Browse Articles
Discover research articles across all indexed journals
Failure-to-rescue after major complications in endometrial cancer: A National Inpatient analysis (NIS, 2016–2023).
e17637 Background: Although endometrial cancer is often considered low risk, postoperative outcomes may be driven more by hospital rescue capacity than tumor biology. Failure-to-rescue (FTR), defined as death after a major complication, is a key surgical quality metric that remains poorly characterized nationally in gynecologic oncology. This study examines the burden, drivers, and system-level variation in FTR after surgery for endometrial cancer. Methods: A retrospective, survey-weighted analysis of the National Inpatient Sample (2016–2023) was performed. Adult hospitalizations with uterine or endometrial cancer (ICD-10-CM C54–C55) undergoing hysterectomy (ICD-10-PCS 0UT*) were identified, excluding admissions with metastatic disease or palliative care coding. Major complications were defined as sepsis (A40/A41, R65.20/R65.21), shock (R57*), acute kidney injury (N17*), respiratory failure (J96*), or invasive mechanical ventilation (5A1935Z/5A1945Z/5A1955Z). FTR was defined as in-hospital death among hospitalizations with major complications. Survey-weighted analyses and multivariable logistic regression evaluated associations with mortality and FTR, adjusting for demographics, payer, socioeconomic status, hospital characteristics, year, and complication type. Results: The weighted cohort included 96,675 hysterectomy hospitalizations for endometrial cancer nationally. Major postoperative complications occurred in 9.9% of admissions, while overall in-hospital mortality was low (0.22%). Mortality was highly concentrated among patients with major complications, yielding a failure-to-rescue (FTR) rate of 1.83%. The most common complications were acute kidney injury (7.2%), respiratory failure (2.8%), sepsis (1.4%), invasive mechanical ventilation (1.0%), and shock (0.8%). Admissions complicated by major events were associated with substantially greater inpatient utilization, with mean length of stay increasing from 3.3 to 8.4 days. In adjusted analyses restricted to patients with major complications, in-hospital mortality increased with age (adjusted odds ratio [aOR] 1.10 per year, 95% CI 1.04–1.16) and was strongly associated with severe organ failure, including invasive mechanical ventilation (aOR 14.03, 95% CI 2.34–84.26), shock (aOR 9.41, 95% CI 1.87–47.28), sepsis (aOR 8.78, 95% CI 2.61–29.48), and respiratory failure (aOR 8.68, 95% CI 1.55–48.59). Conclusions: Among patients undergoing hysterectomy for endometrial cancer, postoperative mortality is rare but largely attributable to failure-to-rescue after major complications. Variation in FTR by complication severity and hospital characteristics underscores rescue capacity, rather than baseline risk, as the primary driver of outcomes.
A clinical study of a prototype DAA/TAA vaccine targeting MUC1 for immune interception and prevention in ductal carcinoma in situ.
TPS2699 Background: Hypoglycosylated tumor MUC1 is a transmembrane glycoprotein, recognized by human T-cells and antibodies as a tumor-associated antigen. It is overexpressed in premalignant precursor lesions, including ductal carcinoma in situ (DCIS), serving as a potential potent DCIS rejection target. Vaccine-induced immune response to MUC1 may halt DCIS recurrence or progression to invasive disease and offer a future strategy for disease prevention in high-risk individuals. The hypothesis is MUC1 peptide vaccine is safe and immunogenic in patients with DCIS and the elicited systemic immune response will affect changes in the microenvironment of the DCIS from pro- to anti-tumor. Eligibility: Female, 18 years or older with biopsy proven ER+ DCIS with surgery planned as part of definite local therapy. Design: This single institution, open label, randomized phase I clinical trial (NCT06218303) is seeking 50 women with untreated ER+ DCIS confirmed on core needle biopsy (CNB). Patients are randomized 2:1 to the vaccine group. The vaccine is composed of a 100aa long MUC1 peptide corresponding to 5 tandem repeats of 20 amino acids from the MUC1 variable number of tandem repeats region (VNTR), admixed with the poly-ICLC adjuvant Hiltonol. The vaccine group receives the MUC1 peptide vaccine series pre-op (at 0, 2, and 10 weeks) with optional tamoxifen or an aromatase inhibitor (AI). The control group receives only optional tamoxifen or AI. All have surgery at 12 weeks. Research blood is drawn in the vaccine group at baseline and 2 weeks after each vaccine, and in the control group at baseline and at week 12. Tissue from the pre-treatment CNB and the post-treatment surgery are collected. An optional booster is available to vaccine responders 6 months post-surgery. Aims: The primary objective is to assess the immunogenicity of the MUC1 vaccine in ER+ DCIS patients prior to surgery. The secondary objective is to assess the safety and feasibility of the MUC1 vaccine in ER+ DCIS patients prior to surgery. The exploratory objective is to characterize peripheral MUC1-specific effector T-cells, regulatory T-cells and myeloid-derived suppressor cells (MDSC) at baseline and after vaccination. Changes in features of the tumor microenvironment and peripheral immunity at baseline and after vaccination may also be explored. Methods: Sample size/power: Assuming a one-sided type I error α ≤ 0.05 and the rate of patients having a ≥ 2× change in anti-MUC1 IgG 1 is 35% in the vaccine arm and 2% in the control arm, a sample sizes of n=32 and 18 respectively in each arm will yield 88% power. With a dropout of up to 3 patients in each arm, the remaining sample size will still yield power of 82%. Statistical analysis: Fisher’s exact test will be performed at the one-sided α=0.05 for the primary immunogenicity endpoint for comparing the experimental arm to the control arm. Clinical trial information: NCT06218303 .
Combining SBRT with GM-CSF and Peg-IFNα to induce abscopal effects in previously treated patients with metastatic thymic tumors: A single-institution, phase II trial.
