e16288 Background: PD-1/PD-L1 inhibitors combined with targeted agents or HAIC (Hepatic Arterial Infusion Chemotherapy) can reduce the burden of micrometastasis and improve clinical outcomes, which has shown promising efficacy as neoadjuvant therapy for resectable hepatocellular carcinoma (HCC). Combination regimens including immunotherapy are available for wider application due to the ease of implementation. This study aims to compare the safety and efficacy of lenvatinib combined with sintilimab versus HAIC as neoadjuvant therapy for resectable stage IB solitary hepatocellular carcinoma with high recurrence risk. Methods: This prospective, randomized, two-cohort, exploratory, phase II study (NCT05621499) enrolled patients with histologically confirmed resectable HCC. Patients with ECOG 0-1, Child Pugh A, and no prior treatment had at least one measurable lesion according to RECIST v1.1. The high risk of recurrence was defined by the investigators before surgery as stage Ib and solitary tumor with the largest diameter > 5cm. Eligible patients were randomly assigned to receive HAIC-FOLFOX for 2 cycles (Group 1, G1) or sintilimab 200 mg Q3W and lenvatinib 12 or 8 mg QD for 2 cycles (Group 2, G2), and surgery was performed within 2-3 weeks after treatment. The primary endpoint was 1-year disease-free survival (DFS) rate. Secondary endpoints were incidence of microvascular invasion, pathological complete response (pCR) rate, objective response rate (ORR, by RECIST v1.1), 2-year DFS rate, 2-year overall survival (OS) rate, and safety. Results: As of December 31, 2024, 54 patients were enrolled and assigned to G1 (n = 27) and G2 (n = 27). At the cutoff date, 5 patients in G1 and 6 patients in G2 were still receiving treatment. Two patients in G2 were excluded due to poor liver function. Radiological evaluations were conducted for 40 patients (G1, n = 21; G2, n = 19). According to RECIST v1.1 and mRECIST, the ORR was 9.52% and 23.81% in G1, 31.58% and 57.89% in G2, respectively. 39 patients (G1, n = 21; G2, n = 18) underwent radical resection and all achieved R0 resection. In terms of pathological outcomes, no patients achieved complete response in G1, the pCR rates were 33.33% (6/18) in G2. No grade≥3 adverse events were observed during neoadjuvant treatment period in either group. Conclusions: Sintilimab in combination with lenvatinib was well tolerated and showed promising efficacy compared with HAIC as a neoadjuvant therapy for high recurrence risk resectable stage IB solitary HCC. The data of DFS was immature, and follow-up is continuing. Clinical trial information: NCT05621499 .

Glioblastoma is a lethal brain tumor that is unresponsive to current cancer immunotherapeutic approaches, including immune checkpoint blockade (ICB). This suggests that initial priming of T cells, rather than their expansion and licensing as effectors, is a restricting feature in this tumor setting. To overcome the limited initiation of CD8 + T cell responses, we employed a strong heterologous prime-boost vaccination with the simian adenovirus ChAdOx1 and poxvirus modified vaccinia Ankara (MVA). Vaccination conferred therapeutic efficacy against orthotopic, immune checkpoint-blockade (ICB)-refractory SB28 murine glioblastoma. Vaccination was effective against both the murine tumor antigen, P1A, and a newly identified glioblastoma-associated antigen, Gpr149. Additional treatment with ICB provided no additional benefit. Systemic ChAdOx1/MVA vaccination induced robust infiltration of antigen-specific T cells in tumor-challenged brains, the majority of which exhibited a CD103 + CD69 + CD8 + tissue-resident memory (TRM)-like phenotype. These cells were polyfunctional, durable in brains with sustained tumor control, and mediated tissue-specific immunological memory. Moreover, intracranial adoptive transfer of glioblastoma-derived antigen-specific TRM-like cells was sufficient to protect naïve recipients from subsequent orthotopic tumor challenge. Together, these findings establish that viral vector vaccination can generate tumor-specific TRM-like cells that mediate effective anti-glioblastoma immunity, providing a rationale for clinical evaluation of ChAdOx1/MVA-based strategies in glioblastoma.

