Kim, Minji

Background: Tumor immune microenvironment (TIME) organization fundamentally shapes cancer progression and immunotherapy outcomes. Traditional 2D spatial molecular approaches inadequately capture the complex three-dimensional architecture and cellular interactions within TIME. We developed an AI-driven platform integrating multiple spatial molecular datasets, enabling comprehensive 3D visualization for therapeutic discovery and patient selection. Methods: Our platform combines holotomography imaging, providing 3D refractive index maps, with diverse spatial molecular data: in situ molecular imaging, spatial proteomics, multiplex immunohistochemistry, and spatial transcriptomics. AI algorithms align datasets within a unified 3D coordinate system, while Poisson implicit surface reconstruction generates subcellular-resolution models. Platform validation used clinical samples from immunotherapy-treated patients. Results: The platform integrated multimodal spatial data into high-resolution 3D TIME representations, revealing novel relationships between protein expression, transcriptional patterns, and cellular architecture. 3D surface protein analysis with subcellular morphology identified patients with target protein expression outside tight junctions, indicating higher risk for adverse effects from antibody-drug conjugates and cellular therapies. This insight, unavailable through 2D analysis, enabled precise patient selection. Conclusions: This platform provides detailed subcellular-resolution TIME architecture analysis, offering actionable insights for therapeutic targeting, biomarker discovery, and patient selection. AI-driven 3D spatial multimodal TIME analysis advances precision medicine across multiple therapeutic approaches, including immunotherapy, targeted therapy, and drug delivery optimization. AI Disclosure: Original abstract content was written by the authors. Claude was used for editorial assistance and formatting without altering the scientific content or context of the work. Citation Format: Inyeop Jang, Yoonho Choi, Minji Kim, Jeonghwan Park, Jiyeon Sung, Byungsoo Ann, Clemenceau R. Jean, TaeHyun Hwang. AI-enabled 3D multimodal integration platform reveals subcellular architecture of tumor immune microenvironment for therapeutic target and biomarker discovery and patient selection [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 3648.

Abstract Background: Understanding the spatiotemporal dynamics of the tumor microenvironment (TME) during treatment is crucial for improving outcomes in pancreatic ductal adenocarcinoma (PDAC). This study presents a pre-planned spatiotemporal analysis from a phase II trial evaluating neoadjuvant mFOLFIRINOX in resectable PDAC, aimed at elucidating treatment resistance mechanisms and identifying predictive biomarkers. Methods: This multi-institutional, single-arm phase II trial enrolled treatment-naïve patients with histologically confirmed resectable PDAC (NCT05066802) who underwent six cycles of neoadjuvant mFOLFIRINOX followed by surgery. The primary endpoint was the R0 resection rate. Matched pre-treatment tumor biopsies and post-treatment surgical specimens were analyzed from selected responders (R) and non-responders (NR) using high-resolution spatial transcriptomics (10X Genomics Visium). This approach enabled comprehensive mapping of TME evolution and treatment-induced changes. Results: Between May 2020 and February 2022, 27 patients were enrolled. Of these, 24 patients underwent surgery after neoadjuvant treatment (resection rate: 88.9%), and 23 achieved R0 resection (85.2%), meeting the primary endpoint. However, 20 resected patients (83.3%) recurred. Spatiotemporal analysis was performed on matched samples from four R and four NR. In pre-treatment samples from NR group, clusters dominated by fibroblasts (enriched for ECM, EMT, and CAF phenotypes) and B cell subtypes were distinct compared to R group, which exhibited elevated antibody responses and robust T cell and macrophage activity. NR samples showed increased IgM expression but lacked corresponding Fc receptors, suggesting limited immune effector function. Post-treatment NR surgical specimens displayed elevated expression of MUC5AC and CEACAM6 in malignant ductal cells, alongside genes promoting tumor growth. Neoadjuvant mFOLFIRINOX increased fibroblast, while reducing immune cell components. NR tumors formed distinct "resistance niches" after chemotherapy, marked by ECM production, CAF activation, and RTK signaling. Conversely, R tumors showed enhanced epithelial organization with increased type-I hemidesmosome complexes facilitating epithelial adhesion to the basement membrane. Conclusion: Three months of neoadjuvant mFOLFIRINOX was feasible and tolerable for resectable PDAC but associated with high recurrence rate, emphasizing the need for better patient selection. Spatiotemporal analysis revealed critical differences in TME dynamics according to treatment response, including the emergence of resistance niches and distinct pre-treatment spatial signatures. These findings support the integration of spatial biology into future pancreatic cancer trials to improve stratification and therapy. Citation Format: Choong-kun Lee, Changjin Hong, Ho Kyoung Hwang, Min Hwan Kim, Jean Rene Clemenceau, Seock-Jin Chung, Isabel Barnfather, Sumanth Reddy Nakkireddy, Jeong Hwan Park, Ji-Youn Sung, Minji Kim, Yooho Choi, Inyeop Jang, Sunho Park, Sung Hyun Kim, Chang Moo Kang, Hye Jin Choi, Tae Hyun Hwang. Spatiotemporal evolution of the tumor microenvironment predicts treatment response in phase II trial of neoadjuvant chemotherapy for resectable pancreatic cancer [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 771.

