Browse Articles
Discover research articles across all indexed journals
Assessment of long-term persistence and phenotypic features of BCMA CAR-T cells in patients with relapsed or refractory multiple myeloma treated with ciltacabtagene autoleucel.
e19521 Background: Ciltacabtagene autoleucel (Carvykti, cilta-cel) is a BCMA-directed CAR-T associated with deep and durable responses in relapsed or refractory multiple myeloma (RRMM). CAR-T cells rapidly expand, peaking 10–14 days post-infusion, after which circulating levels decline and frequently fall below the detection limits (LoD) of spectral flow cytometry (SFC) and next-generation sequencing (NGS). While early expansion correlates with initial response, the relationship between remission durability, and long-term persistence and state of therapeutic cells after response remains poorly understood. This retrospective study evaluated long-term circulating persistence and phenotypic characteristics of CAR-T cells in RRMM patients treated with cilta-cel using CB-Scout, a novel AI-driven nanophotonic imaging ultra-sensitive single-cell assay (Limit of Blank (LoB) < 4.7×10⁻⁸; LoD < 6.1×10⁻⁷). Methods: Five RRMM patients treated with cilta-cel at Yale Cancer Center with available pre-infusion and at least 2 post-infusion (range 2-8) cryopreserved peripheral blood mononuclear cell (PBMC) samples were included in this study. All 5 patients had at least a very good partial response to therapy. Median duration of follow up was 211 days (17-883) with 3 patients having at least 6 months of follow up (211-883 days). CAR-T cells were evaluated using CB-Scout (CellsBin; Vega 1.2 system) and compared to SFC. A shared immunophenotyping panel included CAR detection, memory/stemness markers (CD45RA, CCR7, CD27, CD28, CD95, CD57), activation (CD25, 4-1BB), and exhaustion markers (PD-1, TIGIT, LAG-3, CD39). Results: No CAR-T cells were detected in pre-infusion samples, consistent with the CB-Scout LoB. During early post-infusion time points, median peak CAR-T expansion was 1.3% of PBMCs, concordant with SFC, including ≥80% agreement across key phenotypic metrics. By Day 33, CAR-T frequencies declined below 0.01% of PBMCs in 4/5 patients, falling below SFC reliability limits. In patients with long-term follow-up, CAR-T cells were detected at rare frequencies (~0.0001%–0.01% of PBMCs) using CB-Scout but not by SFC and despite low abundance, phenotypic characterization remained feasible. Detailed CAR-T phenotypic depiction of all 5 patients will be shared during the meeting. Conclusions: Longitudinal cell-therapy tracking with CB-Scout enables assessment of CAR-T phenotypic persistence beyond SFC and NGS. The ability to quantify and phenotypically characterize CAR-T cells at extended time points post infusion may be useful in predicting remission durability and guiding subsequent therapy selection.
Modulation of PD-1 and PD-L1 expression on peripheral blood T-cells by novel rotavirus variants.
2658 Background: Promising oncolytic viruses are typically evaluated based on their direct cytotoxic effects on cancer cells. However, the therapeutic benefit of virotherapy may also stem from its immunomodulatory actions, particularly through influencing immune checkpoint molecules like PD-1 and PD-L1. This study aimed to investigate the effect of unclassified apathogenic rotavirus strains RVK100 and RVK228 on the expression of PD-1 and PD-L1 on T-cells derived from the peripheral blood of patients with breast cancer. Methods: Mononuclear cells of peripheral blood were isolated on a ficoll gradient, cultured in RPMI 1640 (Gibco, USA) without serum at 37 °C, 5.0% CO2 in 4 variants of the experiment: 1) negative control without viruses; 2) positive control of activation with the addition of PHA; 3) experience with the addition of 107 particles per 1 ml of the RVK100 strain; 4) experience with the addition of 107 particles per 1 ml of the RVK100 strain RVK228. After 24 and 72 hours of cultivation, the expression of PD-1 (CD279) and PD-L1 (CD274) was determined on T cells by flow cytometry. The study used antibodies conjugated with fluorochromes: anti-CD4 (PE), anti-CD8 (APC Cy7), anti-CD279 (FITC), anti-CD274 (PerCP-Cy5–5) (Becton Dickinson, USA). Results: After 24 hours, we observed increases in PD-1 expression on CD4+ (PHA—40.5%, RVK100—42.3%, RVK228—37.5%; vs. control—18.1%) and CD8+ cells (PHA—41.7%, RVK100—46.4%, RVK228—42.6%; vs. control—27.7%). Similarly, PD-L1 expression rose on CD4+ cells (RVK100—67.0%, RVK228—58.6%, PHA—75.1%; vs. control—44.8%) and CD8+ cells (RVK100—63.4%, RVK228—58.4%, PHA—52.8%; vs. control—46.2%). At 72 hours, PD-1 levels decreased significantly in CD4+ cells exposed to RVK100 (from 42.3% to 21.6%) and CD8+ cells (from 46.4% to 17.4%). Conversely, PD-L1 expression increased across all groups on CD8+ cells, reaching 67–79%, while minimal change occurred on CD4+ cells except for a minor decline in the RVK100 group. Conclusions: Both strains, like the non-specific T-mitogen PHA, caused the stimulation of the expression of immune checkpoint receptors PD-1 and PD-L1 on T-helpers and CTL after 24 hours of cultivation. After 72 hours of cultivation, RVK100, unlike RVK228, was revealed ability to reduce the expression of PD-1 on these cells.
Pan-cancer <i>KRAS</i> atlas of 9,056 Chinese patients: Subtype-specific immuno-genomic profiling to guide G12C, G12D, and pan- <i>KRAS</i> strategies.