4596 Background: The prognostic value of FGFR3 pathway activation in patients with muscle-invasive bladder cancer (MIBC) treated with neoadjuvant chemotherapy (NAC) has not been established. Activation of FGFR3 signaling drives a subset of urothelial bladder cancers and classically results from pathogenic gene mutations (mut) or fusions (fus). Additionally, an RNA signature (FGFR-PRS) representing activation of the FGFR3 pathway independent from pathogenic FGFR3 mut or fus has been identified (Eisner et al. Clin Can Res 2024). We previously reported that high tumor mutational burden (TMB) may be associated with improved pathological complete response (pCR) and survival in a cohort of MIBC patients treated with cisplatin-based NAC (Burgess et al. GU ASCO ‘26). In the current analysis, we assessed the impact of FGFR3 pathway activation detected by either FGFR3 gene mut/fus or FGFR-PRS (+) on clinical outcomes in the previously reported cohort. Methods: 91 patients with MIBC who received cisplatin-based NAC followed by radical cystectomy underwent genomic analysis of diagnostic transurethral resection of bladder tumor specimens with the Tempus xT platform. Sample size was prespecified by statistical design. A subset of 61 patients had sufficient RNA for FGFR-PRS analysis. Correlation of FGFR3 molecular status with TMB and clinical outcomes was performed using logistic regression, Cox proportional hazards models, and Kaplan-Meier techniques. Results: Median follow up for the cohort was 63.6 months. pCR was achieved in 31.9%. 41.8% of patients relapsed. FGFR3 mut were found in 8 (8.8%) and included p.R248C (n = 2), p.S249C (n = 5), and p.G370C (n = 1). One (1.1%) patient had a FGFR3-TACC3 fus. In the subset with RNA analysis (n = 61), FGFR-PRS (+) was detected in 33 (54.1%), of which 28 (84.8%) were FGFR3 wild type (wt) without detected mut/fus. pCR occurred in 12.5% and 33.7% with and without FGFR3 mut [OR 0.281, p = 0.428]. pCR was similar regardless of FGFR-PRS status [OR 1.372, p = 0.602]. All tumors with FGFR3 mut/fus (n = 9) had TMB < 10 Mut/Mb. Nineteen (57.6%) tumors with FGFR-PRS (+) had TMB < 10 compared to 13 (46.4%) of FGFR-PRS (-) tumors. Six (75%) patients with FGFR3 mut [OR 3.188, p = 0.064] and 10 (30.3%) with FGFR-PRS (+) [OR 1.087, p = 0.999] relapsed. Patients with FGFR3 mut had inferior relapse-free survival (RFS) [HR 2.58, p = 0.028], but those with FGFR-PRS (+) did not [HR 1.37, p = 0.451]. Conclusions: The presence of FGFR3 mutations was associated with an inferior relapse rate and RFS in this MIBC cohort treated with cisplatin-based NAC. FGFR3 pathway activation by RNA signature (FGFR-PRS+) was not associated with inferior outcomes. Whether peri-operative FGFR3 inhibition improves prognosis in MIBC patients with FGFR3 mutations should be studied.

Angiogenesis fuels colorectal cancer (CRC) growth and metastasis by supplying oxygen and nutrients to expanding tumors. Here, we define a TGM2-mediated serotonylation axis in tumor endothelial cells as a driver of angiogenesis and a key mediator of epigenetic reprogramming within the tumor microenvironment. Across CRC single-cell datasets, TGM2 is broadly expressed in endothelial cells, and endothelial TGM2 correlates with the H3Q5Ser mark. Endothelial-specific deletion of Tgm2 in mice markedly suppresses angiogenesis and slows allograft tumor growth. In human umbilical vein endothelial cells, TGM2 knockdown diminishes H3Q5Ser, proliferation, migration, and tube formation, which can be rescued with nuclear-localized wild-type TGM2 but not a transamidase-inactive mutant. Mechanistically, TGM2-catalyzed serotonylation promotes LDHA transcription via H3Q5Ser at the LDHA promoter, thereby upregulating glycolysis. Hypoxia induces TGM2 expression via HIF-1α signaling. Moreover, endothelial cells lack canonical serotonin transporters and serotonin biosynthesis, and exosome-mediated transfer of 5-HT provides a source that fuels endothelial serotonylation. Clinically, high endothelial TGM2 and H3Q5Ser levels predict poorer prognosis. Collectively, the TGM2-serotonylation axis in endothelial cells represents a promising therapeutic target to disrupt tumor angiogenesis and CRC progression, with potential to synergize with immunotherapy by normalizing tumor vasculature and enhancing anti-tumor immunity.

