The full text of this preprint has been withdrawn by the authors due to author disagreement with the posting of the preprint. Therefore, the authors do not wish this work to be cited as a reference. Questions should be directed to the corresponding author.

Human epidermal growth factor receptor 2 (HER2) is frequently overexpressed or amplified in biliary tract cancer (BTC). Although NCCN clinical practice guidelines recommend HER2-targeted agents as subsequent-line therapy, drug resistance often restricts the clinical benefits of existing regimens. Here, we demonstrate that HER2 inhibitors have heterogeneous effects on the proliferation of BTC cells. Further bioinformatic analysis and functional experimental validation revealed that HER2 inhibitors significantly activated fibroblast growth factor receptor (FGFR) signalling pathway in HER2high BTC. Notably, the combination of lenvatinib and a HER2 inhibitor exerted potent antiproliferative effects on HER2high BTC models both in vitro and in vivo. In the clinical cohort, the combination therapy achieved an objective response rate (ORR) of 58.1% and a disease control rate (DCR) of 86.0%. The median progression-free survival (mPFS) was 11.27 months, and the median overall survival (mOS) reached 19.50 months. For patients with HER2high expression, the ORR and mOS were 71.4% and 27.67 months, respectively. The overall safety profile was manageable, with no treatment-related deaths observed. These findings demonstrate that the activation of FGFR signalling confers resistance to HER2 inhibitors in HER2high BTC. The combination of a HER2 inhibitor with lenvatinib, particularly when combined with immune checkpoint inhibitors, has exhibited both promising antitumour responses and good tolerability in patients with advanced BTC.

Abstract Background and objective Conversion or downstaging therapy for intermediate and advanced unresectable hepatocellular carcinoma (HCC) has emerged as a major focus of clinical practice and research in recent years. This study aims to investigate the current clinical application of conversion therapy in China and determine the factors that physicians consider when selecting eligible patients for conversion therapy and choosing the appropriate conversion modality. Methods Physicians who met predefined inclusion criteria were invited to complete an online questionnaire between January and July 2024. The collected data were subsequently pooled and analyzed descriptively. Results A total of 120 valid questionnaires were gathered, mainly from surgical (n = 83, 69.2%) and interventional (n = 37, 30.8%) departments. The survey revealed that approximately 51% of CNLC stage Ib-IIIa patients were selected for conversion or downstaging treatment. Three primary factors were prioritized by physicians for the determination of conversion therapy: portal vein tumor thrombus (PVTT) type (116/120, 97%), future liver volume (108/120, 90%), and Child–Pugh classification (108/120, 90%). Currently, the predominant conversion therapy approach involves a combination of local and systemic therapies, which is utilized in approximately 73% of cases. Among systemic treatments, the combination of lenvatinib and immunotherapy is the most widely adopted. Besides, a higher objective response rate (ORR) was the foremost consideration for 83% (99/120) of physicians, followed by rapid response (82/120, 68%), adherence to guidelines and consensus (76/120, 63%), and lower tumor progression rate (70/120, 58%). Conclusion This survey demonstrated the current status of conversion therapy for HCC in China. Over half of the newly diagnosed HCC patients were eligible for treatment modalities aimed at achieving surgical resection through conversion therapy, and the most popular indications were the presence of PVTT, insufficient FLR and Child–Pugh classification.

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.

The marked heterogeneity of primary liver cancer (PLC), together with the dynamic influence of the tumor microenvironment (TME), remains a major challenge to precision oncology. As currently available therapies provide limited benefit for many patients and robust preclinical platforms for rapid, patient-specific treatment stratification are lacking, there is an urgent need for models that faithfully preserve individual tumor architecture and cellular diversity. We established three-dimensional bioprinted primary liver cancer (3DP-PLC) models comprising 61 patient-derived monocultures and 34 co-culture models with patient-matched cancer-associated fibroblasts (CAFs). To explore the correlation between ex vivo models and clinical drug response, we developed a clinically anchored hybrid stratification framework for response classification. In parallel, the 3DP-PLC/CAF co-culture biobank was used to investigate stromal regulation of therapeutic response. These 3DP-PLC constructs preserved the histological features, biomarker expression patterns, mutational landscapes, and transcriptomic profiles of their corresponding parental tumors. Drug-sensitivity profiling across the biobank revealed substantial intertumoral heterogeneity in therapeutic responses. The clinically anchored hybrid stratification framework integrating ex vivo pharmaceutical profiles with external clinical benchmark data enabled interpretable response classification and demonstrated clinical relevance in patients receiving neoadjuvant or adjuvant targeted therapy. In parallel, the 3DP-PLC/CAF co-culture biobank, combined with single-cell transcriptomic analysis, recapitulated fibrous ring-like architecture and revealed CAF-associated drug-resistant states. These findings support 3DP-PLC as a high-fidelity and scalable platform for personalized therapeutic stratification and mechanistic investigation of tumor–stroma interactions in PLC.

