Browse Articles

Discover research articles across all indexed journals

Development of a multidimensional diagnostic model integrating cfDNA methylation and clinical parameters for colorectal cancer and precancerous lesions using artificial intelligence.

Journal of Clinical Oncology Li-Yue Sun, Hao-Chuan Ma, Jiao-Jiao Yang et al. Jul 01, 2026 DOI: 10.1200/jco.2026.44.19_suppl.94

94 Background: Early detection of colorectal cancer (CRC) and precancerous lesions is critical for improving patient outcomes. This study aimed to develop a diagnostic model integrating cfDNA methylation markers with multidimensional clinical parameters using artificial intelligence (AI) algorithms. Methods: A total of 1,373 participants were enrolled, including 261 patients with CRC, 312 with precancerous lesions, 400 high-risk individuals, and 400 healthy controls. Clinical data were systematically collected and preprocessed to ensure uniformity. Both qualitative variables (e.g., sex, fecal immunochemical test results) and quantitative variables (e.g., age, tumor markers, cfDNA methylation indicators) were included. Fifteen AI algorithms were evaluated, including logistic regression, support vector machine, random forest, XGBoost, and AdaBoost. Model performance was assessed using area under the receiver operating characteristic curve (AUC). The optimal algorithm was selected based on AUC, sensitivity, and specificity. Feature importance was analyzed using XGBoost. Results: AdaBoost was identified as the optimal algorithm. Key features selected included CA19-9, smoking index, white blood cell count, fasting blood glucose, CEA, age, CA724, triglycerides, total cholesterol, and neutrophil to lymphocyte ratio. Two rapid and low-cost cfDNA detection methods (electrochemical cfDNA adsorption and colloidal gold based cfDNA absorbance) were integrated into the model. The final combined model demonstrated excellent diagnostic performance, with an AUC of 0.999 for CRC and 0.843 for precancerous lesions. Sensitivity and specificity for CRC were 98.1% and 99.8%, respectively, and for precancerous lesions were 89.1% and 62.6%, respectively. Conclusions: This study successfully developed a high-performance diagnostic model integrating cfDNA methylation markers with multidimensional clinical data using AI. The model shows exceptional accuracy in detecting both CRC and precancerous lesions, offering a promising tool for early screening and risk stratification. Incorporation of low-cost cfDNA assays enhances its potential for widespread clinical application, particularly in resource-limited settings.

Short-course radiotherapy and chemotherapy with or without PD-L1 blockade (envafolimab) for MSS locally advanced rectal cancer: Interim analysis of the randomized phase III PRECAM-R trial.

Journal of Clinical Oncology Yiming Lv, Fei Wang, Yanbin Shen et al. Jul 01, 2026 DOI: 10.1200/jco.2026.44.19_suppl.92

