Browse Articles

Comprehensive molecular and clinical characterization of early-onset gastrointestinal cancers: Integrating genomic landscapes with allostatic load in a safety-net cohort.

e22621 Background: The incidence of early-onset gastrointestinal cancers (EOGIC), diagnosed in individuals underage 50, is rising globally. While hereditary syndromes account for a minority of cases, the"exposome"- cumulative environmental and psychosocial stressors remains a critical butunderdefined driver, particularly in marginalized populations. This study characterizes theclinical, genomic, and Allostatic Load (AL) profiles of EOGIC patients at a minority-serving safety-net institution to elucidate the intersection of biological aggression and social determinants. Methods: We conducted a retrospective cohort study (2015–2025) of patients aged 18–50yr withhistologically confirmed GI malignancies at the University of Illinois Chicago. Allostatic Load wasquantified using a validated 10-biomarker index, and was graded by dichotomizing each markerinto high-risk quartiles and aggregating points into a cumulative score (High AL > 2). Genomic profiling was performed using the Tempus xT and xF next-generation sequencing assay. Descriptive statistics are being presented for AL burden, demographics, and somatic mutation landscapes. Results: The cohort (N = 75) was ethnically diverse (40.0% Black, 28.0% Hispanic) and socioeconomicallyvulnerable, with 68.0% relying on Medicaid/Medicare or uninsured. Patients predominantlypresented with advanced disease (52.0% Stage IV) and aggressive, poorly differentiatedhistology (37.3%). High AL at diagnosis was found in 41.3% of this cohort. The most common mutations areoutlined in table 1 and included TP53, APC, KRAS, ARID1A, and PIK3. The median CPS was < 1, MSI-Hrate was 2.7%, and median tissue TMB was 3.95 mut/MB with rare extreme TMB outliers ( > 50mut/MB) accounting for 4.5%. Conclusions: EOGIC patients in this urban safety-net setting are characterized by a dual burden of biologicallyaggressive disease and elevated physiological stress at diagnosis. Nearly half the patients exhibited high AL, reflecting substantial cumulative exposure to chronic stress in early life. Concurrently, tumors frequently presented at advanced stage with poor differentiation and molecular profiles that included frequent TP53, APC, and KRAS mutations. Additionally, these tumor exhibited an immunologically cold landscape marked by low median CPS, low MSI-H prevalence, and modest median TMB. Further study is planned on the impact of AL on molecular biology and clinical outcomes. Ten most common mutations, arranged by percentage. Gene Frequency Percent TP53 78 14.16 APC 71 12.89 KRAS 38 6.90 ARID1A 22 3.99 PIK3 10 3.44 SMAD4 17 3.09 CDKN2A 14 2.54 LRP1B 12 2.18 RB1 10 1.81

Supportive care gaps during breast cancer treatment in Vietnam: Determinants and quality-of-life correlates in a resource-constrained setting.

e12733 Background: In low- and middle-income countries (LMICs), breast cancer outcomes are increasingly shaped not only by treatment access but also by unmet supportive care needs. However, evidence clarifying w hich domains are most affected and which patients are at highest risk in resource-constrained settings remains limited. We assessed unmet supportive care needs and their associations with clinical and treatment-related factors and quality of life (QoL) among Vietnamese patients receiving breast cancer treatment. Methods: We conducted a cross-sectional survey of 182 breast cancer patients treated at a university hospital cancer center in Hanoi between November 2024 and May 2025. Unmet supportive care needs were assessed using the patient-reported Vietnamese Supportive Care Needs Survey–Short Form (SCNS-SF34), with domain scores standardized to a 0–100 scale. QoL was measured using the EORTC QLQ-C30 and QLQ-BR23. Internal consistency was high (Cronbach’s α = 0.937; domain α = 0.78-0.93). Group differences in mean domain scores were examined using t-tests for binary factors and one-way ANOVA for variables with three or more categories, with variance checks and appropriate robust methods when assumptions were violated. Associations between unmet needs and QoL outcomes were evaluated using correlation analyses.. Results: The dominant unmet-need domain was health system & information (mean±SD 51.27±19.55), followed by physical/daily living (34.78±20.33) and psychological needs (34.27±26.23); sexuality-related needs were lowest (10.58±20.78). A frequently reported high unmet need was receiving clear updates that the cancer was controlled/remitting (39.6%). Higher unmet needs in the information, physical, and psychological domains were consistently associated with indicators of treatment burden, including greater pain, treatment-related anxiety, more frequent recent hospitalizations, and chemotherapy (all p < 0.05). Unmet needs were closely linked to poorer QoL, with care/support needs correlating with lower role functioning (r = -0.495) and physical functioning (r = -0.429), and higher symptom burden including nausea/vomiting (r = 0.290) and pain (r = 0.260) (all p < 0.001). Conclusions: In this LMIC setting, unmet supportive care needs concentrate on health system communication and information gaps, and extend to physical/daily living and psychological needs that track with treatment burden. These patient-reported needs are strongly associated with functional impairment and symptom burden, underscoring a high-impact, low-resource opportunity to improve quality of care through routine needs screening and targeted nurse-led education, symptom management, and supportive navigation during intensive treatment.

The role of pretreatment vitamin D insufficiency in racial disparity in taxane-induced peripheral neuropathy: Results from the SWOG S0221 trial.