BACKGROUND:Dual-phenotype hepatocellular carcinoma (DPHCC) is a recently defined subtype of hepatocellular carcinoma (HCC) characterized by the simultaneous hepatocellular and biliary epithelial marker expression. This study aimed to elucidate the clinicopathological features of DPHCC following curative liver resection and its relationship with prognosis. METHODS:We analyzed 1493 patients with HCC who underwent curative liver resection at the Peking Union Medical College Hospital from January 2013 to December 2023. All patients were divided into two groups according to immunohistochemical marker expression, with 487 and 1006 cases in the DPHCC and non-DPHCC groups, respectively. Propensity score matching was performed to reduce the deviation caused by baseline characteristics. RESULTS:After 1:2 matching (DPHCC/non-DPHCC group = 453/771), patients were comparable regarding all baseline characteristics. Compared to patients with non-DPHCC, those with DPHCC were significantly associated with poorer differentiation, microvascular invasion, satellite nodules, and bile duct tumor thrombus (P < 0.05). Patients with DPHCC also exhibited significantly worse recurrence-free survival (P = 0.009) and overall survival (P = 0.021). Furthermore, multivariate analysis revealed that DPHCC was an independent risk factor for recurrence-free survival (HR = 1.28, 95% CI: 1.05-1.52, P = 0.019) and overall survival (HR = 1.33, 95% CI: 1.15-1.51, P = 0.023). CONCLUSIONS:DPHCC, a newly proposed subtype of HCC, is associated with poorer clinicopathological features and adverse prognosis, providing important clinical guidance.

Background Biliary tract cancer (BTC) is an aggressive malignancy with limited treatment options and a poor prognosis. Although immune checkpoint inhibitors combined with chemotherapy have improved patient outcomes, their toxicity remains concerning. This phase II multicenter trial evaluated the efficacy and safety of pembrolizumab plus lenvatinib with a reduced-dose gemcitabine and oxaliplatin (GEMOX) regimen as a first-line therapy for advanced BTC.Methods 60 patients with unresectable or metastatic BTC were enrolled from five centers in China. Patients received pembrolizumab (200 mg, every 3 weeks), lenvatinib (8 or 12 mg daily), and modified GEMOX (gemcitabine 1000 mg/m² and oxaliplatin 85 mg/m² on day 1 of each 3-week cycle) for 6–8 cycles, followed by maintenance with pembrolizumab and lenvatinib. The primary endpoint was objective response rate (ORR), and the secondary endpoints included progression-free survival (PFS), overall survival (OS), and safety.Results At a median follow-up of 16.0 months, the ORR (complete response 5.0%, partial response (PR) 50.0%) and disease control rate were 55.0% and 93.3%, respectively. The median PFS and OS were 12.5 months (95% CI 7.93 to 16.3), and 19.5 months (95% CI 17.97 to not estimable), respectively. Elevated baseline CA19-9 (>37 U/mL) and carcinoembryonic antigen levels (>5 ng/mL) were independently associated with poor OS and PFS, respectively. The regimen showed manageable toxicity, with 95% of patients experiencing treatment-emergent adverse events (AEs), mostly grades 1–2; grade 3–4 AEs occurred in 65% of patients, with no treatment-related deaths. Immune-related AEs occurred in 11.7% of the patients and were predominantly mild.Conclusions Pembrolizumab plus lenvatinib with reduced-dose GEMOX demonstrated promising efficacy and a favorable safety profile in advanced BTC, suggesting that chemotherapy de-escalation may optimize the efficacy–toxicity balance. Further randomized studies are warranted to confirm these findings and refine biomarker-based treatment selections.

