We present a structural pharmacology framework for cancer drug discovery based on perturbation analysis of multi-layer signaling networks. Across six cancer types (KRAS-NSCLC, CRC,PDAC, GBM, BRAF-melanoma, EGFR-NSCLC), we compute the delta-r of each signaling protein — the change in inter-layer Pearson correlation upon node removal — and classify proteinsas structural dampers (delta-r > 0) or anchors (delta-r 2.0 under both null types); (2) MEK1’s regime-switch betweenODS and CLASS 6b cancers survives degree correction in CRC and BRAF-melanoma (thetwo networks with clearest structural regimes), while it is near-neutral in KRAS-NSCLC; (3)SOS1 is the top degree-corrected structural damper in EGFR-mutant NSCLC (zδr = +2.47under Null1, +1.43 under Null2), validating it as the primary structural drug discovery gap;(4) drug pairs targeting structurally complementary positions show significantly greater Blisssynergy in the NCI-ALMANAC screen (p = 0.0003, a pharmacological result unaffected by thenull model correction); (5) MEK1 is a high-variance structural node whose degree-correctedclassification is topology-sensitive, resolving its apparent instability as expected behavior for apathway-convergence protein with both local and global influence; and (6) cross-cancer pharmacogenomic dissociation experiments (D8–D9) establish the structural pharmacology boundarycondition: zδr-based predictions hold when the drug-target is directly co-pathway with the cancer’s dependency-driver, but fail for indirect pathway activation (STAT3 via EGFR in GBM,p = 0.55) and off-pathway structural targets (CDK4 in PDAC, p = 0.30). MEK inhibitorsshow the largest validated pharmacogenomic dissociation (d = −0.62, p < 10−10) across BRAFmelanoma vs. GBM — confirming the anchor/damper regime-switch as a pharmacologicallydetectable structural difference; (7) structural fingerprint clustering (D10) reveals that samedriver cancers are less structurally similar (mean cosine +0.046) than different-driver cancers1(mean cosine +0.185), quantifying the Driver-Topology Independence Law; PDAC and BRAFmelanoma share a p53-pathway structural convergence (cosine +0.516) despite opposite driversand regimes; the consequent MDM2 inhibitor repurposing prediction (D11, pre-reg 71ee0d6)reveals a third empirical failure mode — state-dependent pharmacologic accessibility (SDPA):PDAC’s ∼70% TP53 mutation rate collapses signal execution capacity, overriding the strongeststructural damper signal in the atlas (z1 = +1.01); BRAF-melanoma, where three conditionsalign (CDKN2A loss, MAPK hyperactivation, intact apoptotic execution), confirms the prediction via idasanutlin (d = −0.35, p < 10−10); and (8) DepMap CRISPR essentiality testing (D22–D23) establishes and replicates the DAMPER-Essentiality Decoupling principle:SOS1 is the top structural DAMPER in KRAS-MT LUAD (zδr = +4.184, D21) yet is NOTCRISPR-essential in KRAS-MT cells (d = +0.522, p = 0.29), while KRAS itself is stronglyCRISPR-essential (d = −1.336, p = 0.0002); RAF1 is less essential in BRAF-V600E melanoma(d = +1.473, p = 0.021), confirming BRAF-V600E monomer signaling bypasses the RAF1 heterodimerization requirement; D23 confirms the ANCHOR=essential mapping in the RTK-drivenregime: EGFR (strongest structural ANCHOR, zδr = −2.047) is CRISPR-essential in EGFRMT LUAD (d = −0.962, pone = 0.098; all 4 pre-registered hypotheses confirmed by pre-specifiedlarge-effect/directional criteria appropriate for small n, not conventional p < 0.05), SOS1 remains non-essential in the same EGFR-MT context (d = −0.306, Decoupling replicates), andKRAS is less essential in EGFR-MT LUAD (d = +1.142) — the mirror of its strong essentiality in KRAS-MT (d = −1.336, D22) — demonstrating bidirectional ANCHOR-essentialitysymmetry across driver regimes; structural DAMPER status predicts pharmacological networkdisruption but not genetic survival dependence, while ANCHOR status predicts CRISPR essentiality across both GTPase-driven and RTK-driven cancer contexts — providing a structuralcriterion for distinguishing pharmacological inhibition (DAMPER targets) from genetic dependency (ANCHOR targets) in drug discovery. The complete boundary condition framework (6empirical cases, 3 failure modes, all pre-registered) constitutes the first systematic, pre-registeredspecification of the scope and limits of structural network pharmacology.