Triangular motif enrichment as a structural signature of oncogenic signalling: a comparative network analysis across seven cancer types

Therapeutic resistance is a central challenge in oncology, yet the structural basis of signalling network robustness remains poorly characterised. Here we applyTopostability, a graph-theoretic framework based on per-edge triangle counts, to compare seven oncogenic signalling networks spanning distinct driver mutations(HER2 amplification, EGFR mutation, KRAS G12D/V, BRAF V600E, WNT/APC loss, TGF-β EMT switch, mTOR/TSC activation) against three inflammatory pathological networks and fourteen neural circuits. We find that oncogenic networks are significantly enriched in triangular motifs relative to degree-preserving random graphs (6/7 networks, p < 0.05, dual null model), while inflammatory networks show no equivalent enrichment (0/3). The two groups are perfectly separated on all three triangular metrics (Cliff’s δ = ±1.00, Cohen’s d > 3.5), a distinction not captured by classical graph metrics such as betweenness centrality or node degree.Within-network perturbation analysis reveals that combinatorial therapies targeting structurally disconnected triangular modules exhibit structural supra-additivity,providing a topological rationale for observed combination synergies. These results constitute a candidate structural signature of oncogenic signalling architecture, orthogonal to existing pathway-specific descriptions.