3128 Background: KRAS therapy is shifting from allele-specific inhibitors to pan- KRAS strategies. However, pivotal trials are largely Western, and Asian subtype-specific immunogenicity and co-mutation benchmarks remain limited. We present the largest pan-cancer Asian KRAS clinogenomic atlas to define these benchmarks and inform treatment selection and trial stratification for the pan- KRAS inhibitor era. Methods: We analyzed 9,056 Chinese patients with KRAS mutations across over 30 tumor types, including pancreatic 27.4%, colorectal 25.1%, NSCLC 20.8%, and biliary tract 10.0%. All underwent ~600 gene next generation sequencing profiling in a CAP/CLIA certified laboratory including all classes of genomic alterations with integrated analysis of clinical variables, treatment history, co-mutations, TMB, MSI, and PD-L1. Results: KRAS mutations clustered at four hotspot codons: G12 (80%), G13 (9%), Q61 (6%), and A146 (3%). Alleles were dominated by G12D (33.3%), G12V (22.8%), and G12C (11.8%), followed by G13D (7.8%) and G12R (4.6%), together covering 80.3% of patients. Overall, 55% carried non-G12C/D variants. More than one KRAS point mutations occurred in 1.8% of the patients and KRAS amplification in 2.5%. Co-mutations were common, led by TP53 (61.7%), APC (22.2%), SMAD4 (17.9%), CDKN2A (17.1%), PIK3CA (12.5%), ARID1A (11.9%), and STK11 (7.7%). TMB in 98.5% (median 4.0 mut/Mb) with 19.2% TMB-high (≥10 mut/Mb), MSI in 97.4% with 3.5% MSI-H, and PD-L1 in 58.0% with 49.5% positive expression. Subtype patterns were distinct. G12C had the highest TMB-high rate (42.4%) with rare MSI-H (0.8%), supporting targeted inhibition with rational IO combination evaluation. In contrast, G13D showed an Asian-enriched immunogenic phenotype with MSI-H 16.1%, higher than Western benchmarks (7.9%, P<0.001), and elevated TMB-high (30.8%). Clinically relevant stratification emerged in NSCLC G12C, with 82.5% male predominance versus about 40% in Western populations (P<0.0001) and an age gradient consistent with smoking (67% male <50 years rising to 89% at ≥70 years). Adverse co-mutations were enriched in G12C, including STK11 (14.4%) and KEAP1 (9.1%). Multi- KRAS mutations were strongly associated with MSI-H (OR 5.48, P<0.0001). Conversely, G12R showed an immune-cold profile (TMB-high 3.2%, MSI-H 0.2%), supporting reliance on targeted inhibition rather than IO-based approaches. Nineteen percent of patients were treatment-naive at profiling, most commonly in NSCLC (30%). Versus post-treatment cases, treatment-naive patients were enriched for G12C (15.9% vs 10.4%, P<0.0001) and showed higher TMB and PD-L1positives. Conclusions: This largest Asian KRAS atlas defines actionable benchmarks for precision therapy and provides a molecular rationale for region-specific stratification in global pan- KRAS trials.
Zolbetuximab in CLDN18.2-positive biliary tract cancer after prior systemic therapy: A prospective observational pilot cohort study of early outcomes, safety, and patient-reported quality of life.
e16174 Background: CLDN18.2 is a validated therapeutic target in advanced gastric and gastroesophageal junction cancer; however, its role in biliary tract cancer (BTC; intrahepatic or extrahepatic cholangiocarcinoma, or gallbladder carcinoma) remains poorly defined. Immunohistochemical studies suggest that a subset of BTCs express CLDN18.2, but prospective clinical and quality-of-life (QoL) - focused data in BTC are lacking. Patients with advanced, pretreated BTC have limited treatment options after gemcitabine-based therapy. We conducted a prospective observational pilot cohort study to assess feasibility of prospective data capture, early clinical outcomes, safety, and patient-reported QoL among CLDN18.2-positive BTC patients receiving zolbetuximab in the salvage setting. Methods: This single-center, prospective, observational study included adults with unresectable or metastatic BTC treated between January 1, 2025, and December 15, 2025. Eligibility required CLDN18.2 expression in ≥50% of tumor cells (moderate/strong membranous staining) after progression on ≥1 prior systemic therapy. Patients received zolbetuximab in the salvage setting; monotherapy or combination with chemotherapy and dosing schedule (commonly every 3 weeks) were determined by the treating physician. Treatment decisions were at physician discretion; the study did not assign treatment or mandate dosing. The primary endpoint was change in EORTC QLQ-C30 global health status/QoL from baseline to week 6. Secondary endpoints included objective response rate (ORR), disease control rate (DCR), treatment duration, and treatment-emergent adverse events (TEAEs). Results: Seven eligible patients (ECOG 0-2) were included (median age 63 years, range 52-73); all had received ≥2 prior systemic lines. High CLDN18.2 expression (≥75%) was observed in 5/7 patients. ORR was 14% (1 partial response) and DCR was 57%. Median treatment duration was 3.4 months. Mean global health status/QoL improved from 52 to 63 at week 6 (+11 points); 4/7 patients achieved a clinically meaningful ≥10-point improvement. Pain (-15 points) and appetite loss (-12 points) improved most. Any-grade TEAEs occurred in 6/7 patients; grade ≥2 TEAEs occurred in 2/7. No treatment-related deaths occurred. Conclusions: In this prospective, biomarker-selected observational pilot cohort of heavily pretreated BTC, zolbetuximab use was feasible, showed encouraging disease control, and was associated with clinically meaningful QoL improvement. These hypothesis-generating data support further investigation of CLDN18.2-directed strategies in BTC.
Impact of high-risk genomic alterations on survival outcomes in patients with stage III non–small cell lung cancer treated with chemoradiation and consolidation durvalumab.
e20042 Background: Concurrent chemoradiation (CCRT) followed by programmed death ligand-1 (PD-L1) checkpoint inhibition with durvalumab (D) is standard of care for patients with unresectable stage III non-small cell lung cancer (NSCLC). A subset of patients with resectable stage III disease also receive this regimen when surgery is not pursued. Several genomic alterations have been associated with worse outcomes in NSCLC patients treated with PD-L1 pathway inhibitors. We sought to determine whether certain high risk genomic alterations impact overall survival (OS) and progression-free survival (PFS) in these patients. Methods: We retrospectively reviewed the outcomes of patients receiving care at the University of Minnesota with stage IIIA-C NSLC who were treated with CCRT+D between 2017 and 2023. A subset of these patients had next generation sequencing (NGS) of their tumors prior to initiation of chemotherapy. PFS and OS were calculated for patients from the time of chemotherapy initiation. High risk genomic alterations (HRGA) were defined as pathogenic variants of KRAS , TP53 , KEAP1 , STK11 , ALK , EGFR , or ROS1 . Kaplan-Meier analysis was used to calculate PFS and OS for patients with a HRGA compared to patients without these mutations. Results: Among 115 patients in the cohort, 79 had NGS data available from biopsy samples or ctDNA samples obtained prior to starting chemotherapy and were considered in this analysis. 30 patients had at least one HRGA (16 KRAS , 6 TP53 , 1 KEAP1 , 1 STK11 , 1 KEAP1 + STK11 , 1 KRAS + TP53 , 1 KRAS + KEAP1 , 1 EGFR , 1 EGFR + TP53 , 1 ALK ). PFS was 887 days in the HRGA group and 587 days in the non-HRGA group (p-value=0.4). OS was 1,110 days in the HRGA group and 938 days in the non-HRGA group (p-value=0.3). Conclusions: In our cohort of stage III NSCLC patients treated with CCRT+D, we did not observe a statistical difference in OS or PFS among patients with a HRGA compared to those without. While our relatively small sample size limits the ability to draw firm conclusions, these data suggest that presence of these genomic alterations in stage IIIA-C NSLC patients should not preclude treatment with CCRT+D. Baseline characteristics and survival outcomes of patients in our cohort. Characteristic Overall (N = 79) 1 No mutations (N = 49) 1 Any mutations (N = 30) 1 p-value 2 PFS (days) 663 (787) 587 (690) 887 (994) 0.4 OS (days) 1,025 (900) 938 (879) 1,110 (855) 0.3 Age (years) 69 (14)Unknown = 2 70 (12)Unknown = 0 68 (12)Unknown = 2 0.4 PD-L1 status 0.13 PD-L1 <1 PD-L1 >= 1 PD-L1 unknown 24 (34%)46 (66%)9 17 (41%)24 (59%)8 7 (24%)22 (76%)1 Histology <0.001 Squamous cell carcinoma 33 (42%) 28 (57%) 5 (17%) Adenocarcinoma 41 (52%) 18 (37%) 23 (77%) Other NSCLC subtype 5 (6.3%) 3 (6.1%) 2 (6.7%) 1 Median (IQR); n (%). 2 Wilcoxon rank sum test; Fisher's exact test; Pearson's Chi-squared test.