Background: Pancreatic ductal adenocarcinoma (PDAC) is characterized by a dense, collagen-rich stroma that impedes drug delivery. We investigated how iRGD-mediated nanoparticle delivery modifies collagen architecture using humanized mouse models. Computational analysis of Picro-Sirius Red-stained whole-slide images (WSIs) using curvelet transform-based fiber extraction method (CT-FIRE) was performed to quantitatively examine changes in collagen features, thereby understanding treatment-induced alterations in the tumor microenvironment (TME). Methods: Humanized PDAC mice were prepared with either the human PANC-1 cell line or 0779 primary human PDAC cells and treated with or without iRGD monotherapy for two months (n=12). Primary tumors were sectioned and stained with Picro-Sirius Red to visualize collagen structures. WSIs of Picro-Sirius Red-stained tissues were analyzed with CT-FIRE to extract six morphological features of collagen, including average fiber length, waviness, and density to capture structural changes. WSIs were divided into 256×256 pixel tiles with consistent sampling through bootstrapping, followed by permutation testing for statistical validation (1,000 permutations, p<0.01). Results: Computational analysis revealed significant iRGD-induced collagen remodeling with quantifiable changes in key structural and morphological features. Permutation testing showed that, compared to controls, average fiber length decreased by 3.27% (p<0.01), waviness reduced by 3.25% (p<0.01), and density diminished by 36.3% (p<0.01), compared to controls. The structural changes were more notable in 0779-derived tumors compared to PANC-1, with reductions in average fiber length (1.03%, p<0.01), waviness (8.78%, p<0.01), and collagen density (24.27%, p<0.01), demonstrating cell-line dependent stromal remodeling. Conclusions: Our findings suggest that long-term iRGD therapy alters collagen morphology in PDAC. Further investigation is warranted to understand how these alterations contribute to enhanced penetration of drugs and immune cells observed in iRGD-treated PDAC tumors. The observed cell-line dependent variations in stromal remodeling suggest tumor heterogeneity influences treatment response, highlighting the potential for personalized therapeutic approaches in stromal-rich tumors like PDAC. Citation Format: Yoon Ho Choi, Minji Kim, Hotaka Kawai, Yukihito Kuroda, Norio Miyamura, Seock-Jin Chung, Kazuki N. Sugahara, Tae Hyun Hwang. Altered collagen morphology in pancreatic cancer treated with the iRGD tumor-penetrating peptide [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 2698.

