Endometrial cancer(EC),the second most common gynaecologic malignancy,faces challenges in precision treatment due to limited predictive models for therapy selection.Patient-derived organoids,which recapitulate tumour heterogeneity and microenvironment,offer a transformative platform for drug sensitivity testing and personalised therapy.However,standardised protocols for establishing EC organoids,biobanking and clinical translation remain lacking consensus.This expert consensus proposes guidelines for EC organoid culture optimisation,characterisation and clinical validation.By harmonising technological advances with clinical needs,this consensus aims to accelerate the integration of organoid models into EC precision medicine,ultimately improving therapeutic outcomes.

BackgroundOvarian clear cell carcinoma (OCCC) is a rare aggressive, and chemo-resistant subtype of epithelial ovarian cancer. Current limitations in precisely characterizing its molecular features have resulted in restricted availability of clinical targeted therapies and significant therapeutic challenges.MethodsTo address this unmet need, we conducted an integrative multi-omics study of 82 OCCC cases, incorporating whole-exome sequencing (WES), bulk RNA sequencing, and single-cell RNA sequencing (scRNA-seq).ResultsOur analysis uncovered recurrent mutations in multiple epigenetic regulators including ARID1A, EP300, and SETD2B, reinforcing chromatin remodeling as a hallmark of OCCC pathogenesis. Strikingly, FOXA2 mutations were absent in early-stage tumors but specifically enriched in advanced-stage cases (19% frequency), with functional validation demonstrating their role in driving malignant progression. Copy number alteration profiling revealed frequent amplifications in chromosomal arms such as 17q, which contains the ERBB2 oncogene that potentially regulates OCCC progression. Chromosomal translocations were detected in 35.59% of cases, including a novel FGFR2/RPAP3 fusion with therapeutic implications. Notably, scRNA-seq delineated immune-rich subsets characterized by abundant cytotoxic T-cell and B-cell infiltration, suggesting immunotherapeutic opportunities in a patient subset. Moreover, molecular subtyping identified ERBB2 amplification/overexpression as a high-risk feature strongly associated with poor survival. Patient-derived xenograft (PDX) models and a retrospective analysis of two clinical cases demonstrated that HER2-targeted antibody-drug conjugates (HER2-ADCs) significantly suppressed tumor growth and progression in OCCC patients with HER2 expression.ConclusionsIn summary, our study establishes the comprehensive molecular atlas and a targeted therapeutic subtyping framework, revealing therapeutic vulnerabilities and providing novel insights for advancing precision oncology in OCCC management.

Supplementary Figure S2. Low serum ApoA1 impairs CD8 + T-cell tumor infiltration and cytotoxicity (related to Figure 2)

The utility of Large Language Models (LLMs) in high-stakes clinical reasoning is often hindered by a lack of deep domain knowledge and the ambiguity of real-world data. Taking Endometrial Cancer FIGO staging as a challenging exemplar, general models struggle with evolving guidelines and extreme class imbalance. To address this, we propose PathoLLM, a specialized framework featuring a systematic two-stage fine-tuning strategy: (1) Domain Knowledge Injection from expert literature, followed by (2) a novel Logic-aware Human-in-the-Loop (HIL) Refinement process. Crucially, this HIL strategy employs a 3-tiered feedback mechanism to rectify not only the model’s final output but also its underlying reasoning pathways, constructing a high-fidelity, logic-enriched dataset for task adaptation. This process is further enhanced by iterative prompt optimization. Validated on a hold-out test set, PathoLLM achieves an Overall Accuracy of 0.961 and a Macro-F1 of 0.8192. Notably, despite using only 32B parameters, it demonstrates performance competitive with massive state-of-the-art general models, establishing a superior balance between accuracy and computational efficiency for complex clinical decision support, offering a solution for logic-dependent tasks in resource-constrained environments.

Supplementary Figure S3. ApoA1 exerts its antitumor activity in a CD8 +T cell–dependent manner (related to Figure 3)

Figure S6. ApoA1 stabilizes HIF1α protein by reducing HIF1α protein ubiquitination (related to Figure 6)

Figure S5. ApoA1 potentiates CD8 + T-cell immune function via the HIF1α–mediated glycolysis pathway (related to Figure 5)

Supplementary Figure S1. Serum lipid level in ovarian cancer and endometrial cancer patients (related to Figure 1)

