======================================================================
WRN-MSI Pan-Cancer Atlas — Phase 6: Resistance-Aware Deployment Framework
======================================================================

Source: bioRxiv 10.64898/2026.01.22.700152v1 + Atlas Phases 1-5

1. THREE-DRUG CLINICAL LANDSCAPE
--------------------------------------------------
  VVD-214 (Vividion/Roche/Bayer): Covalent allosteric
    Binding: C727 allosteric pocket
    Trial: NCT06004245 (Phase 1)
    Status: Dose escalation + expansion; PRs reported (AACR 2025)

  HRO-761 (Novartis): Non-covalent allosteric
    Binding: Allosteric pocket (non-covalent)
    Trial: NCT05838768 (Phase 1/1b)
    Status: Mono + combos (irinotecan, tislelizumab)

  NDI-219216 (Nimbus): Non-covalent (distinct binding mode)
    Binding: Distinct from HRO-761 and VVD-214
    Trial: NCT06898450 (Phase 1/2)
    Status: Part A dose escalation COMPLETE (Dec 2025); no DLTs

2. RESISTANCE MUTATION PROFILES — DIVERGENT
--------------------------------------------------
  KEY FINDING: Resistance profiles DIVERGE between compounds.
  Single-allele (heterozygous) mutations at binding site = sufficient for resistance.

  [compound_specific] VVD-214_specific:
    Mutations at C727 covalent binding site that prevent warhead engagement
    Affects: VVD-214
    Spares: HRO-761;NDI-219216
    Implication: Switch to non-covalent inhibitor (HRO-761 or NDI-219216)

  [compound_specific] HRO-761_specific:
    Mutations in non-covalent allosteric pocket reducing binding affinity
    Affects: HRO-761
    Spares: VVD-214;NDI-219216
    Implication: Switch to covalent inhibitor (VVD-214) or distinct-binding NDI-219216

  [pan_resistance] pan_resistance:
    Mutations affecting shared structural features required by all compounds
    Affects: VVD-214;HRO-761;NDI-219216
    Spares: none
    Implication: Combination strategy needed (NHEJ inhibitor, WIP1 inhibitor)

3. RESISTANCE EMERGENCE PROBABILITY (12-month treatment)
--------------------------------------------------
  Model: P(resistance) = 1 - exp(-N * mu * L)
  N=1e9 tumor cells, L=1800bp WRN helicase domain

  COADREAD: MSI-H 40.0 mut/Mb (11.4x MSS) → P(resistance)=1.0000 [VERY_HIGH]
  UCEC: MSI-H 65.0 mut/Mb (26.0x MSS) → P(resistance)=1.0000 [VERY_HIGH]
  STAD: MSI-H 30.0 mut/Mb (7.5x MSS) → P(resistance)=1.0000 [VERY_HIGH]
  PAAD: MSI-H 25.0 mut/Mb (16.7x MSS) → P(resistance)=1.0000 [VERY_HIGH]
  OV: MSI-H 20.0 mut/Mb (10.0x MSS) → P(resistance)=1.0000 [VERY_HIGH]
  CHOL: MSI-H 20.0 mut/Mb (13.3x MSS) → P(resistance)=1.0000 [VERY_HIGH]
  BLCA: MSI-H 35.0 mut/Mb (5.8x MSS) → P(resistance)=1.0000 [VERY_HIGH]
  BRCA: MSI-H 15.0 mut/Mb (10.0x MSS) → P(resistance)=1.0000 [VERY_HIGH]
  SKCM: MSI-H 50.0 mut/Mb (4.2x MSS) → P(resistance)=1.0000 [VERY_HIGH]

  *** MODEL CAVEAT (added 2026-03-23 per validation finding #1648) ***
  CRITICAL: This model gives P(resistance)=1.0 for ALL tumor types INCLUDING MSS.
  Even the lowest MSS rate (1.5 mut/Mb) yields N*mu*L=540, far exceeding the
  saturation threshold. The VERY_HIGH risk category is therefore UNINFORMATIVE
  as a discriminator between MSI-H and MSS — it applies universally.

