Citations

378 Background: RATIONALE-305 (NCT03777657) enrolled the highest number of pts with PM among similar trials and stratified pts with advanced GC/GEJC by presence of PM. We used statistical modeling to evaluate the impact of varying PM levels on outcomes with TIS+CT. Methods: We randomized adults with GC/GEJC (1:1) to TIS or placebo (PBO) + CT. Primary endpoints: OS in intent-to-treat (ITT) population and pts with programmed death-ligand 1 (PD-L1) Tumor Area Positivity (TAP) score ≥5%. We weighted pts using the method of moments to match a hypothetical population to ITT, with the same distribution of sex, primary tumor location, PD-L1 (TAP score <5% vs ≥5%), number of metastatic sites (0-2 vs 3+), and liver metastasis. PM representation was set from 20% to 45%. Weighted OS, PFS, and objective response rate (ORR) were estimated. Results: Of 997 pts enrolled, 434 (43.5%) had PM. Among pts with PD-L1 TAP scores ≥5% (54.8%), 39.7% had PM, and among pts with scores ≥1% (88.9%), 43.6% had PM (data cutoff Feb 28, 2024). Fewer pts with (+) PM had PD-L1 high tumors (TAP score ≥5%: 50.0%) and more had primary stomach tumors (86.6%) compared to those without (-) PM (58.4% and 75.5%). TIS+CT improved OS, PFS, and ORR in ITT and in pts +/- PM vs PBO+CT (Table). Across 20%–45% PM, TIS+CT showed consistent benefit, with stable HRs for OS and PFS (20% and 45% shown). TIS+CT showed survival benefits in pts with PM and PD-L1 TAP score ≥5% (OS: HR=0.71; PFS: HR=0.65) and ≥1% (OS: HR=0.79; PFS: HR=0.79) vs PBO+CT. ORR was higher with TIS+CT vs PBO+CT in pts + PM and PD-L1 TAP scores ≥5% (47.3% vs 30.8%) and ≥1% (44.2% vs 33.2%). Conclusions: Our analysis showed that lowering PM representation marginally improved OS and ORR, underscoring the poor prognosis of peritoneal disease. In RATIONALE-305, TIS+CT provides clinically meaningful response and survival benefits irrespective of PM, with higher responses in PD-L1+ GC/GEJC. Clinical trial information: NCT03777657 . Arm(95% CI) ITT + PM - PM 20%+ PM 45%+ PM OS, months TIS+CT 15.0 (13.6, 16.5) 12.3 (10.6, 14.3) 17.3 (15.0, 20.3) 15.4 (13.9, 18.0) 15.0 (13.5, 16.5) PBO+CT 12.9 (12.1, 14.1) 11.8 (10.5, 13.0) 14.0 (12.6, 16.0) 13.1 (12.3, 14.5) 12.8 (12.1, 14.1) HR 0.79 (0.69, 0.91) 0.78 (0.64, 0.96) 0.79 (0.65, 0.95) 0.81 (0.69, 0.95) 0.79 (0.69, 0.90) PFS, months TIS+CT 6.9 (5.7, 7.2) 5.8 (5.6, 7.3) 7.0 (5.7, 8.4) 6.9 (5.7, 7.3) 6.9 (5.7, 7.2) PBO+CT 6.2 (5.6, 6.9) 5.7 (5.3, 6.9) 6.9 (5.7, 7.1) 6.5 (5.6, 7.0) 6.2 (5.6, 6.9) HR 0.79 (0.68, 0.91) 0.80 (0.64, 0.98) 0.77 (0.64, 0.94) 0.79 (0.67, 0.94) 0.79 (0.69, 0.91) ORR, % TIS+CT 47.3 (42.9, 51.8) 43.6 (37.0, 50.5) 50.2 (44.2, 56.2) 48.0(42.9, 53.1) 47.3 (42.8, 51.8) PBO+CT 40.5 (36.2, 45.0) 32.2 (26.0, 39.0) 46.8 (40.9, 52.8) 43.1(38.1, 48.3) 40.3(36.0, 44.8) ORR difference, % 6.8 (0.6, 12.9) 11.4 (2.3, 20.3) 3.4 (-4.9, 11.6) 4.9 (-2.2, 11.9) 6.9 (0.8, 13.0)

