Browse Articles

Genomic characterization of brain metastasis in colorectal cancer to reveal patterns of clonal evolution and selection of driver mutations.

e15590 Background: Colorectal cancer (CRC) is a malignant tumor of the digestive system, known for its high incidence and mortality rates worldwide. Although colorectal cancer brain metastases (CRCBM) are uncommon, they are extremely deadly and greatly affect patient survival and prognosis.This research seeks to explore the genomic changes and clonal evolution in CRCBM, focusing on driver mutations, CNVs, and their potential effects on treatment strategies. Methods: We performed a comprehensive analysis on a cohort of 26 CRC patients and a cohort of 55 patients with CRCBM, including non-synchronous matched primary samples from 13 individuals. This analysis encompassed the characteristics of various mutations, including genomic alterations, distribution and frequency patterns.We also examined driver genes and CNVs across the three cohorts. For the 13 matched cases with both primary and metastatic samples, we identified putative driver genetic events resulting from clonal evolution that could facilitate metastasis and examined the clonal evolution process. Additionally, we compared HLA typing, neoantigens, and pathways affected by genetic alterations in patients with brain metastases to enhance our understanding of the mechanisms that may determine the specific site of metastasis. Results: Compared to primary CRC, CRCBM demonstrated increased microsatellite instability (MSI), an elevated tumor mutational burden (TMB), and a greater number of predicted neoantigens. Nevertheless, an analysis of neoepitope-major histocompatibility complexbinding affinity indicated that the potential efficacy of immunotherapy in patients with CRCBM may be limited. Within the CRCBM cohort, we identified novel driver mutations, including those in AMER1, TCF7L2, and MAP2K7. Clonal evolution analysis revealed early mutations in genes such as LRP1B, CSMD3, and ZFHX3 in primary CRC tumors. In contrast, the brain metastases exhibited a distinct molecular trajectory characterized by early mutations in KMT2D, ZEB1, and TCF7L2, suggesting a unique molecular evolution pathway for CRCBM. Conclusions: This study offers significant insights into the clonal evolution of primary colorectal cancers and their brain metastases. The findings underscore the complexity of tumor evolution, with early mutations initiating tumorigenesis and subsequent subclonal mutations contributing to metastasis and treatment resistance. These results emphasize the importance of clonal diversity in the progression of CRC and suggest that targeting specific clones may present novel therapeutic strategies for CRCBM. Further investigation into clonal evolution and its functional implications in CRCBM could enhance our understanding of metastasis and improve treatment outcomes.

Validation of a 10-item frailty index based on a comprehensive geriatric assessment (FI-CGA-10) for predicting survival in older adults with cancer.

1656 Background: The FI-CGA-10 is a content- and construct-validated measure for quantifying frailty from a CGA [Oncologist, 26, e1751 (2021)]. For criterion validation, we evaluated the predictive validity of the FI-CGA-10 for survival in older adults with cancer. Methods: This prospective cohort study included 1,630 consecutive older adults with cancer who underwent CGA prior to their cancer treatment decision at a geriatric oncology service in Japan from September 2018 through September 2024. Fitness and frailty were evaluated using the FI-CGA-10, which comprises ten domains: cognition, mood, communication, mobility, balance, nutrition, basic and instrumental activities of daily living (ADLs/IADLs), social support, and comorbidity. Deficits in each domain were scored 0 (no problem), 0.5 (minor problem), or 1.0 (major problem). FI-CGA-10 scores were calculated as the sum of domain scores divided by 10 and categorized as fit (<0.20), pre-frail (0.20–0.35), and frail (>0.35). The primary outcome was overall survival (OS) at 1 year from the date of CGA. Two Cox proportional hazards models were fitted to predict 1-year overall survival (OS): a base model (age, sex, cancer type, and stage) and the base model plus the FI-CGA-10 as a categorical variable. Model performance was compared using the likelihood ratio test and Harrell’s c-index. We also examined associations between each FI-CGA-10 domain and 1-year OS, adjusting for the same covariates. Results: Median age was 80 years (range, 51–99); 61% were male, 46% had gastrointestinal cancer, and 41% had distant-stage disease. Overall, 22% were fit, 38% pre-frail, and 40% frail; 29% died within one year. By the likelihood ratio test, adding the FI-CGA-10 significantly improved model fit compared with the base model (LR χ² = 140.7; P < .001). Adding the FI-CGA-10 also significantly improved discrimination (c-index 0.70 vs 0.76; P < .001). Kaplan–Meier curves for the three groups showed clear separation, demonstrating the FI-CGA-10’s discriminatory ability (log-rank P < .001). After adjustment for covariates, pre-frail (HR 2.26; 95% CI 1.61–3.19) and frail (HR 5.06; 95% CI 3.64–7.04) patients had higher hazards of death versus fit patients. Estimated 1-year OS probabilities were 90% (fit), 78% (pre-frail), and 58% (frail). In multivariable analyses adjusted for age, sex, cancer type, and stage, major problems in all domains—and minor problems in cognition, mobility, balance, nutrition, IADL, and comorbidity—were significantly associated with worse survival compared with no problems. Conclusions: These findings support the predictive validity of the FI-CGA-10 for 1-year OS in older adults with cancer. Together with prior work demonstrating its content and construct validity, the FI-CGA-10 satisfies core components of measurement validation that can help clinicians care for older cancer patients.