Understanding the impact of an established oncology program for adolescents and young adults: The role of school and work transitions counseling.
e24111 Background: Adolescent and young adult (AYA) oncology programs provide tailored, multidisciplinary care addressing concerns common to this population, including fertility, sexual health, body image, mental health, peer support, and wellness. Many AYAs experience cancer during critical educational or career transitions, making school and work transition (SWT) counseling an important component of supportive care. At Princess Margaret Cancer Centre (PM), AYAs are defined as individuals aged ≤39 years. This study examines the utilization and role of SWT counseling within an established AYA oncology program. Methods: All patients referred to the PM-AYA program between January 1 and June 30, 2025, were identified. Demographic and clinical characteristics were summarized descriptively. Associations between patient characteristics and receipt of SWT counseling were assessed using chi-square tests for categorical variables and ANOVA for continuous variables. Multivariable logistic regression was used to evaluate associations between SWT counseling and additional supportive care services, adjusting for age, gender, and diagnosis. Results: Among 229 patients included, median age at diagnosis was 29 years (IQR 23–35), and 133 (58%) were female. At referral, 96 (46%) were employed, 58 (28%) were students, and 55 (26%) were unemployed. Although SWT prompted referral in only 22 (9.6%) patients, 77 (34%) ultimately received SWT counseling. Compared with employed patients, students (RR 3.59, 95% CI 2.14–6.03) and unemployed patients (RR 2.96, 95% CI 1.72–5.12) were more likely to receive SWT counseling. Younger age was also associated with receiving SWT support (χ² = 16.152, p < 0.0001). After adjusting for age, gender, and diagnosis, SWT counseling was not associated with receiving palliative care or advanced practice nursing support. Conclusions: SWT counseling is a key component of comprehensive AYA oncology care and addresses needs that may arise beyond the initial referral reason. SWT services are reaching patients at highest risk of educational and employment disruption, particularly students and unemployed individuals. Lower uptake among employed patients may reflect under-recognition of work-related challenges. Continued investment in systematic SWT support is warranted to optimize AYA cancer care.
Can we omit surgical axillary staging in early-stage invasive lobular carcinoma?: Results from a large multicenter retrospective cohort study.
588 Background: The SOUND, INSEMA and BOOG 2013-08 randomized trials demonstrated that, in selected patients with clinically node-negative (cN0) early-stage breast cancer, omission of sentinel lymph node biopsy (SLNB) results in non-inferior disease-free and invasive disease-free survival. However, invasive lobular carcinoma (ILC) was under-represented in these three studies, and the safety of this approach in ILC remains uncertain due to concerns about a potentially higher axillary nodal burden and risk of undertreatment. We evaluated the axillary nodal burden in ILC patients who met the trials eligibility criteria. Methods: This retrospective multicenter study included patients with cT1-2 cN0 ILC, undergoing upfront surgery between 2012 and 2025 across three high-volume Italian institutions. All patients underwent preoperative axillary ultrasound (US), with needle biopsy performed for US-suspicious lymph nodes. Uni- and multivariable logistic regression analyses identified factors associated with advanced nodal disease (pN2/3). Results: A total of 1886 patients (median age 65 years) were included. The median radiologic tumor size was 16 mm (IQR 11-23). Most patients underwent SLNB (84%), while the remaining were treated with axillary dissection. The overall rate of pN2/3 disease was 2.5% (48/1886). On univariable analyses, a palpable mass on physical exam (OR 3.59; p < 0.001), the presence of US-suspicious axillary nodes (OR 8.19; p < 0.001), tumor size > 2 cm (OR 2.96; p = 0.002), multicentric disease (OR 4.5; p < 0.0001), high tumor grade (OR 2.15; p = 0.03), the presence of lymphovascular invasion (LVI) (OR 3.25; p = 0.001) and HER2 positivity (OR 3.81; p = 0.022) were significantly associated with pN2/3 disease. On multivariable analysis, tumor size > 3 cm (p = 0.003), the presence of US-suspicious axillary nodes (OR 6.29; p = 0.002), the presence of LVI (OR 3.65; p = 0.006), multicentric distribution (OR 5.19; p < 0.0001), and HER2 positivity (OR 3.86; p = 0.041) were independently associated with pN2/3 disease. Conclusions: In this large real-world cohort of SOUND/INSEMA eligible patients with ILC, the rate of pN2/3 disease was 2.5%, higher than that reported in the aforementioned trials. Omission of axillary surgical staging in patients harboring high risk features may result in understaging, with potential implications for adjuvant treatment decisions and oncologic outcomes. Meticolous selection is critical to identify patients who are not at risk of potential undertreatment.
Phase 1, FIH study of cabotamig, a TCE CDH17 X CD3, in CDH17-expressing GI malignancies.
3603 Background: Cadherin-17 (CDH17) is a specific cancer target; aberrant expression is up to >95% in GI tumors (stomach, pancreas, bile duct, rectum, small bowel and colon). Cabotamig (ARB202) is a humanized, IgG4, T-cell engaging antibody (TCE) that binds to CDH17 and CD3 T-cells forming synapse to aid activation. This study evaluated the safety, pharmacokinetics (PK), biomarkers, and anti-tumor activity of cabotamig in patients (pts) with CDH17+ gastrointestinal (GI) cancers. Methods: Pts ≥18 years with locally advanced or metastatic CDH17+ GI cancers refractory to therapies, measurable disease and ECOG 0/1 were enrolled. MABEL calculations per FDA TCE guidance were used (0.0003 mg/ml to 1.0 mg/ml by ½ log increments). For safety given multi log difference in CDH17 density, an initial dose of 1/10 was given 3 days before the remaining 9/10. In the 4th cohort, patients with stable disease were given additional doses (7/10 cohort 4-6). Cohort was split into colorectal cancers (CRC) and non-CRC groups given the differences in CDH17 levels and symptoms. After the 7th cohort dosing proceeded to dose optimization. Data cutoff 02142025. Results: Pts (CRC 15/22 total) received cabotamig at 1/10 dose (0.00003-0.032 mg/kg) followed by 9/10 dose (0.0003 to 0.32 mg/kg) across 7 cohorts with the max. of 37.8(3.8/34) mg. The max. tolerated dose was not reached. Dose responsive cytokine release syndrome (CRS) symptoms were evident with more symptoms between the 1/10 and 9/10 dose and with increasing intensity with escalation. Treatment related AEs (CRS 17, fever 16, nausea 7, diarrhea 5, vomiting 5 fatigue 3 chills 2) were reported with 1 st set or 2 nd infusions and occurred in all patients without pre-corticosteroids in the 4 th cohort and intensified even with premed up to the 7 th cohort. PK data indicated dose proportionality and faster-than-typical IgG4 clearance. Anti-tumor biomarker activity, primarily CEA, but when present included other CA markers, showed a dose response (increase then a decrease to levels below baseline, consistent with target cell lysis) with a peak response between the 5 th and 6 th cohorts. Conclusions: This first-in-human (FIH) study (NCT05411133) targeting CDH17 in GI cancer pts showed a tolerable safety profile for cabotamig up to 0.32mg/kg. Step-up dosing appeared to mitigate dose responsive on-target tumor CRS and the lack of persistent GI toxicities indicated minimal on-target normal GI mucosal effects. Dose responsive biomarker anti-tumor activity of cabotamig was observed. Clinical trial information: NCT05411133 .