12136 Background: Taxane-induced peripheral neuropathy (TIPN) affects up to 70% of patients receiving taxanes and can lead to severe, irreversible symptoms that impair daily functioning and quality of life. Approximately 25% of patients require taxane dose modification or discontinuation to prevent TIPN, compromising treatment effectiveness and survival. Self-reported Black patients experience nearly two times greater incidence of TIPN compared to White patients, contributing to treatment alterations and worse cancer outcomes. Vitamin D insufficiency, which is more common among Black patients, has been implicated as a modifiable risk factor for TIPN. This study assessed the extent to which vitamin D insufficiency mediates the racial disparity in TIPN. Methods: This secondary analysis included self-reported Black or White female patients with early-stage breast cancer treated with paclitaxel-based chemotherapy on the SWOG S0221 phase III trial. The primary endpoint was clinician-assessed grade ≥3 sensory neuropathy per NCI CTCAE attributed to paclitaxel. The mediator was pretreatment vitamin D insufficiency, defined as total 25-hydroxyvitamin D ≤20 ng/mL. Causal mediation analysis decomposed the total effect of race on TIPN into its direct effect and an indirect effect mediated through vitamin D insufficiency, adjusted for age and paclitaxel regimen (weekly or every other week). Log-linear model with modified Poisson regression and weighted logistic regression were used for the outcome model and the mediator model, respectively. Confidence intervals were computed via bootstrapping, and significance was set at α=0.05. Results: A total of 1,106 participants (mean age 51.2 [SD 10] years) were included, of whom 109 were Black and 997 were White. Black patients had a higher prevalence of vitamin D insufficiency (77% vs. 28%, p <0.001) and a higher incidence of grade ≥3 sensory TIPN (29% vs. 14%, p <0.001). The adjusted risk ratio (RR) for TIPN comparing Black to White patients was 2.24 (95% CI, 1.54-3.19; p <0.001). Vitamin D insufficiency statistically mediated this association, with an indirect effect RR of 1.18 (95% CI, 1.01-1.38; p = 0.048), accounting for 27% of the racial disparity in TIPN. The direct effect of race on TIPN, independent of vitamin D status, remained significant (RR 1.90; 95% CI, 1.30-2.85; p <0.001). Conclusions: Vitamin D insufficiency contributes to, but does not entirely explain, the racial disparity in TIPN. While other unaccounted factors may interact with or modify this association, vitamin D supplementation in insufficient patients could partially reduce racial disparities in TIPN and improve pharmacoequity.

A new era of immunotherapy: Comparison of real-world outcomes in limited-stage small cell lung cancer.

e20141 Background: Concurrent chemoradiation (CCRT) followed by consolidative anti-PD-L1 immunotherapy (IO) with durvalumab (ADRIATIC) significantly improved both progression-free survival (PFS) and overall survival (OS) in limited-stage SCLC (LS-SCLC), leading to regulatory approval as the new standard of care. Real-world outcomes of patients receiving durvalumab following CCRT for LS-SCLC have not yet been reported. Methods: We retrospectively reviewed the charts of patients diagnosed with LS-SCLC who received CCRT from January 1 to December 31, 2024 in the Indiana University (IU) Health System. Platinum selection (cisplatin or carboplatin), radiation schedule (XRT, once-daily for 60-66 Gy or twice-daily for 45 Gy) and receipt of prophylactic cranial irradiation (PCI) was collected. Patients were stratified by receipt of durvalumab consolidation after CCRT. Fisher’s exact and Wilcoxon rank sum tests were applied to categorical and continuous variables, respectively. Univariate Cox analysis for PFS by IO status was performed. Results: Among the 416 individuals with SCLC, 25 patients had LS-SCLC treated with CCRT (Table 1), including 14 patients (56%) who received consolidative IO. The groups (no IO vs IO) were well balanced, with no significant difference observed in age at diagnosis, sex, ECOG performance status (PS), pre-existing diagnoses of COPD or autoimmune disease, or tobacco use between groups. There was no difference in receipt of PCI between cohorts. With the addition of IO consolidation, a significant improvement in median PFS (≥9.7 vs 6.1 months, P = 0.007) was observed; median OS was numerically higher but did not reach statistical significance (18.2 vs 15.2 months, P = 0.07) with <24 months of median follow-up. Conclusions: In our cohort, the addition of consolidative durvalumab significantly improved PFS and trended towards higher OS in real-world setting for patients with LS-SCLC who received CCRT, while we await data maturation. Carboplatin and once-daily XRT for 60-66 Gy were commonly used in our real-world cohort, while PCI was administered in a minority of patients studied. Characteristic No IO (n = 11) +IO (n = 14) Overall (n = 25) P value Age (Q1, Q3) 67 (64, 74) 69 (64, 78) 69 (64, 75) 0.6 Sex (female) 5 (45%) 10 (71%) 15 (60%) 0.2 ECOG PS (0-1) 7 (64%) 12 (86%) 19 (76%) 0.13 Current tobacco use 7 (64%) 11 (79%) 18 (72%) 0.7 Former tobacco use 4 (36%) 3 (21%) 7 (28%) 0.7 Autoimmune history 1 (9.1%) 2 (14%) 3 (12%) >0.9 +PCI 3 (27%) 3 (21%) 6 (24%) >0.9 Platinum agent (carboplatin) 6 (55%) 8 (57%) 14 (56%) >0.9 XRT schedule (once-daily, 60-66 Gy) 8 (73%) 11 (79%) 19 (76%) 0.4

Factors associated with early cachexia in metastatic NSCLC: A nationwide Veterans Health Administration analysis.