8121 Background: After the failure of multi-line treatment, patients with metastatic thymic tumors have a poor prognosis and few therapeutic options. Combining stereotactic body radiotherapy (SBRT) with granulocyte-macrophage colony-stimulating factor (GM-CSF) and Pegylated interferon-α (Peg-IFNα) may induce abscopal effects and improve prognosis. Methods: We conducted this open-label, single-arm, phase Ⅱ trial to evaluate SBRT plus GM-CSF and Peg-IFNα in previously treated patients with metastatic thymic tumors. A 21-day treatment cycle consisted of SBRT delivered to one metastatic lesion with 30 Gy in 5 fractions from day 1, synchronous subcutaneous injection of GM-CSF 125 μg/ m 2 once daily for 14 days, and subcutaneous injection of Peg-IFNα 90 μg on day 8. If the patient has more than two metastatic lesions, another treatment cycle was repeated. After the completion of 1 or 2 treatment cycles, Peg-IFNα therapy was maintained for at least half a year with a subcutaneous injection of 90 μg once a month. The two primary endpoints were the proportion of patients with abscopal effects and the objective response rate (ORR). The secondary endpoints included overall survival (OS), progression-free survival (PFS), and therapeutic safety. Results: A total of 32 patients from March 2021 to December 2025, were enrolled in this trial, with 2 (6.25%) type A thymoma, 4 (12.50%) type B1 thymoma, 6 (18.75%) type B2 thymoma, 2 (6.25%) type B3 thymoma, 16 (50.00%) thymic squamous cell carcinoma and 2 (6.25%) thymic neuroendocrine tumor. One patient with type B3 thymoma died of cardiac arrest amid the COVID-19 pandemic, rendering tumor evaluation unfeasible. Out of the remaining 31 patients, 9 (29.03%) had abscopal effects, and the ORR was 38.71%. At a median follow-up of 17.20 months, the median OS had not been attained yet. The median PFS was 6.37 months for the whole group. We observed that patients with abscopal effects tended to have longer PFS than those without abscopal effects (13.03 vs. 4.40 months； p = 0.003). 5 patients (16.13%) experienced Grade 3 treatment-related adverse events (CTCAE version 5.0), among which cardiac insufficiency compelled 1 patient (3.23%) to drop out of treatment. Conclusions: Combining SBRT with GM-CSF and Peg-IFNα was well tolerated with acceptable toxicity and may represent a promising salvage therapy for previously treated patients with metastatic thymic tumors. The occurrence of abscopal effects is likely to improve patient outcomes. Clinical trial information: NCT04517539 .
Cervical cancer patient outcomes in a safety-net healthcare system.
e17514 Background: Cervical cancer is largely a preventable cancer. Although cervical cancer rates are decreasing in the United States following the introduction of screening and vaccination, incidence remains high among Hispanic and Black populations. Once diagnosed with cancer, enrollment into clinical trials is disproportionately lower for these patient populations. Safety-net health systems disproportionately care for racial/ethnic minorities, positioning them as critical access points for improving clinical trial availability and enrollment . We aimed to characterize patient population with cervical cancer at a large safety-net health system in North Texas. Methods: We identified patients diagnosed with cervical cancer from 2012 to 2023 from JPS Tumor Registry, the institutional registry data of a Comprehensive Community Cancer Program within JPS Health Network and linked to their electronic health records. Kaplan Meier curve was used to assess the difference in survival time across cancer stages from an index date of diagnosis with censoring at date of last contact up to June 18, 2025. This study was approved by the North Texas institutional review board. Results: Our sample comprised of 386 women with a median age at diagnosis of 49 years, 27% were under 40 years. The majority (79%) were squamous cell carcinomas. Non-Hispanic White comprised 39%, Hispanics 37% and non-Hispanic Blacks 20%. Over half (57%) were uninsured, 21% had Medicare/Medicaid, 14% had private insurance, 7% had other coverage. 35% were positive for HPV whereas 41% were not tested. 70% tested negative for HIV, 28% were not tested and 2% tested positive for HIV infection. Stage breakdown is as follows: Stage 1: 123 (32%), 2: 76 (20%), 3: 95 (25%), 4A: 19 (5%), 4B: 60 (16%), only one patient was stage 0 and the remaining were unstaged. 73% of stage 1 patients underwent hysterectomy. 73% of stage 4B patients received systemic treatment. The overall 5-year survival for the cohort was 49% (95% CI: 44%, 55%). Overall survival at 5 years for stage 1 patients is 78% (95% CI: 70%, 87%) as compared to 9.7% (95% CI: 3.8%, 25%) of stage 4B. Conclusions: Despite a high proportion of racial/ethnic minorities, stage distribution of cervical cancer at our safety-net health system is similar to national reported statistics. National statistics report 68% relative survival at 5 years. Our median and stagewise survival is slightly lower than national statistics, this could be due to socio-economic reasons and comorbidities of the patient population seen in safety-net settings since a quarter of advanced stage patients did not receive systemic treatment. Access to necessary care to prevent cervical cancer and promote early detection will continue to be important. Availability of clinical trial options could improve outcomes for patients with advanced stage.
Spino-RT: A multicenter, randomized controlled phase III study of adjuvant radiotherapy versus surveillance in patients with cutaneous squamous cell carcinoma (cSCC) at high risk of recurrence.
TPS9614 Background: cSCC is the second most common skin cancer, predominantly affecting elderly individuals. Most cSCCs are diagnosed at an early stage and cured with surgery. Recurrences are usually detected during routine clinical examinations or imaging (lymph-node ultrasound, cervicothoracic CT scan). In patients considered at high-risk, local recurrence rates may reach up to 30%. To date, no randomized trial has established the optimal management of cSCCs at high risk of recurrence after complete resection, and indications for adjuvant radiotherapy are only based on local practices or retrospective studies showing variable and contradictory results. Methods: This ongoing randomized phase III trial (NCT06692556) compares the efficacy and safety of two commonly used strategies (adjuvant radiotherapy versus surveillance) in patients with cSCC at high risk of recurrence. Key eligibility criteria include age ≥ 18 years, histologically confirmed localized cSCC, complete surgical resection and high-risk of recurrence, defined as: 1) microscopic peri nerval involvement (PNI) with or without one additional risk factor, or 2) the presence of two or three risk factors (excluding microscopic PNI): immunosuppression, tumor diameter > 20 mm, specific location (lip/ear/temple), deep invasion (thickness > 6 mm or invasion beyond the subcutaneous fat), poor differentiation or desmoplasia. The primary endpoint is recurrence free-survival, defined as the time from the date of randomization to the date of first documented relapse (local, regional or metastatic) or the date of death. Secondary endpoints include local and metastatic recurrence-free survival, overall survival, safety and quality of life (EORTC QLQ-C30 and QLQ-ELD14). Exploratory objectives include progression-free survival after treatment for local recurrence in the surveillance arm. A total of 120 events will provide 80% power to detect a significant improvement of the recurrence-free survival in favor of the radiotherapy arm (HR=0.6) at a 2-sided alpha risk of 5%. Considering a two-years recurrence-free survival rate of 60% in the control arm, an accrual period of 18 months, 36 months follow-up for the last patient, and 10% of non-evaluable patients, a total of 266 patients is required (133 par arm). Since the start of recruitment (February 2025), 38 patients have been randomized. 35 French health institutions will contribute to recruitment. Clinical trial information: NCT06692556 .
