA-Asian subsets, with UAE nationals remaining distinct from both TCGA strata. Conclusions: BC in the UAE demonstrate higher reported TMB and a predominantly luminal driver landscape, accompanied by distinct extracellular matrix and innate immune transcriptional programs. These patterns persist in UAE-Nationals relative to TCGA race strata and are not recapitulated within TCGA between White and Asian subgroups, supporting population-associated differences in tumor and microenvironmental biology.

Protocol for RETRO-TBI: A prospective cohort study of mild traumatic brain injury in older adults

Older adults experience the highest rates of traumatic brain injury (TBI) related hospitalizations and deaths of any age group, yet TBI remains understudied in this population. To improve understanding of recovery over the year following TBI among older adults, we designed the RETRO-TBI study. This manuscript reports the protocol for RETRO-TBI, a prospective cohort study of older adults (65 years and older) with mild TBI (mTBI) with planned enrollment of 250 participants. The study is designed to evaluate recovery across four key domains: physical function, cognitive function, psychological function, and sleep quality. Participants are followed for 12 months after injury, with in-home study visits conducted at approximately 2 weeks and 3, 6, and 12 months post-injury. Blood samples are collected at all visits. The specific aims are to: (1) identify trajectories of recovery in physical function and predictors of poorer physical recovery; (2) identify trajectories of recovery in cognitive function, psychological function, and sleep quality and predictors of poorer recovery in these domains; and (3) examine associations among recovery trajectories across domains. The RETRO-TBI study represents an important step in addressing the knowledge gap on recovery following TBI among older adults and is expected to result in identification of sub-groups of individuals more likely to have poor recovery, informing individualized treatment plans and development of future domain-based rehabilitation strategies. The study has several strengths including its focus on older adults, evaluation of recovery across four domains of function, and longitudinal assessments will permit evaluation of heterogeneity in recovery trajectories.

Antisymmetric planar Hall effect in Ta/Py bilayers—Evidence for perpendicular interfacial magnetization tuned by in-plane bulk magnetization

Antisymmetric planar Hall effect (APHE) is observed in Ta/Py bilayers, which exhibits a sinusoidal dependence on the in-plane magnetization orientation. The APHE resistance is independent of both the amplitude and direction of the applied current, excluding contributions from Joule heating and spin–orbit torque effects. Additionally, the effect is insensitive to the magnitude of the in-plane magnetic field once the magnetization is practically saturated. The APHE amplitude and phase vary with the choice of seed layer and with post-deposition annealing, suggesting involvement of strain. Results suggest the presence of strain-modulated interfacial Dzyaloshinskii–Moriya interaction, which couples the bulk and interfacial spins and induces perpendicular magnetization at the Ta/Py interface tuned by the orientation of the in-plane bulk magnetization: a miter-gear-like combination. This mechanism offers a plausible explanation for the antisymmetric Hall response and points to a new pathway for inducing and manipulating perpendicular interfacial magnetization, with potential implications for novel spintronic devices.

Unlocking the bioactive potential: ZnO NPs synthesized from Chaetomorpha linum aqueous extract

Universal In Situ Flowrate Monitoring for Piezoelectric Microfluidics via Triboelectric Self‐Sensing

ABSTRACT Conventional flowrate monitoring methods for piezoelectric micropumps suffer from limited integration capability, contamination risks, and laboratory dependence. Here, we propose a universal in situ flowrate monitoring method through triboelectric sensing that overcomes these fundamental limitations. By integrating a triboelectric sensor directly within the piezoelectric micropump cavity, we achieve real‐time flowrate monitoring by detecting the piezoelectric vibrator deformation through triboelectric signals. Experimental validation across three distinct micropump architectures demonstrates the remarkable universality of two flowrate monitoring methods (voltage amplitude and frequency pulse), achieving linear fitting coefficients exceeding 0.99 and average monitoring errors below 2.49%. We further develop a microcomputer‐driven system for drug injection applications, achieving flowrates up to 2.2 mL/min with maintained monitoring accuracy of 2.298%. This work establishes a paradigm for fully integrated microfluidic devices and paves the way for intelligent drug delivery systems.