e20681 Background: Cancer-associated cachexia is a frequent and often irreversible driver of morbidity in non-small cell lung cancer (NSCLC), affecting approximately 40% of patients (pts) and directly contributing to death in up to 20%. Yet effective preventative and therapeutic strategies remain lacking. We leveraged data from a nationwide Veterans Health Administration (VHA) cohort to identify clinical, demographic and disease-related factors associated with cachexia at the time of metastatic NSCLC diagnosis. Methods: We conducted an observational cohort study of Veterans with metastatic NSCLC diagnosed and treated within the VHA from January 1, 2010, to September 30, 2025. Cachexia was defined as ≥5% weight loss within 6 months prior to metastatic diagnosis, or ≥2% weight loss within 6 months prior to metastatic diagnosis in pts with body mass index 20 kg/m 2 . Multivariable logistic regression was used to identify factors independently associated with cachexia. Results: Among 23,107 Veterans with metastatic NSCLC (median age 69; 97.5% male; 19.1% non-Hispanic Black; 55.1% adenocarcinoma), cachexia was present in 9,921 pts (42.9%) at diagnosis. Cachexia occurred more frequently among non-Hispanic Black pts (22% vs. 17%, p < 0.001), those with high comorbidity burden (Charlson comorbidity index (CCI) ≥7: 40% vs. 28%, p < 0.001), and poorer performance status (ECOG ≥2: 53.7% vs. 40.9%, p < 0.001). Higher rates of cachexia were also observed in pts with liver metastases (22% vs. 19%, p < 0.001) and bone metastases (41% vs. 37%, p < 0.001), but not among those with brain metastases or lung-only metastases. In multivariable analysis, Black race (OR 1.38, 95% CI: 1.27-1.50, p < 0.001), squamous histology (OR 1.14, 95% CI: 1.06-1.23, p < 0.001), liver metastases (OR 1.16, 95% CI: 1.07-1.26, p < 0.001), and bone metastases (OR 1.18, 95% CI: 1.1-1.26, p < 0.001) were independently associated with increased odds of cachexia. Results were consistent in analyses restricted to pts with a single metastatic site. Cachexia was independently associated with worse survival after adjustment for metastatic sites, race, and comorbidity burden (OR 1.38, 95% CI: 1.33-1.42, p < 0.001). Conclusions: In this large national cohort of pts with metastatic NSCLC, cachexia was highly prevalent at diagnosis and independently associated with clinical and disease-related factors including race, histology, comorbidity burden, and specific sites of metastases. Importantly, cachexia remained independently associated with worse survival after adjustment for metastatic sites and patient characteristics. These findings highlight cachexia as a common, early, and clinically consequential condition in metastatic NSCLC and support the need for risk-stratified approaches to early cachexia identification, prevention, and therapeutic investigation.

Inpatient outcomes and resource utilization by complication clustering phenotypes among colorectal cancer hospitalizations in the United States.

e23215 Background: Acute complications are common drivers of hospitalization in colorectal cancer (CRC), yet most inpatient studies evaluate individual complications in isolation. National data describing outcomes associated with co-occurring complication patterns remain limited. Methods: A serial cross-sectional analysis was performed using the 2018–2022 Healthcare Cost and Utilization Project National Inpatient Sample. Adult hospitalizations with a principal diagnosis of CRC were identified. Acute complications were ascertained from secondary diagnoses and categorized as mechanical (bowel obstruction and/or perforation), gastrointestinal bleeding, or sepsis. Hospitalizations were classified into mutually exclusive complication clustering phenotypes: none, mechanical only, bleeding only, systemic only (sepsis), mixed mechanical + sepsis, and other mixed patterns. Survey-weighted analyses estimated national prevalence, in-hospital mortality, length of stay (LOS), hospitalization cost, and discharge disposition. Survey-weighted multivariable logistic regression assessed associations between complication phenotypes and in-hospital mortality, adjusting for demographics, payer, neighborhood income quartile, hospital characteristics, and calendar year. Results: An estimated 608,685 CRC hospitalizations were identified nationally. Any acute complication occurred in 29.3% of hospitalizations. Phenotype prevalence was 70.7% with no complications, 17.7% mechanical only, 7.0% bleeding only, 1.3% systemic only, 1.6% mixed mechanical + sepsis, and 1.7% other mixed patterns. In-hospital mortality increased stepwise across phenotypes, from 1.4% without complications to 24.4% among mixed mechanical + sepsis hospitalizations. In adjusted analyses, odds of in-hospital death were markedly higher for systemic-only phenotypes (adjusted odds ratio [aOR] 17.19, 95% CI 14.74–20.05) and mixed mechanical + sepsis phenotypes (aOR 20.58, 95% CI 17.89–23.67), compared with no complications. Resource utilization increased substantially with clustering severity: mean LOS rose from 5.2 days without complications to 20.1 days in mixed mechanical + sepsis hospitalizations, while mean hospitalization cost increased from $24,054 to $83,376. Mixed phenotypes were also associated with higher rates of non-home discharge. Conclusions: Among CRC hospitalizations, complication clustering identified a small subset with markedly higher inpatient mortality and resource utilization. Although mixed and systemic phenotypes accounted for fewer than 3.5% of admissions, they contributed a disproportionate share of inpatient deaths and costs. These nationally representative findings provide benchmarking data for complication clustering as a marker of inpatient risk in colorectal cancer.

A first-in-human (FIH), phase I/II open-label, dose-escalation and -expansion study of ILKN421H, an LNP mRNA encoding an IL2Rβγ selective IL-2v, as monotherapy and in combination with pembrolizumab, in patients with advanced solid tumors.