Scalable knowledge distillation for thyroid cancer diagnosis: ResNet-based ultrasound classification with global deployment feasibility.
e20013 Background: Thyroid nodules are highly prevalent, while thyroid cancer accounts for 7–15% of cases, most commonly papillary thyroid carcinoma. Ultrasound is the first-line diagnostic modality, yet interpretation remains subjective and prone to interobserver variability despite standardized TI-RADS frameworks. Fine-needle aspiration is the diagnostic gold standard but is invasive and frequently overutilized. Although deep learning has demonstrated strong performance for thyroid nodule classification on ultrasound, high-capacity models are computationally intensive and limit clinical deployment. Knowledge distillation enables transfer of diagnostic capability from large mentor networks to lightweight mentee models while preserving accuracy. We evaluated a ResNet-based distillation framework for scalable thyroid cancer classification. Methods: We analyzed 2,247 thyroid nodules from the thyroidAI dataset, including 1,320 malignant and 927 benign nodules with histopathologic confirmation. Ultrasound images were curated and independently annotated by two board-certified radiologists using longitudinal and transverse views, then divided into training, validation, and external testing cohorts. A ResNet101 mentor model, leveraging deep residual connections for hierarchical feature extraction, was trained for benign versus malignant classification. A compact ResNet18 mentee model was trained using temperature-scaled knowledge distillation with regularized cross-entropy loss to transfer probabilistic decision structure while reducing computational complexity. Performance was assessed using accuracy, sensitivity, specificity, F1 score, and AUROC. Deployment feasibility was evaluated by expert reviewers across multiple geographic regions. Results: The ResNet101 mentor achieved high diagnostic accuracy ( > 99%) for malignant thyroid nodules. The distilled ResNet18 preserved clinically meaningful performance, achieving approximately 94% accuracy with balanced sensitivity and specificity and AUROC exceeding 0.94 on validation and external datasets. Knowledge distillation reduced model parameters by over 80% and substantially decreased inference latency, enabling real-time execution on standard hardware. Expert reviewers reported consistent clinical utility for risk stratification and biopsy triage across institutions. Conclusions: Knowledge distillation enables a lightweight ResNet18 model to retain near–ResNet101 diagnostic performance for thyroid cancer classification on ultrasound while markedly reducing computational requirements. This scalable approach addresses key translational barriers and supports global deployment of AI-assisted thyroid nodule risk stratification. Prospective evaluation is warranted to assess impact on biopsy utilization and diagnostic consistency.
Safety and clinical impact of totally implantable venous access ports during trabectedin therapy in advanced L-type sarcomas: A multicenter SEPSaC-01 analysis.
e23564 Background: Totally implantable venous access ports (TIVAPs) are routinely used to secure long-term central venous access for repeated chemotherapy. Nevertheless, device use may be accompanied by mechanical problems as well as catheter-related thrombosis or infection, which can compromise treatment continuity. This issue is particularly relevant for trabectedin (T), delivered as a 24-hour continuous infusion via a central line because of its vesicant potential and the risk of delayed, severe tissue injury if extravasation occurs. We assessed how often port-related adverse events occurred during T therapy and evaluated their consequences for subsequent treatment administration in patients with L-type sarcomas. Methods: We conducted a retrospective analysis of the South-Eastern Poland Sarcoma Collaboration (SEPSaC)-01 database from five oncology centers. Patients with advanced L-type sarcomas who started T between August 2011 and September 2025 and had available information on TIVAP use were eligible. The data cut-off was October 2025. Collected variables included demographics, histopathology, date/indication for port implantation, treatment initiation delay > 2 weeks due to awaiting implantation, port-related complications, port removal, and reimplantation. Data were summarized descriptively as counts/percentages and median (range). Results: A total of 128 patients were included (66.4% women), median age at T start 59.6 years (range 24–78). Histopathology was available for 126 patients: leiomyosarcoma in 78 (60.9%) and liposarcoma in 48 (37.5%). TIVAP was implanted specifically for planned T therapy in 65 patients (50.8%). Waiting for implantation resulted in a > 2-week delay in treatment initiation in 6 patients (4.7%). Overall, 14 patients (10.9%) experienced TIVAP-related complications: infection in 7 (5.5%), occlusion in 2 (1.6%), thrombosis in 1 (0.8%), and other complications in 4 (3.1%). One patient discontinued T due to a port-related complication. Ports were removed because of complications in 10 patients (7.8%), and 3 (2.3%) required reimplantation. Conclusions: In patients with L-type sarcomas receiving T, TIVAPs were associated with a low frequency of clinically meaningful complications and supported safe drug administration. The observed complication (10.9%), removal (7.8%), and reimplantation (2.3%) rates were comparable to, or below, those reported in broader oncology populations using TIVAPs. No extravasation events occurred despite trabectedin’s vesicant properties, consistent with previous reports of rare extravasation when central venous access is used. Infection and thrombosis rates were within expected ranges. Implantation-related delays > 2 weeks were uncommon (4.7%), underscoring the importance of early planning of central venous access to prevent treatment postponements.
Circulating tumor DNA (ctDNA) monitoring of high-risk breast cancer patients: A retrospective descriptive analysis in a community setting.
e12606 Background: Circulating tumor (ct)DNA testing has emerged as a key prognostic biomarker for disease recurrence in breast cancer, with the potential to monitor treatment responses. Changes in ctDNA levels before and after neoadjuvant chemotherapy correlate with residual disease burden. This study investigates ctDNA molecular residual disease (MRD) monitoring in high-risk breast cancer patients. Methods: We conducted a retrospective chart review of electronic health records (EHR) from breast cancer patients at a Stanford satellite site between April 2023 and June 2025. Clinical and demographic data were extracted from the EHR. Tumor-informed ctDNA was assessed using the Signatera 16-plex mPCR-NGS assay. Inclusion criteria required a baseline ctDNA test prior to treatment and at least one follow-up assessment after 3–4 months of systemic treatment. Results: We identified 67 patients with matching ctDNA results, primarily non-Hispanic white (76%), followed by Hispanic (10%), non-Hispanic Black (8%), and Asian (6%). Six percent were male, and ten percent had confirmed germline mutations. Most patients had invasive ductal carcinoma, with some diagnosed with other histologies, including inflammatory breast cancer (6%). High-risk features, such as positive axillary lymph nodes and elevated genomic assay scores, were common. Of the cohort, 20 patients (30%) tested ctDNA-positive at one or more time points. Fourteen of these patients (70%) were ctDNA-positive before treatment. While 13 patients (93%) cleared ctDNA during therapy, one patient who was non-compliant developed oligo-brain metastasis despite clearance. In three patients (15%), ctDNA positivity during therapy prompted changes in management. In one case, rising ctDNA positivity in a stage IV patient led to re-imaging and identification of further disease progression. Two non-compliant patients resumed therapy—one with an aromatase inhibitor and the other with a CDK 4/6 inhibitor. Conclusions: In this real-world cohort, ctDNA testing was applied in various clinical contexts, including surveillance, treatment monitoring, and decision-making. Findings highlight substantial variability in ctDNA application in community practice and emphasize the need for prospective studies to define its clinical utility, optimal use cases, and impact on outcomes in ctDNA-guided breast cancer management. This study underscores the clinical implications of ctDNA monitoring and its potential to influence provider-patient discussions and improve treatment adherence.