Assessing the geospatial distribution and association of microplastics with water quality in the lakes of Vellore, Tamil Nadu, India

Bacillus cereus PelADA is a polysaccharide de-N-acetylase required for pel-dependent biofilm formation

Predictors of in-hospital mortality and palliative care utilization in metastatic renal cell carcinoma: A nationwide analysis of 2018–2022 U.S. hospitalizations.

e16567 Background: Despite transformative advances in systemic therapy, metastatic renal cell carcinoma (mRCC) remains associated with high symptom burden, frequent hospitalizations, and inpatient mortality. National data characterizing inpatient outcomes and palliative care utilization in the contemporary immunotherapy era are limited. We sought to identify patient, disease, and hospital-level predictors of in-hospital mortality and palliative care use to inform earlier, risk-aligned serious-illness care. Methods: We analyzed the 2018–2022 National Inpatient Sample to identify adult hospitalizations with renal cell carcinoma and documented metastatic disease. Survey-weighted analyses generated nationally representative estimates. The primary outcome was in-hospital mortality; secondary outcomes included palliative care consultation, length of stay (LOS), and total hospital charges. Multivariable survey-weighted logistic regression models evaluated independent predictors of mortality and palliative care utilization, adjusting for demographics, socioeconomic status, hospital characteristics, comorbidities, acute complications, and metastatic sites. Results: We identified an estimated 24,570 weighted hospitalizations for mRCC during 2018–2022 (mean age 71.9 years, 39.2% female, 80% White, 72% Medicare, 82% urban teaching hospitals). Metastatic involvement was present in 13% of RCC hospitalizations, most commonly lung (7%), bone (5%), and liver (5%). In-hospital mortality was 2.65%, mean LOS was 5 days, and mean total charges were $87,000 per hospitalization. Palliative care consultation occurred in only 6.3% of mRCC hospitalizations but was dramatically higher among decedents (56.9% vs 4.7%, p < 0.001). In adjusted multivariable analyses, mortality was independently associated with palliative care consultation (OR 12.0), congestive heart failure (OR 2.3), protein–energy malnutrition (OR 1.8), electrolyte disorders (OR 3.0), and liver metastases (OR 2.7), while elective admission was strongly protective (OR 0.07). Palliative care utilization was independently associated with older age, urban teaching hospital status and metastatic burden. Conclusions: In this large, nationally representative analysis, patients hospitalized with mRCC experienced substantial inpatient mortality, while palliative care consultation was infrequent and largely confined to patients at extreme risk of death. The strong clustering of palliative care with advanced metastatic burden and acute physiologic decompensation suggests predominantly late, reactive referral patterns. These findings identify a critical opportunity for earlier, systematic integration of inpatient palliative care in mRCC to improve goal-concordant care, resource utilization, and equity across hospital settings.

Association of treatment-induced decrease of tumor chromosome Y and prognosis.