OnkoBank: A clinically annotated cancer biobank supporting translational and precision oncology research.

e15190 Background: High-quality, clinically annotated biobanks are essential infrastructure for translational cancer research and biomarker discovery. OnkoBank was established as a centralized, multi-institutional initiative to systematically collect biological specimens from patients with malignant disease and non-malignant controls, together with standardized clinical and pathological data. We present a descriptive analysis of the current OnkoBank cohort, focusing on tumor anatomical distribution and histopathological characteristics. Methods: This analysis included samples from 2,268 patients enrolled in OnkoBank. Collected biospecimens comprised fresh-frozen tumor tissue, fresh-frozen non-malignant control tissue, and blood-derived samples. Histopathological diagnoses were established for each sample by board-certified pathologists. Clinical and pathological variables were recorded in pseudonymized form using a REDCap-based data management system. Descriptive statistics were used to summarize tumor localization, histological categories, disease stage, and sample availability. Results: Among the 2,268 registered patients, tumors most frequently originated from the central nervous system (25.7%), melanoma or other skin tumors (18.8%), kidney (8.8%), colorectum (7.8%), and uterine corpus (7.6%), followed by ovarian (5.9%), pancreatic (3.0%), prostate (2.5%), bladder (2.2%), and breast cancers (1.8%). Less frequent tumor types included cervical (1.3%), liver (1.1%), vulvar (1.0%), and other localizations (9.1%), demonstrating broad representation across malignant diseases. Pathological tumor stage is reported in 45.6% of cases, with pT1–pT3 disease accounting for most staged tumors (pT1 14.8%, pT2 9.9%, pT3 15.9%), while pT4 disease is less common (3.1%). Distant metastatic disease is infrequent, with metastases present in 1.4% of patients, absent in 67.1%, and identified synchronously with the primary tumor in 24.7% of cases with available data. Paired tumor–normal specimens are available in 43.6% of cases, enabling matched molecular analyses. Conclusions: OnkoBank represents a large, uniformly annotated cancer biobank with comprehensive histopathological coverage and broad tumor-type representation. The availability of high-quality biospecimens, paired tumor–normal samples, and standardized clinical data provides a robust platform for large-scale translational research, biomarker discovery, and precision oncology studies.

Performance of PREMMplus in a racially and ethnically heterogenous population undergoing germline multigene panel testing.

e22640 Background: PREMMplus is a risk prediction model that estimates the probability of carrying pathogenic germline variants (PGVs) across multiple hereditary cancer susceptibility genes. Prior development and validation cohorts demonstrated high performance but were composed predominantly of non-Hispanic White individuals. PREMMplus performance in more heterogeneous clinical populations has not been well characterized. Methods: We evaluated PREMMplus performance in the Hereditary Cancer Panel (HCP) cohort, a large prospective clinical population that met clinical testing criteria and undewent germline multigene panel testing. Participants were recruited from three clinical sites, including a large safety-net hospital serving the greater Los Angeles area, resulting in substantial racial, ethnic, and socioeconomic heterogeneity. The parent study was designed to evaluate the safety, diagnostic yield, and clinical utility of multiplex genetic testing with standardized genetic counseling.. PREMMplus scores were assessed at the ≥2.5% cutoff. Performance was evaluated for PGVs in 11 Category A (APC, BRCA1/2, CDH1, EPCAM, MLH1, MSH2, MSH6, biallelic MUTYH, PMS2, TP53) and 8 Category B genes (ATM, BRIP 1, CDKN2A, CHEK2, PALB2, PTEN, RAD51C, RAD51D), using sensitivity (SE), specificity (SP), positive predictive value (PPV), negative predictive value (NPV), number needed to test (NNT), and area under the receiver operating characteristic curve (AUC). Subgroup analyses examined performance by race, ethnicity, and personal/family cancer history. Results: Among 1,896 participants, 41.5% self-identified as Hispanic and 11.9% as Asian. Overall, performance results were comparable to prior validation studies (Table). Compared with non-Hispanic individuals, Hispanic participants demonstrated lower specificity and discrimination (AUC 0.58 vs 0.68 for Category A genes). Sensitivity and PPV remained comparable across racial and ethnic groups. Performance differences were most pronounced among participants reporting a personal history of cancer but no family history. Conclusions: In a large, multi-site, heterogenous clinical cohort, PREMMplus retained high sensitivity for PGV detection but demonstrated reduced specificity and discrimination in Hispanic and other non-White populations. These findings highlight the importance of accurate family history ascertainment and the impact of under-reporting on the interpretation of risk prediction models in real-world clinical settings. PREMMplus performance (≥2.5% cutoff) by gene category and ethnicity. Gene category Cohort SE % SP % PPV % NPV % NNT AUC A Entire 88.3 19.6 6.9 96.1 14.5 0.64 Non-Hispanic 91.4 24.0 6.2 98.1 16.1 0.68 Hispanic 85.5 13.1 7.8 91.3 12.9 0.58 A + B Entire 86.6 19.7 9.7 93.6 10.3 0.59 Non-Hispanic 86.0 24.0 8.7 95.3 11.5 0.60 Hispanic 87.2 13.3 11.0 89.4 9.1 0.56

The importance of family history in indicating HBOC genetic testing in a resource-limited setting.