Background Patient-reported outcomes (PROs) are essential for assessing symptomatic adverse events (AEs) from a patient perspective, which significantly impact the quality of life and clinical outcomes in patients with glioma. However, no validated patient-reported outcome measures (PROMs) exist to quantify symptomatic AEs in adult-type diffuse gliomas. Methods The study was conducted in two parts. First, we developed a customised Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) scale for adult-type diffuse gliomas using the Simplified Chinese PRO-CTCAE® item library, informed by initial item screening, patient pilot testing, and a two-round Delphi survey. Delphi experts were recruited through the National Glioma Multidisciplinary Team (MDT) Alliance (NGMA) and invited by email in June 2022 (1st round) and August 2022 (2nd round). We subsequently conducted a multicentre, prospective, observational cohort study (VERONICA) at 13 glioma treatment centres in China between September 2022 and March 2025. Eligible participants were adults aged 18 years or older with a diagnosis of adult-type diffuse glioma, who were able to understand and complete the questionnaires; patients with severe cognitive impairment, severe language dysfunction, or other conditions precluding questionnaire completion were excluded. The primary outcome was the psychometric performance of the customised PRO-CTCAE scale, including test-retest reliability, convergent validity, known-groups validity, and responsiveness, evaluated longitudinally across repeated study visits. VERONICA is registered with ClinicalTrials.gov, NCT05486923. Findings For the Delphi survey, all seven invited experts from six centres participated in 1st round (response rate 100·0%), with moderate agreement in symptom rankings (Kendall's W = 0·415; p < 0·001). In 2nd round, 16 of 20 invited experts from 14 centres participated (response rate 80·0%), with consistent agreement in expert ratings (Kendall's W = 0·351; p < 0·001). The final version of the customised PRO-CTCAE scale comprised 53 items covering 31 symptoms, together with one open-ended free-text item. For VERONICA, 450 participants were enrolled across 13 glioma treatment centres. Mean age was 49·1 years (SD 12·8), and the mean Karnofsky Performance Status (KPS) at baseline (Visit 2) was 72·2 (SD 17·1). 424 provided data eligible for at least one prespecified psychometric analysis. Test-retest reliability was acceptable (intraclass correlation coefficient [ICC] ≥0·70 for 47 of 53 items). Convergent validity was supported by correlations in the expected direction with matched European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire-Core 30 (EORTC QLQ-C30) domains, with predominantly moderate-to-strong associations (25 items with r ≥ 0·50). Known-groups validity was supported by discrimination between KPS <70 and ≥70 (Cohen's d ≥ 0·20 for 49 of 53 items; p < 0·05 for 43 of 49 items). In Global Impression of Change (GIC)-anchored responsiveness analyses, 37 items showed standardised response means (SRMs) ≥0·20 among participants reporting worsened overall status. Interpretation The customised PRO-CTCAE scale showed robust psychometric performance for adult-type diffuse gliomas. Remote, longitudinal administration supports low-burden quantification of patient-reported symptomatic AEs in clinical trials and routine neuro-oncology practice. Future work should assess implementation in routine care and clinical trials, and extend translation, cultural adaptation, and validation across different languages. Funding Beijing Medical Award Foundation; Shanghai Municipal Health Commission; Department of Science and Technology of Ningxia Hui Autonomous Region; Huashan Hospital, Fudan University (Clinical Research Project).