4142 Background: Chemotherapy (chemo) combined with or without PD-(L)1 inhibition is the standard first-line treatment for advanced BTC. After first-line treatment fails, patients (pts) lacking targetable genetic alterations (FGFR2, IDH1, or HER2 etc.) have limited options. For the majority, chemo (FOLFOX as the preferred regimen) is the default but suboptimal option. To date, the therapeutic potential of multi-target ICIs and anti-angiogenesis combination in the post-line regimen of BTC remains unexplored and constitutes a significant clinical demand. Additionally, early preclinical and clinical evidences support the combination blockade of TIGIT, TGF-β and PD-(L)1 plus anti-VEGF therapy in several solid tumors. AK130 is a TIGIT/TGF-β bispecific fusion protein. Ivonescimab, the first approved PD-1/VEGF bispecific antibody, has demonstrated promising efficacy in solid tumors, including BTC. Here, we aim to evaluate the efficacy and safety of AK130 plus ivonescimab in post-line BTC and the primary results from phase 1b is reported. Methods: This was an open-label, multi-center phase 1b/2 study. Pts who had progressed on prior chemo combined with or without a PD-(L)1 inhibitor were enrolled and treated with AK130 and ivonescimab. Phase 1b included dose escalation and expansion parts. In the escalation part, escalating doses of AK130 (10, 30, 45 mg/kg Q3W) were administered using a "3+3+3" design, with ivonescimab at 20 mg/kg Q3W (previously approved dose by NMPA). Dose-limiting toxicities (DLTs) were assessed. Based on safety profile, the expansion proceeded at selected doses, with up to 15 evaluable pts per dose. The primary endpoint of phase 1b was the incidence of AEs and DLTs. The secondary endpoint was ORR. Results: As of Jan 2026, a total of 23 pts was enrolled. 12 pts were enrolled in dose escalation across 3 dose levels (10, 30, 45 mg/kg AK130 Q3W; n = 3, 3, 6). No DLTs were observed, and no maximum tolerated dose (MTD) was established. All pts in the 10 mg/kg and 30 mg/kg groups experienced stable disease (SD). The 45mg/kg group was further expanded to 17 pts, including 15 pts that progressed on prior PD-(L)1 inhibition plus chemo.15 pts from 45mg/kg group were efficacy-evaluable with an ORR of 20.0% (3/15) and a DCR of 73.3% (11/15). 87.5% of the SD pts (7/8) reached shrinkage from baseline. 40.0% pts (6/15) had only one time evaluation and are continuing to be followed. Grade 3 or higher treatment-related adverse events (TRAEs) occurred in 39.1% (9/23) pts. The most common TRAEs were anemia and atopic dermatitis, each reported in 2 pts (8.7%). AK130 45 mg/kg Q3W is planned as the recommended phase 2 dose (RP2D), whose safety and efficacy will be further evaluated in phase 2. Conclusions: AK130 plus ivonescimab demonstrated potential anti-tumor activity with a manageable safety profile as a post-line option for advanced BTC, supporting its further development in this setting. Clinical trial information: NCT06938321 .