e12582 Background: Genetic screening for hereditary breast and ovarian cancer (HBOC) is guided by clinical criteria such as early-onset disease and family history. In resource-limited settings with restricted access to next-generation sequencing (NGS), optimizing patient selection is essential to maximize diagnostic yield. This study evaluates the predictive value of these criteria by comparing the distribution of pathogenic/likely pathogenic variants and variants of uncertain significance (VUS) between early-onset patients without family history and those with a documented family history of breast and/or ovarian cancer. Methods: We retrospectively analyzed 62 patients referred for suspected HBOC who met at least one inclusion criterion: breast or ovarian cancer diagnosed before age 40 and/or a positive family history. Genetic testing was performed using an extended 28-gene HBOC multigene panel. Variants were classified according to ACMG/AMP guidelines. Results: Among the 41 patients in the “young without family history” group, pathogenic or likely pathogenic variants were identified in 4 individuals (9.75%), involving BRCA1 (n = 2), ATM (n = 1), and MUTYH (n = 1). Variants of uncertain significance were detected in 14 patients (34.1%), affecting ATM , BRCA1 , RAD50 , MRE11A , NBN , CDKN2A , MLH1 , MUTYH , and CDH1 .In contrast, among the 21 patients with a documented family history of breast and/or ovarian cancer, pathogenic or likely pathogenic variants were identified in 15 patients (71.42%), predominantly involving high-penetrance genes, namely BRCA1 (n = 7) and BRCA2 (n = 6), with additional variants in ATM (n = 1) and MUTYH (n = 1). VUS were identified in 6 patients (28.5%), affecting ATM , NBN , CHEK2 , APC , and NF1 . Overall, VUS were relatively more frequent in the early-onset group, whereas patients with family history showed a markedly higher yield of clinically actionable pathogenic variants. Conclusions: Family history is a substantially stronger predictor of pathogenic variants than young age at diagnosis alone in HBOC genetic screening. In resource-limited settings, prioritizing testing based on family history may optimize the use of NGS and improve identification of high-risk individuals. The higher burden of VUS observed in early-onset patients without family history should be viewed not as a limitation of testing, but as a key challenge in understanding cancer susceptibility in young individuals lacking pathogenic variants in high-penetrance genes, raising questions about the role of moderate-penetrance genes, polygenic risk, and regulatory or non-genetic factors.

Baseline characteristics of NSCLC first-line immunotherapy super-responders in three European lung cancer centers: An ENDEAVOUR-IMIGO project.

8578 Background: Immune-checkpoint inhibitors (ICI) are part of first-line treatment concepts for most patients with metastatic non-small cell lung cancer (NSCLC) without targetable driver mutations. Still, long-term response is rare and data on predictive biomarkers limited. Methods: As part of the Daiichi Sankyo ENDEAVOUR programme supporting young LC researchers, the ENDEAVOUR-IMIGO working group aims to explore novel predictive biomarkers for ICI response with a focus on imaging data. We present the baseline characteristics of first-line ICI “super-responders” with PFS ≥24 months, derived from three independent cohorts from Austria, Italy and the Netherlands. Also, an exploratory single-center analysis compared ICI super- to non-responders (PFS<6M) matched for age, sex, histology, PD-L1 positivity, stage and ECOG performance status. Results: The Linz (Austria) cohort reported 50 (12.4%) super-responders out of a total of 402 patients, the Milan cohort (Italy) 45 out of 283 (15.9%), and the Groningen cohort (Netherlands) 46 (10.7%) out of 428. As shown in table 1, all cohorts had a high prevalence of adenocarcinomas, with KRAS mutations present in nearly half of patients and a mean PD-L1 expression of approximately 50%. In the Austrian cohort, matched baseline data showed a higher frequency of systemic steroids among non-responders, and lower absolute neutrophil count as well as neutrophil-to -lymphocyte ratio in super-responders. Conclusions: Baseline characteristics of ICI super-responders were similar in three European cohorts, with a majority of high-PD-L1 expressing adenocarcinomas. Exploratory matched analyses from one cohort showed implications of baseline systemic steroid treatment as well as of differential blood cell count. Linz Milan Groningen PFS < 6M n = 50(matched) PFS ≥ 2Y n = 50 p value < 6M vs. ≥ 2Y PFS ≥ 2Y n = 45 PFS ≥ 2Y n = 46 Male sex, n (%) 24 (48) 25 (50) 0.841 34 (75.5) 22 (48) Age <70, n (%) 14 (28) 15 (30) 0.826 15 (33.3) 34 (74) ECOG, n (%) 0 25 (50) 24 (48) 0.768 21 (47) 18 (39) 1 16 (32) 19 (38) 23 (51) 23 (50) 2+ 9 (18) 7 (14) 1 (2) 5 (11) Histological subtype, n (%) Adenocarcinoma 36 (72) 36 (72) 0.341 30 (67) 33 (72) Squamous cell carcinoma 12 (24) 14 (28) 6 (13) 7 (15) Other 2 (4) 0 (0) 9 (20) 6 (13) PD-L1 expression %, mean (SD) 41.9 (36.8) 48.9 (38.8) 0.403 52.3 (35.4) 54.0 (39.0) Brain metastases, n(%) 16 (32) 13 (26) 0.509 14 (31) 7 (15) KRAS mutation, n (%) 12 (35) 18 (50) 0.214 19 (42) 23 (50) Treatment, n (%) Chemo-ICI combination 29 (58) 38 (76) 0.056 18 (40) 19 (41) ICI-monotherapy 21 (42) 12 (24) 27 (60) 27 (59) Systemic steroid treatment, n(%) 17 (34) 6 (12) 0.009 12 (27) 8 (17) Absolute neutrophil count (G/L), mean (SD) 11.5 (8.7) 7.7 (2.9) 0.015 6.8 (3.2) 7.3 (2.6) Absolute lymphocyte count (G/L), mean (SD) 1.7 (3.7) 1.4 (0.7) 0.071 1.9 (0.7) 1.6 (0.7) Neutrophil-to-lymphocyte ratio, mean (SD) 13.4 (11.0) 7.2 (5.2) 0.006 4.4 (3.6) 5.9 (4.0)

Mitochondrial DNA (mtDNA) expression as used to define metabolic and immune states in colorectal cancer (CRC).