Safety-aligned evaluation of large language models for oncology clinical decision support across disease subtypes.
e13701 Background: Large language models (LLMs) are increasingly evaluated for oncology clinical decision support; however, reported performance varies widely, and safety failures such as hallucinations and guideline misalignment remain poorly characterized across disease contexts. We conducted a multi-subtype, clinician-adjudicated evaluation to assess how evidence-source constraints influence safety-aligned performance. Methods: We curated 216 oncology clinical vignettes using a standardized tumor-board format spanning leukemia, breast cancer, gastrointestinal (GI) cancers, CNS metastases, and gynecologic oncology. Each vignette was evaluated using three systems: an unconstrained LLM (Output 1), an NCCN-anchored retrieval-augmented generation (RAG) configuration (Output 2), and a literature-anchored system (Output 3). Two board-certified oncologists independently scored each output using a modified Generative Performance Score (mGPS; range −1 to +1), incorporating guideline concordance and hallucination penalties. Readability and rationality were rated separately (Likert 1–5) and used for contextual interpretation. Overall disparity severity was conservatively assigned as the maximum severity across hallucination and guideline axes. Results: Across all vignettes, the NCCN-anchored RAG system achieved higher mean mGPS and lower hallucination penalties compared with unconstrained and literature-anchored systems. Safety performance varied substantially by disease subtype. Leukemia outputs demonstrated predominantly low to intermediate disparity with rare hallucination-driven high-risk events. Breast cancer outputs showed low-intermediate risk, with high-disparity cases driven primarily by biomarker-dependent guideline misalignment. GI cancers exhibited intermediate-to-high disparity, reflecting multidisciplinary complexity and biomarker omission. CNS metastases and gynecologic oncology represented the highest-risk domains, with frequent high-disparity classifications driven by combined hallucination and guideline failures despite fluent presentation. Readability was consistently moderate to high across systems but did not independently mitigate safety risks. Conclusions: Safety-aligned performance of oncology LLMs is highly disease-dependent and strongly influenced by evidence-source constraints. Guideline-anchored retrieval significantly reduces hallucination-related risk but does not fully mitigate failures in complex, multidisciplinary settings. Multi-axis, disease-specific evaluation frameworks are essential prior to clinical deployment of LLM-based decision support.
A phase 3, international, multicenter, randomized, controlled trial of selective index lymph node resection versus therapeutic lymph node dissection after neoadjuvant immunotherapy for stage IIIB-D melanoma (MSLT-3/S2601/EORTC2519).
TPS9609 Background: Neoadjuvant immunotherapy (NAT) with immune checkpoint inhibitors (ICI) is a standard therapy for resectable stage III melanoma, with randomized trials demonstrating superior outcomes compared to adjuvant therapy alone. The phase 2 SWOG-1801 trial showed 72% EFS at 2 yrs for NAT & 49% for adjuvant PEMBRO (P=0.004), while the phase 3 NADINA trial confirmed superiority of NAT IPI/NIVO over adjuvant NIVO (2-year RFS 83.7% vs 57.2%, HR 0.32). Pathological response, particularly major pathological response (MPR; ≤10% viable tumor), strongly correlates with improved outcomes, with the International Neoadjuvant Melanoma Consortium (INMC) pooled analysis (N=610 ICI patients) demonstrating 3-yr RFS of 93% for MPR patients (pts) vs 41% for those with no pathological response. The PRADO trial demonstrated that selective index lymph node (ILN) resection (RES) can safely de-escalate surgery in MPR pts, with only 4/60 (6.7%) MPR pts recurring after median follow-up of 28.1 months, all locoregionally. ILN RES significantly reduced surgical morbidity & improved quality of life compared to therapeutic lymph node dissection (TLND). However, PRADO was a single-arm proof-of-concept study. An international survey of 117 melanoma experts showed 71% believe a phase 3 randomized controlled trial is needed to change practice. The MSLT-3 trial will definitively establish whether ILN RES is non-inferior to standard TLND for MPR pts. Methods: This phase 3, international, multicenter, randomized, non-inferiority trial will enroll approximately 1,574 pts with resectable stage IIIB-D cutaneous melanoma to identify 496 with MPR following NAT. Eligible pts must have cytologically/histologically confirmed resectable stage IIIB-D melanoma with at least one macroscopic lymph node in groin, axilla or neck basins. All pts are randomized 1:1 to ILN RES or TLND, stratified by continent/region, AJCC stage, & NAT regimen. Prior to NAT, all pts undergo radiological placement of a marker in the largest metastatic lymph node. Pts receive NAT per institutional standard of care (minimum one PD-(L)-1 checkpoint inhibitor, maximum 6 weeks duration) followed by surgery at weeks 6-9 according to randomized arm. Pathological response is assessed per INMC criteria. Pts with MPR in the ILN arm undergo surveillance only; non-MPR pts in the ILN arm proceed to TLND within 3 weeks. Primary endpoint is 2-year recurrence-free survival in MPR pts (non-inferiority margin -5%). Secondary endpoints include escalation to TLND for isolated nodal recurrence, salvage therapy rates, distant metastasis-free survival, event-free survival, overall survival, surgery-related AEs, quality of life (QLQ-C30, EQ-5D-5L, FACT-M), concordance of imaging/ctDNA with pathology & health economics. Clinical trial information: NCT07049276 .
Monitoring blood-based biomarkers as early predictors of progression-free survival in a randomized Bria-ABC phase 3 trial for advanced metastatic breast cancer: An ongoing analysis.