Cholangiocarcinoma is a heterogeneous biliary malignancy characterized by late diagnosis, stromal desmoplasia, immune exclusion, and frequent resistance to chemotherapy, immunotherapy, and molecularly targeted therapy. Although genomic stratification has enabled biomarker-directed treatment for selected patients, cancer cell-intrinsic alterations alone do not fully explain disease progression or therapeutic failure. Recent advances in single-cell RNA sequencing, single-cell multi-omics, spatial transcriptomics, spatial proteomics, multiplex imaging, and pathomics have redefined the CCA tumor microenvironment as an active ecosystem composed of malignant cholangiocytes, cancer-associated fibroblasts, tumor-associated macrophages, regulatory T cells, exhausted CD8 + T cells, natural killer cells, B cells/plasma cells, endothelial cells, extracellular matrix, and tertiary lymphoid structures. These approaches have identified clinically relevant cellular states and spatial niches, including CD146 + vascular cancer-associated fibroblasts, LGALS1 + fibroblasts, SPP1 + macrophages, MEOX1 + regulatory T cells, POSTN+ fibroblast-rich invasive fronts, and CAF-TAM-vascular interaction units. Major communication axes such as IL-6/IL-6R, TGF-β/TGFBR, SPP1/CD44, CXCL12/CXCR4, PD-1/PD-L1, VEGF/VEGFR, and POSTN/integrins connect cellular heterogeneity with immune escape, invasion, angiogenesis, and therapy resistance. This review synthesizes current single-cell and spatial atlases of CCA and proposes a conceptual translational framework in which TME-derived cellular states, spatial biomarkers, and druggable communication axes may inform patient stratification and rational combination therapy. Because many proposed TME phenotypes and composite biomarkers remain atlas-derived, correlative, or preclinical, this framework should be viewed as a roadmap for prospective validation rather than as a clinically established classification.

Background: Immunotherapy has revolutionized hepatocellular carcinoma (HCC) treatment, increasingly being incorporated into various stages of treatment, including preoperative therapy. However, preoperative immunotherapy alters both physiological status and tumor biology, potentially impacting the perioperative risk. We aimed to explore the association between the implementation details of preoperative immunotherapy and perioperative risk. Methods: This study retrospectively enrolled HCC patients who received immune checkpoint inhibitor (ICI)-based immunotherapy prior to radical surgery. A comprehensive dataset encompassing baseline characteristics and immunotherapy-specific parameters was collected. Univariate and multivariate logistic regression analyses were performed to identify independent risk factors, followed by validation through machine learning algorithms to evaluate the predictive performance of the derived determinants. Results: This multicenter retrospective study enrolled 303 HCC patients who received preoperative immunotherapy across 26 participating centers. Based on specified perioperative risk stratification criteria, 182 patients (60.07%) were stratified into the low-risk cohort, while 121 patients (39.93%) constituted the high-risk cohort. Multivariate logistic regression analysis identified several independent risk factors of elevated perioperative risk: ICI duration exceeding four cycles [odds ratio (OR), 2.00; 95% confidence interval (CI) 1.20-3.32]; the occurrence of grade >= 3 immune-related adverse events (irAEs) (OR, 2.36; 95% CI: 1.13-4.92); the presence of major vein thrombus (OR, 2.05; 95% CI: 1.25-3.38); multiple tumor nodules (OR, 1.74; 95% CI: 1.02-2.98); and maximal tumor diameter >5 cm (OR, 2.02; 95% CI: 1.10-3.73). Subgroup analyses consistently demonstrated these associations. Machine learning algorithms quantified feature importance for perioperative risk predictors, with extended ICIs duration (mean importance rank =2.88) and grade >= 3 irAEs (mean importance rank =2.38) emerging as top-ranked variables. Conclusions: This multicenter retrospective study reveals ICIs duration and high-grade irAEs were associated with elevated perioperative risk in HCC patients undergoing preoperative immunotherapy, providing actionable insights for optimizing perioperative management protocols.

BACKGROUND:Early allograft dysfunction (EAD) is a common complication following liver transplantation. Early identification of high-risk recipients is crucial to improving postoperative outcomes. METHODS:A predictive model for EAD was developed based on a meta-analysis of 22 cohort studies involving 17 582 liver transplantation (LT) donor-recipient pairs. Risk factors and their odds ratios were extracted, and only those statistically significant in the pooled analysis were included in the model. The validation cohort, consisting of both deceased donor liver transplantation and living donor liver transplantation recipients from China, used a manually set cutoff value of 22.5. Risk of bias was assessed using the Newcastle-Ottawa Scale. RESULTS:Significant risk factors included donor after circulatory death, donor age, donor body mass index, cholestatic liver disease, cold ischemia time, intraoperative fresh frozen plasma, and intraoperative red blood cell transfusion. The final model achieved a mean area under the receiver operating characteristic curve of 0.744 in external validation. CONCLUSION:This prediction model reliably estimates the risk of EAD in adult LT recipients and may serve as a practical clinical tool. Cholestatic liver disease may play a role in the development of EAD, and further studies are warranted to validate this association.