Table S3 shows the anti-tumor activity in cell lines of multiple cancer types.

HLA-DPB1*1508:01 shows one nucleotide substitution at position 668 G>A in exon 4 compared with HLA-DPB1*36:01.

Introduction: Cholangiocarcinoma (CCA) is an aggressive malignancy with poor prognosis and limited treatment options. Molecular heterogeneity and the tumor immune microenvironment play crucial roles in CCA progression and therapeutic response, but comprehensive stratification remains insufficient for guiding treatment decisions. Methods: This study performed integrated multi-omics profiling as well as immunohistochemical validation on 424 CCA patients. Multidimensional data, including genomic, epigenomic, transcriptomic, and immune profiling, were analyzed to identify clinically relevant molecular subtypes. Results: Four distinct molecular subtypes of CCA were identified, each exhibiting unique clinical and immunological features: C1 (proliferation): driven by TP53/KRAS mutations and CpG island methylator phenotype (CIMP+) hypermethylation, showing T helper 17 cells infiltration and poor outcomes; C2 (Immune-Suppressed Macro_LYVE1 [ISM_LYVE1]): stroma-rich with LYVE1+ macrophages and epithelial-mesenchymal transition (EMT) activation; C3 (immune-activated Macro_C1QC (IAM_C1QC)): enriched in C1QC+ macrophages, CD8+ T-cells, and metabolic pathways, highly responsive to immune checkpoint blockade (ICB; 75% overall response rate); C4 (immune exclusion): FGFR2-altered and IDH1-mutant, with an immunologically cold phenotype. Conclusion: We validated ATP2B1 as a novel prognostic biomarker and developed a 160-gene classifier for subtype prediction. The C3 subtype's exceptional ICB response, independent of conventional biomarkers (PD-L1/microsatellite instability/tumor mutational burden), highlights the clinical utility of this classification system in guiding precision immunotherapy for CCA.

Colorectal cancer (CRC) is a leading cause of cancer mortality worldwide, with liver metastasis representing the principal cause of death in CRC patients. Current primary treatment options for colorectal cancer liver metastases (CRLM) encompass chemotherapy, radiotherapy, targeted therapy, and immunotherapy. However, the survival rate for CRLM patients remains below 40%. The unique immune-tolerant microenvironment of the liver has recently been recognized as a central driver of metastatic colonization, immune evasion, and therapeutic resistance. This review examines the development of CRLM, with particular emphasis on recent therapeutic advances. It also explores promising research and emerging treatment strategies. A deeper understanding of CRLM pathogenesis and therapeutic progress will enable clinicians to optimize treatment approaches and improve patient outcomes.

Hepatocellular carcinoma (HCC) is a common malignancy and a leading cause of cancer-related mortality. Current guidelines and staging systems provide coarse categories, but often miss within-stage heterogeneity and the clinical context in electronic medical records (EMRs). We present HCC-STAR (Hepatocellular Carcinoma Staging, Treatment And pRognosis), a clinically aligned large language model that reads routine EMR narratives and jointly outputs risk score-based staging, ranked guideline-consistent treatments with evidence-based rationales, and individualized survival estimates. We curated about 30,000 HCC cases from SEER and expanded them into EMR-style narrative training data using a clinician-validated, prompt-based augmentation workflow. On this corpus, we developed a knowledge-aligned reasoning framework optimized with a step-verifiable composite reward, moving beyond text-level memorization of clinical guidelines. In a multi-center cohort of 6,668 patients from 12 hospitals in China, HCC-STAR achieved state-of-the-art performance in treatment recommendation and risk stratification compared with clinical guidelines and competitive models, including GPT-5 and Gemini-2.5 Pro. Hypothetical overall-survival analysis showed a median survival of 51 months under adherence to HCC-STAR recommendations, compared with 29 and 32 months under BCLC and CNLC. In clinician-centric evaluations, blinded hepatobiliary specialists rated HCC-STAR's reasoning and evidence-based justifications as trustworthy. The model surpassed resident and attending physicians in treatment accuracy and helped physicians make more accurate decisions faster when used as an assistant. These findings support HCC-STAR as a reliable and verifiable decision-support system for risk stratification and precision therapy in HCC.