92 Background: Microsatellite stable (MSS) locally advanced rectal cancer (LARC) is refractory to immune checkpoint inhibitors. Our prior Phase II trial demonstrated remarkable efficacy, suggesting short-course radiotherapy (SCRT) plus chemotherapy acts as a potent immunogenic primer for PD-L1 blockade (via subcutaneous envafolimab). We conducted this multicenter, randomized Phase III trial (PRECAM-R) to validate if this patient-centric strategy, combining SCRT, chemotherapy, and immunotherapy, improves outcomes versus standard care. Methods: Patients with resectable, stage II/III (cT3-4a/N+) MSS LARC were randomized (1:1) to the Experimental Group (SCRT 5×5 Gy, then only 2 cycles CAPEOX plus subcutaneous envafolimab) or Control Group (SCRT then CAPEOX). Total mesorectal excision (TME) was performed following neoadjuvant therapy. The primary endpoint was pCR (ypT0N0). Secondary endpoints included tumor regression grade (TRG), Neoadjuvant Rectal (NAR) score, and safety. This pre-specified interim analysis was triggered after 58 patients completed surgery. Results: Fifty-eight eligible eligible patients were randomized (n=29 per arm). Baseline characteristics were well-balanced; 77.6% had Stage III disease. The Experimental Group achieved a pCR rate of 44.8% (13/29), significantly tripling that of the Control Group (13.8% [4/29]; P = 0.0195). Major pathological response (TRG 0-1) rates were 72.4% vs. 51.7% (P = 0.176). The Experimental Group showed superior distribution of tumor downstaging based on NAR scores ( P = 0.036). Notably, early distant metastasis occurred in 13.8% (4/29) of patients in the Control Group (liver/lymph nodes) compared to 0% in the Experimental Group. The addition of envafolimab was well-tolerated, with Grade 3-4 adverse events comparable to the control arm (3.4% vs. 0%, P > 0.99). Crucially, this strategy did not compromise surgical safety, showing no increase in operative complexity or postoperative complications (10.3% vs 13.8%, P > 0.99). Conclusions: This regimen, featuring a patient-centric short-course backbone, shows promise in overcoming resistance to PD-L1 blockade in MSS LARC. This strategy yields high pCR rates and tumor regression with high compliance and a favorable safety profile. The early signal for systemic control positions this convenient regimen as a compelling option for organ preservation. Clinical trial information: NCT05752136 . Interim efficacy and safety outcomes. Endpoint Experimental (n=29) Control (n=29) P Value Primary Endpoint pCR (ypT0N0), No. (%) 13 (44.8) 4 (13.8) 0.0195 Secondary Endpoints MPR (TRG 0-1), No. (%) 21 (72.4) 15 (51.7) 0.176 Median NAR Score (IQR) 8.43 (0.94-8.43) 8.43 (6.66-14.98) 0.036 Distant Metastasis, No. (%) 0 (0) 4 (14.8) 0.112 Grade 3-4 Adverse Events, No. (%) 1 (3.4) 0 (0) 0.99 Postoperative Complications, No. (%) 3 (10.3) 4 (13.8) 0.99

High-rate single-crystalline Li-rich layered oxide with diversified surface-phase cation ordering for Li-ion batteries

Nature Communications Jingxi Zhang, Shile Chen, Yuting Xia et al. Jul 01, 2026 DOI: 10.1038/s41467-026-75101-w

Enhancing Magneto-Optical Activity via Coordination Distortion in Chiral Er <sub>4</sub> M <sub>8</sub> Clusters

Journal of the American Chemical Society Jia-Nan Chen, Jiaye Chen, Ming-Qiang Qi et al. Jul 01, 2026 DOI: 10.1021/jacs.6c05602

Motivational and psychological variables related to unethical uses of artificial intelligence across multiple life domains (academics and online interactions)

PLoS ONE Kyle J. Messick, Ash Bass, Matthew Fulmer et al. Jul 01, 2026 DOI: 10.1371/journal.pone.0352419

Past studies about unethical uses of generative artificial intelligence (AI) have focused within a specific life domain (e.g., how AI is used for cheating in academics). An online survey was conducted to investigate how AI users utilized AI across multiple domains, how unethical behaviors in one domain related to behaviors in others, and which motivational and psychological variables corresponded with those uses. The goal was to help create a foundational understanding of how people use AI more broadly, including what the largest motivations for unethical use are and if unethical use had psychological consequences for the user. Creating this foundation can help develop successful interventions for unethical AI use. Findings supported that performing potentially unethical behaviors in one area of life (e.g., academics) were positively related to unethical behaviors in another area of life (e.g., social media interactions). Using AI unethically was positively related to knowing that the use was unethical. Unethical AI use was not related to intrinsic or extrinsic motivation, but was positively related to results pressure (prioritizing outcomes), external pressure (unethical use is normative and unlikely to be penalized), and time pressure (saving time by completing a task more quickly). Unethical AI use was also positively related to a desire for social media popularity and narcissism. Unethical uses of AI were not related to self-esteem, self-efficacy, or loneliness, but were positively related to satisfaction with life, making social comparisons, internalizing the perspective of others about their own body, prioritizing physical appearance over other attributes, and comparing one’s physical appearance with others. This study provided evidence that despite ethical misuse, generative AI is being used across multiple domains of life and is largely associated with positive consequences for the user. Unethical AI use was applied across multiple areas of life and was associated with variables related to elevating status, achieving good outcomes, behaving within broader cultural norms, and saving time. Future applications could explore if a more wholistic approach that targets motivations is more effective than a domain-specific approach when attempting to create interventions to curb unethical AI use.