Autoimmune liver diseases (AILDs), including primary biliary cholangitis (PBC), primary sclerosing cholangitis (PSC), and autoimmune hepatitis (AIH), pose significant diagnostic and therapeutic challenges due to poorly understood mechanisms. While most studies focus on absolute protein levels, protein–protein ratios (PPRs), reflecting the relative abundance of paired plasma proteins, emerge as critical yet understudied biomarkers for decoding disease-specific network perturbations. To harness this potential, we integrated protein quantitative trait loci (pQTLs), ratio QTLs (rQTLs), and mediation Mendelian randomization (MR) to map causal proteomic networks, aiming to unravel pathogenic networks and identify therapeutic targets in AILDs. Using two-sample MR, we analyzed 2821 plasma PPRs and 2923 individual proteins from the UK Biobank Pharma Proteomics Project. The primary analysis employed the inverse-variance weighted (IVW) method, complemented by MR-Egger regression, weighted median, simple mode, and weighted mode methods, all within a random-effects model. Sensitivity analyses were performed to validate the findings, including Cochran’s Q test, MR-Egger intercept analysis, MR-PRESSO, and Steiger filtering. Cross-trait linkage disequilibrium score regression (LDSC) quantified genetic correlations, while the MR approach based on Bayesian model averaging (MR-BMA) prioritized independent causal PPRs. Two-step mediation MR identified mechanistic pathways. Functional enrichment and protein–protein interaction (PPI) networks were constructed using STRING and the clusterProfiler package. Finally, we investigated the associations between the causal PPRs and AILD-related symptoms/complications, as well as the influence of modifiable lifestyle factors on these PPRs. CD74 exhibited dual roles in AILDs. Elevated plasma CD74 levels were associated with an increased risk of PSC (OR = 1.54, 95

Immune checkpoint inhibitors (ICIs) plus chemotherapy serve as the primary treatment for advanced intrahepatic cholangiocarcinoma (ICC), but they present limited effectiveness. This multicenter study intended to evaluate the feasibility of LRT plus ICIs, chemotherapy, and lenvatinib in patients with advanced ICC. Patients receiving LRT-ICIs-Chemotherapy-Lenvatinib (LRT-ICI-Chemo-Len group, N = 66) or ICIs-Chemotherapy (ICI-Chemo group, N = 81) between April 2020 and May 2025 were enrolled. The study’s outcomes included progression-free survival (PFS), overall survival (OS), tumor response, and adverse events (AEs). The LRT-ICI-Chemo-Len group showed notable enhancement in median PFS (9.9 months vs. 7.6 months, P = 0.004), median OS (20.9 months vs. 13.7 months, P = 0.004), objective response rate (ORR: 59.1

