STRC Research

STRC mRNA-LNP Strategy B Audiogram Rescue

8 min read

STRC mRNA-LNP Strategy B Audiogram Rescue

Closes the clinical-endpoint layer for Strategy B (STRC mRNA-LNP Strategy B Full-Length) symmetrically with STRC mRNA-LNP Audiogram Rescue (Strategy A). Same three-layer composition: per-OHC therapeutic flag × tonotopic LNP distribution × ABR transfer function. Two findings: (1) once per-OHC threshold passes, Strategy B’s audiogram projection is numerically identical to Strategy A at matched eff_mean — the tonotopic × ABR framework is mechanism-agnostic (for Misha at 20% OHC targeting, 21 dB basal recovery either way); (2) the decisive asymmetry is the DFNB16 biallelic null scenario, where Strategy A cannot rescue any OHC (no pre-mRNA substrate) while Strategy B achieves 28 dB weighted improvement at 20% targeting — this is the patient population that unambiguously requires Strategy B. For Misha compound het, both strategies produce the same audiogram when per-OHC passes, but Strategy B needs ~30× more extracellular dose per OHC.

Method

Three layers, identical framework to Strategy A audiogram rescue:

  1. Per-OHC absolute criterion (new to Strategy B): transfected-OHC total STRC ≥ 1× WT (functional) or ≥ 2× WT (robust). Computed as s_endo_frac + s_exog_tx_trough / s_WT. Absolute threshold, not fold-over-baseline — this is the meaningful substitution relative to Strategy A.

  2. Tonotopic LNP distribution: basal 2× / mid 1× / apical 0.5× × eff_mean, clamped at 1.0. Same parametric model as Strategy A (round-window injection basal bias).

  3. ABR transfer function: ABR_dB = A − B·log10(functional_fraction + C), same 8-point calibration (Bredberg 1968, Schuknecht 1993, DB-OTO OTOF). DFNB16 untreated baseline = 85 dB; WT baseline = 20 dB (normal); Misha sloping 65/55/40.

Reference regimens per (scenario, threshold) from strc_mrna_strategy_b_pkpd.json:

Scenario1× WT threshold2× WT threshold
WT referenceQ12W × 200 (trivially passes)m1ψ Q2W × 2,000
DFNB16 nullm1ψ Q2W × 2,000 (2.6M/yr extra/OHC)m1ψ Q3W × 5,000 (4.25M/yr extra)
Misha (s_endo=0.15)m1ψ Q2W × 2,000 (2.6M/yr extra/OHC)m1ψ Q3W × 5,000 (4.25M/yr extra)

Floor logic (same as Strategy A): if baseline audiogram in a zone is better than rescue projection, post = baseline (no change).

Results — DFNB16 biallelic null 85 dB baseline

Strategy A: cannot treat — the null alleles lack pre-mRNA substrate for RBM24 splicing correction; per-OHC rescue fails at every LNP efficiency. Weighted improvement = 0 dB.

Strategy B (m1ψ Q2W × 2,000 mol/OHC intra reference, 1× WT threshold):

LNP scenarioBasalMidApicalWeighted postWeighted impResponder?
Untargeted 0.8%83 dB85 dB85 dB84.5 dB0.5 dBno
Cochlear-tropic 5%69 dB76 dB81 dB75.6 dB9.4 dBno
OHC-targeted 20%44 dB58 dB69 dB57.0 dB28.0 dByes (basal)
Hypothetical 50%23 dB39 dB54 dB38.7 dB46.3 dByes (all)
Anc80L65 67%23 dB33 dB48 dB34.6 dB50.4 dByes (all)

This is the table Strategy A cannot produce. DFNB16 biallelic null homozygotes are the patient population for whom Strategy B is not just an alternative but a requirement.