8556 Background: ILKN421H is an LNP mRNA encoding an IL2Rβγ selective IL-2v derivative fused with Human Serum Album. Preclinically, the novel LNP mRNA platform allowed expression and therefore extended plasma half-life of IL2v in animal models, which is not achieved with other protein-based IL-2 drugs. ILKN421H stimulated significant and sustained expansion of CD8 T cells for more than three weeks, with minimal expansion of regulatory T cells. Moreover, ILKN421H demonstrated superior efficacy than IL-2 based drugs in multiple cancer models and has an excellent safety profile in rodents and non-human primates. Therefore, the novel design of ILKN421H has the potential to fully unleash the antitumor effects of the IL-2 pathway while mitigating unwanted toxicity. Methods: This first-in-human, open label phase 1 study evaluates the safety, tolerability, and initial efficacy of ILKN421H with or without pembrolizumab in patients with advanced solid tumors. The design includes two portions of the study: Parts A, ILKN421H monotherapy in which patients receive intravenous ILKN421H once every 3 weeks (Q3W) and Part B in which patients receive ILKN421H in combination with pembrolizumab (200 mg on day 1), both intravenous Q3W. Both portions of the study consist of 3+3 escalation cohorts to define maximum tolerated dose (MTD). If response signal is observed in a given dose cohort, additional 10 patients will be enrolled in the expansion portion of the study to further evaluate the safety and efficacy. The trial is ongoing and data as of 5/7/2024 is presented in this abstract. Results: A total of 47 participants were enrolled so far. In Part A, 9 patients in three dose cohorts received ILKN421H. All patients demonstrated high and prolonged plasma level of IL-2, and remarkable elevation of CD8 T cells (up to 5 folds) and NK cells (up to 25 folds). No dose-limiting toxicities (DLT) were observed. In Part B, 38 patients in three dose cohorts were enrolled, 24 of which are 1L advanced NSCLC. All patients showed robustly increased CD8 T cells and NK cells, and well tolerated safety profile (TESAE=26%). At the cut-off date (Nov 25th, 2025), of the 22 efficacy evaluable 1L NSCLC patients regardless of the PDL1 expression levels, 17 were evaluated to be PR (ORR=77.2%), and 50% PFS was not reached and was projected to be more than 14 month. In PDL1 negative patients (TPS <1%), 3 out of 6 patients achieved PR (ORR=50%). Conclusions: ILKN421H was generally well tolerated at the current dose level, which already demonstrated high and prolonged plasma level of IL-2, and robust immune stimulatory activity in promoting CD8 T cells and NK cells proliferation. Initial signs of antitumor efficacy were seen in combination with pembrolizumab. Clinical trial information: NCT05978102 .

Longitudinal genomic characterization of pediatric acute myeloid leukemia at diagnosis and relapse.

10020 Background: Relapse remains the leading cause of treatment failure and mortality in pediatric acute myeloid leukemia (pAML). In AML, relapse is common and is increasingly recognized as a consequence of therapy-driven clonal evolution rather than mere persistence of the diagnostic clone. This study aimed to delineate mutational trajectories from diagnosis to relapse and identify relapse-associated genomic alterations of potential clinical relevance. Methods: This prospective study was conducted from July 2021 to December 2025. Bone marrow or peripheral blood samples were collected from children ≤18 years at diagnosis and relapse. Genomic DNA from paired samples was sequenced exomically to identify somatic mutations. Longitudinal comparative analyses assessed clonal persistence, emergence of relapse-specific mutations, and pathway-level enrichment. Results: A total of 100 children with AML were enrolled. Paired longitudinal analysis (n=18) identified 133 mutated genes. Seventy-nine (59.4%) were shared between diagnosis and relapse, forming a conserved truncal genomic backbone. Fifty genes (37.6%) were exclusive to diagnosis, while only four (3.0%) were exclusive to relapse, suggesting strong therapeutic selection pressure rather than widespread genomic diversification. At diagnosis, missense variants predominated (88.4%), followed by in-frame insertions/deletions (6.5%), nonsense (2.4%), and frameshift alterations (2.5%). Relapse samples showed further enrichment of missense variants (90.5%) and a relative depletion of truncating events, consistent with selection for functionally adaptive mutations. Truncal alterations were consistently observed in epigenetic regulators TET2 and KMT2C, supporting their role as early leukemogenic events. In contrast, relapse samples showed enrichment of signaling and transcriptional regulators, including FLT3 and PTPN11, along with relapse-exclusive emergence of WNT pathway components (FZD7, LEF1) and the PI3K regulator PIK3R1. Functional enrichment analysis revealed that diagnostic samples were primarily enriched for DNA damage response, p53 signaling, apoptosis, and cell-cycle checkpoint pathways (p<0.05). At relapse, these core programs persisted alongside marked enrichment of oncogenic signaling pathways (WNT, PI3K–AKT, FLT3) and metabolic processes, including glycolysis and ATP generation. Reactome analysis highlighted convergence on FLT3- and AKT-driven signaling, TP53-mediated DNA repair, and escape from apoptosis. Conclusions: Longitudinal genomic profiling demonstrates that pAML relapse is driven by the persistence of early truncal mutations, coupled with the selective expansion of signaling- and metabolism-adapted subclones. Enrichment of WNT, PI3K–AKT, and FLT3-associated pathways at relapse reveals potential targets for relapse-directed precision strategies.

Brain metastases in HER2- positive metastatic gastrointestinal cancers: Incidence, timing, and implications for brain surveillance.