Numerical investigation of the behaviour of shallow foundations under reverse fault rupture using concrete damage plasticity model

Scientific Reports Amin Monshizadeh Naeen, Ali Akhtarpour Jul 01, 2026 DOI: 10.1038/s41598-026-60152-2

Listening in on the human brain cells that produce speech

Nature Jul 01, 2026 DOI: 10.1038/d41586-026-02004-7

Physics‐Enhanced Deep Learning Optimized Semitransparent Organic Photovoltaics for Building‐Integrated Sustainable Energy

Advanced Materials Baozhong Deng, Xiaokai Zhang, Zhouyi Lu et al. Jul 01, 2026 DOI: 10.1002/adma.73762

ABSTRACT Global energy challenges establish building‐integrated photovoltaics as a pivotal decarbonization frontier, where semitransparent organic photovoltaics (ST‐OPVs) represent a promising technology for simultaneous power generation and daylight transmission. However, their widespread application is constrained by a fundamental efficiency and transparency trade‐off governed by complex photon management. Herein, we introduce a physics‐enhanced deep learning (PDL) framework that embeds optical physical priors into neural network, significantly reducing the reliance on extensive experimental datasets while enhancing predictive accuracy beyond conventional simulation and purely data driven methods. Building on a novel halogen‐additive engineering strategy, that enables opaque devices with a power conversion efficiency exceeding 20%, our PDL‐guided optimal optical design delivers corresponding ST‐OPVs with a record light utilization efficiency of 6.09%. When scaled to large‐area manufactured modules, multi‐scale building energy modeling demonstrates that the nationwide deployment of such ST‐OPVs could meet up to one‐fifth of China's total energy demand, highlighting their transformative potential in advancing sustainable energy systems and supporting global carbon neutrality goals.

A revised model for PHF20L1 Tudor function: DNA binding overrides methylation selectivity on nucleosomes

Journal of Biological Chemistry Xiaolei Huang, Qin Xiao, Xin Liu et al. Jul 01, 2026 DOI: 10.1016/j.jbc.2026.113181

Systemic immune-inflammation index as a prognostic biomarker across T-stage, biliary obstruction, and treatment setting in cholangiocarcinoma: A retrospective cohort study.

Journal of Clinical Oncology Naiyarat Prasongsook Jul 01, 2026 DOI: 10.1200/jco.2026.44.19_suppl.137

137 Background: Cholangiocarcinoma (CCA) carries a dismal prognosis, and low-cost biomarkers applicable across disease stages and treatment settings remain an unmet need. The systemic immune-inflammation index (SII = neutrophils × platelets ÷ lymphocytes), calculated from standard CBC, combines three values into a composite inflammatory measure. While SII has been studied in surgical iCCA, its relationship with T-stage invasion depth, biliary obstruction, and prognosis in non-surgical CCA—representing &gt;70% of patients—remains uncharacterized. Methods: Retrospective cohort of 93 CCA patients at Phramongkutklao Hospital, Bangkok (January 2020–July 2025); protocol approved by institutional Ethics Committee. Inclusion: histopathological or radiological CCA; SII, NLR, and CA 19-9 from peripheral blood within 30 days of diagnosis. Exclusion: concurrent malignancy, active cholangitis, immunosuppressive therapy. Markers analyzed as continuous variables (Mann-Whitney U; Kruskal-Wallis). Missing values excluded per variable. Potential confounders (subtype, stage, jaundice) noted descriptively. Three pre-specified analyses: (1) SII/NLR across T-substages (T1a–T4); (2) SII/NLR as outcome discriminators in non-surgical patients (n=68); (3) SII/NLR by jaundice (TB &gt;2 mg/dL) across CCA subtypes. No correction for multiple comparisons; findings are hypothesis-generating. Results: Median age 60 years (29–91); 59% male; iCCA 74%, pCCA 14%, eCCA 12%; Stage IV 45%; surgery 27% (n=25). NLR rose stepwise from T1a (2.6) to T3/T4 (5.6); SII from 910 to 1,634. In non-surgical patients, SII was 3.7× higher in deceased versus alive-with-disease patients (1,680 vs. 451); NLR separated by 1.8× (5.1 vs. 2.8). Jaundice (TB &gt;2 mg/dL) associated with markedly higher SII across all CCA subtypes (1,379 vs. 841); ALP quartile Q4 vs. Q1 showed 2.75× SII difference (see Table). Conclusions: SII tracks stepwise with T-stage depth, separates survival outcomes in the non-surgical majority more effectively than NLR, and is amplified by biliary obstruction across CCA subtypes. These data suggest SII captures both local tumor invasiveness and obstructive cholestatic inflammation, broadening its prognostic applicability beyond resected cohorts. Prospective validation incorporating multivariable survival analysis is warranted. SII and NLR by T-stage, non-surgical outcome, and biliary obstruction (n=93). Finding Group n NLR median SII median T-stage gradient T1a 18 2.6 910 T-stage gradient T2 54 3.9 1047 T-stage gradient T3/T4 12 5.6 1634 Non-surgical: outcome Died of disease 21 5.1 1680 Non-surgical: outcome Alive with disease 5 2.8 451 Biliary obstruction TB &gt;2 mg/dL 30 5.0 1379 Biliary obstruction TB ≤2 mg/dL 63 3.5 841 ALP burden (Q1 vs Q4) Low vs High ALP 23 vs 24 2.7 vs 4.1 460 vs 1266

