DNV-TC: a developmental network vulnerability-based classification system for precision oncology.

INTRODUCTION: Current tumor classification systems based on histology and molecular subtypes inadequately capture the functional complexity of cancer biology. Tumors frequently reactivate embryonic developmental programs, becoming dependent on master regulators that govern pluripotency, differentiation and tissue morphogenesis. This developmental reactivation creates architectural vulnerabilities in gene regulatory networks that can be therapeutically exploited. METHODS: We propose DNV-TC, a four-dimensional classification system that stratifies tumors according to the reactivated developmental program, network architectural vulnerability, cascade expansion phenotype and metastatic propensity. The system integrates developmental biology with network medicine through three quantitative indices computed on multi-layer gene regulatory networks: the tumor developmental regulatory impact (T-DRI), tumor disease network vulnerability (TD-NV) and cascade expansion index-tumor (CEI-T). DNV-TC was applied retrospectively to representative tumor types using published molecular and clinical data. RESULTS: Tumors with high network vulnerability showed marked responses to targeted therapies (87% response rate) but rapidly developed resistance (median 6-8 months), while tumors with reactivated pluripotency programs displayed cancer stem-cell characteristics and therapeutic resistance. Network vulnerability class retained independent prognostic value after adjustment for stage, grade and age (HR 2.3, 95% CI 1.8-2.9, p < 0.001), and metastatic propensity class outperformed TNM staging for 5-year metastasis-free survival prediction (AUC 0.842 versus 0.712, p < 0.001). DISCUSSION: DNV-TC provides a mechanistically grounded classification system that bridges developmental biology and precision oncology, thus enabling rational selection of therapeutic targets and combination strategies. The present analyses are retrospective and in silico, and prospective external validation is required.