Metastatic brain disease occurs in up to 30% of patients with lung, melanoma and breast cancers, and the median survival time remains less than a year. Treating these patients is a challenge because surgical approaches are limited and most chemotherapeutic drugs and immunotherapies are ineffective at crossing the blood-brain barrier (BBB). Given the unique abilities of macrophages to cross the BBB and exert their phagocytic function on tumour cells, we genetically engineer macrophages that express a chimaeric antigen receptor (CAR) targeting mesothelin (MSLN). To specifically target metastatic brain tumours, we fused the cells with the immune signalling molecule MyD88. This chimaeric antigen receptor macrophage (CARMA) penetrates the BBB and decreases brain metastasis growth in a humanized mouse model. MSLN-CARMA shows antigen-specific phagocytosis activity against tumour cells and exhibits a bystander effect by releasing TNF to act on surrounding tumour cells lacking the tumour antigen. These features of CARMA represent advantages over other immune therapies and CARMA may serve as a promising therapeutic tool for the treatment of brain metastasis.

Table S2 showes the anti-tumor effect of SDP-LIV1 with different DAR values in MCF-7 xenograft model.

Importance:Immune checkpoint inhibitor (ICI) monotherapy for patients with recurrent, unresectable, or metastatic (RUM) head and neck squamous cell carcinoma (HNSCC) is associated with poor outcomes. Low-dose, hypofractionated, quad-shot radiotherapy (QSRT) may enhance the efficacy of ICIs. Objective:To evaluate the efficacy and toxic effects of ICIs in combination with QSRT for patients with RUM HNSCC that is amenable to QSRT. Design, Setting, and Participants:This single-arm, phase 2, nonrandomized clinical trial was conducted at a single, academic, National Cancer Institute-designated Comprehensive Cancer Center and included patients with RUM HNSCC who were going to receive treatment with pembrolizumab monotherapy and had a targetable site in the head and neck that was amenable to QSRT (based on location, prior radiotherapy dose, and tolerance). Patients were enrolled and treated between February 2021 and May 2024. Interventions:Pembrolizumab (200 mg/m2) was administered every 3 weeks until disease progression, intolerance, or a maximum duration of 2 years. QSRT comprised 14.8 Gy in four 3.7-Gy fractions that were delivered twice daily on consecutive days. QSRT courses were delivered cyclically between pembrolizumab cycles. Main Outcomes and Measures:The primary outcome was overall response rate per Response Evaluation Criteria in Solid Tumors, version 1.1, in patients with at least 1 response assessment. Secondary outcomes were progression-free survival (PFS), overall survival (OS), and clinician-rated (Common Terminology Criteria for Adverse Events, version 5.0) and patient-reported (patient-reported outcomes version of the Common Terminology Criteria for Adverse Events, version 1.0) adverse effects (AEs). Results:Of the 29 individuals who were screened, 21 enrolled, and 15 patients (median [range] age, 61 [33-86] years) were eligible for response analysis: 5 women (33.3%), 10 men (66.7%), 9 individuals with oropharyngeal/oral cavity primary tumors, and 14 current/former smokers (93.3%). Fourteen patients had received prior systemic therapy, and 11 had received prior RT. Among evaluable patients, the overall response rate was 47% (95% CI, 21%-73%). Median PFS and OS among the treated patients (n = 20) were 9.2 months (95% CI, 3-17) and 14.9 months (95% CI, 5 to not estimable), respectively. For the 15 patients who were evaluable for the primary end point, median PFS was 11.5 months (95% CI, 5.1 to not estimable), and median OS was 15.3 months (95% CI, 6.3 to not estimable). Grade 3 and 4 AEs occurred for 6 and 2 patients, respectively. Patient-reported AEs did not substantially change over time. Conclusions and Relevance:This nonrandomized trial demonstrated the feasibility of pembrolizumab plus QSRT for patients with RUM HNSCC and provides preliminary evidence of a potential benefit to this novel combination that warrants investigation in future randomized clinical trials. Trial Registration:ClinicalTrials.gov Identifier: NCT04454489.