Machine learning-based prediction of ICU mortality in patients with pancreatic cancer: A MIMIC-IV retrospective cohort study.

Journal of Clinical Oncology Shankar Biswas, Yashasvi Srivastava, Ayman Hamadttu Jul 01, 2026 DOI: 10.1200/jco.2026.44.19_suppl.142

142 Background: Pancreatic cancer patients requiring ICU admission face high short-term mortality, yet no disease-specific prediction tools exist. General severity scores like SOFA may not capture pancreatic cancer-specific risk factors including biliary sepsis and malnutrition. We aimed to develop and validate ML models to predict ICU mortality in pancreatic cancer patients using MIMIC-IV. Methods: This retrospective study used MIMIC-IV (v3.1, 2008-2022). Adults with pancreatic cancer (ICD-9: 157.x; ICD-10: C25.x) and ICU stays ≥6 hours were included; neuroendocrine tumors were excluded. Features included demographics, tumor characteristics, comorbidities, SOFA score, disease-specific complications, labs, and vitals from the first 24 hours. Primary outcome was ICU mortality. Four models (logistic regression, LASSO, random forest, XGBoost) were trained (2008-2016) and temporally validated (2017-2022). Performance was compared against SOFA alone. SHAP analysis identified key predictors. Results: 736 ICU stays from 591 patients were analyzed (training: 519; validation: 217). Mean age was 68.0±11.4 years; 57.5% male. Tumors were predominantly pancreatic head (63.0%) and 60.6% had metastatic disease. Biliary sepsis was the leading admission reason (38.3%). ICU mortality was 12.6%, hospital mortality 22.8%, and 28-day mortality 36.3%. All ML models outperformed SOFA alone (Table 1). Logistic regression achieved the highest AUROC (0.856) while random forest had the best calibration (Brier 0.084). SHAP analysis identified SOFA score, mean arterial pressure, SpO2, lactate, and respiratory rate as top predictors, with biliary obstruction and albumin among key disease-specific contributors. Risk stratification yielded a low-risk group (82.9%, 6.1% mortality) versus high-risk group (12.0%, 46.2% mortality). Conclusions: ML models incorporating pancreatic cancer-specific features significantly outperformed the SOFA score for predicting ICU mortality. Biliary sepsis dominated ICU admissions, and disease-specific factors including biliary obstruction and hypoalbuminemia contributed meaningfully to prediction. These models may support clinical decision-making regarding ICU triage and goals-of-care discussions in this high-risk population. Model AUROC AUPRC Brier Score Logistic Regression 0.856 0.518 0.092 LASSO 0.854 0.504 0.093 Random Forest 0.855 0.552 0.084 XGBoost 0.853 0.425 0.101 SOFA Score Alone 0.791 0.393 —

Activating in-plane dead zones of anode catalyst layers in proton exchange membrane water electrolyzers

Nature Communications Zicheng Zhao, Xiao Liang, Nan Lin et al. Jul 01, 2026 DOI: 10.1038/s41467-026-75223-1