3136 Background: Loss of chromosome Y (LOY) in cancer has been linked to increased mortality. Since tumors that are not completely eradicated often harbor additional molecular changes which may alter biological behavior, here we sought to examine if therapy has any impact on tumor chromosome Y content and if so, the relevance this has on patient outcome. Changes in chromosome Y content with therapy would impact clinical decision making in patients with persistent or recurrent tumors since tumors with LOY have also been shown to be more responsive to specific therapies. Methods: Male patients with >=2 sequential tumor samples profiled at Caris Life Sciences (Phoenix, AZ) who received systemic therapy between collections were included. RNASeq was performed using NovaSeq platform. Chromosome Y score (YChr score) was calculated using ssGSEA (log-rank normalization) based on a previously published 9-gene signature ( Nature 2025). Change in YChr score between pre and post treatment paired samples (ΔYChr) was computed per patient. A decrease in YChr score in post sample compared to pre sample was defined as ΔYChr ≥0.029 (cohort mean) and called progressive LOY (pLOY) while the remainder were called no-pLOY. Real-world clinical data were obtained from insurance claims. Overall survival (OS) was defined from first sample collection to last contact. Hazard ratios (HRs) were estimated using Cox proportional hazards models, with log-rank p values reported. Results: Among 1,343 patients with paired samples, pLOY was associated with shorter OS compared with no-pLOY (median OS [mOS]: 33.2 vs. 36.0 months; HR 1.168, 95% CI 1.028–1.326; p=0.0167). Tumor-specific analyses demonstrated significantly worse OS in pLOY versus no-pLOY for colorectal cancer (CRC; N=124 vs. 157; mOS 38.9 vs. 50.6 months; HR 1.535, p=0.0038), bladder cancer (N=26 vs. 41; mOS 25.7 vs. 37.7 months; HR 1.992, p=0.0204), NSCLC (N=100 vs. 134; mOS 27.6 vs. 32.7 months; HR 1.379, p=0.0379), and gastric cancer (N=13 vs. 16; mOS 17.6 vs. 28.5 months; HR 2.603, p=0.0223), with a trend observed in melanoma (N=17 vs. 27; mOS: 31.4m vs. inf, HR: 2.231 [0.922-5.4], p=0.075). No association was observed in other tumor types. Treatment distributions for mentioned cancer types did not differ between pLOY and no-pLOY groups (p>0.05). Among CRC patients with primary-to-metastatic paired samples (57%), pLOY remained significantly associated with worse OS (mOS 39.5 vs. 56.3 months; HR 1.647; p=0.0112). pLOY was not associated with specific therapies or combinations across the full cohort. Conclusions: In a large real-world clinico-genomic cohort, longitudinal decrease in Chromosome Y score is independently associated with poor prognosis across multiple cancer types supporting further investigation into the biological and clinical relevance of Y chromosome loss and providing insights on novel treatment selection strategies.

Comparison of prognoses in pancreatic cancer patients receiving neoadjuvant therapy versus adjuvant therapy.

e16451 Background: Pancreatic cancer remains one of the most lethal malignancies, with limited long-term survival despite advances in systemic therapy and surgical techniques. The optimal sequencing of chemotherapy— whether administered in the neoadjuvant or adjuvant setting—remains controversial. Comparing survival outcomes between neoadjuvant and adjuvant treatment strategies may help refine treatment algorithms and improve prognostic stratification. In this study, we aimed to evaluate overall survival outcomes by chemotherapy administration timing. Methods: This retrospective cohort study was conducted using the Surveillance, Epidemiology, and End Results (SEER) database to identify patients diagnosed with pancreatic cancer who received either neoadjuvant or adjuvant therapy between 2000 and 2022. Demographic, clinicopathological, and treatment related variables were extracted for analysis. Results: The study was analyzed using data from 5681 patients who met the study criteria. The proportion of patients aged 60 or older was higher (78.5%). Pancreatic cancer most frequently originated from the head of the pancreas (68.9%). The stage distribution of the patients included in the study at the time of diagnosis was stage 1-29.3%, stage 2-41.8%, and stage 3-28.9%, respectively. All patients had undergone primary tumor surgery. All patients had received perioperative chemotherapy, and no patient had received radiotherapy (neoadjuvant or adjuvant). The rate of patients receiving adjuvant treatment was 54.7%, while the rates of patients receiving neoadjuvant or neoadjuvant + adjuvant treatment were 25.5% and 19.8%, respectively. 1988 patients (35%) died during the follow-up period. The median survival time was 34 months (95% CI, 32.3-35.6%). 1, 3, and 5-year survival rates were 88.4%, 48.8%, and 30%, respectively. Factors affecting survival were evaluated using multivariate analysis, and the results were as follows: age (p = 0.020), gender (p = 0.535), tumor site (p = 0.071), initial stage (p < 0.001), and type of perioperative chemotherapy (p = 0.059). Overall survival was significantly better in patients receiving neoadjuvant + adjuvant chemotherapy compared to those receiving only adjuvant treatment (p = 0.024, HR: 1.14, 95% CI 1.01-1.29) or only neoadjuvant treatment (p = 0.044, HR: 1.15, 1.0- 1.32) in a one-to-one comparison. Conclusions: In this study, we evaluated the effect of optimal chemotherapy timing on overall survival in patients with operable pancreatic cancer. While the p-value for perioperative chemotherapy type was borderline in the multivariate analysis, neoadjuvant+adjuvant chemotherapy was found to be superior to either neoadjuvant or adjuvant therapy in a direct comparison. Furthermore, this study found that overall survival varied according to initial tumor stage and age at disease onset.