Lanreotide for advanced gastroenteropancreatic neuroendocrine tumors (GEP-NETs) in a Korean population: A multicenter prospective observational study (AIM-NETs).
4170 Background: Gastroenteropancreatic neuroendocrine tumors (GEP-NETs) are a heterogeneous group of malignancies. Somatostatin analogues (SSAs) are the standard of care for patients with unresectable or metastatic well-differentiated G1/G2 GEP-NETs. However, most pivotal data including the CLARINET trial were derived from Western cohorts. We prospectively evaluated the efficacy and safety of lanreotide in a real-world Korean cohort to bridge this geographic data gap. Methods: AIM-NETs is a prospective, observational, multicenter study of patients with unresectable or metastatic GEP-NETs treated with lanreotide. Patients received lanreotide 120 mg deep SC every 4 weeks (dose reduction to 90 mg permitted). Tumor response evaluation was done every 8 to 12 weeks according to the RECIST v1.1. The primary endpoint was the 2-year progression-free survival (PFS) rate. Secondary endpoints included median PFS and overall survival (OS), and safety. Results: Between Feb 2021 and Feb 2023, 71 patients were enrolled in five tertiary referral cancer centers in Korea. Median age was 59 years (range 25-82); 53.5% (n = 38) were male. Primary sites included pancreas (n = 40, 56.3%), rectum (8, 11.3%), colon (3, 4.2%) and stomach (3, 4.2%). Only two patients (2.8%) had functioning tumors. Tumor grades were G1 (n = 20, 28.2%), G2 (50, 70.4%) and G3 (1, 1.4%). The 2-year PFS rate was 50.6% (95% CI, 38.2%-63.0%) and median PFS was 26.0 months (95% CI, 15.2-32.7). The 2-year OS rate was 89.9% (82.9%-97.0%); median OS was not reached. The objective response rate was 23.5% (95% CI, 14.1%-35.4%). Higher tumor grade was significantly associated with worse PFS (p < 0.0001) and OS (p < 0.0001). High somatostatin receptor expression (Krenning score of 3 or higher on baseline Ga-68 DOTATOC scan) did not significantly correlate with PFS (p = 0.358) or OS (p = 0.199). No new safety signal of lanreotide were identified. Conclusions: Lanreotide provides robust disease control and a manageable safety profile in Asian patients with advanced GEP-NETs. The observed efficacy, including a favorable ORR, supports lanreotide as a standard of care in this population, aligning with previous clinical trial outcomes. Clinical trial information: NCT04696042 .
Exploiting ribosomal biogenesis dependence through RNA polymerase I inhibition in medulloblastoma.
e14106 Background: Medulloblastoma (MB) is the most common malignant pediatric brain tumor. Despite multimodal therapy, high-risk subgroups such as group 3 MB have 5-year overall survival rates of less than 50%, with frequent relapse and significant long-term treatment-related toxicities. Rapid tumor proliferation requires increased ribosome biogenesis, making RNA polymerase I a potential therapeutic target. CX-5461 is a selective inhibitor of RNA polymerase I that has demonstrated antitumor activity in several adult malignancies, but its therapeutic potential in medulloblastoma has not been fully defined. Methods: Group 3 and SHH medulloblastoma cell lines were treated with increasing concentrations of CX-5461. Cell viability, apoptosis, and cell cycle distribution were assessed using standard biochemical and flow cytometric assays. Nucleolar stress and DNA damage response signaling were evaluated by immunoblotting and imaging-based analyses. Therapeutic efficacy was tested using subcutaneous and intracranial xenograft models. Patient-derived xenograft (PDX) studies are currently ongoing. Results: CX-5461 reduced viability of group 3 and SHH medulloblastoma cells with IC50 values ranging from 75 to 150 nM. Treatment induced significant cell cycle arrest and increased apoptosis, accompanied by nucleolar disruption and activation of DNA damage response pathways. In subcutaneous xenograft models, CX-5461 significantly inhibited tumor growth without evidence of major systemic toxicity. In an orthotopic intracranial model, CX-5461 treatment increased survival by up to three weeks compared with control. PDX studies are currently underway to further evaluate therapeutic efficacy across patient tumors. Conclusions: These studies demonstrate that RNA polymerase I inhibition using CX-5461 has potent antitumor activity in high-risk medulloblastoma models, including survival benefit in an intracranial setting. These findings identify ribosome biogenesis as a therapeutic vulnerability in medulloblastoma and support continued translational development of CX-5461 toward future clinical evaluation in children with high-risk medulloblastoma.
Feasibility of an integrated aging assessment for cancer patients in oncology clinics.
e13593 Background: Older adults living with cancer navigate complex medical, psychosocial, and environmental needs – which are challenging to identify and address in clinical care. Patient-completed screening tools pose an opportunity to proactively uncover geriatric syndromes, high-risk behaviors, and patient concerns. We evaluated the feasibility and potential impact of one such screening tool given prior to an initial oncology visit - the Integrated Aging Assessment questionnaire (IA3). Methods: In a pilot feasibility study, older patients (≥65 years) presenting for a new patient visit in the University of Colorado Breast or Genitourinary oncology clinics completed the IA3. This web-based tool includes validated screening items and provides risk assessment and structured feedback for oncologists and patients across a spectrum of “whole person health” domains: geriatric frailty, health behaviors (diet, activity, sleep, tobacco, alcohol), mental health, and social needs (e.g., transportation). We used co-creation partner engagement strategies to work with patients, clinicians, operational leaders, and staff in developing workflows to implement the IA3. Post-visit questionnaires assessed patient-reported acceptability in terms of incorporation of IA3 into the visit, discussion of patient-prioritized concerns, perceptions of whole-person care; and oncologist-reported utility. Results: Thirteen patients completed the IA3 tool. The most common elevated risks (medium or high) were diet-related concerns (11/13), excessive daytime sleepiness (8/12), and low physical activity (7/12), followed by tobacco or alcohol use (5/13) and mental health concerns (3/13). Few patients screened as medium or high risk for loneliness or social isolation (2/12), financial insecurity (1/12), or housing instability (1/12). Only 1 of the participants screened as potentially frail in the geriatric health screen. Despite low overall risk profiles, all patients screened positive in at least one actionable domain (13/13). Most participants reported that the IA3 was incorporated into the clinic visit (11/12). Specifically, patients reported oncologists considered multiple dimensions of health (11/11) and discussed the patient’s prioritized IA3 issues (12/12). Oncologists indicated that IA3 influenced the care plan in 30.7% of visits and rated the IA3 as modestly helpful to the visit (5.3/10 [SD1.4]). Conclusions: The IA3, a patient completed health assessment, performed prior to an initial visit was acceptable to patients and feasibly integrated into clinic workflows. Oncologists noted modest helpfulness and did not object to its use. Future work may focus on clinic integration as the IA3 has the potential to enhance patient centeredness of oncology visits, and to connect vulnerable older patients with prioritized resources early in their cancer care trajectory. Clinical trial information: NCT05871008 .