HLA-C*07:1144 differs from HLA-C*07:02:01:01 by one single nucleotide substitution in exon 2.

Abstract Background: Racial disparities in non-small cell lung cancer (NSCLC) outcomes persist, with Black American (BA) patients having inferior survival compared to White American (WA) patients, even after adjusting for stage, treatment, and socioeconomic factors. We hypothesize that distinct molecular and cellular features in the tumor microenvironment (TME) contribute to ancestry-specific heterogeneity and these disparities. Methods: We performed 10x scRNA-seq on treatment-naïve, early-stage NSCLC from 42 patients: 29 adenocarcinomas (15 WA, 14 BA) and 13 squamous (10 WA, 3 BA). Unsupervised clustering and annotation of 220,591 cells identified six major cell types: epithelial, endothelial, fibroblasts, myeloid, mast, and lymphocytes-each further subdivided into lineage-specific phenotypes. Analyses stratified by ancestry include: differential gene expression and cell abundance, gene set enrichment analysis, trajectory inference, and ligand-receptor analysis. Results: Malignant epithelial cells showed significant pathway enrichment of OXPHOS, fatty acid metabolism, and G2/M checkpoint in BA tumors (q≤0.01) and TNF-α signaling via NF-κB, IFN-γ response, and heme metabolism in WA tumors (q=0.004, 0.008, and 0.02, respectively). Endothelial cells showed enrichment in IL-2/STAT5 signaling and fatty acid metabolism in BA tumors (q=0.01 and 0.03, respectively) and in proliferation pathways in WA tumors (Myc Targets V1, q=0.005).BA and WA tumors exhibited distinctive immunosuppressive features. BA tumors showed higher numbers of FAP- cancer associated fibroblasts (CAF; q=0.07) and M2 macrophages (q=0.004) but low effector CD8+ (p=0.02) and γδ T cells (p=0.018). BA tumor associated macrophages (TAM) showed preserved activation (CD40, CD80) but impaired antigen presentation (HLA-DM/DQ/DR, all q<0.05) and were enriched for hypoxia and TGF-β signaling gene signatures (q<0.01). In contrast, WA tumors contained more FAP+ CAFs and macrophages displayed M0/M1-like features with prominent antigen presentation and IFN-I/II responses (q<0.01). Also, BA tumors showed active innate immune networks, particularly SPP1-CD44 signaling (p<0.001), whereas WA tumors exhibited stronger adaptive stimulatory (MHCII-TCR) and inhibitory (CD86-CTLA4) signaling (p<0.001). Furthermore, CD8+ T cells in BA tumors adopted an early exhaustion phenotype associated with TGF-β signaling (q<0.01), while in WA tumors they showed progressive exhaustion (high PDCD1, HAVCR2, LAG3) in an IFN-α/γ-driven environment (q<1E-5). Conclusions: This first large-scale, ancestry-stratified scRNA-seq atlas of NSCLC reveals molecular mechanisms underlying distinctive immunosuppressive programs in BA and WA patients and highlights molecular and cellular TME features that may contribute to racial disparities in NSCLC. Citation Format: Liang Liu, Pengbo Zhang, Enzo Palma, Ashley Ballard, Lance Miller, Gregory Hawkins, Cristina M. Furdui, Edward Levine, Alberto de Hoyos, Wencheng Li, Fang-Chi Hsu, Ralph D'Agostino, David Foureau, Wei Zhang. Single-cell RNA sequencing reveals ancestry-associated molecular and cellular immunosuppressive features in non-small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 2092.