BACKGROUND:Precise evaluation is pivotal in managing relapsed ovarian cancer, particularly as poly adenosine diphosphate-ribose polymerase inhibitor (PARPi) maintenance therapy can render some recurrent lesions difficult to detect. We aimed to investigate the clinical applications and underlying mechanisms of 68Gallium-labelled FAP-inhibitors (68Ga-FAPI) compared with 18F-fluorodeoxyglucose positron emission tomography/computerized tomography (18F-FDG PET/CT) in recurrent ovarian cancer. METHODS:Between January 2022 and July 2023, patients with suspected recurrent ovarian cancer at Fudan University Shanghai Cancer Center underwent both PET/CT imaging modalities. To minimize heterogeneity, the final analysis included only patients with platinum-sensitive high-grade serous ovarian cancer. Clinical characteristics, treatment strategies, and pathological findings were collected. Additional analyses included immunohistochemistry, single-nucleus RNA sequencing (snRNA-seq), and patient-derived organoid (PDO) models experiments. RESULTS:Eighty-nine eligible patients were prospectively enrolled. Concordant imaging findings were observed in only 37 (41.6%) patients. Thirty-three patients underwent surgery, and 28 (84.8%) achieved complete resection. The overall diagnostic accuracy was 96.3% for FAPI compared with 86.9% for FDG. Patients receiving PARPi maintenance therapy were more likely to have additional FAPI-positive (FAPI+) lesions, while bevacizumab appeared to influence FAPI uptake. Patients with additional FAPI-detected lesions experienced shorter progression-free survival, particularly those with prior PARPi maintenance therapy. In PDO models, FAPI-positive/FDG-negative (FAPI+/FDG-) tumors after PARPi maintenance therapy were resistant to PARPi re-challenge. Immunohistochemistry staining and snRNA-seq analyses revealed hypoglycolytic tumor features and activation of cancer-associated fibroblasts (CAFs) in these lesions. CONCLUSION:68Ga-FAPI demonstrated superior lesion detection and diagnostic accuracy compared with 18F-FDG PET/CT for recurrent ovarian cancer, especially in the PARPi era. Hypoglycolysis and CAF activation may underlie PARPi resistance in FAPI+/FDG- lesions.

Sepsis, a life-threatening condition, remains a leading cause of global mortality. It is frequently accompanied by systemic inflammatory response syndrome (SIRS), which triggers excessive release of inflammatory factors and may lead to a cytokine storm. The sensitive detection of procalcitonin (PCT) is therefore critical for early prediction and clinical management of sepsis. In this work, an electrochemical immunosensor was developed based on polyaniline-encapsulated NiCo alloy (NiCo@PANI) nanoflowers for the ultrasensitive detection of PCT. The NiCo alloy ensures rapid electron transfer, and the in-situ polymerization of PANI on the surface of NiCo nanoflowers not only enhances their antioxidant stability but also offers an efficient matrix for antibody immobilization. During the detection process, the specific binding of PCT to the antibody on the electrode surface increases the electron transfer resistance and the steric resistance, resulting in a decrease in current value. By fitting the linear relationship between the attenuated signal and the PCT concentration, the quantitative identification of PCT can be achieved. The fabricated immunosensor achieves ultrasensitive detection of PCT within 20 min, exhibiting a broad linear range from 0.1 pg/mL to 100 ng/mL and a detection limit as low as 0.015 pg/ mL. Moreover, it exhibited high selectivity, excellent stability, and satisfactory reproducibility, along with accurate PCT detection in real human serum samples. This work provides a promising platform for ultrasensitive PCT monitoring, with potential applications in early disease diagnosis and point-of-care health assessment.

BackgroundThe trueness of serum estradiol (E2) measurement is critical for monitoring follicle development in assisted reproductive technology (ART). This study aimed to evaluate E2 assay standardization, assess the trueness of harmonized E2 results across platforms, and investigate the relationship between estimated E2 levels and follicle diameters on human chorionic gonadotropin (hCG) trigger days.MethodsSerum samples from 90 individuals were analyzed using assays from four manufacturers and LC-MS, which served as the reference method. A Bland-Altman plot-based harmonization algorithm (BA-BHA) was applied to harmonize E2 results. Trueness was assessed by mean percent difference and 95% limits of agreement (LoA). Harmonized results were compared to identify high-trueness kits. Additionally, 237 serum samples with corresponding follicle data from ART patients on hCG trigger days were analyzed. Multiple linear regression was applied to establish the relationship between harmonized E2 levels and follicle diameters.ResultsBefore harmonization, mean percent differences from LC-MS ranged from−2.3% to 17.4%. LiCA-E2 demonstrated the best performance. Following harmonization using the BA-BHA, LiCA maintained superior trueness, as indicated by a mean percent difference of 0.1% and a sum of 95% LoA of 41.4%. Multiple linear regression revealed a positive correlation between estimated E2 levels and follicle diameters. LiCA exhibited the strongest linear correlation between estimated E2 levels and corresponding diameters of follicle, with a linear regression equation and a Pearson's correlation coefficient r = 0.8077.ConclusionsHarmonization effectively improves E2 assay comparability. LiCA's superior performance and strong E2-follicle correlation offer a reliable tool for predicting follicle maturation, guiding clinical decisions in ART.