Safety and efficacy of CS2009, a first-in-class PD-1/VEGF/CTLA-4 trispecific antibody, in patients with advanced non-small cell lung cancer: Results from a phase 1/2 study.

8558 Background: CS2009 is a novel PD-1/VEGF/CTLA-4 trispecific antibody. An ongoing phase 1/2 study is evaluating its safety, tolerability, PK, PD, and anti-tumor activity as a single agent and in combination with systemic treatment in patients (pts) with advanced solid tumors. This study comprises a phase 1 dose escalation and phase 2 dose expansion. Here, we report the initial efficacy and safety data from advanced non-small cell lung cancer (NSCLC) pts treated in first-line and later-line settings. Methods: The study enrolled both treatment-naïve (1L) and heavily pre-treated later-line (≥2L) NSCLC pts. The ≥2L group included previously treated pts without known actionable oncogenic alterations (AGA) who had progressed on at least one prior line containing an anti-PD-1/L1 antibody and platinum-based chemotherapy; these pts received CS2009 at 10, 20, 30, or 45 mg/kg, i.v., Q3W, until disease progression or intolerance. The 1L cohort enrolled treatment-naïve pts with PD-L1 TPS ≥1% NSCLC without known AGA; these pts were randomized 1:1 to receive CS2009 at 20 or 30 mg/kg, i.v., Q3W, until disease progression or intolerance. Safety was assessed in all treated pts. Efficacy was assessed in pts who had at least one post-baseline tumor assessment per RECIST v1.1. Results: As of Jan. 4, 2026, 40 pts with ≥2L NSCLC were treated with CS2009 across four dose levels (10 mg/kg, n=3; 20 mg/kg, n=12; 30 mg/kg, n=20; 45 mg/kg, n=5). Median age was 67 (range 37-78) years; 72.5% were Asian, 27.5% were White; 77.5% were male; 77.5% had ECOG PS 1 at baseline. Among 30 efficacy-evaluable pts, ORR was 20.0% (6 PRs); in 16 pts treated at 30 mg/kg, ORR was 25.0% (4 PRs). Any-grade and grade ≥3 treatment-related adverse events (TRAEs) occurred in 27 (67.5%) and 8 (20.0%) pts, respectively. The most common TRAEs were proteinuria (12.5%), hypertension (12.5%), bilirubin conjugated increased (10.0%), and fatigue (10.0%), which were mostly grade 1 or 2, except for grade ≥3 hypertension in two pts. No treatment-related deaths were reported. TRAEs led to discontinuation in 3 (7.5%) pts. More mature ≥2L efficacy data with additional evaluable pts will be presented at the conference. As of Jan. 4, 2026, 19 pts with 1L NSCLC (PD-L1 TPS ≥1%) were randomized to receive CS2009 at 20 mg/kg (n=10) or 30 mg/kg (n=9). Median age was 69 (range 52-82) years; 84.2% were male; 89.5% had ECOG PS 1 at baseline. Preliminary encouraging efficacy and favorable safety signals were observed. Updated efficacy and safety data in approximately 50 1L pts will be disclosed at the conference. Conclusions: CS2009, as a first-in-class trispecific antibody targeting PD-1, VEGF, and CTLA-4, demonstrated a favorable safety profile and clinically meaningful anti-tumor activity in advanced NSCLC pts without known AGA. These findings warrant further investigation of CS2009 in this population. Clinical trial information: NCT06741644 .