Abstract The utilization of iridium-based anode catalysts in proton exchange membrane water electrolyzers is largely limited by the presence of electrochemically inactive “dead zones” within the catalyst layer. Here, by combining in-situ visualization of gas bubble evolution with quantitative conductivity measurements and electrochemical analysis, we establish that the limited in-plane electronic conductivity, dictated by the ionomer disrupting the conductive network of IrO 2 nanocatalysts, is the dominant factor. A sequential spray-coating strategy is further developed, which decouples the deposition of a pristine conductive catalyst layer from the ionomer and thereby preserves continuous electron transport pathways. This approach effectively activates the in-plane dead zones, resulting in membrane electrode assemblies that exhibit over 30% higher activity (4.2 A cm −2 @2.0 V@80 °C; membrane: Nafion 115) than those prepared by the conventional one-step method based on IrO 2 /ionomer mixed inks. Crucially, this approach achieves both low iridium loading (0.25 mg cm –2 ) with standard catalysts and demonstrates extended stability over 8300 hours under practical current densities.

Construction of an Interpretable Regression Model for Yield Prediction and Mechanistic Insight Enabled by Automated Reaction Path Exploration

Journal of the American Chemical Society Takahiro Doba, Yu Harabuchi, Yuuya Nagata et al. Jul 01, 2026 DOI: 10.1021/jacs.6c05203

Spatial association of seabirds and aquatic birds with highly pathogenic avian influenza (H5N1) outbreaks in Brazil: A nationwide ecological and statistical modelling approach

PLoS ONE Byron Abdel Hernández-Ortiz, Thiago Araujo dos Santos, Camila Michele Appolinário et al. Jul 01, 2026 DOI: 10.1371/journal.pone.0350505

Identifying wild bird species associated with highly pathogenic avian influenza (HPAI) is essential for optimizing surveillance and mitigating spillover risks. This study analyzes Brazil’s nationwide HPAI surveillance data (up to July 2025), comprising 1,153 records across 127 bird and mammal species from 525 municipalities. Using a multi-model framework—including chi-square association tests, binary logistic regression, and spatial Generalized Additive Models (GAMs). Species–outbreak associations were significative and positive for Thalasseus maximus (χ² = 237.34, p &lt; 0.0001), Thalasseus acuflavidus (χ² = 216.12, p &lt; 0.0001), Sterna hirundo (χ² = 83.88, p &lt; 0.0001), and Sterna hirundinacea (χ² = 77.56, p &lt; 0.0001). An optimized logistic regression model (model 3) highlighted T. acuflavidus as the strongest predictor of HPAI (OR = 80.74; 95% CI: 21.85–298.39; p &lt; 0.001), achieving good predictive performance (AUC = 0.85; Pseudo R² = 0.48). To account for spatial dependence, we fit a binomial GAM incorporating a bivariate longitude–latitude spatial smoother, significantly improving model fit (AUC = 0.96; Pseudo R² = 0.58) and effectively accounting for residual spatial autocorrelation (Moran’s I = 0.0399, z = 2.10, p = 0.044). Outbreaks were concentrated along Brazil’s southeastern coast, overlapping with high-density poultry zones, while inland spread remained sporadic, suggesting migratory routes as key transmission pathways. These results underscore the critical role of seabirds—particularly T. acuflavidus —in HPAI H5N1 dynamics in Brazil. The enhanced predictive power of the spatial GAM supports its utility in risk mapping. We recommend integrating biodiversity data with spatial modeling to guide targeted surveillance in high-risk coastal areas, reducing spillover threats to poultry and wild populations.

A study on the stability of amphiphilic cellulosic materials and the CO2-responsiveness of their interpenetrating network derivative with polyacrylamide

Scientific Reports Funsho Afolabi, Zulhelmi Amir, Muhamad F. A. Patah et al. Jul 01, 2026 DOI: 10.1038/s41598-026-55038-2

Efficient Osmotic Energy Conversion Enabled by Self‐Standing COF Membranes With Varied Sulfonic Acid Group Density

Advanced Materials Xi Ma, Xiaoxiao Cheng, Tamara Fischer et al. Jul 01, 2026 DOI: 10.1002/adma.73758