2652 Background: Circulating Tumor Cells (CTCs) are prognostic for poor outcomes in metastatic Breast Cancer (mBC), however CTCs are uncommon in mBC (<20%) and many pts without CTCs often progress. Cancer associated macrophage-like cells (CAML) are prognostic inflammatory pro-tumorigenic PD-L1 expressing macrophages common in mBC pt blood (>90%). In a previous randomized phase II trial, CTC & CAML decreases post Bria-IMT induction correlated with clinical benefit. Bria-IMT is an allogenic whole cell vaccine engineered to express tumor associated antigens & GM-CSF, promoting adaptive & innate immune responses. The ongoing Bria-ABC (NCT06072612) phase 3 study compares Bria-IMT to physician’s choice (TPC) in late stage mBC. We present interim results, without treatment arm comparison, for Progression Free Survival (PFS) by CTC & CAML changes as the exploratory part of the trial. Methods: This still blinded ongoing multicenter randomized open label Phase 3 trial evaluates Bria-IMT+ checkpoint inhibitor (CPI) vs TPC in mBC pts lacking approved therapies. Pts are randomized 1:1:1 to Bria-IMT+CPI, TPC, or Bria-IMT monotherapy (discontinued after 150 pts). The Bria-IMT consists of cyclophosphamide, irradiated SV-BR-1-GM cells, micro-dose pegylated α IFN at each inoculation site. CPI is administered day -3 to 3. TPC followed standard of care. Blinded anonymized blood was taken at baseline (BL), prior to therapy & 2nd (T1) taken at cycle 3 (~4 weeks post initiation). CTCs & CAMLs, and PD-L1 expressions, were quantified using LifeTracDx liquid biopsy with analysis of PFS by censored univariate analysis. Results: At time of analysis, >250 consented, >170 randomized, 119 had BL and 78 had T1. Median age 56 yrs [34–83], median 6 [2–13] prior lines of therapy, 31% TNBC, 62% ER+/PR+, & 15% HER2+. ≥1 CTCs were found in 25% (30/119) at BL & 22% (17/78) at T1. ≥1 CAMLs were found in 93% (111/119) at BL & 95% (74/78) at T1. At BL, ≥1 CTC was not significant for PFS (HR=1.7, CI95% 1.0-2.9, p=0.0513), but ≥2 CTCs was significant for worse PFS (HR=1.8 CI95% 1.1-3.1, p=0.0480). At T1, ≥1 CTCs nor ≥2 CTCs correlated with PFS (HR=0.9, p=0.8392) & (HR=1.6, p=0.1079), respectively. Further, a decrease in CTCs was seen in 11 pts but did not correlate with PFS (HR=0.7, p=0.6552). ≥1 CAML at BL nor T1 correlated with PFS (HR=1.2 CI95% 0.7-2.0, p=0.6811) or (HR=0.9 CI95% 0.7-2.5, p=0.5307), respectively. However, 51 pts (65%) had a decrease or stable CAML counts between BL & T1 which did significantly correlate with better PFS (HR=2.2 CI95% 1.2-3.9, p=0.0154). Conclusions: In an ongoing analysis of a heavily treated mBC pts, we observed that in the entire blinded population, 65% of pts had stability/drop in CAMLs significantly correlated with better PFS. Treatment arm specific comparisons will not be unblinded until completion of the designated milestone (144 mortalities). Clinical trial information: NCT06072612 .
Multimodal AI prediction of head and neck cancer treatment outcomes with whole slide imaging.
e18000 Background: Current risk stratification and adjuvant treatment decisions for head and neck cancer following surgical resection rely primarily on pathological risk factors. The growing adoption of digital pathology presents an opportunity to leverage histopathological image features to enhance risk stratification accuracy. This study investigated the performance of machine learning models for treatment outcome prediction using image features and traditional clinical variables, and identified optimal strategies for combining these modalities. Methods: We analyzed data from 645 patients in the publicly available HANCOCK dataset for model development and testing. All patients had head and neck cancers treated with primary surgery with or without adjuvant therapy. For each patient, H&E-stained whole slide images of the primary tumor and nine clinical variables served as model inputs. Slide-level image embeddings were extracted using a pretrained vision-language pathology foundation model (TITAN). Clinical variables included tumor site, pT classification, pN classification, histologic grade, perineural invasion, lymphovascular invasion, extranodal extension, margin status, and smoking history. Models were trained to predict survival, with performance measured by Harrell's concordance index (C-index), using an machine learning-based Cox proportional hazards model (DeepSurv). We compared unimodal models (using either clinical variables or image features alone) with multimodal models utilizing different fusion strategies (concatenation, late fusion, and cross-attention). Five-fold cross-validation with early stopping was implemented during training. Results: Unimodal models achieved C-indices of 0.62 +/- 0.04 (clinical variables) and 0.65 +/- 0.07 (image features). Multimodal models demonstrated progressive improvement: late fusion (0.63 +/- 0.05), concatenation (0.67 +/- 0.07), and cross-attention (0.69 +/- 0.07), with cross-attention achieving the highest performance. Conclusions: H&E-stained whole slide images from resected head and neck cancers contain significant prognostic information. Multimodal AI models integrating histopathological images with clinical variables, particularly using cross-attention fusion, enhance prognostic prediction and may improve risk stratification for adjuvant therapy decisions. Harrell's concordance index for each type of model input and feature fusion methods. Model input C-index (mean +/- std) Clinical variables alone 0.62 +/- 0.04 Image features alone 0.65 +/- 0.07 Multimodal (late fusion) 0.63 +/- 0.05 Multimodal (concatenation) 0.67 +/- 0.07 Multimodal (cross attention) 0.69 +/- 0.07
Inpatient outcomes in acute myeloid leukemia: A contemporary national study.
e18615 Background: Acute myeloid leukemia (AML) is an aggressive hematologic malignancy associated with high morbidity, frequent hospitalizations, and substantial healthcare resource utilization despite therapeutic advances. Contemporary real-world data on inpatient outcomes among adults with AML remain limited. We evaluated inpatient hospitalization outcomes and resource utilization in this population. Methods: A retrospective study using the National Inpatient Sample identified adult AML hospitalizations from 2021-2022. Sociodemographic variables included age, sex, race/ethnicity, and insurance. Primary outcomes were in-hospital mortality, length of stay (LOS), and costs; secondary outcomes included sepsis, acute kidney injury, respiratory failure, mechanical ventilation, and dialysis. Multivariable regression identified factors independently associated with inpatient outcomes. Results: Among 126,940 hospitalizations, the median age was 66 years; patients were predominantly male and White (71.0%). Medicare (51.7%) and private insurance (32.1%) were the most common payers. Median LOS was 6 days, with 41.8% lasting >8 days. During hospitalization, 20.7% developed sepsis, 18.2% respiratory failure, and 4.8% required invasive mechanical ventilation. The median cost per hospitalization was $75,000, and in-hospital mortality was 8.6%. In multivariable analyses, older age was associated with higher inpatient mortality (adjusted OR, 1.46; P <0.001), shorter length of stay (−2.40 days; P <0.001) and lower total hospital charges (−$50,121; P <0.001). Female sex, compared with male sex, was associated with lower odds of inpatient mortality (adjusted OR, 0.88; P <0.001), slightly shorter hospitalizations (−0.16 days; P =0.047), and lower hospitalization charges (−$11,839; P <0.001). Compared with White patients, Black patients experienced significantly higher inpatient mortality (adjusted OR, 1.48; P <0.001) despite shorter length of stay (−0.75 days; P <0.001) and lower hospital charges (−$39,542; P <0.001). Insurance type and median household income were independently associated with all outcomes; higher income quartiles were associated with lower inpatient mortality (adjusted OR, 0.96 per quartile; P <0.001) but longer hospitalizations and higher costs (both P <0.001). Across all models, critical illness severity was the strongest predictor of adverse outcomes, conferring markedly increased inpatient mortality (adjusted OR, 12.84; P <0.001), longer length of stay (+4.61 days; P <0.001), and higher hospitalization charges (+$108,392; P <0.001). Conclusions: AML hospitalizations are characterized by high morbidity and mortality, with illness severity as the primary driver of outcomes. However, independent associations between race, socioeconomic factors, and inpatient outcomes indicate opportunities to improve risk stratification and optimize inpatient management strategies.