Results — Misha’s sloping audiogram (65/55/40 dB)

Strategy B (m1ψ Q2W × 2,000 mol/OHC intra, 1× WT threshold — passes because s_endo=0.15 + 1.49 = 1.64):

LNP scenarioBasal 65→Mid 55→Apical 40→Weighted imp
Untargeted 0.8%65 dB55 dB40 dB0.0 dB
Cochlear-tropic 5%65 dB55 dB40 dB0.0 dB
OHC-targeted 20%44 dB55 dB40 dB6.9 dB (21 dB basal, responder)
Hypothetical 50%23 dB39 dB40 dB19.2 dB
Anc80L65 67%23 dB33 dB40 dB21.4 dB

Numerically identical to Strategy A’s Misha audiogram. Once per-OHC threshold passes (Strategy A: 2× fold, Strategy B: s_endo + s_exog ≥ 1), the zone-rescue projection is determined entirely by the tonotopic distribution × ABR transfer function — both mechanism-agnostic layers.

The difference is in dose, not audiogram:

MetricStrategy A (Misha)Strategy B (Misha, 1× threshold)
Reference regimenm1ψ Q6W × 200m1ψ Q2W × 2,000
Annual intra per OHC1,800 mol52,000 mol
Annual extra per OHC90,000 mol2.6×10⁶ mol
Per-OHC passes2.17× fold (= 0.33× WT absolute for Misha)1.64× WT absolute
Audiogram @ 20% LNP21 dB basal, 6.9 dB weighted21 dB basal, 6.9 dB weighted

Interpretation: Strategy A’s 2.17× fold for Misha actually delivers only 0.33× WT absolute (0.15 × WT endo × 2.17 = 0.33× WT). This is below the 1× WT functional threshold — per-OHC rescue for Misha under Strategy A is almost certainly subtherapeutic in absolute terms despite meeting the fold-based criterion. Strategy B’s 1.64× WT absolute is above threshold. The Strategy A audiogram numbers for Misha may be optimistic; Strategy B’s are on firmer footing.

Results — WT reference (normal hearing 20 dB baseline)

All scenarios give 0 dB improvement (floored at 20 dB baseline). Expected: nothing to rescue in normal-hearing controls. Included for completeness.

Interpretation

  1. DFNB16 biallelic null is the definitive Strategy B indication. 28 dB basal-band recovery at 20% OHC-targeted LNP, climbing to 46-50 dB weighted at 50-67% LNP. Strategy A produces zero improvement for this population because it has no mRNA substrate to splice. This is a categorical distinction, not a magnitude one.

  2. For compound heterozygotes (Misha), Strategy A and Strategy B converge on audiogram at matched LNP eff_mean — the ABR transfer function doesn’t know the mechanism. The three-layer composition filters out mechanism once per-OHC passes. What differs is (a) which patients can be treated, (b) the dose burden, (c) whether per-OHC actually passes in absolute terms.

  3. The per-OHC absolute-vs-fold reframing matters for Misha. Strategy A’s “2.17× fold trough” sounds therapeutic but means 0.33× WT absolute for Misha because his endogenous baseline is 0.15× WT hypomorphic. Strategy B reaches 1.64× WT absolute at modest dose — unambiguously above the 1× WT functional floor. The fold-based framing in the Strategy A audiogram model was flattering; the absolute framing in Strategy B is stricter.

  4. Strategy B’s dose burden is 30× Strategy A’s for matched audiogram (2.6M vs 90K mol/yr extracellular per OHC). A 3-dose cochlea-scale manufacturing comparison:

    • Strategy A Q6W × 200 mol/OHC × 12,000 OHC × 17 doses/yr × 1.5 kb mRNA ≈ 0.15 mg mRNA/cochlea/yr
    • Strategy B Q2W × 2,000 mol/OHC × 12,000 OHC × 26 doses/yr × 6 kb mRNA ≈ 3.6 mg mRNA/cochlea/yr
    • Both practically manufacturable; Strategy B just needs more mass.

  5. The LNP tropism ladder is the same for both strategies: 20% = minimum-therapeutic gate (Misha basal-band responder), 50% = AAV-parity (full-band responder, weighted imp ~19 dB), 67% = Anc80L65 reference. Engineering target for LNP research is invariant.

  6. Unmodelled Strategy B advantage: non-cell-autonomous secretion. Stereocilin is a secreted extracellular protein — if it diffuses from transfected to adjacent untransfected OHCs (within 10 μm radius = immediate neighbors), the effective functional fraction doubles or triples. This would break the audiogram ceiling at low eff_mean — e.g., 5% cochlear-tropic LNP with 3× secretion bonus = 15% effective functional coverage, reaching 1.0 weighted improvement that the mechanism-agnostic model currently says is flat zero.