2647 Background: CRC exhibits metabolic reprogramming driven by the Warburg effect; however, active mitochondria and oxidative phosphorylation (OXPHOS) often remain crucial for tumor growth. mtDNA encodes critical OXPHOS components and influences the balance between OXPHOS and glycolysis, with effects on tumor microenvironment and response to immune checkpoint inhibitors (ICIs). We evaluated whether mtDNA gene expression predicts metabolic phenotype, immune contexture and benefit from ICIs. Methods: 30,887 CRC cases with DNA/RNA sequencing were analyzed from Caris Life Sciences. Expression of mtDNA-encoded OXPHOS genes ( MT-ND1–6, MT-ND4L, MT-CO1–3, MT-ATP6, MT-CYB ) was summarized as a composite Z-score due to correlation (r>0.9). Tumors were stratified into quartiles (n=7,722 each), mtDNA-high (MT-H, top quartile) and mtDNA-low (MT-L, bottom quartile) cohorts. Overall Survival (OS) was calculated in months (m) from first treatment to last contact. Hazard ratios (HRs) were calculated using Cox proportional hazards models and p-values by log-rank tests. Gene set enrichment analysis (GSEA) was performed to evaluate pathway differences. Results: MT-H tumors were enriched for Consensus Molecular Subtype (CMS) 2 compared to MT-L (44.2% vs 21.9%) and CMS3 (24.5% vs 8.7%), whereas CMS4 was markedly enriched in MT-L tumors (MT-H 16.5% vs MT-L 52.2%); all p<0.001. GSEA showed a trend toward higher OXPHOS activity in MT-H tumors (NES 1.19, FDR q=0.472), while glycolysis was significantly downregulated (NES -2.43, FDR q=0.002), with low immune/inflammatory signaling (interferon signaling and inflammatory response, among others) and reduced immune cell infiltration. These associations persisted in microsatellite stable (MSS) CRC, including CMS (CMS2/CMS3 46.3%/24.9% in MT-H vs CMS4 55.6% in MT-L) and immune signatures (all q<0.05). MT-L was prognostic for improved OS vs MT-H (median OS [mOS] 30.1 vs 27.2 m; HR 0.87, 95% CI 0.84-0.91, p<0.001). In ICI-treated patients (pts), MT-L showed amplified effect (mOS 24.3 vs 13.5 m; HR 0.69, 95% CI 0.58-0.83, p<0.001), including in CMS1 (mOS 39.3 vs 26.0 m; HR 0.71, 95% CI 0.53-0.96, p=0.024), CMS4 (mOS 19.3 vs 10.7 m; HR 0.61, 95% CI 0.40-0.94, p=0.024), MSS (mOS 13.8 vs 9.02 m; HR 0.67, 95% CI 0.54-0.84, p<0.001) and in pts with liver metastases (mOS 13.4 vs 7.93 m; HR 0.65, 95% CI 0.44-0.97, p=0.034). Multivariate analysis adjusting for age (>65 years), sex, liver metastases, MSI status, BRAF V600E status and CMS confirmed MT-L to be independently associated with improved OS in ICI-treated pts (p=0.01). No survival association was observed in pts treated with other therapies. Conclusions: mtDNA-encoded OXPHOS expression defines biologically distinct CRC subsets with distinct metabolic states and immune infiltration, with MT-L linked to improved OS and enhanced benefit from immunotherapy, including in MSS and liver metastases where ICI sensitivity is limited.

Surgical resection versus stereotactic body radiation therapy in stage IV non–small cell lung cancer: A large, real-world, propensity-matched survival study.

8075 Background: The role of definitive local thoracic therapy in metastatic non–small cell lung cancer (NSCLC) remains controversial. While stereotactic body radiation therapy (SBRT) is increasingly utilized, comparative survival outcomes with surgical resection in real-world populations remain limited. We evaluated survival and pulmonary outcomes associated with surgery versus SBRT in patients with stage IV NSCLC. Methods: We conducted a retrospective cohort study using the TriNetX Research Network including adults (≥18 years) with stage IV NSCLC treated between January 2013 and December 2023 who underwent definitive local thoracic therapy with either surgical resection or SBRT. Patients with small cell lung cancer, prior lung cancer diagnoses, prior thoracic radiation, or incomplete survival data were excluded. Propensity score matching (1:1) was performed to balance demographics, comorbidities, smoking history, and exposure to systemic therapy including chemotherapy and immunotherapy within 6 months of index treatment. Primary outcomes included landmark 1-year all-cause mortality and overall survival. Secondary outcomes included pneumonia and respiratory failure. Kaplan–Meier analysis and Cox proportional hazards regression were performed. Results: After matching, 1,934 patients were included (967 per cohort). At 1 year, surgical resection was associated with significantly lower mortality compared with SBRT (17.5% vs 31.4%; RR 0.56, 95% CI 0.47–0.66; p < 0.001), corresponding to an absolute risk reduction of 13.9% and an estimated number needed to treat of approximately 7. Kaplan–Meier analysis demonstrated superior survival in the surgical cohort (log-rank p < 0.001). Surgical resection remained associated with significantly improved long-term survival, with a 64% lower hazard of death (HR 0.36, 95% CI 0.30–0.44; p < 0.001) and prolonged median survival (3,942 vs 1,182 days). Pulmonary outcomes favored surgery. Pneumonia incidence was significantly lower in the surgical cohort at both 1 year (12.9% vs 29.3%; RR 0.44; p < 0.001) and 3 years (5.9% vs 14.2%; RR 0.42; p < 0.001). Rates of respiratory failure did not differ significantly between treatment groups. Conclusions: In this large real-world propensity-matched study, surgical resection was associated with significantly improved survival compared with SBRT in carefully selected patients with stage IV NSCLC. Surgery was additionally associated with lower long-term pulmonary infectious complications without increased respiratory failure risk. These findings suggest a potential role for aggressive local therapy in selected metastatic patients and support prospective trials to refine patient selection strategies.