e16001 Background: Brain metastases (BM) are considered rare in gastrointestinal (GI) malignancies, and routine brain imaging is not recommended, though often required as eligibility in HER2-positive GI clinical trials. Though HER2-positive GI cancers have a higher risk of BM, there is limited data on their timing and natural history. Methods: We retrospectively analyzed 432 patients with HER2-positive metastatic GI cancers treated between 2000 and 2024. HER2 positivity was defined as immunohistochemistry (IHC) 3+ or ERBB2 amplification identified by next-generation sequencing. BM were confirmed radiographically. Time to BM development was analyzed using a competing-risk framework, with death without BM treated as a competing event. Cumulative incidence functions (CIF) were estimated, and Fine–Gray subdistribution hazard models were used to evaluate factors associated with BM development. Overall survival (OS) was analyzed using Kaplan–Meier methods and multivariable Cox proportional hazards regression. Results: Among 432 patients with HER2-positive metastatic GI cancers, 45% had gastroesophageal cancers, 39% colorectal cancer, 13% biliary tract cancers, and 3% other gastrointestinal cancers. Median OS for the overall cohort was 20.4 months (95% CI 16.1–24.0). Overall, 35 patients (8.1%) developed BM. Using a competing-risk framework, BM was observed within the initial months following metastatic diagnosis, with cumulative incidence estimates of 3.9% by 2 months and 5.3% by 12 months, increasing to 8.9% at 24 months, 10.2% at 5 years, and 14.2% by 8 years. The hazard of BM peaked within the first 2–3 months following metastatic diagnosis and remained non-zero throughout follow-up. In multivariable competing-risk analysis, tumor type was evaluated for association with BM development (gastroesophageal vs colorectal cancer: HR 2.10, 95% CI, 0.97–4.51; P = 0.059), whereas biliary tract cancer was not significantly different from colorectal cancer (HR 0.358, 95% CI, 0.046-2.79, P = 0.3269). Conclusions: In HER2-positive metastatic GI cancers, cumulative incidence of BM at 2 years was 8.9%. Critically, the majority of this risk occurred early with cumulative incidence of 3.9% at 2 months. These findings challenge the perception of BM as a late or rare event in this population and support consideration of brain imaging especially at the diagnosis of metastatic disease.

A prospective clinical and dosimetric study on the feasibility and adaptability of MR-Linac in the treatment of cervical cancer: Initial experience from a tertiary institution.

e17527 Background: Cervical cancer is a leading cause of cancer burden globally, with a disproportionate burden in LMICs like India. The pelvis is a highly dynamic anatomical region; tumour regression, variable bladder and rectal filling, and small bowel displacement produce substantial inter- and intra-fraction changes that challenge a single pre-treatment plan. We treated cervical cancer patients with standard protocol, with reduced anisotropic margins and daily online adaptive radiotherapy, evaluated the feasibility, dosimetric impact, OAR sparing, tumour regression and bladder dynamics & target coverage with reduced margins. Methods: 7 patients with FIGO IB–IIIC2 cervical cancer received EBRT to 45 Gy in 25 fractions with concurrent weekly cisplatin, followed by MRI-guided brachytherapy. CT - MR based delineation was done for GTV, CTV and OAR's with an anisotropic PTV margins of 5–13 mm. Daily HR 3D MRI was acquired on Unity prior to each fraction. Fractions with minor changes were treated with Adapt-to-Position (ATP), and those with significant anatomical variation (>5 mm shift, >10% volume change, bowel/bladder deformation) with Adapt-to-Shape (ATS). After adaptation, Iso-centre MR scan was acquired immediately pre-beam in 123/175 fractions to assess intra-fraction changes and re-optimise plans when required. Dosimetry of scheduled and adapted plans was compared for PTV, bladder, rectum and small bowel. Also, the MR plans were compared with the plans generated with Standard PTV margis as per the EMBRACE 2 protocol. Serial GTV volumes were observed for tumour regression. Results: All patients completed chemoradiation with 175/175 fractions. Mean total treatment time was 35.2 ± 5.8 minutes per fraction. Baseline median GTV, CTV and PTV were 29.03, 121.52 and 291.13 cm³ respectively. Mean PTV D95% was 98.53 ± 1.26%, with 100% of fractions achieving ≥95% coverage despite reduced margins. All patients demonstrated marked tumour regression: baseline GTV 12.51–52.40 cm³ decreased to 0–2.35 cm³, with mean GTV regression of 96.8 ± 2.4%. Inter-fraction variability was dominated by bladder and bowel motion: mean inter-fraction bladder volume difference between scheduled and adapted scans was 128.92 cm³. Intra-fraction analysis universal significant bladder filling: 100% fractions had increased intra fraction bladder volume [mean increase 128.92 ± 70 cc], 91.1% showing >50 cc change. Conclusions: Daily online adaptive MRgRT for cervical cancer is feasible with a ~35-minute workflow, achieves excellent target coverage with reduced PTV margins. The combination of near-complete tumour regression, substantial small bowel sparing, and profound inter- and intra-fraction bladder and bowel motion establishes the same. This supports MR-LINAC–based adaptive radiotherapy as a promising precision platform in cervical cancer.

Post-treatment ctDNA monitoring using a tumor-informed whole-genome sequencing-based assay in homologous recombination–proficient ovarian cancer.

5592 Background: Ovarian cancer (OC) has high mortality due to advanced stage at diagnosis and frequent, incurable recurrences. Homologous recombination–proficient (HRP) disease comprises approximately half of high-grade serous OC and is associated with limited benefit from platinum chemotherapy and PARP inhibitors, leading to poorer outcomes. Circulating tumor DNA (ctDNA) is a validated biomarker for molecular residual disease and early relapse; this study evaluates it’s ability to stratify HRP patients (pts) into clinically meaningful risk groups. Methods: We performed a retrospective analysis of clinically annotated residual samples from commercial ctDNA testing in pts diagnosed with OC. Signatera Genome assays (Natera, Inc.) were designed from the respective pts’ matched tumor and normal whole genome sequencing (WGS) data to detect and quantify ctDNA in serially collected post-operative plasma. Adjuvant chemotherapy (ACT) decisions and cadence of testing were at the provider’s discretion. HR status was determined using available clinical records and/or genomic data. Descriptive statistics were used to compare association of ctDNA across stages and at various timepoints post-surgery. Correlation between ctDNA status and progression-free survival (PFS) was assessed using Cox regression analysis. Results: Study cohort comprised 164 Stage I-IV OC pts, with 1,256 plasma samples for ctDNA testing (median: 7 tests/ pt, range 1–24). The median age at diagnosis was 66 (range 19–89) years. A total of 157 cases were HRP; among these, clinical outcome data was available for 44 pts with corresponding ctDNA timepoints for analysis. ctDNA-positivity within two weeks post-debulking surgery (before starting ACT) was associated with significantly worse PFS compared to ctDNA-negativity (N=20; HR: 14.01, 95%CI: 0.6–308.16; p=0.01). Furthermore, ctDNA-positivity post-ACT was also associated with worse PFS (N=26; HR: 6.6, 95%CI: 1.7–25.3; p=0.002) compared to ctDNA-negativity. Among ctDNA-positive pts, 50% (5/10) had progression <6 months post-ACT, whereas no events were observed <12 months post-ACT in ctDNA-negative pts (N=12). Clinical follow-up and additional analyses are ongoing. Conclusions: Tumor-informed ctDNA monitoring identifies pts with HRP OC who are at high risk for recurrence post-surgery including treatment refractory/resistant cases. Post-operative and post-adjuvant ctDNA positivity was strongly associated with inferior PFS. These findings support the use of ctDNA as a clinically important biomarker to inform post-surgical/treatment risk stratification, and consideration of alternative or escalated therapeutic strategies in this high-risk population. These results showcase the utility of ctDNA to further risk stratify HRP OC pts, which supports it’s integration into prospective clinical trials and clinical care.