Clinical and socioeconomic prognostic factors in chronic lymphocytic leukemia with normal FISH.
e19025 Background: Chronic lymphocytic leukemia (CLL) is the most common leukemia, with an estimated 23,690 new cases in 2025. Fluorescent in situ hybridization (FISH) provides important prognostic information and is routinely incorporated alongside the Rai staging systems to guide risk stratification and treatment decisions. Approximately 20% of patients with CLL have normal FISH results. Outcomes in this subgroup remain heterogeneous, and data are limited regarding the clinical, cytogenetic, and socioeconomic factors that may influence prognosis, including the potential impact of underlying karyotypic complexity not captured by standard FISH panels. Herein, we aim to evaluate clinical and socioeconomic prognostic factors associated with outcomes in patients with CLL who have normal FISH results. Methods: This is a single center retrospective study which included patients over 18 years of age with a diagnosis of CLL and a normal standard FISH panel. Primary end points were time to treatment initiation (TTI) and overall survival (OS). Descriptive statistics were used to summarize baseline characteristics. Time-to-event outcomes (TTI and OS) were estimated using Kaplan–Meier methods. Overall survival was summarized by reporting 5-year survival rates. Univariate and multivariate Cox regression models were used to analyze predictors of TTI and OS. Results: We identified 163 patients with CLL and normal FISH results between 2007-2025. Median age was 66.2 years, 82 were male (50.3%) and majority were non-Hispanic whites (89%), had Medicare insurance (66.2%) and were IgHV mutated (66.2%). With a median follow up of 65 months, the median TTI was 32 months, and 5-year OS was 92.1%. On multivariable analysis, older age (HR 1.06, 95 % CI 1.03-1.08, p<0.001), commercial insurance (HR 3.23, 95% CI 1.91-5.50, p <0.001), Medicaid (HR 5.58 95% CI 1.90-16.39, p < 0.002), Other/non-insured (HR 3.86, 95% CI 2.11-7.05, p<0.001) were associated with decreased survival. While older age (HR 1.04, 95% CI 1.02-1.06, p<0.0001), commercial insurance (HR 2.2, 95% CI 1.32-3.68, p=0.0025), Medicaid (HR 4.36, 95% CI 1.52-12.52, p=0.0063) and other/uninsured (HR 2.11, 95% CI 1.17-3.82, p=0.0137) were associated with shorter time to treatment initiation. IgHV status did not affect survival (HR 1.08, CI 0.76-1.52, p=0.67) or TTI (HR 0.93, 95% CI 0.66-1.31, p=0.67). Conclusions: In CLL patients with normal FISH, age and insurance status were significantly associated with survival and time to treatment initiation, highlighting the importance of demographic and socioeconomic determinants of outcome. In contrast, IgHV status did not demonstrate prognostic significance in this subgroup in our cohort.
Characterization and management of gastrointestinal (GI) adverse events (AEs) with zanidatamab + chemotherapy (CT) ± tislelizumab in first-line (1L) HER2-positive (HER2+) locally advanced or metastatic gastroesophageal adenocarcinoma (mGEA): Analysis from HERIZON-GEA-01.
4042 Background: In HERIZON-GEA-01, replacing 1L trastuzumab (tras) + CT with zanidatamab + CT ± tislelizumab significantly improved progression-free survival and, with tislelizumab, yielded a statistically significant overall survival benefit in HER2+ mGEA. The safety profile was manageable; diarrhea was the most common AE. Here we further characterize GI AEs and diarrhea management. Methods: Eligible patients (pts) with previously untreated HER2+ mGEA were randomized 1:1:1 to zanidatamab (1800 mg [<70 kg]/2400 mg [≥70 kg] IV Q3W) + tislelizumab (200 mg IV Q3W) + capecitabine/oxaliplatin (CAPOX) or 5-FU/cisplatin (FP); zanidatamab + CAPOX or FP; or tras + CAPOX or FP. CT could be discontinued per physician preference after cycle 6. Tras dose reductions were not permitted. Diarrhea prophylaxis (loperamide 4 mg orally BID) was mandatory in zanidatamab-containing arms for the first 7 days of cycle 1. Results: Median treatment duration was 43.1 wk with zanidatamab + tislelizumab + CT, 31.0 with zanidatamab + CT, and 30.0 with tras + CT. Median number of CT cycles was 6, 6, and 7, respectively. Diarrhea was the most common GI AE in all arms; other GI AEs were generally similar across arms. Among pts who experienced diarrhea, most had first onset in cycle 1 (76% in ≤3 wk) with median duration <2.5 wk (Table). Few pts had their first onset of diarrhea occur after cycle 6 when pts could discontinue CT. HER2-targeted therapy discontinuations (4.1%, 1.3%, 0.3%, respectively) and dose reductions (9.9%, 10.8%, NA) or delays (13.6%, 13.4%, 7.6%) due to diarrhea were infrequent. Diarrhea was more often managed with CT dose reduction (21.8%, 23.6%, 14.2%, respectively). Immune-mediated colitis occurred in 2.7% of pts with zanidatamab + tislelizumab + CT. Additional incidence and management data will be presented. Conclusions: In zanidatamab-treated pts, most diarrhea events were grade 1/2, and first-onset events tended to occur in cycle 1 and resolved in <3 wk. Diarrhea rarely led to zanidatamab discontinuation. The safety of zanidatamab-containing regimens appears favorable given the survival benefits; diarrhea should be managed with prophylactic loperamide and CT dose modifications as needed. Clinical trial information: NCT05152147 . Diarrhea Zanidatamab + Tislelizumab + CT n = 294 Zanidatamab + CT n = 305 Tras + CTn = 302 Any-grade, n (%) 244 (83.0) 241 (79.0) a 161 (53.3) Grade 1 79 (26.9) 80 (26.2) 84 (27.8) Grade 2 92 (31.3) 99 (32.5) 38 (12.6) Grade ≥3 73 (24.8) 61 (20.0) 39 (12.9) Time to first onset, n (%), wk ≤3 188 (77.0) 192 (79.7) 108 (67.1) >3 to ≤6 25 (10.2) 27 (11.2) 22 (13.7) >6 to ≤9 7 (2.9) 14 (5.8) 11 (6.8) >9 to ≤12 4 (1.6) 4 (1.7) 2 (1.2) >12 to ≤18 8 (3.3) 3 (1.2) 7 (4.3) >18 12 (4.9) 1 (0.4) 11 (6.8) Duration of first onset, median (95% CI), wk 2.0 (1.6, 2.6) 2.4 (1.9, 2.9) 1.4 (1.0, 2.1) a Grade missing for 1 pt.