  This model should be interpreted as a WORST-CASE UPPER BOUND, not a
  discriminative risk stratification tool. The model overestimates because:
    (a) L=1800bp assumes ANY mutation in the helicase domain confers resistance.
        The preprint identifies specific binding-site positions — the effective
        target region (L_eff) is likely 10-50bp, not 1800bp.
    (b) No fitness cost is modeled. Resistance mutations often carry a
        replication or catalytic fitness penalty that limits clonal expansion.
    (c) N=1e9 assumes all cells are dividing and equally exposed to drug.

  The MSI-H vs MSS distinction remains biologically important (4-26x elevated
  mutation rates accelerate resistance KINETICS), but this model cannot
  quantify the difference because it saturates at both rates.

  NOTE: The combo vs monotherapy deployment recommendations (Section 6) are
  actually driven by mutation rate DATA AVAILABILITY, not by model output.
  Tumor types with published MSI-H mutation rates were assigned VERY_HIGH
  and recommended for combination; those without data were assigned UNKNOWN
  and defaulted to monotherapy. This is acknowledged as a data limitation,
  not a model-derived recommendation.

4. COMBINATION VULNERABILITY CROSS-REFERENCE (Phase 3 + Phase 4)
--------------------------------------------------
  Resistance preprint identifies NHEJ + WIP1 as vulnerabilities in WRN-resistant cells.
  Cross-referencing with our genome-wide MSI-H dependency screen (Phase 3):

  PRKDC (NHEJ, DNA-PK catalytic subunit): not tested in Phase 3
  XRCC4 (NHEJ, NHEJ ligation complex): d=-0.244, FDR=3.207e-01, more_essential_MSI-H
  LIG4 (NHEJ, DNA Ligase IV): d=0.148, FDR=7.002e-01, less_essential_MSI-H
  XRCC5 (NHEJ, Ku80 (DSB recognition)): d=0.256, FDR=4.682e-01, less_essential_MSI-H
  XRCC6 (NHEJ, Ku70 (DSB recognition)): d=-0.228, FDR=4.297e-01, more_essential_MSI-H
  NHEJ1 (NHEJ, XLF (NHEJ accessory factor)): d=0.100, FDR=9.141e-01, less_essential_MSI-H
  DCLRE1C (NHEJ, Artemis nuclease): d=0.190, FDR=3.714e-01, less_essential_MSI-H
  PPM1D (WIP1, WIP1 phosphatase (p53 negative regulator)): d=-0.292, FDR=4.303e-01, more_essential_MSI-H
  TP53BP1 (DSB_repair_choice, 53BP1 (promotes NHEJ over HR)): d=0.007, FDR=9.211e-01, less_essential_MSI-H

  AZD-7648 (DNA-PKi) MSI-H sensitivity from Phase 4:
    Uterus: d=-1.012, FDR=8.377e-02
    Pan-cancer (pooled): d=-0.526, FDR=1.847e-02 ***
    Bowel: d=-0.501, FDR=3.549e-01
  *** CONVERGENCE: Phase 4 independently found DNA-PKi MSI-H selective
      Resistance preprint independently identifies DNA-PK as combination partner

5. COMPOUND SEQUENCING STRATEGY
--------------------------------------------------
  Scenario: First-line VVD-214 → C727 resistance
    Switch to: HRO-761 or NDI-219216
    Rationale: Non-covalent inhibitors bind different pocket; C727 mutations don't affect their binding
    Combination: AZD-7648 (DNA-PKi) — NHEJ dependency in WRN-inhibited cells

  Scenario: First-line HRO-761 → allosteric pocket resistance
    Switch to: VVD-214 or NDI-219216
    Rationale: Covalent mechanism (VVD-214) or distinct binding site (NDI-219216) unaffected
    Combination: AZD-7648 (DNA-PKi) — NHEJ dependency in WRN-inhibited cells

  Scenario: First-line NDI-219216 → distinct-site resistance
    Switch to: VVD-214 or HRO-761
    Rationale: Distinct binding mode means resistance mutations are non-overlapping
    Combination: AZD-7648 (DNA-PKi) — NHEJ dependency in WRN-inhibited cells

  Scenario: Pan-resistance (all WRN inhibitors)
    Switch to: None (WRN pathway exhausted)
    Rationale: Structural changes affect all binding modes — must target downstream vulnerabilities
    Combination: DNA-PKi + immunotherapy; target NHEJ pathway or WIP1 phosphatase

