Differential network analysis is essential for revealing patterns of network rewiring across various conditions and understanding the biological mechanisms underlying complex diseases. Existing methods often rely on limited modalities and do not adequately capture temporal dynamics in gene expression. To address these limitations, we propose a novel framework for constructing multiple dynamic differential protein interaction networks, named Multi-DPIN, which combines temporal gene expression data with protein-protein interaction topology. Initially, active proteins and their interactions are identified using the 3 -sigma rule to construct dynamic disease networks and background networks. Subsequently, common topological structures shared with the background network are eliminated from each disease network. Finally, consistent structures across all disease networks are identified to establish the multiple dynamic differential protein interaction networks. We evaluated Multi-DPIN on publicly available cancer datasets (breast cancer and acute myeloid leukemia) and compared it with five baseline methods. Experimental results demonstrate that Multi-DPIN outperforms existing methods in identifying known oncogenes.