Characteristics and clinical outcomes of newly diagnosed acute myeloid leukemia with MAPK pathway mutations: A referral-center retrospective analysis.
e18506 Background: Somatic mutations in the MAPK signaling pathway are frequently identified and implicated in the disease biology in patients with newly diagnosed acute myeloid leukemia (NDAML), but the real-world clinical experience on treatment response and long-term survival remains incompletely characterized. Methods: We conducted a retrospective analysis of adult patients with NDAML harboring mutations in the MAPK pathways seen at the University of Iowa Health Care between March 2015 and December 2024. The primary endpoint was overall survival (OS), and the secondary endpoint was composite complete remission (CRc), defined as complete remission (CR), CR with incomplete hematologic recovery (CRi), or CR with partial hematologic recovery (CRh) following induction therapy. Intensive induction regimens (IIR) were anthracycline-based, while non-intensive regimens (NIR) were hypomethylating agent with or without venetoclax. The effect of patient, disease, and treatment characteristics on the odds of CRc or allogeneic stem cell transplant (HSCT) was evaluated using logistic regression and OS was evaluated using Cox regression. Results: Among the 74 patients identified (median age 66, 35% female), the following MAPK pathway mutations were found: NRAS only (N=45), KRAS only (N=19), NRAS/KRAS (N=5), BRAF (N=4), and HRAS (N=1). G12D was the most common variant in both KRAS and NRAS . 39 patients received IIR (median age 52), 23 patients received NIR (median age 73), and 12 did not receive treatment (NT) (median age 74). At a median follow-up of 7.9 months, 49 patients had died. Median OS for IIR, NIR and NT were 58.7, 7.5, and 0.2 months, respectively. On multivariable analysis (MVA) among those who received treatment (N=62), European LeukemiaNet (ELN) 2022 adverse risk (Hazard Ratio (HR): 3.56, 95% Confidence Interval (CI): 1.52-8.32) and KRAS mutation (HR: 2.14, 95% CI: 1.10-4.17) conferred inferior OS, while age and treatment intensity were not significantly associated with OS. The CRc rate was 53%. On MVA, abnormal karyotype (Odds Ratio (OR): 0.25, 95% CI: 0.07-0.89) and increasing age (OR: 0.97, 95% CI: 0.94-0.99) were associated with decreased odds of achieving a CRc while ELN 2022 risk was not. 22 patients underwent HSCT, and on MVA, only increasing age (OR: 0.93, 95% CI: 0.89-0.97) was associated with decreased odds of receiving a HSCT whereas treatment intensity was not. Conclusions: In this real-world clinical experience of MAPK-mutated NDAML, KRAS mutation conferred an inferior effect on OS but not CRc. Treatment intensity also did not have a significant effect on OS and the odds of CRc. We speculate that the adverse effect of KRAS mutation may be confounded by older age, abnormal cytogenetics, and adverse ELN risk. Larger, multi-center analyses are needed to confirm these findings.
Burden of liver metastases among pancreatic cancer patients in the United States.
e16160 Background: Liver metastasis is a common and severe complication in pancreatic cancer, contributing significantly to morbidity and mortality. Despite therapeutic advances, the burden remains substantial. This study examines mortality trends, disparities, and geographic variation using CDC WONDER data. Methods: We accessed the CDC WONDER database (1999–2023) to identify individuals who died from liver metastasis (ICD-10: C78.7) among pancreatic cancer patients (ICD-10: C25.0 to C25.9). Mortality disparities were examined by year, sex, race/ethnicity, and geography. Joinpoint regression was used to estimate the average annual percentage change (AAPC) and annual percentage change (APC), with 95% confidence intervals (CIs). Results: A total of 65,939 deaths were related to liver metastasis among pancreatic cancer patients. Most deaths occurred in the 65–85+ year age group (45,323) and at the decedent’s home (51.8%). Overall, the AAMR increased from 0.97 (95% CI: 0.92 to 1.01) in 1999 to 1.97 (95% CI: 1.92 to 2.02) in 2023 (AAPC: 3.00; 95% CI: 2.33 to 3.66; p<0.001). The trend initially declined from 1999 to 2008 (APC = -3.28; 95% CI: -3.91 to -2.65; p<0.001), followed by a significant upward trajectory until 2023 (APC = 5.68; 95% CI: 4.93 to 6.42; p<0.001). The increase in mortality was slightly more pronounced in women than in men (AAPC: 3.03% vs. 2.89%). The highest incidence rates were observed among non-Hispanic (NH) Black or African Americans, followed by NH Whites and Hispanic/Latino populations. Geographic disparities were evident, with the South experiencing the greatest impact and the Northeast the lowest. Rural regions consistently showed higher AAMRs compared to urban areas. From 1999 to 2023, California recorded the highest number of deaths (7,765), while Alaska reported the lowest. Conclusions: Mortality from liver metastasis in pancreatic cancer patients has increased over the past two decades, with sharper increases in men. Persistent racial, geographic, and rural–urban disparities highlight the need for targeted interventions to achieve equitable outcomes and equity-focused healthcare strategies. Deaths and age-adjusted mortality rates (AAMRs) per 100,000 for trends related to liver metastasis with pancreatic cancer from 1999 to 2023. Variable Deaths AAMR (95%CI) 1999 AAMR (95% CI)2023 Overall 65,939 0.97(0.92 to 1.01) 1.97(1.92 to 2.02) Male 35,479 1.17(1.10 to 1.25) 2.31 (2.22 to 2.39) Female 30,460 0.81 (0.76 to 0.87) 1.70 (1.63 to 1.77) NH Whites 50,964 0.98 (0.93 to 1.03) 2.07 (2.01 to 2.14) NH Blacks 7,533 1.32 (1.14 to 1.51) 2.20 (2.03 to 2.37) South 26,210 0.99 (0.91 to 1.06) 2.10 (2.02 to 2.19) Northeast 9,227 1.05 (0.94 to 1.15) 1.39 (1.28 to 1.49) Urban 40,915 (2020) 0.95 (0.90 to 1.00) (2020) 1.66 (1.61 to 1.72) Rural 9,299 1.03 (0.92 to 1.14) 1.85 (1.72 to 1.98)
Organ-specific metastatic progression in non-small cell lung cancer treated with pembrolizumab versus atezolizumab: A TrinetX network analysis.