ABSTRACT The Gibbs free energy generated from the mixing of seawater and freshwater across a salinity gradient is considered one of the most significant yet underutilized renewable energy sources. Membrane‐based reverse electrodialysis (RED) enables direct electricity generation from osmotic energy by harnessing the net ion flux driven by concentration gradients across ion‐selective membranes. However, entropy generation caused by non‐selective ion mixing significantly limits the power density of RED systems. Therefore, enhancing membrane ion selectivity is critical. 2D covalent organic frameworks (COFs) demonstrate remarkable potential for osmotic energy conversion due to their aligned 1D nanochannel, high porosity, and organized ionic groups. Herein, we present a strategy leveraging electrostatic repulsion to controllably fabricate TpPa‐(SO 3 H) X COF ( X = 0.5, 1, 1.5, 2) membranes with varied ionic group density. Via stoichiometric modulation during COF synthesis, we achieved variation in sulfonic acid group density within nanochannels, enabling optimized charge‐governed ion selectivity. Under salinity gradients mimicking seawater/freshwater conditions (0.5  m /0.01  m , NaCl), the device delivered an exceptional power output density of 24.53 W m −2 , representing a 4.9‐fold enhancement over commercial benchmarks (5 W m −2 ). This study presents a novel method and strategy for the design and application of ion‐selective membranes in mass transport and efficient energy conversion.

Reversible, Photo‐Induced Lattice Distortions in Halide Perovskites

Advanced Materials Mansha Dubey, Bekir Turedi, Andrii Kanak et al. Jul 01, 2026 DOI: 10.1002/adma.202521800

ABSTRACT The unique characteristics of soft‐lattice halide perovskites (HPs) have motivated extensive research into their application in semiconductor devices. Mechanically and chemically modulated strain in monocrystalline HPs improves stability and phase purity. However, a comprehensive understanding of the photoexcitation‐driven lattice distortions arising from the strong electron‐phonon coupling paired with the dynamics of the A‐site cation remains underdeveloped. Here, we present the transient lattice distortions occurring in single crystal MAPbBr 3 , FAPbBr 3 , and CsPbBr 3 in response to above‐bandgap light excitation. Using an X‐ray probe, we uncover reversible and hysteresis‐free photoinduced lattice distortion. We find that the photoinduced distortion in the HPs is a function of the pump power, with CsPbBr 3 showing the highest resilience against lattice deformation with a 0.062% change in its out‐of‐plane lattice parameter. Conversely, the organic HPs unveil a stronger interaction with photocarriers resulting in more significant yet elastic distortion, with MAPbBr 3 exhibiting up to 0.3% change. We demonstrate the modality of this distortion by varying the excitation power over 20 distinct states and cycles, highlighting the suitability of HPs as building blocks for electrostriction devices. Our findings represent a key step toward establishing HPs as ideal platforms for optical and strain‐driven switchable photonic devices.

The retinol-metabolizing enzyme DHRS3 coordinates antigen presentation, endothelial stability, and cholesterol metabolism to suppress hepatocellular carcinoma progression

Journal of Biological Chemistry Hongtao Hu, Lidan Zhang, Jiawen Gu et al. Jul 01, 2026 DOI: 10.1016/j.jbc.2026.113168

Learning from failure: A multifactorial analysis of locoregionally advanced laryngeal cancer (LALC) recurrence and the impact of a standardized SOP in a resource-limited setting.

Journal of Clinical Oncology Bhaskar Chakraborty Jul 01, 2026 DOI: 10.1200/jco.2026.44.19_suppl.231