First-in-class Trop2-targeted PET imaging with <sup>68</sup> Ga-MY6349 for diagnostic accuracy and altered management in recurrent/metastatic thyroid cancer compared to standard <sup>18</sup> F-FDG PET/CT.
6041 Background: Accurate, non-invasive detection of recurrent and metastatic thyroid cancer remains an unmet clinical need due to the limitations of conventional imaging and biopsy. Trop2 is a tumor-associated antigen that frequently overexpressed in thyroid cancer, it represents a promising molecular target for imaging. This study aimed to evaluate the diagnostic accuracy of Trop2-targeted PET/CT using the novel nanobody tracer, 68 Ga-MY6349, the results were compared with those of 18 F-FDG PET/CT. Methods: In this prospective, single-center, single-arm trial conducted at the First Affiliated Hospital of Xiamen University (Xiamen, China), adults with suspected or confirmed recurrent or metastatic thyroid cancer were enrolled between January and December 2024. Each participant underwent both 68 Ga-MY6349 and 18 F-FDG PET/CT scans within one week. Three independent, blinded readers assessed the PET/CT images. The primary endpoints were patient-based sensitivity and specificity, using histopathology or clinical follow-up as the reference standard. The full analysis set included patients with evaluable PET/CT imaging and a confirmed final diagnosis. The trial is registered with ClinicalTrials.gov, NCT06465017 and is closed to enrollment. Results: Of 161 screened participants, 143 were finally included in the primary analysis. Papillary thyroid cancer (PTC) was the most common pathological subtype (115/143, 80%). The median clinical follow-up duration was 19 months (IQR: 14-22 months). In the overall cohort, the sensitivity and specificity of 68 Ga-MY6349 PET/CT was 90% (95% CI 83-94) and 91% (95% CI 76-97), respectively. Among participants with PTC, its sensitivity and specificity were 94% (95% CI, 87–98) and 96% (95% CI, 79–100), respectively. Additionally, 68 Ga-MY6349 PET/CT demonstrated superior diagnostic accuracy in participants with thyroglobulin-elevated negative iodine scintigraphy (TENIS), with sensitivity of 92% (95% CI 82-97) and accuracy of 93% (95% CI 84-97), respectively. Further details are presenting in the Table. No grade 2 or higher adverse event was observed during study. Conclusions: 68 Ga-MY6349 PET/CT is a safe and highly accurate imaging modality for detecting recurrent and metastatic thyroid cancer, particularly in PTC and patients with TENIS. These findings suggest it could significantly influence clinical practice and may become a standard imaging option in this setting. Further multicenter studies are warranted to validate its diagnostic accuracy and assess its long-term impact on patient management and outcomes. Clinical trial information: NCT06465017 .
Later-line treatment of PD-1 and CTLA-4 blockade combined with liposomal irinotecan plus leucovorin and fluorouracil for advanced biliary tract cancer: A prospective, phase 2 trial.
e16178 Background: For advanced biliary tract cancer (BTC) patients who failed first-line chemotherapy, second-line options are very limited, especially for those without specific genetic mutations. This study aimed to evaluate the efficacy and safety of the combination of cadonilimab with liposomal irinotecan plus fluorouracil and leucovorin for second or more line treatment of advanced BTC. Methods: This is an interim analysis (data cutoff: December 26, 2025) of a prospective, dual-cohort, phase II trial. Fifty-one locally advanced or metastatic BTC patients who failed at least first-line systemic gemcitabine based chemotherapy with or without PD-L1/PD-1 inhibitor (Cohort 1, prior ICI exposure and Cohort 2, ICI-naïve), were planned to be enrolled. Eligible participants will receive cadonilimab at a dosage of 6 mg/kg combined with intravenous liposomal irinotecan at a dosage of 70 mg/m 2 plus leucovorin at a dosage of 400 mg/m 2 and fluorouracil at a dosage of 2400 mg/m 2 for 46 h every 2 weeks, up to 12 cycles. Patients will receive maintenance cadonilimab up to 24 months if disease stable or remission after combination therapy.The primary endpoint is overall response rate (ORR). The secondary endpoints include overall survival (OS), progression-free survival (PFS), adverse event (AE) incidence rate. Results: Between August 30, 2024 and December 26, 2025, a total of 24 patients were enrolled. Of these,17 patients (Cohort 1: n = 10; Cohort 2: n = 7) with tumor assessment were included in this analysis. The median follow-up was 10.9 months (range, 1.9-11.4). In Cohort 1, the ORR was 30.0% (3/10) and the DCR was 60.0% (6/10). In Cohort 2, the ORR was 28.6% (2/7) and the DCR was 71.4% (5/7). For the overall population (n = 17), the ORR was 29.4% and DCR was 64.7%. The median progression-free survival (PFS) for the combined population was 4.7 months (95% CI, 2.2-NA). By cohort, median PFS was 3.0 months (95% CI, 2.2-NA) in Cohort 1 and was not reached in Cohort 2. The median overall survival (OS) for the combined population was 9.1 months (95% CI, 7.5-NA). By cohort, median OS was 7.7 months (95% CI, 6.4-NA) in Cohort 1 and was not reached in Cohort 2. Treatment-related adverse events (TRAEs) occurred in 100% of the 17 assessed patients, with Grade ≥3 TRAEs reported in 58.8%. The most common Grade ≥3 TRAEs were leukopenia (5/17, 29.4%), neutropenia (5/17, 29.4%), and anemia (2/17, 11.8%). Immune-related adverse events (irAEs) were noted in 52.9% of patients, including rash (5/17, 29.4%), hypothyroidism (3/17, 17.6%), and cardiac events (2/17, 11.8%). Conclusions: In this interim analysis, the combination of cadonilimab with liposomal irinotecan plus fluorouracil and leucovorin shows preliminary signs of efficacy and tolerable toxicity for later-line treatment in advanced BTC. These findings warrant further investigation in larger cohorts. Clinical trial information: NCT06438822 .