11182 Background: Immune checkpoint inhibitors (ICIs) targeting the PD-1 (Pembrolizumab) and PD-L1 (Atezolizumab) have revolutionized the first-line management of metastatic non-small cell lung cancer (mNSCLC). While these agents are often considered clinically interchangeable, the differences in molecular binding mechanism and tissue-specific penetration may lead to distinct therapeutic patterns. Since the real-world evidence on the propensity for organ-specific metastatic spread remains limited, we aim to compare organ-specific metastatic progression patterns between the two therapies. Methods: We conducted a retrospective cohort study utilizing the TriNetX US Collaborative Network. After identifying adults with stage IV mNSCLC on initial Pembrolizumab or Atezolizumab monotherapy, patients with pre-existing metastases at the specific organ site of interest were excluded. Propensity score matching (1:1) was performed to balance cohorts across demographic and clinical variables. Our endpoints were metastases to the CNS, mediastinum, bone, liver, adrenal glands, and skin over 365 days. Hazard ratios (HR) were estimated using Cox proportional hazards models. Results: Atezolizumab was associated with a significantly higher risk of CNS metastases compared with Pembrolizumab (1-year incidence: 16.9% vs 6.1%; HR 3.08, 95%CI: 2.73-3.46; p < 0.001). A similar pattern was seen in the mediastinum (HR 1.62, 95% CI: 1.32–1.99), bone (HR 1.53, 95% CI: 1.38–1.70), liver (11.5% vs 7.8%; HR 1.55, 95% CI: 1.37–1.75), and adrenal glands (HR 1.43, 95% CI: 1.20–1.75). In contrast, an inverse relationship was observed in the skin where Atezolizumab showed a significantly lower risk of metastases (1.0% vs 1.7%; HR 0.59, 95% CI: 0.44 - 0.79; p = 0.01). Conclusions: Markedly distinct patterns of organ-specific metastasis were seen, where Atezolizumab consistently demonstrated an elevated risk of progression in the CNS and visceral organs, whereas Pembrolizumab showed a greater propensity for cutaneous progression. These observations indicate that the selection between PD-1 and PD-L1 inhibition may influence the location of the subsequent site of disease progression. Metastatic Site Atezolizumab Cohort (Events / N) Pembrolizumab Cohort (Events / N) Hazard Ratio* (95% CI) P-Value CNS 994 / 5,890 376 / 6,138 3.08 (2.73 – 3.46) <0.001 Mediastinum 228 / 7,043 150 / 7,153 1.62 (1.32 – 1.99) 0.02 Bone 824 / 4,997 624 / 5,374 1.53 (1.38 – 1.70) 0.43 Liver 602 / 5,250 490 / 6,262 1.55 (1.37 – 1.75) 0.35 Adrenal Gland 287 / 6,821 213 / 6,912 1.43 (1.20 – 1.75) 0.068 Skin 71 / 7,306 123 / 7,177 0.59 (0.44 – 0.79) 0.01 Note: N varies by organ site due to the exclusion of patients with pre-existing metastases at that specific site. *Hazard Ratio > 1.0 indicates increased risk with Atezolizumab; HR < 1.0 indicates increased risk with Pembrolizumab.
A randomized controlled trial of a multilingual digital genetic cancer risk assessment tool in a diverse patient population.
10619 Background: Most individuals with hereditary cancer risk remain unidentified, particularly racial, ethnic, and linguistic minorities and those publicly insured. Patient-facing digital tools may improve personal/family history collection and genetic cancer risk assessment (GCRA). In this randomized controlled trial (RCT), we evaluated a multilingual digital GCRA tool in a diverse, predominantly publicly insured population. Methods: An IRB-approved RCT (NCT05562778) was conducted among new patients at four urban academic gynecology clinics. Eligible patients (≥ 18 years old, English-, Spanish-, Cantonese-, or Mandarin-speaking, no prior cancer genetic testing (GT)) were randomized to a digital GCRA tool versus usual care. The digital tool collected patient history and generated National Comprehensive Cancer Network (NCCN)-based GT eligibility reports for providers. Usual care relied on clinician assessment of personal/family history. The primary outcome was clinician identification of high-risk patients and recommendation for GT. Results: From 10/2023 to 10/2025, 210 patients enrolled in the study; 104 (50%) were randomized to GCRA via digital tool and 106 (50%) via clinician interview. Eighty (38%) patients self-identified as Hispanic, 39 (19%) non-Hispanic White, 40 (19%) non-Hispanic Black, and 51 (24%) Asian; 159 (76%) were English-speaking, 40 (19%) Spanish-speaking, and 11 (5%) Mandarin/Cantonese-speaking. Most were publicly insured (133, 63%), (Table 1). In the intervention arm, 99 (95%) completed the digital tool. Among all patients, 36 (17%) were identified as meeting NCCN criteria for high risk of hereditary cancer and thus eligible for GT (intervention: 14; control: 22). Use of the digital GCRA tool was associated with a higher likelihood of high-risk patients being identified and recommended GT (13 [93%] vs. 3 [14%]; P = < 0.001). Digital tool completion and high-risk identification did not differ by race, ethnicity, or language. Conclusions: In a diverse, predominantly publicly insured population, a multilingual digital GCRA tool increased identification and recommended GT for those at elevated risk for hereditary cancer. Integrating such tools may improve detection of at-risk individuals and reduce disparities in genetic service utilization. Clinical trial information: NCT05562778 . Participant characteristics. Characteristic Total Control Intervention Age in years (median, (IQR)) 45.5 (34-59) 46.5 (34-58) 43.4 (33.5-59) Race and Ethnicity (N (%)) Hispanic Non-Hispanic White Non-Hispanic Black Asian 80 (38%)39 (19%)40 (19%)51 (24%) 40 (38%)20 (19%)21 (20%)25 (24%) 40 (38%)19 (18%)19 (18%)26 (25%) Primary Insurance (N (%)) Public Private No Coverage/Other 133 (63%)46 (22%)31 (15%) 64 (60%)26 (25%)16 (15%) 69 (66%)20 (19%)15 (14%) IQR: interquartile range. Public: Medicaid, Medicare, or other government insurance. Private: employment-based or directly purchased.
Burnout among caregivers of hospitalized medical oncology patients: A cross-sectional survey study.
