Longitudinal and Multi-Omics Analysis of Metastatic Melanoma Reveals Dynamic Tumor Evolution and Multi-Level Heterogeneity

Abstract

Melanoma is the most lethal cutaneous cancer. Advanced-stage (Stage III-IV) melanoma has a 10-year survival rate of less than 30%. Despite recent advances in targeted and immunotherapies, clinical outcomes of melanoma patients remain highly variable, largely due to its highly heterogeneous nature. This heterogeneity is observed in 3 levels: intratumoral, intertumoral, and inter-patient, complicating clinicians' efforts to identify consistent biomarkers and therapeutic targets. Therefore, traditional approaches that focus on individual genes or pathways are insufficient to capture the complexity of metastatic melanoma progression and treatment response. To address this limitation, integrative multi-omics strategies are required to systematically characterize tumor evolution and tumor-immune interactions across different biological scales. In this study, we analyzed two complementary melanoma patient cohorts. A longitudinal in-transit metastasis (ITM) cohort with whole-exome sequencing (WES) and bulk RNA-seq data was used to reconstruct phylogenetic trees and investigate tumor evolutionary dynamics over time. A second cohort from the Human Tumor Atlas Network (HTAN), comprising WES, single-nucleus RNA sequencing (snRNA-seq), spatial transcriptomics (MERFISH), and spatial proteomics data, enables the characterization of tumor cell states and immune interactions across defined clinical response groups. We integrated genomic, transcriptomic, and spatial-omics analyses to identify recurrent patterns of tumor progression, including clinically relevant clonal expansion, biological or functional pathways, mutational signatures, and melanoma state transitions. Our results reveal that melanoma progression is characterized by dynamic clonal evolution that is associated with therapeutic interventions. During melanoma progression, we observed a decreased genomic heterogeneity but increased phenotypic and mutational process diversities over time. Non-progressor tumors exhibited enhanced antigen presentation and interferon signaling, highlighting the importance of tumor-immune interactions in treatment outcomes. We also identified biologically interpretable tumor programs that capture coherent tumor-intrinsic states beyond single-gene signals, which can be validated across modalities and applied clinically. Together, this study demonstrates that melanoma progression is driven by coordinated genomic evolution, phenotypic plasticity, and tumor-immune interactions, which underscores the value of multi-modal integration for understanding complex cancers.