Preclinical efficacy and brain uptake of an insulin-mediated nanoparticle co-delivering trastuzumab and pertuzumab (AxS007) in HER2+ breast cancer brain metastasis models.

2027 Background: Anti-HER2 antibodies are standard therapies for metastatic HER2+ breast cancer but demonstrate limited intracranial activity due to restricted penetration across the blood–brain barrier (BBB). AxS007 is an insulin-mediated nanoparticle conjugated to trastuzumab and pertuzumab designed to enhance BBB transport and increase brain exposure. Methods: Brain uptake and distribution were assessed in C57BL mice using biodistribution analysis (5 min–90 days, n=25) and two-photon microscopy, comparing AxS007, a non-insulin-conjugated nanoparticle control, and free trastuzumab. Brain uptake was also evaluated in non-human primates (NHPs). Based on confirmed BBB penetration and brain exposure, AxS007 antitumor activity was evaluated in vitro in HER2+ cell lines (BT474, SKBR3, N87) and in vivo in BT474 subcutaneous (SC; n=20) and intracranial xenografts. Tumor growth inhibition (TGI) was assessed by MRI, and survival was compared between vehicle, AxS007 (IV, 10 mg/kg), and free trastuzumab + pertuzumab (IV, 10 mg/kg each). Safety was further investigated through single ascending dose (SAD) and multiple ascending dose (MAD) toxicological studies. Results: In mice, AxS007 demonstrated measurable brain uptake within 30 min and widespread parenchymal distribution at 24 h, with markedly higher brain exposure than free trastuzumab; in contrast, the non-insulin control and free antibodies showed minimal BBB crossing. In NHPs, AxS007 increased brain antibody exposure ~50-fold versus free antibodies at 48 h. In vitro, AxS007 showed antitumor activity comparable to or greater than free antibodies (SKBR3: 90% vs 50%; BT474: 50% vs 64%). In vivo, AxS007 achieved near-complete tumor elimination in SC tumors and demonstrated 82% TGI versus vehicle (p<0.01) and 60% TGI versus free trastuzumab +pertuzumab (p<0.05) in intracranial tumors, as assessed at day 60 post-inoculation. In an overall survival study, AxS007 extended survival approximately 3-fold compared with free trastusumab + pertuzumab. AxS007 exhibited a good safety profile with no drug related side effects. Conclusions: AxS007 showed enhanced BBB penetration and increased brain antibody exposure, with improved preclinical efficacy versus free antibodies in HER2+ breast cancer brain metastasis models. These findings support further translational advancement of AxS007 toward first-in-human clinical testing.

Phase 1 dose escalation of CTX-8371, a novel PD-1×PD-L1 bispecific antibody, in patients with advanced malignancies post checkpoint inhibition.

2629 Background: CTX-8371 is a novel bispecific antibody targeting both programmed cell death 1 (PD-1) and programmed cell death ligand 1 (PD-L1). In addition to potent PD-1 and PD-L1 blockade, CTX-8371 uniquely enables CD80-CD28 engagement and facilitates the cleavage of PD-1 from the surface of activated T-cells. Preclinically, CTX-8371 exhibited greater potency than singular blockade with either anti-PD1 or anti-PD-L1 alone. Methods: In this first-in-human study (NCT06150664) with a 3+3 dose-escalation design, patients (pts) with metastatic melanoma, non-small cell lung cancer (NSCLC), head and neck squamous cell carcinoma (HNSCC), Hodgkin lymphoma (HL), and triple-negative breast cancer (TNBC) in whom standard treatments including prior immune checkpoint inhibitors (ICI) had failed, were eligible. CTX-8371 was administered intravenously every two weeks at 5 dose levels ranging from 0.1 to 10.0 mg/kg. The primary objectives were to evaluate the safety and tolerability of CTX-8371 and determine doses to be evaluated in expansion. Secondary objectives included assessment of anti-tumor activity, pharmacokinetics, pharmacodynamics, and immunogenicity of CTX-8371. Results: Between 15 Apr 2024 and 12 Aug 2025, 17 pts were enrolled; 15 pts completed the dose-limiting toxicity (DLT) period and had post-baseline imaging: NSCLC n=7, melanoma n=3, HNSCC n=2, TNBC n=2, HL n=1. All 15 pts had received a minimum of 2 prior therapies including checkpoint inhibitors. Median duration on study was 9.8 months; CTX-8371 treatment is ongoing in 3 pts. CTX-8371 was well tolerated with no DLTs. The maximum tolerated dose (MTD) was not reached. All treatment-related adverse events (AEs) were Grade 1/2, except one Grade 3 AE of asymptomatic lipase increase which resulted in the only dose interruption. There were no dose reductions. Dose-dependent increases in CTX-8371 serum concentrations (AUC and Cmax) were observed. In the 15 pts with post-baseline imaging, the overall response rate (ORR) was 20% (one pt each with NSCLC [irPR], TNBC and HL) and the disease control rate (DCR) was 60%. The responses were 2.3 months NSCLC; 6.4+ months TNBC; 3.6+ months HL with greater depth and longer durability at the recommended doses for further study (≥ 3 mg/kg). The two ongoing responders are in pts treated at 3 mg/kg (TNBC: >90% reduction in overall tumor volume) and 10 mg/kg (HL: metabolic partial response). Conclusions: CTX-8371, a novel bispecific antibody targeting both PD-1 and PD-L1, was well tolerated and demonstrated promising clinical activity as a monotherapy in advanced pts resistant to prior ICI. Dose expansion in pts with NSCLC and TNBC in 2 dosing cohorts (3 and 10 mg/kg) is underway. Future development in treatment refractory HL is planned. Clinical trial information: NCT06150664 .