AgrOmicSo: A client-server interface for accessible large-scale analysis of next-generation sequencing data

The analysis of large-scale next-generation sequencing (NGS) data requires substantial computational power, often necessitating the use of high-performance computing (HPC) environments. However, the command-line interfaces for these resources create a significant barrier for many researchers. To bridge this gap, we developed AgrOmicSo (Agri-bio Omics Solution), a software solution designed as a user-friendly interface to a powerful server-side analysis engine. AgrOmicSo’s client-server architecture allows researchers to manage and execute complex, large-scale NGS data analysis pipelines on a remote server directly from an intuitive graphical user interface on their local computer. The software integrates a comprehensive suite of bioinformatics tools for quality control, read mapping, variant calling, and annotation. Notably, it supports three distinct variant calling algorithms—GATK, DeepVariant, and VarScan—offering users flexibility for their specific research needs. AgrOmicSo provides both a “One-Step” mode for rapid, automated batch processing and a “Step-by-Step” mode for detailed, customized analyses. This paper describes the architecture, implementation, and utility of AgrOmicSo as an interface for large-scale genomic analysis, highlighting its potential to advance research by making powerful computational resources more accessible, efficient, and reproducible for a broader scientific community. The client and server program of AgrOmicSo are freely available at https://agromicso.com .

Green biosynthesis of silver nanoparticles from Citrus aurantium leaves plant as corrosion inhibitor for carbon steel in sulfuric acid

Interface‐Triggered Bulk Electrical Coupling in SnO ₂ via Mesoscopic Chemical Reconfiguration for Scalable, High‐Efficiency Perovskite Photovoltaics

ABSTRACT The efficiency and stability of n‐i‐p PSCs are often hindered by energy losses and defect‐induced interfacial failure, fundamentally stemming from mesoscopic surface chemical mismatch in SnO 2 nanocrystals. Herein, we propose a surface reconstruction strategy utilizing in situ generated [Al(OH) 4 ] − ions to rebuild the surface electrical double layer. This approach not only yields monodisperse, highly stable SnO 2 colloids but also synergistically passivates deep‐level defects arising from hydroxyl groups and oxygen vacancy clusters. The resulting amorphous Sn‐AlO x layer induces a surface electrostatic potential, significantly enhancing electrical coupling within the ETL bulk and establishing a robust carrier transport pathway. Consequently, we achieved champion efficiencies of 26.70% (certified 26.64%) for small‐area cells and 24.56% (certified 24.22%) for mini‐modules (21.50 cm 2 ). Notably, fully blade‐coated large‐area modules (65 cm 2 ) reached a high efficiency of 21.91%. By eliminating reactive sites at the buried interface, the unencapsulated devices retained 93% of their initial efficiency after 1100 h of MPPT (ISOS‐L‐2) and demonstrated a T 90 lifetime exceeding 700 h at 85°C (ISOS‐D‐2). This study provides novel insights for the large‐scale production application of SnO 2 sol‐gel in n‐i‐p perovskite solar modules.

Group-aware and position-interactive learning for point cloud robust segmentation in subterranean environments

Antibodies blocking PlGF or VEGF interactions with the NRP1 receptor mediate antiproliferative effects

A mediation analysis of neurocognitive deficits and patient-reported mood among patients with lung cancer with brain metastases.

12044 Background: Advances in lung cancer therapeutics have led to a growing population of patients with lung cancer who are facing the challenges of living with the sequela of brain metastases and the therapies used to treat them. Patients experience neurocognitive deficits, difficulties with functioning, and mood symptoms. However, the mechanisms by which neurocognitive deficits contribute to mood symptoms in this patient population remain to be fully elucidated. In this study, we aimed to evaluate patient-reported functioning as a mediator between neurocognitive deficits and mood symptoms. Methods: Using a prospectively collected single-site registry, we conducted an exploratory cross-sectional study among patients with lung cancer with brain metastases at the time of evaluation for stereotactic radiosurgery. The registry included data on patient demographic and clinical characteristics, neurocognitive deficits per clinician exam (visual impairment, cranial nerve deficit, sensory deficit, motor deficit, gait impairment/incoordination, decline in ability to concentrate, decline in memory), and patient-reported functioning (self-care, usual activities, mobility) and mood symptoms (anxiety/depression) from the European Quality of Life 5-Dimensions 3-Level Version. First, we selected functional domains significantly associated with mood symptoms in a logistic regression analysis to test as potential mediators. Then, we used a regression-based path analysis to test the selected functional domain as a mediator between neurocognitive deficits and mood symptoms. The mediation model adjusted for age, performance status, tobacco use, extracranial disease status, prior brain surgery, and prior brain radiation. Results: Of 463 patients, 33.9% demonstrated any neurocognitive deficit, 38.7% self-reported any functional difficulty (self-care, usual activities, or mobility), and 15.3% reported mood symptoms. Difficulty with usual activities was associated with increased mood symptoms (OR=1.65, 95% CI=1.36, 1.97), which we then tested as a mediator. Motor deficits (OR=1.66, 95% CI=1.30, 2.38) and gait impairment/incoordination (OR=1.50, 95% CI=1.19, 2.09) had positive indirect effects on mood symptoms (i.e. increased mood symptoms) through difficulties with usual activities. No other neurocognitive deficits showed significant effects on mood symptoms. Conclusions: Among patients with lung cancer with brain metastases, the associations of motor deficits and gait impairment/incoordination with worsened mood are mediated by difficulties with engaging in usual activities. These findings underscore the potential for tailored supportive care, utilizing strategies from cognitive rehabilitation, palliative care and psychology, to improve patients’ mood by helping them adapt to and cope with difficulties in engaging in their usual activities.

Extended follow-up (6 years) of the phase 2 LITESPARK-004 study of belzutifan in participants with von Hippel-Lindau disease–associated neoplasms.

