A spatial PK/PD framework predicts ADC response in solid tumors by integrating drug delivery, target engagement, and microenvironmental readiness

Background: Antibody-drug conjugates achieve their effect only when payload delivery, target accessibility, and microenvironmental readiness align within the tumor. Conventional biomarkers capture fragments of this biology, but not the spatial interplay that ultimately determines response. To address this gap, we conducted a translational spatial PK/PD study to determine whether high-resolution tissue mapping could predict early ADC activity in solid tumors. Methods: Tumor biopsies from patients and PDX models treated with a clinically relevant ADC were analyzed using an integrated spatial PK/PD workflow. Quantitative payload imaging by mass spectrometry imaging was combined with spatial proteomics and transcriptomics to map antigen distribution, stromal architecture, vascular organization, immune niches, and early pharmacodynamic activation. All data modalities were spatially registered and modeled using AI-based neighborhood analysis to generate a continuous delivery-engagement score for each tumor region. Importantly, pharmacodynamic readouts such as cleaved caspase-3 and gammaH2AX were evaluated in a blinded manner. Results: Across fourteen matched tumor samples, the analysis revealed a consistent pattern linking spatial drug behavior with early biological response. Regions showing strong spatial convergence between payload localization and antigen-rich tumor pockets exhibited significantly higher apoptotic signaling. Tumors displaying coherent payload influx along perfused vascular corridors were more likely to show downstream pharmacodynamic activation, regardless of their bulk target expression. In addition, microenvironments enriched for dendritic cells, CD8⁺ T cells, and interferon-responsive stromal states demonstrated enhanced sensitivity to the ADC even before radiographic change. When integrated into a unified spatial competency score, these features correctly classified early responses in eleven of fourteen samples. Conclusion: This study shows that a spatial PK/PD framework can predict ADC response by capturing the coordinated alignment of delivery, target engagement, and microenvironmental readiness within tissue. These findings provide a mechanistically grounded, clinically relevant approach for improving patient selection and advancing predictive biomarker strategies for next-generation ADCs.