Clinicopathologic and molecular features of Krukenberg tumors from gastric cancer (GC): A single-institution experience.

4043 Background: Krukenberg tumors (KTs), ovarian metastases most commonly arising from gastric adenocarcinoma, represent an aggressive and understudied metastatic phenotype, frequently associated with diffuse-type gastric cancer (GC). The molecular determinants underlying this metastatic pattern remain poorly characterized. We aimed to define the clinicopathologic and genomic landscape of GC-associated KTs. Methods: We conducted a retrospective review of 103 patients with metastatic gastric or gastroesophageal junction adenocarcinoma treated at City of Hope (2020–2025). Twenty-two patients with confirmed KTs were identified. Clinicopathologic data were extracted from electronic medical records. Next-generation sequencing (NGS) for KTs was performed on primary or metastatic tumor specimens. Survival outcomes were estimated using the Kaplan-Meier method, with progression-free survival (PFS) and overall survival (OS) reported with 95% confidence intervals (CIs). Results: The median age at diagnosis was 45 years (range, 30–64). HER2 positivity was infrequent (14%), and PD-L1 expression was heterogeneous across tumors. Genomic profiling demonstrated recurrent alterations in TP53 (53%), ARID1A (20%), CDH1 (13%), KRAS (13%), and PIK3CA (13%). Notably, 86.4% of KTs exhibited CLDN18.2 immunohistochemical expression > 75%, and 80% of patients had low PD-L1 CPS scores ( < 5). The CLDN18::ARHGAP26 fusion was detected in 30% of sequenced tumors, representing a highly recurrent structural alteration. Median PFS was 9 months (95% CI, 4–15), and median OS was 17 months (95% CI, 5–30). Conclusions: GC–associated KTs demonstrate a distinct molecular architecture enriched for diffuse-type genomic features, including frequent TP53 mutations and recurrent CLDN18::ARHGAP26 fusions, aligning with the poor prognosis observed in this cohort. Our findings establish CLDN18 fusions as a recurrent molecular hallmark of GC–associated ovarian metastases, providing novel clinicopathologic insight into mechanisms of ovarian metastatic tropism. Importantly, the high prevalence of CLDN18.2 protein expression identifies a biologically actionable therapeutic vulnerability. CLDN18.2-directed strategies may represent an effective biomarker-driven treatment approach capable of achieving disease control without reliance on surgical intervention. These data define a high-risk, biologically distinct metastatic phenotype and support precision-guided therapeutic targeting, in this clinically aggressive patient population.

Prescribing of anti-dementia medications in primary care: A retrospective cohort study in 1489 English General Practices

Objective Evidence suggests that patient-level characteristics such as socio-economic status or ethnicity affect the likelihood of receiving guideline recommended anti-dementia medications. Existing studies often included all-cause dementia, not just the specific subtypes in which medication is indicated. Patterns of prescribing of Acetyl Cholinesterase Inhibitors (AChEIs) and memantine require further exploration, with little evidence about rates of co-prescribing in English primary care. We examined variations in anti-dementia medication prescribing with patient-level characteristics, and over time. Design and setting Retrospective cohort study, using the Clinical Practice Research Datalink Aurum. Data from 1,489 practices, in England between 2006–2024, were included and linked to patient level Index of Multiple Deprivation data (2019). Cox-regression modelling, clustered at practice level, assessed association between patient-level characteristics and receiving AChEIs, and/or memantine. Time-series analyses examined co-prescribing of memantine and AChEIs. Participants 242,007 patients, aged >=18 years, with Alzheimer’s or Lewy-Body Dementia, or mixed dementia including one of these subtypes, were included. Results Among the 242,007 patients, 63.1% were prescribed an anti-dementia medication; co-prescribing of memantine and AChEIs peaked at 4.2%. Those in the most deprived quintile were less likely to be prescribed AChEIs (Hazard Ratio (HR) 0.82,0.78-0.86) compared to the most affluent quintile. People with Asian (HR 0.89,0.84−96), or Black (HR 0.79, 0.73-0.86) ethnicities were less likely to be prescribed memantine compared to white people. Those with learning disabilities were substantially less likely to be prescribed AChEIs (HR 0.46,0.42-0.50) or memantine (HR 0.58, 0.50-0.67) compared to those without. Conclusion Overall rates of prescribing of anti-dementia medications were lower than expected. Rates of co-prescription of AChEIs and memantine were low, despite guideline recommendations. We found inequity in anti-dementia medication prescribing, relating to multiple patient-level characteristics highlighting the need for more equitable access to evidence-based treatments.