The presence of diffuse large B-cell lymphoma (DLBCL) cells in bone marrow (BM) smears is recognised as a key morphological basis for diagnosing BM involvement by DLBCL. BM involvement directly affects disease staging, treatment strategies, and prognosis assessment. Nevertheless, traditional manual identification of DLBCL cells in BM smears under a microscope is time-consuming, subject to observer variability, and lacks standardization between hospitals. In this study, we developed an artificial intelligence (AI) detection system (DLBCL-CQnet) based on deep learning, using BM smears from 117 patients with newly diagnosed DLBCL across four centers. DLBCL-CQnet comprises a region of interest (ROI) classification model based on MobileNetV2, and a cell detection and classification model (CDC-MOD) based on YOLOv8. Performance was evaluated using accuracy, precision, recall, F1-score, and mean average precision at an intersection over union threshold of 0.5 (mAP50). MobileNetV2 achieved the best ROI classification performance (accuracy 0.925, precision 0.922, F1-score 0.919), while YOLOv8 yielded a classification accuracy of 0.890 and an mAP50 of 0.934 for DLBCL cell detection. The system achieved an ROI classification accuracy of 0.925, a cell classification accuracy of 0.890, and a mean average precision (mAP50) of 0.934 at an intersection-over-union (IoU) threshold of 0.5. The optimal cutoff for identifying BM infiltration by DLBCL cells was 2.12%, with a specificity of 0.941 and positive predictive value (PPV) of 0.923, demonstrating high diagnostic value. The recognition speed significantly outperforms manual assessment, substantially reducing time consumption. DLBCL-CQnet enables automated, standardized identification of DLBCL cells in BM smears across institutions, serving as a reliable auxiliary diagnostic tool for assessing BM involvement by DLBCL. Its high specificity and operational efficiency streamline diagnostic workflows, especially in resource-constrained environments.

Supplementary Table S1. Summary of public datasets used in this study. Supplementary Table S2. Genes highly expressed in GBM single cells from both peritumoral and HFC areas. Supplementary Table S3. Clinical information of GLASS GBM cohort. Supplementary Table S4. Clinical information of GBM CGGA cohort. Supplementary Table S5. Correlation between C1QL1 expression and GBM recurrence. Supplementary Table S6. GSC cell line information. Supplementary Table S7. DEGs in C1QL1 KO vs Vector GSCs. Supplementary Table S8. GO enrichment for downregulated genes in C1QL1 KO vs Vector GSCs. Supplementary Table S9. DEGs in neurons cocultured with C1QL1 KO vs Vector GSCs. Supplementary Table S10. GO enrichment for DEGs in neurons cocultured with C1QL1 KO vs Vector GSCs. Supplementary Table S11. Marker genes for microenvironmental cells in snRNA-seq. Supplementary Table S12. DEGs among tumor cells from different groups in snRNA-seq. Supplementary Table S13. GO enrichment of DEGs in tumor cells from different groups in snRNA-seq. Supplementary Table S14. Signature genes of vehicle-treated recurrent tumor cells. Supplementary Table S15. DEGs among excitatory neurons from different groups in snRNA-seq. Supplementary Table S16. GO enrichment for DEGs in excitatory neurons from different groups in snRNA-seq.

Autism spectrum disorder (ASD) is a neurodevelopmental disorder, and maternal immune activation (MIA) is highly implicated in neuropathology and ASD-like phenotypes in offspring. However, the underlying regulatory mechanisms of ASD are multifactorial and remain largely unknown in MIA offspring. Here, we performed spatial transcriptome and single-nucleus RNA sequencing (snRNA-seq) analysis in MIA offspring brain to explore the neurobiological features of ASD. We obtained MIA-induced genes and pathways across multiple key brain regions. We found that senescence-associated APP-CD74 pathway, IGFBP7, and CDKN1A may act as the key pathogenic factors for ASD development. Our further analysis identified broad senescence-associated secretory phenotype (SASP) signature of MIA-induced ASD brain that are regulated in a cell-type specific manner. Moreover, we validated that pharmacological inhibition of IGFBP7 and CDKN1A effectively prevents ASD-like behaviors in MIA offspring. Collectively, our data reveal a senescence-associated regulatory mechanism for ASD, and provide potential intervention strategy for the treatment of ASD.

HLA-A*32:181 differs from HLA-A*32:01:01:01 by one single nucleotide substitution at position 500 C>T in exon 3.