Longitudinal monitoring for post-surgical molecular residual disease in patients with stage I–IIIB melanoma using a whole-genome sequencing–based ctDNA assay.

9577 Background: Circulating tumor DNA (ctDNA) is a novel biomarker for detecting molecular residual disease (MRD) and monitoring recurrence risk in melanoma. Personalized, tumor-informed ctDNA assays may enable highly sensitive and specific longitudinal detection of disease burden following curative-intent surgery. Here, we evaluate the prognostic value of whole-genome sequencing (WGS)-based ctDNA detection in patients with resected stage I–IIIB melanoma. Methods: This was a retrospective analysis of residual samples from a deidentified real-world dataset. Clinicopathologic information was collected from Natera’s commercial database under an IRB-approved protocol (#20-049-ALL). Patients with stage I-IIIB melanoma who underwent surgery between March 2018 to May 2023 were included. Whole genome-based ctDNA assays (Signatera Genome, Natera, Inc.) were designed from the respective patients’ matched tumor and normal WGS data. ctDNA was subsequently tracked in the patients’ longitudinal blood samples, which were collected postoperatively until recurrence/end of follow-up. The correlation between ctDNA status post-definitive treatment and recurrence-free survival (RFS) was assessed using Cox regression analysis with a time-dependent covariate and the Kaplan-Meier method. Results: A total of 103 patients with stage I-IIIB melanoma and residual material available for testing were included with a median of 6 tests per patient (range: 1–23) and a median of 91 days between ctDNA tests. Median follow-up was 25.1 months (range: 8.5–73.4). Stage distribution was: 1.9% (2/103) IA, 3.9% (4/103) IB, 7.8% (8/103) IIA, 39.8% (41/103) IIB, 20.4% (21/103) IIC, 8.7% (9/103) IIIA, and 17.5% (18/103) IIIB, and 38 patients received adjuvant treatment, with the majority 87% (33/38) receiving immunotherapy only. ctDNA results during the post-operative treatment window were available for 95 patients. Among the 15 patients who experienced recurrence after completion of definitive treatment, 12 (80%) were ctDNA-positive prior to or within 2-weeks of clinical recurrence (range: 99 days pre - 12 days post-imaging). Among patients who did not experience recurrence, 97.5% (78/80) were serially ctDNA-negative. ctDNA positivity at any time during post–definitive treatment was associated with significantly shorter RFS compared with persistent ctDNA-negativity (HR 55.3; 95% CI 18.7–163.3; p < 0.0001). Analyses are ongoing to characterize ctDNA dynamics (including quantification) and clinical outcomes in an expanded cohort. Conclusions: This analysis demonstrates the prognostic value of longitudinal post-operative ctDNA monitoring for the detection of recurrence in patients with stage I–IIIB melanoma using a WGS-based, personalized, tumor-informed assay with implications for optimizing surveillance and improving clinical outcomes.

The molecular landscape and clinical nomogram of pancreatic cancers harboring DNA damage repair genes alterations.