3114 Background: Molecular profiling of cholangiocarcinoma (CCA) is highly recommended to guide access to targeted therapies such as FGFR2 inhibitors. 3HP-2827 is a highly selective and potent FGFR2 inhibitor targeting FGFR2 alterations. Here we report the findings of 3HP-2827 in advanced solid tumors harboring FGFR2 alterations. Methods: This is a phase I/II study to evaluate the safety, tolerability, PK, and preliminary efficacy of 3HP-2827 in pts with advanced solid tumors harboring FGFR2 alterations. Pts failed to standard therapy with/without prior pan-FGFR inhibitor (FGFRi) are eligible for enrollment. Bayesian Optimal Interval (BOIN) and back filling design were employed in the dose-escalation stage of phase I, followed with the expansion stage. Treatment-related adverse events (TRAEs), PK, and anti-tumor activity (RECIST v1.1) were assessed. Results: As of Jan 8, 2026, a total of 52 pts (46 with CCA, 6 with others) were enrolled from 12 centers, received 3HP-2827 at doses of 60-180 mg once daily (QD), including 42 with FGFR2 fusion/rearrangement (f/r), 6 with FGFR2 mutation. 19 pts (36.5%) had prior FGFRi. No dose-limiting toxicities (DLTs) were observed at the dose levels of 60 mg, 120 mg, 180 mg QD. 3HP-2827 had favorable PK with doses≥120 mg QD providing FGFR2 occupancy＞90%. The most common TRAEs were FGFR2 on-target toxicities including dry mouth (69.2%, G3 1.9%), nail toxicity (67.3%, G3 3.8%), stomatitis (50%, G3 3.8%), dry eyes (26.9%, G3 1.9%), PPE (19.2%, G3 3.8%). 12 pts (23.1%) experienced at least one G3/G4 TRAE and no G5 TRAEs occurred. 7 pts (13.5%) experienced at least one serious adverse event, 4 (7.7%) of them were considered related to 3HP-2827. No discontinued treatment or death due to TRAEs. Anti-tumor activities were observed from 60 mg QD. Among 15 CCA pts (FGFR2 f/r, FGFRi-naive) who had at least one post-treatment tumor assessment, objective response rate (ORR) was 80% (95% CI: 51.9, 95.7), and disease control rate (DCR) was 100% (95% CI: 78.2, 100). The ORR and DCR in 15 CCA pts (FGFR2 f/r, FGFRi-refractory) was 26.7% (95% CI: 7.8, 55.1) and 86.7% (95% CI: 59.5, 98.3). The ORR and DCR in 4 advanced solid tumor pts (FGFR2 mutation, FGFRi-naive) was 75% (95% CI: 19.4, 99.4) and 100% (95% CI: 39.8, 100). Conclusions: 3HP-2827 was found to be safe and tolerable, with encouraging anti-tumor activity observed in pts with FGFR2 alterations. Further clinical efficacy will be explored in the expansion stage and phase II stage. Clinical trial information: NCT06378593 .

BACKGROUND:The management of initially unresectable hepatocellular carcinoma (HCC) with portal vein tumor thrombus (PVTT) has been revolutionized by conversion therapy combining locoregional and systemic modalities. While successful downstaging followed by salvage surgery offers potential for cure, the optimal duration of postoperative adjuvant targeted-immunotherapy remains poorly defined. This multicenter study aimed to determine the relationship between adjuvant therapy duration and survival outcomes in this specific patient population. METHODS:We conducted a retrospective cohort analysis of 124 patients with initially unresectable HCC and PVTT who achieved successful conversion using combined local-systemic therapy and subsequently underwent R0 resection at 4 tertiary medical centers between September 2019 and December 2022. Patients were stratified into 4 groups according to adjuvant targeted-immunotherapy duration: no adjuvant therapy (0 month, n = 26), short-term therapy (1-3 months, n = 28), medium-term therapy (4-6 months, n = 25), and extended therapy (≥7 months, n = 45). Primary endpoints were overall survival (OS) and progression-free survival (PFS), analyzed using Kaplan-Meier methods and Cox proportional hazards models. RESULTS:With a median follow-up of 28.3 months, significant differences in both OS and PFS were observed among the 4 groups (both P < .0001). Compared to the no-adjuvant group, all treatment durations showed significant survival benefits, with hazard ratios of 0.49 (P = .004) for OS and 0.44 (P = .010) for PFS in the 1-3 month group, improving to 0.22 (P < .001) for OS and 0.15 (P < .001) for PFS in the 4-6 month group. Critically, the medium-term therapy (4-6 months) demonstrated statistical non-inferiority compared to extended therapy (≥7 months) for both OS (P = .85) and PFS (P = .30), establishing a clear efficacy plateau. Treatment-related adverse events were manageable and comparable across all duration groups. CONCLUSIONS:This study provides compelling evidence that adjuvant targeted-immunotherapy duration significantly impacts survival outcomes in converted unresectable HCC patients with PVTT. The 4-6 months adjuvant regimen represents the optimal therapeutic window, maximizing survival benefits while avoiding unnecessary extended treatment. These findings should inform clinical practice and guide the design of future prospective trials.