231 Background: In low- and middle-income countries (LMICs), LALC prognosis is often compromised by treatment heterogeneity. This study aimed to stratify predictors of early versus late radiotherapy (RT) failure and evaluate the efficacy of a newly implemented Institutional Standard Operating Procedure (SOP) on survival outcomes. Methods: This two-phase study was conducted at a tertiary center. Phase I (Retrospective): A cohort of 1,000 patients (2016–2020) was reviewed. Failures were stratified by timing: Early (&lt; 12 months) vs. Late (&gt; 12 months). Univariate and multivariate Cox proportional hazards regression models were utilized to identify independent predictors of recurrence. Phase II (Prospective): Based on Phase I findings, a standardized SOP was implemented (2020–2021) for 100 Stage III/IV patients, prioritizing precise target delineation,stringent monitoring of set up error with adaptive replanning,implementation of radionomics with dose escalation and strict nutritional support. Primary endpoints were Recurrence-Free Survival (RFS) and 3-Year Overall Survival (3YOS), calculated using Kaplan-Meier analysis. Results: Phase I: Recurrence occurred in 20% (200/1,000) of the historical cohort. Failures were predominantly Early (n = 165; 82.5%), characterized by local failure (73%). Multivariate analysis identified geographical miss (HR &gt; 1.5; p &lt; 0.05), low RT dose, and poor nutrition as independent predictors of early failure. Late recurrences (n = 35; 17.5%) were driven by inadequate chemotherapy dose and aggressive tumor biology. Phase II: Following SOP implementation (n = 100), the total recurrence rate dropped to 12% at 4 years. Of the 12 failures, 8 occurred within 1 year and 4 between 12–18 months. The 4-year Recurrence-Free Survival (RFS) was 88%. Total mortality was 11% (7 cancer-related, 4 non-malignant), resulting in a 3-Year Overall Survival (3YOS) of approximately 89%. This represents a statistically significant improvement in disease control compared to historical controls. Conclusions: Historical analysis reveals distinct failure patterns: early recurrence is driven by technical and nutritional factors, while late recurrence is associated with systemic therapy inadequacy. Implementation of a targeted SOP addressing these specific vulnerabilities successfully improved RFS to 88% and OS to 89%, validating this protocol as a feasible standard of care for resource-limited settings.

Structural determinants of diagnosis and prognosis in cervical cancer: Associations between care setting, comorbidity burden, and sociodemographic factors.

Journal of Clinical Oncology Jessica Moore, Jungwun Lee, Sarika Gurnani et al. Jul 01, 2026 DOI: 10.1200/jco.2026.44.19_suppl.209

209 Background: Cervical cancer remains a significant public health issue, with approximately 13,000 new cases and over 4,000 deaths annually in the United States. Understanding how healthcare setting and sociodemographic factors influence diagnostic and presentation pathways is critical to addressing inequities. Methods: A retrospective cohort study at a large, urban, tertiary care safety-net academic medical center was conducted. Patients with a new diagnosis of cervical cancer between January 1, 2016, and December 31, 2023, were included. Socio-demographics, diagnosis location, Carlson Comorbidity Index (CCI), FIGO cancer stage and clinical factors were obtained. Multivariate logistic regression analysis was performed. Results: 126 patients met inclusion; 20 diagnosed inpatient and 106 diagnosed outpatient. Mean age was 52.6 years old, and most were non-white (60.3%), non-Hispanic (72.2%) non-English speaking (59.5%), had a primary care physician (PCP) (68%) and had public insurance (82.3%). The mean number of Papanicolaou (pap) smear screenings prior to diagnosis was 1.89, and 20 patients had 0 pap smears. Inpatient setting at diagnosis was strongly associated with advanced stage (stage 3 vs 1: OR 3.86; stage 4 vs 1: OR 11.85; both p &lt; 0.0001) and higher CCI (OR 1.27, p &lt; 0.0001), while having insurance (OR 0.26, p &lt; 0.0001) and college education (OR 0.37, p = 0.0036) were associated with lower odds of inpatient diagnosis. Having a PCP was associated with lower odds of advanced stage (stage 4 vs 1: OR 0.42; p≤0.0002) and American nationality (OR 0.55; p &lt; 0.0001), but higher odds among English proficiency (OR 2.11, p &lt; 0.0001), insurance (OR 2.12, p &lt; 0.0001), and college education (OR 1.98, p = 0.018). Mortality was independently associated with advanced stage (stage 3 vs 1: OR 1.76; stage 4 vs 1: OR 5.24; both p≤0.0004) and higher comorbidity burden (OR 1.27, p &lt; 0.0001), while demographic factors were not independently associated with death. Conclusions: Inpatient diagnosis and lack of an established PCP were significantly associated with advanced stage at presentation, higher CCI and markers of socioeconomic instability. Although most patients had insurance and a PCP, gaps in preventive care were evident, with a low mean number of prior pap smears or never screened. Advanced stage at diagnosis was independently associated with mortality, highlighting missed opportunities for access to screening, continuity of care, and early detection. Findings suggest that inequities in preventive access and longitudinal care may shift diagnostic pathways toward more frequent, higher acuity inpatient presentations with overall higher mortality outcomes, for structurally vulnerable populations. Efforts to improve equitable access should focus on disparities related to insurance status, educational attainment, nationality, and language access.