e16424 Background: Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal malignancy with 5-year survival below 13% and rising incidence. Most patients present with advanced disease unsuitable for resection. DNA damage repair (DDR) alterations, including BRCA1/2, PALB2, RAD51 paralogs, ATM, ATR, CHEK1/2, and Fanconi anemia genes, occur in 25% of PDAC and confer platinum-sensitivity and PARP inhibitor responsiveness. Olaparib demonstrated significant benefit as maintenance therapy in germline BRCA-mutated metastatic PDAC, establishing biomarker-driven treatment. Here, we characterize the molecular landscape and develop an OS nomogram for DDR-altered PDAC. Methods: We analyzed data from 4,132 PDAC patients with DDR alterations from cBioPortal (2012-2025). We investigated BRCA1/2, PALB2, RAD51B/C/D/54L, BARD1, BRIP1, ATM, ATR, CHEK1/2, FANCA, FANCL, PTEN, NBN, MRE11, STK11, CDH1, CDK12 , and interacting oncogenes ( KRAS , BRAF , PIK3CA ), reflective relevant HRR NGS panels used in clinical practice. We evaluated copy number alterations, mRNA expression, clinicopathological factors, and OS. Transcriptomic profiling identified survival-associated signatures. A nomogram integrating clinical, pathological, and molecular features was constructed for OS prediction. Results: Genomic analysis of 4,024 PDAC samples revealed mutations in BRAC2 (3%), BRAC1 (1.5%), TP53 (68%), STK11 (2%), KRAS (87%), PIK3CA (2%) and ATM (3%). Notably, DDR alterations were associated with inferior OS (log-rank P < 0.001; mOS altered vs wild-type 18.28 vs 26.24 months) and DFS (log-rank P = 0.041; mDFS altered vs wild-type 14.95 vs 27.30 months). Transcriptomic and GSEA profiling linked these alterations to changes of oncogenic pathways, including dysregulated homologous recombination, base excision repair, Fanconi checkpoint control, cell cycle regulation, epithelial-mesenchymal transition, and immune pathways. The prognostic nomogram incorporated age, stage, differentiation, and top 5 most significantly identified transcriptomic signatures. Conclusions: This study characterizes DDR-altered PDAC, revealing distinct genomic and transcriptomic profiles. DDR alterations identify patients benefiting from platinum-chemotherapy and PARP inhibitors, with superior outcomes versus unselected populations. The validated nomogram provides a precision tool for OS prediction, guiding treatment selection, surveillance, and trial stratification. These findings emphasize universal DDR screening importance and support molecular profiling integration to improve outcomes through biomarker-driven strategies.

Circulating tumor DNA (ctDNA) as a biomarker for pathologic response and clinical outcomes in patients with cutaneous squamous cell carcinoma (CSCC) receiving neoadjuvant anti–PD-1 therapy.

9587 Background: ctDNA is a novel biomarker that can be used to evaluate recurrence and treatment response in multiple cancers. The utility of ctDNA in patients with CSCC receiving neoadjuvant immunotherapy (nIT) is unknown. Methods: In this retrospective study, we examined patients with CSCC who received nIT and had ctDNA assay available pre- and post- operatively. Positive ctDNA was defined as a value >0, and ctDNA sensitivity pre-operatively (pre-op) was reported in all patients. Patients with at least two ctDNA measurements pre-op were classified into two cohorts: cohort-A included patients whose ctDNA decreased; Cohort-B included patients whose ctDNA increased. Pathologic response (PR) was defined as complete (pCR) if 0% viable tumor [vt], major (mPR) if vt < or equal to 10%, and no-response (nPR) if vt >10%. The association between ctDNA kinetics during the pre-op period nIT and PR was analyzed using Fisher’s exact test. During the post-op period, we used minimal residual disease (MRD) status defined by positive ctDNA measurement within 3 months post-op. Recurrence-free survival (RFS) (landmark analysis with time starting at post-MRD timepoint until recurrence, death or loss-to follow-up) was compared between patients who had detectable MRD (positive ctDNA within 3 months post-op) versus those who had undetectable MRD. Results: Forty nine patients were included of which 42 (85.7%) were male, 11 (22.4%) were immunosuppressed, and 31 (63.3%) had AJCC stage IV CSCC. The median follow-up time since starting nIT was 12.9 months. The ctDNA sensitivity at baseline (prior to starting nIT) was 79.6%. The median number of ctDNA collections per patient was 6 (range: 1, 21). Of the 49 patients who underwent surgery, 38 were eligible who had at least 2 ctDNA measurements pre-op. Cohort-A (35 patients) had higher cPR and mPR compared to cohort-B (3 patients) (p=0.05). When excluding 3 patients who did not have kinetic ctDNA changes (ctDNA value was 0 at all pre-op time points), ctDNA decrease was significantly associated with cPR and mPR (p= 0.03). For post-op period, patients with detectable MRD (n=5) had significantly lower RFS compared to those with undetectable MRD (median RFS 11.3 months vs. not reached, HR: 11.9; 95% CI: 2.2-65.8; p< 0.01)(Table). Conclusions: ctDNA changes and detectable MRD may be suitable biomarkers for pathologic response assessment and post-op period in patients with CSCC receiving nIT. Further research is essential to confirm these findings. Recurrence-free survival according to MRD status. MRD status N Median after 3-month post-op (95% CI) month 6-month RFS after 3-month post-op (95% CI) months 12-month RFS after 3-month post-op (95% CI) months Hazard ratio (95% CI) Negative 31 NA (NA, NA) 0.96 (0.89, 1) 0.89 (0.74, 1) ref Positive 5 11.3 (2.22, NA) 0.6 (0.29, 1) 0.3 (0.06, 1) 11.9 (2.2, 65.8)

