Mukhopadhyay, D.

Abstract Background While immune checkpoint inhibitors (ICIs) provide durable clinical benefit in a subset of patients, current biomarkers to predict response lack accuracy. Ex vivo cytokine profiling has shown promise for improved accuracy but has been mostly limited to ICIs targeting PD-1 [Voabil et al, 2021]. This poses the question of whether ICIs beyond αPD-1 – such as CTLA4 blockade, which has been reported to act within the draining lymph node – would be amenable to similar approaches. The Elephas platform characterizes response to ICIs in live tumor fragments (LTFs) from core needle biopsies (CNBs) and resections using longitudinal cytokine profiling. Using a panel of 7 clinically relevant ICI regimens, we demonstrate cytokine responses across diverse tumor types, characterize similarities and differences in cytokine response profiles across different ICIs, and explore specimen variability in responses across ICI treatments. Methods CNBs were collected through 3 observational clinical trials (NCT05478538, NCT05520099, NCT0634962) and a biobank collection study, while resections were procured from NCI CHTN. LTFs were generated using an automated cutting device and encapsulated in hydrogel. LTFs were treated with mono (αPD-1, αPD-L1, αCTLA4, αLAG-3) or combination therapy (αPD-1/αCTLA4, αPD-1/αLAG-3, αPD-L1/αCTLA4) using a sequential treatment strategy in which isotype control followed by ICI were added to the same tissue in a single well. This approach enabled the comparison of control and treatment conditions while mitigating the effects of tumor heterogeneity. Production rates for 46 cytokines were measured using a multiplex assay. Remaining T-cell reactivity was also evaluated at the end of the assay to better understand the presence of functional T cells in LTFs. Results Unsupervised hierarchical clustering of cytokine profiling data from CNB LTFs across 129 patients identified defined responder and non-responder clusters. Within the response cluster, 4 subclusters with distinct immune phenotypes were identified. While tight clustering of samples treated with the same ICI was not observed, some enrichment of samples treated with combination ICIs was observed within individual subclusters. Further, cytokine responses to αCTLA4 monotherapy were observed, suggesting that CTLA4 blockade may act beyond the draining lymph node. LTFs from resections enabled the characterization of cytokine response profiles across treatment conditions within individual specimens, including the magnitude of cytokine response and variability in cytokine response profiles. Evaluation of T-cell reactivity provided further insight into the presence of functional T-cells in LTFs and the extent of T-cell activation achieved by ICI treatment. Conclusions This study demonstrates the platform’s ability to recapitulate diverse mechanisms of immune activation ex vivo, enabling the characterization of cytokine responses to clinically relevant ICI mono or combination therapy in live tumor tissue and supporting future efforts to improve ICI response prediction beyond αPD-1. Citation Format: Pichet Adstamongkonkul, Pierre Cunin, Santanu Bhattacharya, Tanmay Kulkarni, Amreen Nasreen, Nathan Marhefke, Nina Chen, T S. Ramasubramanian, Laura Hrycyniak, Julie Zweng, Erika von Euw, Christina Vivelo, Sean Caenepeel, Debabrata Mukhopadhyay, Hinco Gierman, Hilario Ramos. Characterization of αPD1, αPD-L1, αCTLA4, and αLAG3 as mono and combination therapy in live tumor fragments using an ex vivo cytokine profiling platform [abstract]. In: Proceedings of the AACR Immuno-Oncology Conference (AACR IO): Discovery and Innovation in Cancer Immunology: Revolutionizing Treatment through Immunotherapy; 2026 Feb 18-21; Los Angeles, CA. Philadelphia (PA): AACR; Cancer Immunol Res 2026;14(2 Suppl):Abstract nr B032.

Abstract Background: Immune checkpoint inhibition (ICI) in combination with chemotherapy is associated with improved outcomes in a subset of patients with triple negative breast cancer (TNBC). However, biomarkers for predicting patient response, such as PD-L1 expression, have limited predictive efficiency. Voabil et al. has shown that ex vivo cytokine profiling of live tumor tissue resections treated with αPD-1 can predict patient response to ICI (Voabil et al., Nat Med. 2021). However, this method requires large amounts of tissue to address intra-specimen tumor heterogeneity, which is not feasible with standard of care (SOC) diagnostic core needle biopsies (CNBs) for breast cancer, where tissue is scarce. We show for the first time that ex vivo cytokine profiling is possible in live tumor fragments (LTFs) generated from CNBs in breast tissue. Methods: We present a cytokine profiling platform to assess response to ICI treatment in LTFs generated from CNBs. CNBs were collected from patients who were identified as potential candidates for SOC ICI (by treating physician) and enrolled in one of two prospective observational clinical trials (NCT05520099 and NCT06349642). CNBs were cut by an automated, proprietary cutting instrument to generate LTFs ofapproximately 300 µm in thickness. To address tissue heterogeneity and mitigate tissue scarcity in CNBs, we used a sequential treatment strategy in which control and ICI treatment were applied to LTFs in the same well, in a phased approach. This strategy provides a solution for the variability observed in cross-well comparisons. LTFs were encapsulated in hydrogel and treated with IgG control followed by αPD-1 (BioXCell). Cytokine production rates were measured longitudinally using multiplex assays assessing>45 cytokines, including 15 cytokines assayed by Voabil et al., and the fold changes of cytokine production rates in response to ICI over IgG control were calculated. Results: Using non-breast tumor tissue, we validated that our platform can measure changes in cytokine production in LTFs following sequential treatment, and that these measurements are comparable to those observed in a cross-well setting. In a small cohort of TNBC specimens, we show changes in cytokine production rates in response to ex vivo sequential control and ICI treatment. Ten cytokines, including IFN-γ and CXCL10, shown to be predictive of patient response to ICI in Voabil et al. were found upregulated in response to αPD-1 in TNBC specimens. Future determination of correlation with clinical response data (pathological complete response for early stage and RECIST 1.1 for metastatic patients) gathered from NCT05520099 and NCT06349642 will be used to validate platform responders and non-responders. To date, 11 patients with confirmed TNBC diagnoses have enrolled in these clinical studies. Conclusion: We report, for the first time to our knowledge, the ability to detect immune checkpoint inhibitor responses in live TNBC tissue obtained via CNBs. This proof-of-concept lays the groundwork for integrating functional tissue responses with clinical correlation data, with important implications for guiding current ICI therapies and advancing this platform for use in next-generation immuno-oncology treatments for TNBC. Citation Format: P. P. Advani, J. Harvey, J. Polley, M. Weidner, A. Arnold, D. Haley, R. Bollam, R. Rao, S. Chumsri, K. Hugghins, M. Acampora, C. Vivelo, H. Hernan, L. Hrycyniak, J. Zweng, H. Gierman, S. Bhattacharya, D. Mukhopadhyay. Ex vivo cytokine profiling of live triple negative breast cancer patient specimens from core needle biopsies illustrates proof of concept to assess tumor response to immune checkpoint inhibition [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS4-01-08.