Clinical decision framework

Patient profileStrategy A viableStrategy B viableRecommendation
Biallelic null (e.g., two deletion alleles)no (no substrate)yesStrategy B only
Compound het with one null (Misha)partial (maternal only, absolute subtherapeutic)yesStrategy B primary; A as adjunct
Biallelic hypomorphic (missense/missense)yesyesA for dose economy; B if A caps below threshold
Single hypomorphic (e.g., homozygous E1659A)yesyesA preferred (dose-frugal)

Limitations

  • All Strategy A audiogram limitations carry over (tonotopic gradient parametric assumption, uniform rescue within zones, pass/fail per-OHC, no anti-PEG, no secretion bonus).
  • Misha’s s_endo_frac = 0.15 is a construct combining paternal-null + maternal-E1659A-residual assumptions. Actual cochlear STRC measurement is impossible; range 0.1-0.2 is best estimate. Sensitivity to this parameter: at s_endo = 0.2, Misha already passes 1× WT with smaller Strategy B dose.
  • Floor effect means Strategy B’s high-LNP benefit is masked in Misha’s mid/apical zones — his residual function clears the rescue projection there. For a patient with harder mid/apical loss, Strategy B would show more full-band benefit.
  • Same binary per-OHC pass/fail as Strategy A. A continuous functional-rescue model (Hill-sigmoid from absolute STRC level to amplification gain) would discriminate high vs marginal rescue within the “passes” class.
  • Non-cell-autonomous secretion not modelled; could shift Misha’s effective LNP-coverage by 2-3× without dose change.

Next steps

  1. Model non-cell-autonomous STRC secretion: explicit diffusion model around each transfected OHC. First-order approximation: each transfected OHC rescues N neighbors (N ≈ 3-6 from OHC hexagonal packing); effective functional fraction = min(1.0, eff_zone · (1 + N)). Could drop the LNP-tropism engineering target from 20% to 5-8%.
  2. Tonotopic gradient sensitivity sweep: same as flagged in Strategy A audiogram note. Basal-bias ranges from flat (uniform distribution) to steep (3:1:0.3) would bound Misha’s basal-recovery uncertainty.
  3. Continuous per-OHC rescue curve: replace pass/fail at 1× or 2× WT with sigmoid amplifier_gain(total_STRC/WT). Deeper rescue (Strategy B at 2× WT) then outscores marginal (Strategy A at 1× effective) in audiogram dB — restoring the Strategy B preference within compound het patients.
  4. Misha-specific audiogram: replace approximated 65/55/40 with measured PTA. Re-run.
  5. Strategy A+B co-formulation: pack both RBM24 mRNA (for maternal-allele amplification) and full-length STRC mRNA (for de novo from paternal null) in the same LNP. For Misha the additive contribution may exceed either alone.

Replication

cd ~/STRC/models
/opt/miniconda3/bin/python3 strategy_b_audiogram_rescue.py
# outputs: strategy_b_audiogram_rescue.json

Files / Models

  • ~/STRC/models/strategy_b_audiogram_rescue.py — composition script
  • ~/STRC/models/strategy_b_audiogram_rescue.json — per-zone predictions across scenarios × LNP targeting × thresholds
  • ~/STRC/models/strc_mrna_strategy_b_pkpd.json — per-OHC PK/PD input
  • ~/STRC/models/abr_transfer_model.py — shared ABR transfer calibration

Ranking delta

Applied against STRC Hypothesis Ranking (2026-04-21):

  • STRC mRNA-LNP Strategy B Full-Length: no change — still S-tier; this proof confirms clinical-endpoint plausibility (21 dB basal at 20% LNP for Misha, 28 dB weighted for biallelic null).
  • STRC mRNA Therapy Hypothesis (Strategy A): no tier change, but evidence column updated — Strategy A’s Misha audiogram numbers exposed as optimistic (2.17× fold = 0.33× WT absolute, below functional threshold); backburner position reinforced.
  • No other hypotheses affected: ABR transfer function is mechanism-agnostic, so this proof is Strategy-specific.

Connections

Graph View

Backlinks