BACKGROUND:Most patients with biliary tract cancer (BTC) do not derive durable clinical benefit (DCB) from immune checkpoint inhibitors (ICIs), underscoring the urgent need for predictive biomarkers. While urinary proteomics represents a non-invasive approach for biomarker discovery and mechanism exploration, its utility in ICI-treated patients with cancer remains unexplored. OBJECTIVE:We aimed to establish urinary proteomics as a predictive tool for ICI responsiveness and to elucidate its relationship with tumour dynamics and tumour microenvironment (TME) remodelling in BTC. DESIGN:We performed a staged mass spectrometry (MS)-based discovery-validation proteomics workflow in 211 urine samples from 97 treatment-naïve patients with BTC undergoing ICI-based therapy. A machine learning model was developed based on baseline proteomic features for ICI response prediction. Single-cell transcriptomics of 11 pretreatment tumour biopsies and spatial transcriptomics were integrated to explore the link between urinary proteomics and TME. RESULTS:Patients achieving DCB exhibited enrichment of immune activation and systemic inflammatory pathways, whereas non-durable benefit was correlated with protumourigenic processes. Longitudinal urinary proteomic dynamics could mirror TME remodelling and tumour evolution. A machine learning-derived 4-urinary protein panel (protein tyrosine phosphatase non-receptor 13 (PTPN13), SUB1, MICAL-L1, VARS1) robustly predicted DCB and early responses. Subsequent external validation in an independent cohort (n=24) using parallel reaction monitoring-MS further confirms its generalisability. PTPN13+ malignant cells were identified as key regulators of proapoptotic TME states, contributing to sustained ICI responsiveness. CONCLUSIONS:This study pioneers the application of urinary proteomics in immuno-oncology, providing a non-invasive approach to predict and monitor ICI responsiveness, while offering mechanistic insights into TME dynamics in BTC.

The full text of this preprint has been withdrawn by the authors as it was submitted and made public without the full consent of all the authors. Therefore, the authors do not wish this work to be cited as a reference. Questions should be directed to the corresponding author.

4116 Background: Delayed recognition and fragmented management of immune-related adverse events (irAEs) can trigger unnecessary interruption/discontinuation of immune checkpoint inhibitors (ICIs) and compromise real-world outcomes. We built an asynchronous, message-based multidisciplinary care system embedded in a widely used instant messaging platform to support longitudinal management for advanced biliary tract cancer (BTC) on ICIs, and assessed outcome changes from exploratory to protocolized implementation beyond secular trends. Methods: Single-center retrospective cohort of 546 adults with unresectable locally advanced/metastatic BTC receiving ≥1 ICI dose (Jan 2019–Jan 2024). All patients entered a clinician-moderated closed digital care group with structured triage and multidisciplinary coordination. By ICI start date: WGMS-E (before Jan 1, 2021) vs WGMS-P (on/after Jan 1, 2021). Primary endpoint: overall survival (OS). Secondary: progression-free survival (PFS), irAE recognition/management, and ICI rechallenge. Inverse probability of treatment weighting (IPTW) balanced baseline covariates. Interrupted time-series (ITS) analysis disentangled intervention effect from secular improvements in care. Results: Median follow-up 29.1 months; median OS overall 15.0 months. OS improved in WGMS-P vs WGMS-E (21.3 vs 12.4 months; log-rank P < 0.0001). IPTW-weighted Cox: WGMS-P associated with lower mortality (HR 0.55; 95% CI 0.43–0.69; P < 0.001) and higher 1-/2-year survival (~18%/~19% absolute). PFS also favored WGMS-P (HR 0.72; 95% CI 0.59–0.89; P = 0.0019). Overall irAE incidence was similar, but first irAE recognition occurred earlier in WGMS-P (restricted mean difference −1.68 months within 12 months). Permanent ICI discontinuation due to irAEs decreased (11% vs 22%), while rechallenge increased (10% vs 4%) without higher recurrent irAEs. ITS showed reversal of an increasing pre-intervention mortality trend to a decreasing post-intervention trend. Conclusions: A protocolized, asynchronous, message-based multidisciplinary care system integrated into routine oncology workflows was associated with improved survival among patients with advanced BTC treated with ICIs. By enabling earlier irAE recognition, reducing unnecessary permanent discontinuation, and facilitating safer rechallenge, this low-cost digital intervention may represent a scalable strategy to optimize real-world immunotherapy outcomes beyond secular advances in cancer care. These findings suggest that small, low-cost organizational changes in care delivery may translate into disproportionately large survival gains.