Predicting multi-cancer risk from EHR data using multi-agent LLMs.

10501 Background: Accurate multi-cancer risk prediction is essential for optimizing screening strategies and yet remains challenging due to heterogeneous and longitudinal electronic health record (EHR) data. Large language models (LLMs) offer a unified framework for multi-cancer risk assessment without disease-specific model training, but single-agent LLMs struggle to reason effectively over long patient histories. Methods: We developed a multi-agent system (MAS) composed of chained LLM agents with episodic memory to synthesize longitudinal EHR data into comprehensive cancer risk profiles. The system generates 1-year cancer risk scores (1–10) for individual cancer types. Using Truveta Data, we identified cases for 9 cancer types with clinically recognized precursor signs using clinician-curated diagnostic codes. For each cancer type, 500 cases were randomly sampled and matched 1:1 with controls by 10-year age group and sex. All structured EHR data (conditions, laboratory results, observations, medications, and procedures) up to 1 year before diagnosis or index date were included. Performance was evaluated across cancer types; lung cancer was used as a benchmark cohort for comparison with traditional machine learning (ML) models and a single-agent LLM baseline. Results: The MAS demonstrated heterogeneous but clinically meaningful discrimination across cancer types (AUROC range: 0.64 to 0.80), with liver cancer performing the best and colorectal cancer the worst. In the lung cancer cohort, performance (AUROC=0.79) was comparable to that of trained ML models (AUROC: XGBoost 0.82, logistic regression 0.73, k-nearest neighbors 0.58). Beyond risk prediction, the system generated concise patient summaries and interpretable rationales that supported downstream analyses, including cancer real-world evidence generation and exploratory knowledge discovery. Compared with a single-agent LLM, the MAS showed improved temporal coherence and clinical reasoning. Conclusions: A multi-agent LLM system can leverage longitudinal EHR data to estimate short-term multi-cancer risk within a unified framework, without cancer-specific model training. These findings support the potential role of LLM-based systems as scalable tools to inform risk stratification and screening strategies in real-world clinical populations. The MAS’s performance. Lung Ovarian Liver Pancreatic Colorectal Multiple Myeloma Lymphoma Gastric Bladder AUROC [95% CI] 0.79 [0.77, 0.82] 0.68 [0.65, 0.71] 0.80 [0.77, 0.83] 0.66 [0.62, 0.69] 0.64 [0.61, 0.68] 0.70 [0.67, 0.73] 0.69 [0.66, 0.72] 0.69 [0.66, 0.72] 0.71 [0.68, 0.74]

Cabozantinib plus nivolumab (C+N) versus sunitinib (S) in patients with advanced renal cell carcinoma (aRCC) and bone metastasis: Updated subgroup analysis of the phase 3 CheckMate-9ER trial.