Abstract Recent evidence has highlighted the critical requirement for “immune triads” of CD8+ T cells, CD4+ T cells and dendritic cells (DC) for tumor elimination in response to checkpoint inhibition (CPI) immunotherapy. However, the mechanistic underpinnings of this cross-talk are poorly understood. Immune triads are often located in peripherally-distributed DC-rich niches within the tumor stroma. We have hypothesized that stromal remodeling in response to tumor expansion generates homeostatic signals that trigger an adaptive immune response that may ultimately fail through effector exhaustion or apoptosis. In particular, stromal proteolysis of the large matrix proteoglycan versican (VCAN) has been associated with T-cell infiltration across multiple tumor types. The objective of this study was to interrogate a mechanistic link between stromal remodeling, immune triad formation and response to CPI. We used HALO image analysis to determine the spatial relationship of CD4+ and CD8+ T cells in a panel of 95 human non-small cell lung cancers. Each tumor was VCAN proteolysis-scored based on a binning system measuring staining intensity for a neoepitope exposed through site-specific VCAN proteolysis. Tumors displaying VCAN proteolysis intensity of zero (0) had an average of 0.122 CD8+ T cells located with 10um of a CD4+ T cell vs. 0.216 in tumors with any VCAN proteolysis (intensity 1, 2, 3 combined); p=0.04. Similar results were obtained when measuring proximity of CD4+ to CD8+ T cells (0.124 vs 0.223, p=0.02). These data demonstrate that remodeling stroma promotes CD4+:CD8+ crosstalk. Site-specific VCAN proteolysis releases a 441 N-terminal bioactive fragment, versikine. Versikine constitutes a classical “matrikine”, i.e., a bioactive matrix fragment possessing novel functions compared to its parent macromolecule. To determine the mechanisms by which versikine may promote immune triad niches, we generated several transplantable syngeneic tumor cell lines engineered to stably secrete versikine into the TME. One CPI-refractory tumor cell line (Lewis Lung Carcinoma, LLC) and two partially CPI-responsive transplantable tumor lines (CT26 and MC38 colorectal carcinoma) were extensively characterized through flow cytometry, spatial immunofluorescence and scRNAseq. Our results demonstrate that in all contexts, versikine results in expansion of Tfh-like CD4+ T cells expressing Cd40lg and Icos. This expansion is associated with specific patterns of CD8+ T cell exhaustion that will be discussed. Therapeutically, versikine reversed anti-PD-1 pathway refractoriness in the poorly-immunogenic LLC model and demonstrated superior synergy with CPI in the more immunogenic CT26 and MC38 models, leading to tumor eradication with resistance to tumor rechallenge. Our data bolster the mechanistic rationale for exploitation of stromal remodeling signals to promote CD4+T:CD8+T:DC cross-talk for enhanced CPI efficacy. Citation Format: Daniel Lagal, Duncan Hong, Athanasios Papadas, Alicia Gibbons, Yun Huang, Yaling Dou, Alexander Cicala, Elsa Molina, Kristina Matkowskyj, Dustin Deming, Fotis Asimakopoulos. Stromal remodeling promotes immune triad formation in the tumor microenvironment (TME) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2588.

AbstractHeparin-induced thrombocytopenia (HIT) is a serious adverse drug reaction characterized by antibodies that recognize platelet factor 4/heparin complexes (PF4/H) and activate platelets to create a prothrombotic state. Although a high percentage of heparin-treated patients produce antibodies to PF4/H, only a subset also makes antibodies that are platelet activating (PA). A close correlation between PA antibodies and the likelihood of experiencing HIT has been demonstrated in clinical studies, but how PA (presumptively pathogenic) and nonactivating (NA) (presumptively benign) antibodies differ from each other at the molecular level is unknown. To address this issue, we cloned 7 PA and 47 NA PF4/H-binding antibodies from 6 patients with HIT and characterized their structural and functional properties. Findings showed that PA clones differed significantly from NA clones in possessing 1 of 2 heavy chain complementarity-determining region 3 (HCDR3) motifs, RX1-2R/KX1-2R/H (RKH) and YYYYY (Y5), in an unusually long complementarity-determining region 3 (≥20 residues). Mutagenic studies showed that modification of either motif in PA clones reduced or abolished their PA activity and that appropriate amino acid substitutions in HCDR3 of NA clones can cause them to become PA. Repertoire sequencing showed that the frequency of peripheral blood IgG+ B cells possessing RKH or Y5 was significantly higher in patients with HIT than in patients without HIT given heparin, indicating expansion of B cells possessing RKH or Y5 in HIT. These findings imply that antibodies possessing RKH or Y5 are relevant to HIT pathogenesis and suggest new approaches to diagnosis and treatment of this condition.