4550 Background: Belzutifan, a potent and selective hypoxia-inducible factor 2α inhibitor, is indicated for treatment of patients with von Hippel-Lindau (VHL) disease–associated renal cell carcinoma (RCC), CNS hemangioblastomas (HB), or pancreatic neuroendocrine tumors (pNETs) not requiring immediate surgery based on prior results from the open-label phase 2 LITESPARK-004 study (NCT03401788). We report results from LITESPARK-004 after a minimum of 6 years of follow-up. Methods: Participants (pts) with germline VHL alterations, ≥1 measurable nonmetastatic RCC tumor, no tumor >3 cm that required immediate surgery, no metastatic disease, no prior anticancer systemic treatment, and an ECOG PS of 0 or 1 received oral belzutifan 120 mg once daily until disease progression, unacceptable toxicity, or pt withdrawal. The primary end point was objective response rate (ORR) in VHL disease–associated RCC per RECIST v1.1 by independent review committee (IRC). Secondary end points included safety, ORR in non-RCC neoplasms, duration of response (DOR), and progression-free survival (PFS) per RECIST v1.1 by IRC. Results: Overall, 61 pts received ≥1 dose of belzutifan. Median study follow-up was 73.8 months (range, 72.1-82.1). As of the data cutoff date (April 1, 2025), 34 pts (56%) remained on treatment. ORR was 70% for RCC, 52% for CNS HB, and 91% for pNETs. Additional efficacy results are shown in the Table. Among 14 pts (n = 18 affected eyes by ophthalmic evaluation), retinal HBs in 100% (95% CI, 81-100) of eyes showed improvement. Median DOR in pts with retinal HBs was not reached (NR; 95% CI, 8.5-NR). Within 5 years prior to initiating belzutifan, 46 of 61 pts (75%) had ≥1 VHL-related tumor reduction procedure (surgery or radiation therapy; 43 RCC, 39 CNS HB, 12 retinal HB, 3 pNET, 2 other). Since initiating belzutifan treatment, 22 of 61 pts (36%; 14 during treatment and 8 after discontinuing treatment) underwent 25 VHL-related tumor reduction procedures (16 RCC, 3 CNS HB, 5 retinal HB, 1 pNET). Grade 3 treatment-related adverse events (TRAEs) were reported in 12 pts (20%); the most common grade 3 TRAE was anemia (11%). No grade 4 or 5 TRAEs occurred. No additional pts discontinued belzutifan due to TRAEs since the previous analysis. Conclusions: After 6 years of follow-up, belzutifan continues to show durable antitumor activity and a manageable safety profile in pts with VHL-associated RCC, CNS HB, and pNETs who do not require immediate surgery. These results continue to support the use of belzutifan in this patient population. Clinical trial information: NCT03401788 . RCCn = 61 CNS HBn = 50 pNETsn = 22 ORR (95% CI), % 70 (57-81) 52 (37-66) 91 (71-99) Best overall response, n 8 CR35 PR 17 SD0 PD1 NE 7 CR19 PR20 SD3 PD1 NE 14 CR6 PR2 SD0 PD0 NE DOR, median (range), months NR (5.8+ to 69.1+) NR (0.0+ to 74.3+) NR (11.0+ to 71.8+) 48-month DOR rate 78% 84% 90% PFS, median (95% CI), months NR (71.5-NR) NR (NR-NR) NR (NR-NR) 48-month PFS rate 79% 80% 90%

Fruquintinib alternating with bevacizumab plus capecitabine as maintenance therapy after first-line treatment in metastatic colorectal cancer (mCRC): A multicenter, open-label, phase II study.

e15539 Background: Maintenance therapy with bevacizumab (Bev) plus capecitabine (Cap) is recommended to unresectable mCRC patients (pts). Fruquintinib (Fru) is a highly selective TKI that inhibits vascular endothelial growth factor receptor (VEGFR)-1,2,3. This study is to compare the therapeutic potential of alternating treatment with fruquintinib and bevacizumab plus capecitabine as maintenance therapy for mCRC (NCT05659290). Here, we report the updated results. Methods: Eligible mCRC pts aged 18-75 years with stable disease or better after induction treatment with chemotherapy in combination with Bev, ECOG PS 0-2, adequate bone marrow, liver, and renal function were enrolled. Sixty-one patients were included (20 in Phase Ⅱa, 41 in Phase Ⅱb). In Phase IIa, pts were orally administered with Fru (3~5 mg, qd, d1-14, q3w) alternating with Bev (7.5 mg/kg, iv.gtt, d1, q3w) plus Cap (850 mg/m 2 , orally, twice daily, d1-14, q3w). In Phase Ⅱb, pts were randomly assigned (1:1) to either maintenance treatment with Fru (3 mg, qd, d1-14, q3w) alternating with Bev plus Cap or Bev plus Cap. The primary endpoint was progression-free survival (PFS). Secondary endpoints included objective response rate (ORR), disease control rate (DCR) and safety. Results: As of December 2025, in Phase Ⅱa (n = 20), the median PFS was 8.25 months ( 95% CI: 5.78-NA). PFS tended to be longer in patients with liver metastases compared to those without liver metastases (8.28 vs 4.80 months; p = 0.08), and in patients with left-sided tumors versus right-sided tumors (8.28 vs 4.48 months; p = 0.70).In Phase Ⅱb, patients were randomized to receive Fruquintinib + Bevacizumab + Capecitabine (Fru+Bev+Cap, n = 21) or Bevacizumab + Capecitabine (Bev+Cap, n = 20). The median ages were 60 (range 25–73) and 58 years (range 46–75), respectively. Baseline characteristics included left-sided primary tumors (57.1% vs 85.0%), liver metastases (71.4% vs 30.0%), and ≥2 metastatic sites (66.7% vs 60.0%). Thirty-five patients had undergone at least one tumor assessment. Median PFS was not mature in the Fru+Bev+Cap arm and was 8.18 months in the Bev+Cap arm (p = 0.30). The ORR was 6.25% versus 5.26%, and the DCR was 81.3% versus 73.7%, respectively. The most common treatment-emergent adverse events (TEAEs) in the Fru+Bev+Cap arm were hypertension (57.1%), proteinuria (42.9%), and decreased platelet count (23.8%). In the Bev+Cap arm, the most common TEAEs were hypertension (50.0%), hyperuricemia (35.0%), and decreased platelet count (20.0%). Conclusions: Fruquintinib alternating with bevacizumab plus capecitabine as maintenance therapy after first-line treatment in mCRC demonstrated promising antitumor activity and manageable toxicity. The ongoing Phase IIb trial warrants further investigation in this patient population. Clinical trial information: NCT05659290 .