4528 Background: In first-line aRCC, C+N significantly improved progression-free survival (PFS), overall survival (OS), and objective response rate (ORR) vs S in the phase 3 CheckMate 9ER trial (NCT03141177; Motzer et al. Ann Oncol 2026). In an exploratory analysis with median 18.1-mo f-u, C+N improved PFS, OS, and ORR vs S in patients with or without bone metastasis at baseline. We performed an updated exploratory analysis of outcomes by baseline bone metastasis status based on the 5-year update and analyzed subgroups with additional metastatic sites. Methods: 651 patients with clear-cell aRCC were randomized 1:1 to C (40 mg QD) + N (240 mg Q2W) or S (50 mg QD for 4 weeks of 6-week cycles). Median f-u was 67.6 mo. PFS (primary endpoint) and ORR were per RECIST v1.1 by blinded independent central review. Patient subgroups analyzed here include those with or without bone metastasis at baseline and those with liver, lung, lymph node (LN), or adrenal gland metastasis in addition to bone metastasis. Results: 154 patients had bone metastasis at baseline. Baseline characteristics were generally consistent between treatment arms for subgroups with and without bone metastasis. PFS and OS were prolonged with C+N vs S in patients with or without bone metastasis (Table). The ORR was higher and duration of response (DOR) was longer with C+N vs S in both groups. Although outcomes were less favorable in those with vs without bone metastasis, the relative benefit with C+N vs S was consistent across metastatic site subgroups. The safety profile among patients with bone metastasis was generally consistent with the overall population. Conclusions: With additional follow-up, a consistent benefit was maintained for C+N vs S in patients with or without bone metastasis. Among patients with bone metastasis, a consistent benefit was observed regardless of concomitant metastasis in liver, lung, LN, or adrenal gland. Clinical trial information: NCT03141177 . Efficacy by metastatic sites. Subgroup(C+N v S) Bone + Any Site(s) a (n=79 v n=75) No Bone(n=244 v n=253) Bone + Liver(n=15 v n=17) Bone + Lung(n=57 v n=58) Bone + LN(n=33 v n=38) Bone + Adrenal Gland(n=10 v n=8) mPFS, mo 13.8 v 5.3 b 16.6 v 9.5 6.9 v 3.8 10.0 v 4.4 9.0 v 4.1 15.2 v 5.7 PFS HR (95% CI) 0.43 (0.30, 0.64) b 0.60 (0.49, 0.74) 0.77 (0.33, 1.76) 0.45 (0.29, 0.70) 0.50 (0.29, 0.87) 0.62 (0.19, 2.07) mOS, mo 34.8 v 20.7 b 49.5 v 41.0 20.9 v 12.7 31.6 v 22.1 21.3 v 17.5 54.4 v 31.0 OS HR (95% CI) 0.66 (0.45, 0.95) b 0.85 (0.68, 1.06) 0.57 (0.26, 1.24) 0.72 (0.47, 1.10) 0.72 (0.43, 1.20) 0.72 (0.21, 2.50) ORR, % 49 v 9 b 58 v 33 40 v 18 51 v 10 42 v 5 70 v 13 mTTR, mo 2.9 v 7.3 2.8 v 4.3 2.9 v 3.0 2.8 v 5.7 2.9 v 3.4 4.0 v 11.0 mDOR, mo 18.0 v 6.9 22.9 v 15.4 12.6 v 6.4 18.0 v 6.7 16.7 v NC 14.4 v 19.3 a Only 9 patients had bone-only disease (n=7, C+N; n=2, S), precluding meaningful analysis of this subgroup. b Motzer et al. Ann Oncol 2026. m, median; NC, not calculable; TTR, time to response.