Background: Convergent studies have demonstrated that the topological structure of the brain network undergoes significant alterations in Alzheimer's Disease (AD). However, the underlying mechanisms driving these topological changes remain unclear. Network motifs, as fundamental components of brain networks, provide valuable insights into how disconnections may lead to alterations in the macroscale topological structure. Methods: This study focuses on undirected triangle motifs within the brain network, identifying 20 unique undirected triangle motifs based on edge strength and length derived from the regional radiomics similarity network (R2SN). To comprehensively capture the distribution of these motifs, we introduce a measure named Principal Motif Value (PMV) of the 20 motifs using principal component analysis (PCA). Results: Our findings reveal significant spatial heterogeneity in PMV across different brain regions. In addition, we identify reproducible alterations of PMV in AD, which were observed through three independent datasets, particularly in the inferior temporal gyrus, middle temporal gyrus, and parahippocampal gyrus. Notably, PMV demonstrates significant correlations with clinical manifestations and neurological features. Finally, we elucidate that alterations in PMV are associated with gene expression related to neuronal systems and synapsis features. Conclusion: These findings provide novel insights into the relationship between disconnection and the topological structure of the brain network in AD.