Recurrent ovarian clear cell carcinoma (OCCC) remains a therapeutic challenge due to intrinsic chemoresistance and paucity of targeted options. Surufatinib is a multi-target tyrosine kinase inhibitor that may enhance antitumor immunity when combined with PD-1 blockade. We report a 53-year-old female with recurrent OCCC who developed a platinum-sensitive first relapse (platinum-free interval, 9.8 months) after adjuvant platinum-based chemotherapy and achieved a 24-month progression-free survival (ongoing) on surufatinib plus toripalimab. The best response was partial response, and treatment was well tolerated except for Grade 1 hemoptysis and Grade 1–2 proteinuria. Radiologic disease control was durable. However, because biomarker assessment relevant to immune checkpoint inhibition was not performed, the biological basis of response remains uncertain. This case suggests the clinical feasibility of surufatinib plus toripalimab and its association with durable disease control in an individual patient with recurrent OCCC. However, given the platinum-sensitive nature of the relapse, treatment efficacy should be interpreted cautiously. Further prospective studies incorporating biomarker assessment are warranted.

Supplementary Material Original Western blots and statistics

Endometrial cancer (EC) incidence is rising, yet current diagnostics lack precision and scalability. We develop an artificial intelligence (AI)-based platform integrating multi-biofluid omics and clinical data for EC stratification. Using two independent cohorts from different clinical centers (531 participants for model development, 204 for external validation), we collect 1,179 samples (plasma, cervical/uterine secretions) and the corresponding clinical data (age, ultrasound, etc.). Machine learning identifies EC-specific signatures, and the AI framework fuses omics features with clinical factors to enable multilevel risk stratification. On the external validation cohort, the platform achieves 95.65% sensitivity for minimally invasive EC screening and balanced performance with an area under the curve (AUC) value of 0.94 for EC confirmation. The model also shows potential for high-risk subtype detection. Biological plausibility is supported by identified omics signatures. A web tool is developed to support clinical translation. This platform demonstrates the potential of multi-omics and AI in precision oncology.

Epithelial ovarian cancer (EOC) is an aggressive malignancy with limited therapeutic options. Poly(ADP-ribose) polymerase inhibitors (PARPi) have shown remarkable efficacy, especially in BRCA-mutant patients, and are approved as maintenance therapy to prevent recurrence after initial response to chemotherapy. However, the development of PARPi resistance poses a major clinical challenge. This study utilized a whole-genome CRISPR-Cas9 genetic screening to identify genes associated with PARPi sensitivity upon knockout. Based on the screening and validated through further experiments, we confirmed that CLK1 knockdown is synthetically lethal with PARPi in ovarian cancer. The combination of the PARPi Olaparib and CLK1 inhibitor TG003 exhibited potent anti-proliferative effects both in vitro and in vivo. Mechanistically, CLK1 inhibition downregulated the functional ERCC1-202 isoform, resulting in enhanced DNA damage and apoptosis. Our findings reveal a novel mechanism underlying PARPi sensitivity and suggest that targeting CLK1 in combination with PARPi may represent a promising therapeutic strategy for PARPi-resistant ovarian cancer.

Supplementary Figure S4. ApoA1 mimetic peptide L-4F potentiates the antitumor effect of CD8+ T cells (related to Figure 4)