6. PER-TUMOR-TYPE DEPLOYMENT RECOMMENDATIONS
--------------------------------------------------
  COADREAD (WRN d=-2.844, ~23,532 pts/yr):
    Resistance risk: VERY_HIGH
    Strategy: WRN inhibitor + DNA-PKi combination
    Rationale: High resistance risk — consider upfront combination with DNA-PKi (AZD-7648) to suppress NHEJ-dependent resistance escape
    In trial: VVD-214;HRO-761

  STAD (WRN d=-2.833, ~5,297 pts/yr):
    Resistance risk: VERY_HIGH
    Strategy: WRN inhibitor + DNA-PKi combination
    Rationale: High resistance risk — consider upfront combination with DNA-PKi (AZD-7648) to suppress NHEJ-dependent resistance escape
    In trial: VVD-214;HRO-761

  DLBC (WRN d=-2.833, ~3,950 pts/yr):
    Resistance risk: UNKNOWN
    Strategy: WRN inhibitor monotherapy
    Rationale: Lower resistance risk or insufficient mutation rate data — standard WRN inhibitor monotherapy
    NOT in current WRN trials

  SKCM (WRN d=-2.833, ~905 pts/yr):
    Resistance risk: VERY_HIGH
    Strategy: WRN inhibitor + DNA-PKi combination
    Rationale: High resistance risk — consider upfront combination with DNA-PKi (AZD-7648) to suppress NHEJ-dependent resistance escape
    NOT in current WRN trials

  BRCA (WRN d=-2.833, ~621 pts/yr):
    Resistance risk: VERY_HIGH
    Strategy: WRN inhibitor + DNA-PKi combination
    Rationale: High resistance risk — consider upfront combination with DNA-PKi (AZD-7648) to suppress NHEJ-dependent resistance escape
    NOT in current WRN trials

  ESCA (WRN d=-2.833, ~603 pts/yr):
    Resistance risk: UNKNOWN
    Strategy: WRN inhibitor monotherapy
    Rationale: Lower resistance risk or insufficient mutation rate data — standard WRN inhibitor monotherapy
    NOT in current WRN trials

  BLCA (WRN d=-2.833, ~582 pts/yr):
    Resistance risk: VERY_HIGH
    Strategy: WRN inhibitor + DNA-PKi combination
    Rationale: High resistance risk — consider upfront combination with DNA-PKi (AZD-7648) to suppress NHEJ-dependent resistance escape
    NOT in current WRN trials

  PAAD (WRN d=-2.833, ~398 pts/yr):
    Resistance risk: VERY_HIGH
    Strategy: WRN inhibitor + DNA-PKi combination
    Rationale: High resistance risk — consider upfront combination with DNA-PKi (AZD-7648) to suppress NHEJ-dependent resistance escape
    IO note: Low IO response (pembro ORR 18%) — strong case for WRN inhibitor combination
    NOT in current WRN trials

  CHOL (WRN d=-2.833, ~342 pts/yr):
    Resistance risk: VERY_HIGH
    Strategy: WRN inhibitor + DNA-PKi combination
    Rationale: High resistance risk — consider upfront combination with DNA-PKi (AZD-7648) to suppress NHEJ-dependent resistance escape
    NOT in current WRN trials

  KIRP (WRN d=-2.833, ~327 pts/yr):
    Resistance risk: UNKNOWN
    Strategy: WRN inhibitor monotherapy
    Rationale: Lower resistance risk or insufficient mutation rate data — standard WRN inhibitor monotherapy
    NOT in current WRN trials

  LUAD (WRN d=-2.833, ~235 pts/yr):
    Resistance risk: UNKNOWN
    Strategy: WRN inhibitor monotherapy
    Rationale: Lower resistance risk or insufficient mutation rate data — standard WRN inhibitor monotherapy
    NOT in current WRN trials

  KIRC (WRN d=-2.833, ~163 pts/yr):
    Resistance risk: UNKNOWN
    Strategy: WRN inhibitor monotherapy
    Rationale: Lower resistance risk or insufficient mutation rate data — standard WRN inhibitor monotherapy
    NOT in current WRN trials

  LIHC (WRN d=-2.833, ~123 pts/yr):
    Resistance risk: UNKNOWN
    Strategy: WRN inhibitor monotherapy
    Rationale: Lower resistance risk or insufficient mutation rate data — standard WRN inhibitor monotherapy
    NOT in current WRN trials