Mortality burden of hepatocellular carcinoma and diabetes in the United States, 1999–2023.

e16355 Background: Hepatocellular carcinoma (HCC) and diabetes mellitus are major contributors to morbidity and mortality when coexisting. We evaluated national trends in HCC- and diabetes-related mortality from 1999 to 2023 across demographic, geographic, and urbanization strata in the United States. Methods: Mortality data for individuals aged ≥45 years were obtained from the CDC WONDER database (1999–2023). Deaths with HCC and diabetes listed as underlying or contributing causes were identified. Age-adjusted mortality rates (AAMRs) per 100,000 population were calculated using the 2000 U.S. standard population. Joinpoint regression was used to estimate annual percent change (APC) and average annual percent change (AAPC) with 95% confidence intervals, stratified by sex, race/ethnicity, census region, urbanization status, and age group. Results: From 1999 to 2023, 61,358,342 deaths occurred among individuals aged ≥45 years (30,217,781 males [49.24%], 31,140,551 females [50.75%]). HCC- and diabetes-related mortality increased significantly overall (AAPC 3.90%, 95% CI 3.24–4.55; p < 0.000001). Mortality rates were substantially higher in males (AAPC 3.78%, 95% CI 3.20–4.37; p < 0.000001) than females (AAPC 3.53%, 95% CI 2.37–4.69; p < 0.000001). Significant upward trends were observed across all census regions, most prominently in the South and West. Metropolitan areas demonstrated approximately threefold higher mortality rates than non-metropolitan areas from 1999–2020. Increasing trends were observed across racial and ethnic groups, including Hispanic and non-Hispanic Black populations, with the largest absolute increases among non-Hispanic White individuals. Conclusions: Despite overall declines in cancer mortality, age-adjusted mortality related to coexisting HCC and diabetes has risen substantially in the United States over the past two decades, particularly among adults aged ≥45 years, males, and selected racial/ethnic groups. These findings highlight the need for integrated metabolic and oncologic risk management strategies and targeted public health interventions.

Assessing the true burden of immune-related colitis in lung cancer through large language model–enabled clinical phenotyping.

1615 Background: Immune checkpoint inhibitors (ICIs) have revolutionized lung cancer treatment but cause adverse events, including immune-related colitis (ir-colitis). Accurate identification of ir-colitis is critical for improving patient outcomes and clinical research. However, current methods relying on ICD codes have substantial limitations: non-specific categories that fail to capture the cases or distinguish cases from infectious or ischemic etiologies, as well as incomplete coding that leads to missed cases. Large language models (LLMs), with advanced natural language understanding capabilities, offer a scalable approach for automated phenotyping using unstructured clinical notes. We aimed to develop and evaluate an LLM-based approach for identifying ir-colitis in patients with lung cancer receiving ICIs and compare it with ICD code based method. Methods: We identified 5,278 lung cancer patients with ICI administration at Mayo Clinic (2012-2025). To investigate ir-colitis occurrence within two years after ICI initiation, patients were stratified into three ICD-based cohorts: (1) IR-Colitis, with clinician-assigned ir-colitis-specific diagnosis codes; (2) Non-IR Colitis, with colitis diagnosis codes excluding ir-colitis; and (3) No-Colitis, absent of any colitis-related codes. We used three open-source LLMs to process clinical notes : Llama4, MedGemma, and Qwen. Prompts were designed to extract both gold standard evidence (explicit ir-colitis terminology, such as "immune-mediated colitis" and "checkpoint inhibitor colitis") and supportive evidence (such as pathology and lab results). Results: Llama4 demonstrated superior accuracy among three LLMs in manual evaluation and was thus selected for all cohort analyses. Among ICD-based IR-Colitis patients (n = 339), 82.0% had confirmatory evidence while 18% lacked evidence, suggesting potential false-positive coding. In the Non-IR Colitis cohort (n = 297), LLM-based extraction identified 43.8% patients in fact had ir-colitis, demonstrating substantial misclassification. In the No-colitis cohort (n = 4,642, random sample of 996 for analysis), the LLM identified 10.7% with documented ir-colitis, revealing substantial missing by administrative coding. Conclusions: LLM-based phenotyping demonstrates superior sensitivity for identifying ir-colitis compared to ICD codes alone, successfully uncovering cases missed by administrative coding. This scalable approach enables more comprehensive case identification for immunotherapy safety research and clinical surveillance. ICD-Based Cohorts LLM-Identified ir-Colitis Gold Standard Evidence Supportive Evidence Any Evidence ir-Colitis (n = 339) 277/339 (81.7%) 275/339 (81.1%) 278/339 (82.0%) Non-ir-Colitis (n = 297) 129/297 (43.4%) 126/297 (42.4%) 130/297 (43.8%) No Colitis (n = 996) 106/996 (10.6%) 104/996 (10.4%) 107/996 (10.7%)

Association of neighborhood-level vulnerability with health behavior patterns in survivors of childhood cancer: A report from the St. Jude Lifetime Cohort (SJLIFE).