Comparison of prophylactic effects for chemotherapy-induced neutropenia between same-day versus next-day administration of pegteograstim in chemotherapy regimen composed of day 1 intensive myleosuppressive agent: A randomized clinical trial (CONCISE, KCSG PC22-11).
12126 Background: Same-day administration of pegylated G-CSF is frequently used in clinical practice despite guideline recommendations favoring next-day administration. This study evaluated whether same-day administration of pegteograstim is noninferior to next-day administration with respect to the duration of grade (Gr) 4 chemotherapy induced neutropenia (CIN) in multi-day regimens containing day 1-intensive myelosuppressive agents. Methods: This multicenter, open-label, randomized, noninferiority trial (originally designed as a phase III study) enrolled patients receiving adjuvant/neoadjuvant or first-line palliative chemotherapy, including mFOLFIRINOX, ECb, EP, FOLFIRI, and FOLFOX. Patients were assigned in a 1:1 ratio to receive pegteograstim 6mg either within 4 hours after completion of chemotherapy (same-day group) or 24-36 hours (next-day group). The primary endpoint was Gr 4 CIN duration during cycle 1 (C1). A total sample size of 160 patients was calculated to verify noninferiority with a margin of 0.6 days; however, the trial closed prematurely after enrolling 90 patients. Results: Of the 90 enrolled patients, 81 were included in the analyses (37 same-day, 44 next-day). Mean Gr4 CIN duration was 0.43 days (95% CI, 0.13–0.74) in the same-day group and 0.09 days (95% CI, 0.00–0.29) in the Next-day group. The mean difference (Same-day minus Next-day) was 0.342 days (90% CI, 0.037 - 0.646). Because the upper bound exceeded the prespecified noninferiority margin of 0.6 days, noninferiority was not demonstrated (p-value = 0.080 for noninferiority). The incidence of Gr4 CIN was 21.1% (8 patients) of same-day group and 2.3% (1 patient) in the next-day group. During cycles 1-4, febrile neutropenia occurred in 1 patient (2.6%) in the same-day group and 3 patients (6.8%) in the next-day group. Two deaths occurred in each group: one FN-related death occurred in the same-day group. Conclusions: Same-day administration of pegteograstim did not meet the prespecified criterion for noninferiority compared with next-day administration. The incidence of grade 4 CIN, a key secondary outcome, favored next-day administration. These randomized data support maintaining next-day pegteograstim as standard practice in multi-day regimens with day 1–intensive myelosuppression. Clinical trial information: KCT0007694.
The role of GPD1L in colorectal cancer progression: A systematic review of its expression, molecular mechanisms, and prognostic significance.
e15728 Background: Colorectal cancer (CRC) continues to be one of the leading causes of cancer-associated morbidity and mortality globally. Despite advances in screening and treatment, many patients develop advanced disease marked by invasion, metastasis, and poor long-term outcomes. Increasing evidence suggests that metabolic dysregulation and hypoxia-related signaling play important roles in CRC progression. Glycerol-3-phosphate dehydrogenase 1-like (GPD1L)–a metabolism-related gene related to redox homeostasis signaling in cells, has also been proposed to be a targeted tumor suppressor in several tumors. However, its biological and clinical relevance in CRC has not been comprehensively evaluated. Methods: We conducted a systematic review of literature through January 2026 on PubMed, Scopus and Web of Science. Studies evaluating GPD1L expression, functional effects, or prognostic significance in CRC using bioinformatic datasets, clinical samples, or experimental models were included. We extracted data from studies on study design, datasets or cohorts, molecular mechanisms, and clinical outcomes, and findings were synthesized descriptively. Results: Eight studies were included, comprising over 1,500 CRC tumors and more than 150 normal colorectal samples from TCGA, GEO, and institutional cohorts. All studies demonstrated significantly lower GPD1L expression in CRC compared with normal tissue ( p < 0.01). Low GPD1L expression was consistently associated with advanced TNM stage and lymph node metastasis ( p < 0.05). Survival analyses showed that reduced GPD1L expression predicted worse overall survival, with reported hazard ratios ranging from 1.6 to 2.4, and poorer recurrence-free survival (HR 1.5–2.1), remaining significant after multivariable adjustment ( p ≤0.01). Functional studies revealed that GPD1L suppression increased CRC cell proliferation, migration, and invasion, whereas GPD1L overexpression reduced invasive capacity by approximately 40–55% ( p < 0.01). Mechanistically, GPD1L loss was linked to increased HIF-1α stability, upregulation of MMP9, and activation of metabolic and hypoxia-related pathways. Conclusions: GPD1L is consistently downregulated in colorectal cancer and is associated with aggressive disease features and significantly worse survival outcomes. Functional evidence supports a tumor-suppressive role for GPD1L through metabolic and hypoxia-driven mechanisms. These findings support GPD1L as a promising prognostic biomarker and potential therapeutic target in CRC.
A systematic review and meta-analysis on safety and efficacy of Lu-177 PSMA-617 and standardized treatment in mCRPC from randomized phase II/III trials.