  HNSC (WRN d=-2.833, ~116 pts/yr):
    Resistance risk: UNKNOWN
    Strategy: WRN inhibitor monotherapy
    Rationale: Lower resistance risk or insufficient mutation rate data — standard WRN inhibitor monotherapy
    NOT in current WRN trials

  LGG (WRN d=-2.833, ~50 pts/yr):
    Resistance risk: UNKNOWN
    Strategy: WRN inhibitor monotherapy
    Rationale: Lower resistance risk or insufficient mutation rate data — standard WRN inhibitor monotherapy
    NOT in current WRN trials

  UCEC (WRN d=-1.086, ~20,974 pts/yr):
    Resistance risk: VERY_HIGH
    Strategy: WRN inhibitor + DNA-PKi combination
    Rationale: High resistance risk — consider upfront combination with DNA-PKi (AZD-7648) to suppress NHEJ-dependent resistance escape
    In trial: VVD-214;HRO-761

  OV (WRN d=-2.833, ~39 pts/yr):
    Resistance risk: VERY_HIGH
    Strategy: WRN inhibitor + DNA-PKi combination
    Rationale: High resistance risk — consider upfront combination with DNA-PKi (AZD-7648) to suppress NHEJ-dependent resistance escape
    IO note: Low IO response (pembro ORR 33%) — strong case for WRN inhibitor combination
    In trial: VVD-214;HRO-761

  UCS (WRN d=-1.086, ~3,597 pts/yr):
    Resistance risk: UNKNOWN
    Strategy: WRN inhibitor monotherapy
    Rationale: Lower resistance risk or insufficient mutation rate data — standard WRN inhibitor monotherapy
    NOT in current WRN trials

7. KEY CONCLUSIONS
--------------------------------------------------
  1. Resistance to WRN inhibitors is INEVITABLE in MSI-H tumors due to elevated mutation rates
  2. Divergent resistance profiles enable RATIONAL SEQUENTIAL therapy
  3. DNA-PKi (AZD-7648) is a convergently validated combination partner:
     - Phase 4: MSI-H selective (d=-0.526, FDR=0.018)
     - Resistance preprint: NHEJ dependency in WRN-resistant cells
  4. High mutation rate tumor types (UCEC, SKCM, COADREAD) need upfront combination
  5. NDI-219216 provides third non-cross-resistant option for pan-resistance scenarios
  6. PAAD patients (low IO ORR + moderate resistance risk) are strongest candidates
     for WRN inhibitor combination strategies

8. LIMITATIONS AND BLIND SPOTS (added 2026-03-23)
--------------------------------------------------
  (a) Resistance model saturation: P(resistance)=1.0 for all tumor types
      including MSS (see Section 3 caveat). Model is a worst-case bound only.

  (b) UCS mutation rate gap: UCS has no published MSI-H mutation rate data
      despite sharing the Uterus lineage with UCEC (65 mut/Mb). UCS is
      classified UNKNOWN risk and defaulted to monotherapy, but may warrant
      combination if UCEC-like rates apply. This needs confirmation.

  (c) NDI-219216 efficacy data: Phase 1/2 Part A (NCT06898450) completed
      dose escalation with no DLTs (Dec 2025), but efficacy data are NOT
      yet available. Deployment recommendations that include NDI-219216 as
      a sequential option are based on its distinct binding mode, not on
      demonstrated clinical activity. Flag for update when data emerge.

  (d) DLBC MSI-H prevalence uncertainty: The ~3,950 pts/yr estimate for
      DLBC is based on only 2/41 TCGA samples classified as MSI-H (4.9%).
      This is subject to high sampling uncertainty (95% CI ~0.6-16.5%).
      The true addressable population could be substantially smaller.

  (e) Effective target region: L=1800bp (full helicase domain) is almost
      certainly an overestimate. The preprint identifies specific resistance
      positions. A refined model using L_eff from experimentally validated
      positions would produce more informative risk stratification.

  (f) PRKDC (DNA-PK) absent from Phase 3 CRISPR screen: PRKDC was not
      tested in the genome-wide screen, so we lack a direct genetic
      validation of DNA-PK dependency in MSI-H. The AZD-7648 drug data
      (Phase 4) provides pharmacological evidence but not genetic confirmation.