Abstract Background: Pancreatic ductal adenocarcinoma (PDAC) exhibits a complex, immunosuppressive tumor immune microenvironment (TIME), necessitating molecular-level understanding to identify therapeutic targets. We investigated how iRGD peptide therapy remodels the immunosuppressive TIME to an immune-favorable state using single-cell spatial molecular analysis. Methods: Single-cell level in situ molecular imaging was performed using Xenium analyzer with 5K gene panel on tumor samples from humanized PDAC mice (n=7) treated with iRGD monotherapy (n=3) or control (n=4), enabling comprehensive spatial molecular characterization of the TIME. Results: iRGD therapy significantly transformed the immunosuppressive TIME to an immune-favorable state through T-cell composition remodeling. We observed increased populations of CD4+ memory T cells, CD4+ helper T cells, and cytokine-secreting effector T cells, indicating enhanced immune memory and effector functions. Concurrent decreases in regulatory T cells and activated CD4+ T cells reflected reduced immune suppression. Additionally, decreased TGF-β signaling in the iRGD cohort suggested TGF-β antagonism as a mechanism for overcoming immunosuppression. Conclusions: iRGD therapy effectively converted the immunosuppressive PDAC microenvironment to an immune-favorable state by enhancing beneficial T cell populations and reducing immunosuppression through TGF-β pathway modulation. These findings support previous work by Kuroda et al. and provide mechanistic rationale for the ongoing iRGD-based immunotherapy clinical trial (iLSTA:ACTRN12623000223639).Acknowledgments: We thank DeepL for the English language editing. Citation Format: Minji Kim, Yoon Ho Choi, Hotaka Kawai, Yukihito Kuroda, Norio Miyamura, Kazuki N. Sugahara, Tae Hyun Hwang. iRGD peptide therapy transforms immunosuppressive microenvironment to immune-favorable state in pancreatic ductal adenocarcinoma [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 2701.

Abstract Background: Immune checkpoint inhibitors (ICIs) targeting PD-1 or PD-L1 have dramatically improved survival of patients with gastric cancer (GC). However, recent clinical trials, including KEYNOTE-059, KEYNOTE-061, and ATTRACTION-2, show that the objective response rate (ORR) for ICIs is still less than 20% in patients with advanced GC. Therefore, predicting who will respond to ICIs in advance is essential to make better treatment plans for GC patients. There are potential biomarkers to predict ICI response, but most of them can only be applied to a limited population of patients. For example, Kim ST., et al., (PMID: 30013197) show that patients with MSI-high or EBV-positive responded well to ICIs, but these subtypes account for less than 30 percent of GC patients (PMID:25079317). In this study, we propose a prediction model of ICI response using the 32 GC-specific genes that we identified from mutation data (PMID: 26635139) and showed their close association with survival and ICI response in GC (PMID: 35140202; PMID: 36508166). In addition, since the number of available GC-ICI samples is small, we design our prediction model in Bayesian and ensemble learning frameworks which are robust to small sample size problems. Methods: We collected mRNA sequencing data from GC-ICI treated patients from our institutions and publications (PMID: 30013197; PMID: 35254400; PMID: 33846173). We used 108 samples previously included in our previous work (PMID: 36508166) as the training data and 27 samples (13 responders and 14 non-responders) from (PMID: 3384617) as the test data. We built and trained a classifier to separate responders from non-responders using the expressions of the 32 genes. We proposed to use a Deep Gaussian process (GP) model, a multi-layered extension of a GP that is itself a Bayesian method. We optimized the parameters of each GP model in variational inference. To implement ensemble learning, we trained multiple DGP models with different initializations of the model parameters and data splitting (training/validation sets). For each training run, the best model was selected based on the prediction performance on the validation set, measured by AUC. The final prediction for a test sample was made by combining the prediction results of all the DGP models by taking the average. Results: We implemented the DGP-based prediction model using PyTorch in Python, and the codes will be publicly available on the GitHub repository (https://github.com/hwanglab/GC_ICI_response_prediction). We achieved a prediction performance of 0.813 AUC on the test data (sensitivity = 0.769 and specificity = 0.714). Conclusion: We propose the prediction model of ICI response, which can handle small sample size problems in the framework of Bayesian and ensemble learning. The experimental results using the GC-ICI data demonstrate that the 32 genes can be used as predictive markers for ICI response in GC. Citation Format: Sunho Park, Minji Kim, Changjin Hong, Tae Hyun Hwang. A deep Gaussian process (DGP)-based prediction model for immunecheckpoint inhibitor (ICI) response in gastric cancer (GC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2316.