5564 Background: Recurrence surveillance of epithelial ovarian cancer (EOC) is a major clinical challenge, given the suboptimal performance of serum CA125 levels and limitations of radiological assessments. Serum small extracellular vesicle (sEV)-based liquid biopsy provides tumor-enriched and stable biomarkers, with potential to improve the precision of recurrence monitoring and support post-treatment surveillance. This study aimed to develop and preliminarily validate a serum sEV protein-based model to enhance recurrence monitoring in ovarian cancer. Methods: The study was designed with two stages (model development and validation) and plans to prospectively enroll 200 patients with EOC (NCT06925126). This analysis reports preliminary stage I data, and patients were categorized into relapse and non-relapse groups based on imaging-confirmed recurrence status. All relapsed patients were platinum-sensitive with low tumor burden. Quantitative assessment was performed on biomarkers derived from sEVs (specifically, E-CA125, E-HE4, and E-C5a) as well as corresponding serum markers measured in routine clinical practice (namely, H-CA125 and H-HE4), utilizing peripheral blood samples. Biomarker distributions were compared between groups, and predictive performance for recurrence detection was evaluated using receiver operating characteristic (ROC) analysis, including area under the curve (AUC), accuracy, sensitivity, and specificity. Results: A total of 113 patients were analyzed (42 non-relapsed, 71 relapsed), with a median age of 57 years. The majority (95/113) of patients had high-grade serous ovarian cancer, while other histological types were also included. Levels of E-CA125, E-HE4, H-CA125, and H-HE4 were significantly higher in relapsed patients compared with non-relapsed patients (all p < 0.01), while E-C5a showed no discriminative value between the two groups. Among individual sEV biomarkers evaluated, E-CA125 demonstrated the highest performance for recurrence detection (AUC = 0.877), followed by E-HE4 (AUC = 0.689) and E-C5a (AUC = 0.528). Moreover, the sEV protein-based model achieved an accuracy of 77.9% with a sensitivity of 71.8% and a specificity of 90.9%. This performance surpassed that of the conventional serum CA125, which exhibited an accuracy of 50.0%, a sensitivity of 26.8%, and a specificity of 100.0%. These results indicate that the sEV protein-based model may provide added clinical value when serum CA125 monitoring is unreliable after prior therapy. Conclusions: The sEV-derived CA125 and the sEV model demonstrate promising performance for recurrence detection in EOC. These findings support the feasibility of a serum sEV-based surveillance strategy that may improve the reliability of post-treatment monitoring. Prospective validation is ongoing to determine its potential role in guiding clinical surveillance and intervention. Clinical trial information: NCT06925126 .

e17505 Background: The clinicopathological profile and prognostic factors of cervical clear cell adenocarcinoma unrelated to diethylstilbestrol exposure are not well characterized in Chinese population. Methods: This multicenter retrospective study was conducted to analyze the characteristics of cervical clear cell adenocarcinoma patients from 10 medical centers. 97 patients surgically diagnosed with primary cervical clear cell adenocarcinoma were collected from 2010 to 2022. A comprehensive clinicopathological profile of cervical clear cell adenocarcinoma patients was depicted. The association between clinicopathological characteristics and prognostic factors was also assessed. Results: The peak age prevalence occurred in the 46-55 years group (n = 97), with the human papillomavirus (HPV)-negative rate of 85.1%. All surgically treated patients had FIGO stages IA-IIA or IIIC and received adjuvant radiotherapy and/or chemotherapy. With a median follow-up of 52 months (interquartile range, 36-70 months), multivariable analysis identified ovarian metastasis (HR = 3.89, P = 0.027 ) and lymph node metastasis (HR = 4.26, P = 0.004 ) as independent risk factors for overall survival. Multivariable analysis also demonstrated that lymph node metastasis (HR = 3.15, P = 0.005 ) remained a significant predictor for progression-free survival. Conclusions: This retrospective multicenter study characterized clinicopathological features of cervical clear cell adenocarcinoma and identified prognostic factors. Univariate and multivariate analysis of factors associated with overall survival. Variables Univariate HR (95%CI) P-value Multivariable HR (95%CI) P-value Tumor size >4 cm 3.85 (1.15–12.91) 0.029 3.13 (0.80–12.24) 0.102 DSI >2/3 8.30 (1.81–38.02) 0.006 3.36 (0.62–18.22) 0.161 Ovarian metastasis (Yes) 3.09 (1.03–9.39) 0.047 3.89 (1.17–12.98) 0.027 LVSI (Yes) 4.47 (1.80–11.07) 0.001 0.49 (0.26–3.58) 0.481 LNM (Yes) 4.80 (1.94–11.86) 0.001 4.26 (1.58–11.45) 0.004 Surgical approach (ARH) 1.57 (0.36–6.79) 0.548 - - HR, hazard ratio; 95%CI, 95% confidence interval; DSI, depth of stromal invasion; LVSI, lymphovascular space invasion; LNM, lymph node metastasis; ARH, abdominal radical hysterectomy; LRH, laparoscopic radical hysterectomy.

Supplementary Figure S7. Anti-PD-1 therapy has no antitumor effects in tumor-bearing A1KO mice (related to Figure 7)