12086 Background: Caregivers of patients hospitalized in medical oncology wards play a crucial role in daily care and clinical decision-making, yet they are rarely screened for burnout. While caregiver burden has been widely studied in outpatient and palliative care settings, data focusing on inpatient oncology caregivers remain limited. This study aimed to evaluate burnout levels among caregivers of hospitalized medical oncology patients and to identify associated demographic and caregiving-related factors. Methods: This cross-sectional study included 91 caregivers of patients hospitalized in a medical oncology ward. Burnout and psychosocial distress were assessed using a structured 10-item survey developed based on established conceptual domains of burnout and caregiver burden. Items were rated on a 5-point Likert scale, with higher scores indicating greater burnout. Burnout levels were categorized as low (≤2), moderate (2–4), or high (≥4). Internal consistency was assessed using Cronbach’s alpha. Group comparisons were performed using Mann–Whitney U and Kruskal–Wallis tests. Correlations were analyzed using Spearman’s rho. Multiple linear regression analysis was conducted to identify independent predictors of burnout. Results: The median caregiver age was 49 years and 73.6% were female. The median burnout score was 3.0. Overall, 23.1% of caregivers reported low burnout, 57.1% moderate burnout, and 19.8% high burnout. Female caregivers had significantly higher burnout scores compared with males (p = 0.008). Caregivers with additional caregiving responsibilities also reported higher burnout levels (p = 0.026). Burnout scores correlated positively with duration of in-hospital caregiving (ρ = 0.228; p = 0.030). In multivariable regression analysis, female gender remained an independent predictor of higher burnout (β = −0.298; p = 0.003). Table 1 shows demographic and clinical characteristics of caregivers and patients. Conclusions: Caregivers of hospitalized medical oncology patients experience substantial levels of burnout, with nearly one in five reporting high burnout. Female caregivers appear to be at particularly high risk. These findings highlight the need for routine psychosocial screening and targeted supportive interventions for caregivers in inpatient oncology settings. Variable N=91 (%) Caregiver gender (F/M) 67 (73.6)/24 (26.4) CG age (years) (median) 49 (23-75) CG Education level (L/Medium/H) 39 (42.9)/27 (29.7)/25 (27.5) Relationship to patient (Spouse/Child/First degree relatives/Other) 32 (35.2)/29 (31.1)/19 (20.9)/11 (12.9) Additional caregiving responsibility (Y/N) 29 (31.9)/62 (68.2) Burnout level [L/Moderate/H] 21 (23.1)/52 (57.1)/18 (19.1) Patient gender (F/M/Other) 41 (45.1)/46 (50.5)/4 (4.4) Patient age (years) (median) 61 (20-81) Cancer diagnosis groups Lung and H&N 27 (29.7) GI 23 (23.3) GU+breast 22 (24.2) Other 8 (8.8)
Analysis of early access to cancer medicines in the United Kingdom following incorporation of the FDA-coordinated pathway, Project Orbis.
e23043 Background: After its withdrawal from the European Union, the United Kingdom (UK) transitioned to international regulatory collaboration via Project Orbis; an FDA-coordinated framework for concurrent oncology medicines review. While the Medicines and Healthcare products Regulatory Agency (MHRA) grants approval, nationwide patient access depends on reimbursement from the National Institute for Health and Care Excellence (NICE). Additionally, manufacturers may voluntarily apply for the Early Access to Medicines Scheme (EAMS); providing a pre-license "bridge" that grants access until regulatory determination. This study evaluates the impact on patient access as the UK shifts toward international regulatory work-sharing. Methods: A retrospective analysis compared two cancer medicine approval cohorts (n = 30 EAMS 2015-2025; n = 24 Project Orbis 2021-2025). Human impact was modelled by applying clinical trial Overall Survival (OS) gains to the duration of the access window (EAMS) or reimbursement gap (Orbis), weighted by monthly patient enrollment (High: 40; Medium: 15; Low: 5). For drugs with immature survival data, an EAMS-derived benchmark (4.2 months) used for imputation. For EAMS, a Net Access Gap was calculated by subtracting the pre-license access window from the standard reimbursement delay. Results: The annual volume of new oncology EAMS has decreased from a peak of 11 initiations in 2021 to 4 in 2022, 1 in 2023, and no initiations in 2024–2025. Median regulatory lag from FDA to UK approval was 163 days (IQR: 144) for EAMS drugs vs. 192 days (IQR: 94) for Project Orbis. The median reimbursement lag (MHRA to NICE) was 243.5 days (IQR: 310.25; n = 28) with EAMS medicines compared to 269 days (IQR: 500; n = 18) for Project Orbis. After accounting for the pre-license access window, the EAMS Net Access Gap was 127 days (IQR: 261.5). We estimate that pre-license access (EAMS) provided an estimated gain of 1,189.7 Quality-Adjusted Life-Years (QALYs). Utilizing the historical median EAMS window (120.5 days) as a benchmark, the opportunity cost of not utilizing a pre-license bridge for the Orbis cohort is 1,171.3 Life-Years (819.9 QALYs). Conclusions: FDA-coordinated pathways have become integral to UK cancer medicine approval, yet rapid regulatory approval can make pre-licensing schemes like EAMS less feasible to establish. Because the EAMS "bridge" that once mitigated reimbursement delays has diminished, the net duration patients spend without access has increased. For global stakeholders, this data suggests that regulatory acceleration alone is insufficient; without synchronized reimbursement, accelerated pathways may unintentionally result in a net reduction of the early access window once available to patients.
Evaluation of the NorthStar ctDNA assay for monitoring gastrointestinal malignancies: A retrospective analysis.
e15058 Background: Liquid biopsies offer a non-invasive alternative to traditional tissue biopsy for guiding therapy and response monitoring. However, current assays are limited by poor sensitivity and need a prior tissue sample. NorthStar Select and Response are tumor-naïve assays that utilize novel aneuploidy and methylation-based analytics to improve circulating tumor DNA (ctDNA) detection over competing methods. The short half-life of ctDNA is approximately 1-2 hours, making it valuable for real-time response monitoring. This study aims to provide descriptive analysis of Northstar and evaluate its potential for clinical utility in cancer monitoring and therapy selection. Methods: A single-institution, retrospective analysis was conducted on 99 patients with gastrointestinal cancers treated at UCI Medical Center. Blood samples collected between December 2019 and July 2025 were analyzed using the NorthStar platform, which included the Response assay’s methylation-based ctDNA quantification. Clinical and radiographic data were extracted to assess correlations between ctDNA dynamics, imaging findings, and treatment response. R was used for statistical analysis. Results: A total of 99 patients were included in the study, of whom 53 were male and 46 were female. 91 percent of individuals were in the 45-90 age range. 23 percent of patients identified as Hispanic, 74% as non-Hispanic, and 3% as unknown. 56% were White, 18% Asian, 13% mixed race, 2% Black or African American, and 11% other or unknown. The most common cancers were pancreatic (49%), colorectal (23%), unknown primary (10%), and cholangiocarcinoma (10%). At diagnosis, 48 patients had localized disease, while others had metastases to the liver (26), lung (18), peritoneum (9), lymph nodes (3), bone (3), brain (1), or other sites (6). 335 lines of therapy were recorded, including chemotherapy (165), targeted therapy (52), radiation (45), immunotherapy (26), clinical trial (40), and mixed therapies (7). 2235 tumor marker tests were performed: CEA (852), CA 19-9 (1303), CA 125 (36), and other (40). NorthStar ctDNA levels were available for 87 patients (mean 5,865; median 130; range 0–120,000). ctDNA became positive a mean of 153 days before radiographic recurrence. Initial ctDNA positivity rates were 76.2% for localized disease, 76.5% for lung metastases, and 74.1% for any metastatic disease. Conclusions: In this cohort of 99 patients with gastrointestinal cancers, the Northstar test demonstrated broad applicability across age groups, cancer types, and stages, including localized and metastatic disease. Initial ctDNA positivity was high across disease groups and metastatic status. The assays provided quantitative ctDNA measurements that may complement traditional markers and imaging, supporting their potential utility for real-time therapy monitoring and clinical decision-making.