10010 Background: Childhood cancer survivors face elevated risks for chronic health conditions and premature mortality. Engaging in healthy behaviors may mitigate these risks, yet neighborhood-level vulnerability may constrain survivors’ ability to adopt or sustain them. Methods: Among survivors (n=3,303) and community controls (n=652) in SJLIFE, neighborhood-level vulnerability was measured using census tract level overall and domain-specific (i.e. socioeconomic status, household composition, minority status, housing type and transportation) Social Vulnerability Index (SVI) scores, with higher values indicating higher vulnerability, and the U.S. Department of Agriculture’s persistent poverty measure. Health behaviors, including sedentary time/physical activity, smoking, alcohol, and illicit drug/marijuana use, were categorized as healthy, moderately unhealthy (1-2 unhealthy behaviors), and unhealthy (≥3 unhealthy behaviors). Multinomial logistic regression evaluated associations between neighborhood-level vulnerability at enrollment and health behavior patterns at most recent evaluation among survivors, adjusted for sociodemographic and cancer treatment factors as determined by backward selection. Interaction models compared associations between survivors and controls. Results: Survivors were 53.1% male, 82.5% non-Hispanic White, with a median age of 31.0 years (interquartile range: 23.5 – 38.5) at health behavior assessment. The most common childhood cancer diagnosis was acute lymphoblastic leukemia (28.8%). Survivors in the highest vs. lowest SVI tertile had higher odds of reporting moderately unhealthy (Odds Ratio [OR]: 1.46, 95% Confidence Interval [95% CI]: 1.08 – 1.98) or unhealthy (OR: 1.70, 95% CI: 1.13 – 2.56) behavior patterns, compared to a healthy behavior pattern. Socioeconomic and housing type SVI domains drove these associations, with survivors in the highest tertiles more likely than those in the lowest to have an unhealthy rather than healthy behavior pattern (Socioeconomic OR: 1.55, 95% CI: 1.02 – 2.35; Housing type OR: 1.53, 95% CI: 1.03 – 2.27). Persistent poverty was associated with greater odds of a moderately unhealthy behavior pattern (OR: 1.70, 95% CI: 1.09 – 2.66) in survivors. Housing type vulnerability was associated with disproportionately higher odds of an unhealthy behavior pattern among survivors compared to controls (ratio of ORs: 2.80, 95% CI: 1.19 – 6.60). Conclusions: Neighborhood-level vulnerability was associated with higher likelihood of unhealthy behavior patterns in childhood cancer survivors with housing type vulnerability having a greater impact on health behaviors in survivors compared to controls. Effective interventions to improve survivors’ health behaviors should incorporate neighborhood-level factors.

Randomized phase II trial of trifluridine/tipiracil (FTD/TPI) plus ramucirumab (RAM) versus FTD/TPI for previously treated patients with advanced gastric or gastroesophageal junction adenocarcinoma: Final analysis of RETRIEVE study (WJOG 15822G).

4038 Background: Promising activity of trifluridine/tipiracil (FTD/TPI) plus ramucirumab (RAM) has been reported in single-arm phase II trials as third- or later-line treatment for advanced gastric cancer (GC) and gastroesophageal junction (GEJ) adenocarcinoma (Lancet Gastroenterol Hepatol. 2021; 6: 209-217). We previously reported the primary analysis of the RETRIEVE study at ASCO-GI 2026, which demonstrated no significant difference in progression-free survival (PFS; median 2.69 vs 2.07 months) between FTD/TPI plus RAM and FTD/TPI alone. Here, we present the final analysis, including overall survival (OS) as a key secondary endpoint, along with updated results of other endpoints. Methods: RETRIEVE study is a multicenter, prospective, open-label, randomized phase II trial comparing FTD/TPI plus RAM versus FTD/TPI alone in patients with unresectable or recurrent GC or GEJ adenocarcinoma who were refractory or intolerant to fluoropyrimidine, taxane, or irinotecan, and refractory to RAM. Treatment is repeated every 4 weeks until disease progression or unacceptable toxicity. Key eligibility criteria include age of ≥20 years; ECOG performance status of 0 or 1, and at least a measurable lesion per RECIST 1.1. Results: Between January 2023 and June 2024, 111 patients were randomly assigned to receive FTD/TPI plus RAM (n = 56) or FTD/TPI alone (n = 55). At a median follow-up of 18.4 months, median OS was 6.08 months in the combination arm and 7.36 months in the monotherapy arm (hazard ratio [HR], 1.233; 95% CI, 0.817–1.862; P = 0.318). The proportion of patients who received subsequent treatment was 64.3% in the combination arm and 56.4% in the monotherapy arm. In the updated analysis, median PFS was 2.69 months in the combination arm and 2.07 months in the monotherapy arm (HR, 1.010; 80% CI, 0.781–1.307; P = 0.959). Objective response rate (ORR; 5.4% vs 9.1%, P = 0.489) and disease control rate (DCR; 50.0% vs 47.3%, P = 0.850) were consistent with the primary analysis. No novel adverse events were identified from the primary analysis. Any grade of n (49.1% vs 25.5%) and diarrhea (29.1% vs 14.5%) were more frequently observed in the combination arm than in the monotherapy arm. Conclusions: The final analysis demonstrated that FTD/TPI plus RAM did not provide a significant OS benefit compared with FTD/TPI alone in third- or later-line treatment for advanced GC and GEJ adenocarcinoma. Updated results of PFS, tumor response, and safety profile were consistent with the primary analysis. Clinical trial information: jRCTs041220120.