e17069 Background: Lu-177 PSMA-617 delivers targeted β-radiation to prostate cancer cells, offering a distinct mechanism beyond androgen-signaling inhibition in metastatic castration-resistant prostate cancer (mCRPC). Pivotal trials have shown improved progression-free survival with favorable tolerability. As its clinical use expands, comparative evidence with established systemic therapies remains limited. This meta-analysis evaluates the pooled efficacy and safety outcomes. Methods: A systematic review and meta-analysis were performed according to PRISMA guidelines to evaluate Lu-177 PSMA-617 in mCRPC. Of 456 studies screened, seven met eligibility criteria and were included in the final analysis. Data on progression-free survival (PFS), overall survival (OS), and grade ≥3 adverse events (AEs) were extracted and pooled using a random-effects model (REML method). Hazard ratios (HRs) and risk ratios (RRs) with 95% confidence intervals (CIs) were calculated, and heterogeneity was assessed using the I² statistic, and publication bias was evaluated using both Egger’s regression and Begg’s rank correlation. Results: A total of 2,526 patients from seven randomized trials were analyzed, including 1,365 in the Lu-177 PSMA-617 arms and 1,161 in the standard-of-care arms. Lu-177 PSMA-617 significantly improved PFS compared with control (pooled HR = 0.64; 95% CI 0.50–0.81; p < 0.001). No significant difference was observed in OS (HR = 0.91; 95% CI 0.66–1.25; p = 0.55). The pooled RR for grade ≥ 3 AEs was 0.98 (95% CI 0.83–1.14; p = 0.75). Conclusions: Across contemporary randomized trials, Lu-177 PSMA-617 consistently prolongs PFS with no significant difference in AEs. OS benefit remains unconfirmed, likely reflecting post-progression and salvage use of Lu-177 PSMA-617 in control arms in some trials, as well as disease and trial heterogeneity. The data collectively reinforce Lu-177 PSMA-617 as a well-tolerated, effective radioligand option for advanced mCRPC, warranting continued integration into earlier disease settings and combination strategy. Pooled Overall Survival comparing Lu-177 PSMA-617 and standard of care. Study HR OS for Lu-177 PSMA-617 vs standard PSMAFore 0.98 (0.75-1.28) SPLASH 1.11(0.73-1.69) VISION 0.62 (0.52-0.74) TheraP 0.97 (0.70-1.35) Enza-P 0.55 (0.36-0.84) CCT Group 1.64 (1.14-2.35) Pooled Results 0.91 (0.66-1.25)
Comparative real-world safety of nivolumab versus pembrolizumab in malignant melanoma: A propensity-matched cohort study.
e21587 Background: Nivolumab and pembrolizumab are widely used programmed cell death-1 (PD-1) inhibitors approved for the treatment of malignant melanoma. Although both agents demonstrate comparable efficacy, real-world comparative safety data between these two therapies remain limited. We conducted a real-world cohort study to compare immune-related adverse events between nivolumab and pembrolizumab in patients with melanoma. Methods: We performed a retrospective cohort study using the TriNetX Global Collaborative Network, including adult patients (≥18 years) diagnosed with malignant melanoma treated with either nivolumab or pembrolizumab between January 2015 and December 2025. Patients receiving other immune checkpoint inhibitors were excluded. Propensity score matching (1:1) was used to balance baseline demographics and comorbidities, yielding 6,028 patients in each cohort. Outcomes assessed within three years after treatment initiation included pneumonia, inflammatory liver disease, autoimmune thyroiditis, type 1 diabetes mellitus, colitis, myocarditis, and dermatitis/eczema. Risk ratios, hazard ratios, Kaplan–Meier survival analyses, and number of event instances were evaluated. Results: After matching, baseline demographic and clinical characteristics were well balanced between the two cohorts. Nivolumab was associated with a significantly lower risk of pneumonia compared with pembrolizumab (risk ratio [RR] 0.89, 95% CI 0.81–0.99; hazard ratio [HR] 0.89, 95% CI 0.80–0.99). Rates of inflammatory liver disease, autoimmune thyroiditis, type 1 diabetes mellitus, colitis, and dermatitis were similar between groups, with no statistically significant differences observed. However, nivolumab was associated with a significantly higher risk of myocarditis compared with pembrolizumab (RR 2.31, 95% CI 1.29–4.15). Overall burden of immune-related adverse events was lower with nivolumab, with a reduced cumulative incidence of any immune-related adverse event (RR 0.93, 95% CI 0.87–0.99; HR 0.92, 95% CI 0.85–0.99). Conclusions: In this large real-world propensity-matched cohort of patients with malignant melanoma, nivolumab demonstrated a comparable overall safety profile to pembrolizumab, with lower risks of pneumonia and overall immune-related adverse events, but a higher risk of myocarditis. These findings provide clinically relevant real-world evidence to support individualized PD-1 inhibitor selection in melanoma.
Longitudinal immune programs and association with toxicity burden and antitumor response during immunotherapy.
2541 Background: Immune-related adverse events (irAEs) and tumor responses often co-occur during immune checkpoint inhibitor (ICI) therapy. We investigated whether blood immune programs associated with toxicity can be temporally and biologically dissociated from response programs. Methods: ICI-naïve patients with advanced solid tumors (n=37; enriched for melanoma and NSCLC) were profiled at baseline (T0), early on-treatment (T1; week 4), and later timepoints (T2–T3); patients with irAEs had an additional sample at onset (Ttox) before immunosuppression. PBMCs were analyzed by multiparameter flow cytometry and plasma cytokines by 48-plex multiplex assay (log2). We compared timepoints and delta windows (ΔT1–ΔT3), controlled multiple testing with Benjamini–Hochberg (q<0.1), and evaluated baseline predictors from T0 alongside landmark Cox models from T1 for ΔT1 predictors to avoid immortal-time bias; baseline multiple-irAE signatures were tested in penalized multivariable models with clinical covariates. Results: irAEs occurred in 20/37 (54%) patients (grade ≥3: 8/37, 22%); multiple irAEs occurred in 14/37 (38%). Objective response occurred in 15/37 (41%), and 10/15 (67%) responders developed irAEs; median time to first irAE was 86 days (IQR 63–108). Baseline multiple-irAE susceptibility centered on Tfh states (higher Tfh1/Tfh17 PD1+ICOS− and lower Tfh1 PD1−ICOS+), and a MultiTox signature predicted multiple irAEs (OR 5.46; p=0.033). Early dynamics strengthened toxicity prediction: ΔT1 Tfh2 PD1+ICOS− decreased in AnyTox/MultipleTox and predicted subsequent irAEs (HR 0.41; p=0.0019), whereas ΔT1 PDGF-BB (log2) increased risk (HR 2.25; p=0.006). Approaching onset, activated regulatory compartments (activated Treg and Tfr) contracted, followed at Ttox by a surge in IFN-inducible CXCR3 chemokines (CXCL9/MIG p=3.1×10⁻⁵; CXCL10/IP-10 p=0.0021), consistent with a Th1/IFN axis. In contrast, response-associated programs emerged later and reflected a Tfh/B cell–activated CD8 axis (memory B at T1 p=0.036; ΔT2 Tfh2 PD1−ICOS− q<0.1, p=0.0010; ΔT2 eosinophils q<0.1, p=0.0071) and remained independently associated with response (p≈0.02–0.03). Conclusions: Toxicity—particularly high-burden toxicity—appears to reflect a baseline susceptibility that is amplified by early on-treatment immune trajectories and culminates in a chemokine-rich onset state. In contrast, response-associated programs emerge later and are at least partly dissociable from high-burden toxicity. Distinct Tfh states—especially early Tfh2 dynamics—may support timing-informed monitoring and improved benefit–risk stratification during ICI therapy.