Post-treatment with biogenic silver nanoparticles modulates hydration and polymer structure in soft contact lenses: An FTIR-chemometric study

Artificially Designed 2D Composite Material With Tailored Thermal Expansion Toward Temperature‐Stable and Accurate Nanomechanical Pressure Sensors

ABSTRACT Two‐dimensional (2D) materials are suitable for constructing resonant nanoelectromechanical systems (NEMS) toward advanced sensing applications, due to their ultrasmall mass, high Young's modulus, and large strain limit. However, there exists a fundamental limitation: the thermal expansion of 2D materials leads to intrinsic frequency susceptibility of 2D NEMS to temperature variation, which has significantly impaired their deployment in realistic sensing applications. Here, leveraging the layer degree of freedom in 2D materials, we realize a type of artificially designed 2D composite materials, in this case specifically designed graphene–molybdenum disulfide (MoS 2 ) van der Waals heterostructures, with tailored thermal expansion, and successfully suppress temperature‐induced frequency drift by leveraging the near‐zero net thermal expansion. With this approach, we demonstrate temperature‐stable and highly‐accurate NEMS pressure sensors: within 200–400 K range, we improve the temperature stability by 2‐orders‐of‐magnitude compared with individual MoS 2 and graphene devices, and achieve nearly 2000‐fold enhancement of signal‐to‐error ratio over the entire temperature range required for automotive grade sensors. Our strategy offers an effective solution, with clear design guideline, for tailoring the thermal expansion of 2D materials, opening new possibilities in temperature‐stable 2D NEMS sensing applications under realistic environmental settings.

Prediction of the surface roughness of Ti-6Al-4 V alloy during surface grinding using machine learning models

Abstract This study presents a machine learning (ML)-based approach to predict surface roughness, R a during dry grinding of Ti-6Al-4 V alloy using Decision Tree, Random Forest, Gradient Boosting, Extreme Gradient Boosting (XGBoost), Linear Regression, and Polynomial Regression models. Experiments were conducted with Aluminium Oxide (Al₂O₃) and Silicon Carbide (SiC) wheels at varying feed rates (0.2–0.9 mm/rev) and depths of cut (0.02–0.08 mm). Results showed that Al₂O₃ consistently produced lower R a values.Increased feed and depth of cut led to rougher surfaces. XGBoost algorithm achieved the highest prediction accuracy (R² = 0.90), effectively capturing nonlinear dependencies. Feature importance analysis identified feed rate as the most influential factor (Importance Score > 0.85), followed by depth of cut and wheel type. These findings demonstrate the potential of ML, particularly XGBoost, for optimizing grinding parameters and enhancing surface quality in Ti-6Al-4 V machining.

Rab8 and TNPO1 function as the ciliary transport adapters for GPCRs

Tumor microenvironment modulation by losartan in advanced gallbladder cancer (GBC): Results and genomic correlates from the phase 2 LAST study.

4154 Background: Advanced gallbladder cancer (GBC) carries a dismal prognosis, with a real-world median overall survival (OS) of 6.8 months. We hypothesized that adding losartan to standard chemotherapy could improve outcomes by modulating the fibrotic tumor microenvironment (TME). Methods: This was a single-arm phase II study conducted at two academic centers between July 2022 and April 2024. Adult patients with histologically confirmed, unresectable or metastatic gallbladder cancer or cholangiocarcinoma received six cycles of gemcitabine and cisplatin with concurrent oral losartan. The primary endpoint was overall survival (OS). Exploratory analysis by whole-exome sequencing (WES) was done on 18 patients. Results: Thirty patients were enrolled. Clinical characteristics are depicted in Table. Median OS was 10.3 months (range 0.5–38.7), with 6-month and 12-month OS rates of 68% and 36%, respectively. Median PFS was 5 months. An objective response was observed in 88% of patients, including five complete responses. Patients with ECOG performance status 2 and those with cholelithiasis had inferior survival. Losartan was well tolerated. Genomic analysis revealed a clear discretion between cohorts (median PFS >, <=6 months). Responders exhibited a "TME-favorable" signature, characterized by genes associated with anti-inflammatory properties, regulated angiogenesis, and cellular stress response. This group was uniquely enriched for variants in ATRX, CHEK2, DNMT3A, EZH2, IDH1, SMARCA4 , and TSC2. Conversely, non-responders (NR) showed a pro-inflammatory profile and oncogenic mutations in the angiogenesis pathway ( AKT2, BRAF, MET, NRAS, SMAD4 ), potentially mediating resistance to losartan-induced vascular normalization. Both groups shared core GBC drivers, including TP53, KRAS, ARID1A , and ERBB2. Conclusions: Losartan improves survival in GBC patients with a specific molecular buildup. The presence of anti-inflammatory genomic signatures and specific variants like SMARCA4 and ATRX may serve as predictive biomarkers for TME-targeted therapy, whereas pro-inflammatory profiles and primary angiogenesis mutations identify a resistant phenotype. Clinical trial information: CTRI/2023/02/049523. Baseline clinical features (n=30). Clinical features n % Age in years Median (range ) 59 (31-75) Gender Male Female 1119 3664 ECOG PS * 1 2 219 7030 Gallstones 12 41 Cholecystectomy 10 34 Jaundice at baseline Present Absent 822 2674 Site Gallbladder Cholangiocarcinoma Intra-hepatic Extra-hepatic distal CBD ** 2613 87310 Weight in Kgs Mean (range ) 58 (41-82) History of weight loss present 15 52 Co-morbidities Diabetes mellitus Hypertension Hypothyroidism 781 23263 Angiotensin receptor blocker Losartan Telmisartan 282 946 ERBB2 positive Received Trastuzumab 22 6100 Immunotherapy(Durvalumab) 5 17 * PS: Performance status, ** CBD: Common bile duct.