Abstract 768: Decode the antibody-drug conjugate surfaceome in metastatic urothelial cancer (DECODE-ADC)

Background: Antibody-drug conjugate (ADC) therapies like Enfortumab Vedotin (EV), targeting NECTIN4, are transforming cancer treatment, particularly for metastatic urothelial cancer (mUC). While single-agent EV and EV in combination with immunotherapy show promising results, long-term benefits are limited to a subset of patients, underscoring the challenges associated with its non-selective application. All-comer application of EV raises concerns about toxicity and cost, highlighting the necessity for more personalized approaches. ADCs targeting other tumor antigens, such as HER2, HER3 and EGFR, are in late-stage clinical development, emphasizing the importance of understanding spatial molecular target distributions in patient tissues. Spatial multi-omics and single cell technologies applicable to patient tissues like formalin fixed paraffin embedded tissue (FFPE) open the possibility to unlock these spatial molecular profiles which ultimately could be used to identify companion diagnostic biomarkers and providing valuable insights into tumor microenvironment heterogeneity to optimize treatment efficacy and minimize off-target effects. Methods: The DECODE-ADC project integrates single-cell and spatial multi-omics data from FFPE tissues of EV-treated patients, encompassing matched pairs of primary tumors and metastatic biopsy from a cohort of 20 patients. Single-nucleus RNA sequencing and high-resolution spatial transcriptomics were used to profile various ADC targets with additional markers of the tumor microenvironment and drug resistance (over 5000 gene panels). To bridge translational gaps, we employed machine learning (ML) models, such as variational autoencoders, to predict molecular signatures from routine Hematoxylin and Eosin (H&E) stainings, enabling cost-effective clinical application. Results: Using spatial single-cell multiomics, we identified novel spatial domains regulating ADC target expression and delineated tumor microenvironment-specific signals. Unique molecular signatures distinguished luminal from basal urothelial cancer subtypes, revealing conserved microdomains across metastatic sites. Importantly, machine learning models transferred high-resolution molecular insights to H&E images, enabling precise, scalable predictions of ADC target distribution. Conclusions: DECODE-ADC provides unprecedented insights into the spatial dynamics of ADC targets within the tumor microenvironment of mUC. By integrating advanced molecular profiling with ML-driven predictions, this study advances our understanding of ADC target expression and establishes a pathway for rational biomarker selection Our findings propose a foundation for precise ADC biomarker development and stratification strategies to improve treatment ADC responses, which ultimately paves the way for ADC precision oncology. Citation Format: Christoph Kuppe, Niklas Klümper, Osman Goni, Paul Kießling, Emilia Scheidereit, Samaneh Samiei, Thomas Büttner, Friedemann Zengerling, Nadine Gaisa, Christian Bolenz, Marieta Toma, Stefanie Zschäbitz, Viktor Grünwald, Michael Hölzel, Mridula Iyer, Nachiket Kashikar, Markus Eckstein. Decode the antibody-drug conjugate surfaceome in metastatic urothelial cancer (DECODE-ADC) [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 768.