CRD-4730, a CaMKII inhibitor for CPVT1, was halted at 7 patients on a business decision

What was tried

Cardurion Pharmaceuticals ran NCT06005428, a Phase 2A randomized, placebo-controlled, three-period crossover study of CRD-4730, an oral small-molecule inhibitor of calcium/calmodulin-dependent protein kinase II (CaMKII), in adults with catecholaminergic polymorphic ventricular tachycardia type 1 (CPVT1). The masking was quadruple, with investigator and subject blinded and the sponsor unblinded (ClinicalTrials.gov, NCT06005428). The protocol focused on safety, tolerability, pharmacokinetics, and pharmacodynamics rather than a hard arrhythmia endpoint. Primary outcomes were treatment-emergent adverse events, laboratory values, vital signs, physical examination, and electrocardiogram measurements from baseline through Day 22 (ClinicalTrials.gov, NCT06005428). Actual enrollment reached 7 participants (ClinicalTrials.gov, NCT06005428). The study was terminated. The posted reason was administrative: "The early termination of the clinical trial is due to an internal administrative and business decision, the study is not being terminated for safety reasons" (ClinicalTrials.gov, NCT06005428).

The biological hypothesis

CPVT1 is driven by gain-of-function variants in the cardiac ryanodine receptor RYR2 that promote diastolic calcium leak from the sarcoplasmic reticulum under adrenergic stress. CaMKII sits directly on this pathway. When activated by calcium and calmodulin, CaMKII phosphorylates RyR2 and other calcium-handling proteins, increasing open probability and feeding a cycle of leak, delayed afterdepolarizations, and triggered ventricular tachycardia (Cardurion, CaMKII program description). Preclinical work supports the node. Exercise training reduced CaMKII-dependent RyR2 phosphorylation and prevented ventricular tachycardia in a CPVT1 model (Kim et al., Cardiovascular Research, 2016), and reactive oxygen species-mediated CaMKII activation was shown to increase arrhythmogenic calcium release (doi:10.1161/CIRCULATIONAHA.120.048698). The therapeutic logic was that blocking CaMKII would blunt RyR2 hyperphosphorylation and raise the threshold for triggered activity without depending on beta-blockade alone.

What actually happened

The trial stopped after enrolling 7 of a planned crossover cohort, and no results were posted (ClinicalTrials.gov, NCT06005428). The sponsor stated the decision was administrative and business-driven and explicitly not safety-related. No efficacy readout, no pharmacodynamic biomarker, and no arrhythmia signal entered the public record. The openFDA FAERS database held only a single report mentioning CRD-4730, which is insufficient to compute any disproportionality signal (openFDA FAERS, queried 2026-05-31). In Open Targets, the CAMK2D association with CPVT1 scored 0.0485, and the broader CPVT term scored 0.0747, both carried almost entirely by the animal_model datatype with no human genetic evidence for CAMK2D itself (Open Targets v26). The genetic case for CPVT belongs to RYR2, CASQ2, and CALM genes, not to the kinase being drugged.

Failure mechanism, best guess

This was not a mechanistic defeat. The program ended before it could test its hypothesis in humans, so the failure mechanism is organizational rather than biological. The most useful read is that a rare-disease Phase 2A with 7 participants and a safety-and-PK primary set carried very little inferential weight, which made it an easy line item to cut during a portfolio or financing decision. The underlying target biology remained intact and well supported by preclinical models. What the public record cannot resolve is whether CaMKII inhibition would have produced a measurable reduction in RyR2 phosphorylation or arrhythmia burden at tolerated doses. The thin human genetic support for CAMK2D (Open Targets association 0.0485, animal_model only) means the program was leaning on model-system causality, which raises the value of an early pharmacodynamic anchor that the trial design did not capture before it closed.

How to prevent this next time

Two quantitative levers would have protected this program from ending as an uninformative stub.

First, an explicit Bayesian predictive-probability stopping framework. With a 7-participant crossover, the program should have pre-specified the posterior probability of clinical success and required a minimum informativeness threshold before commitment of further capital.

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

A predictive-probability gate forces a decision rule that separates "the drug failed" from "the study could not speak," and it makes the cost of an underpowered design visible before enrollment rather than after termination.

Second, a pharmacodynamic biomarker enrichment and power calculation tied to RyR2 phosphorylation. CaMKII activity has measurable readouts, including phospho-RyR2 at Ser2814 and surrogate calcium-handling markers. Powering the study to detect a target-engagement effect size, rather than only cataloguing adverse events, would have produced a usable signal even at small n. A target-engagement endpoint with a defined minimum detectable effect converts 7 participants from an administrative liability into a mechanistic probe.

The single highest leverage change would have been to anchor the Phase 2A on a pre-specified CaMKII target-engagement biomarker with a Bayesian predictive-probability stopping rule, so that even an early business decision would have left behind an interpretable human pharmacodynamic result.

What this means for similar programs

CaMKII remains a heavily modeled but lightly drugged node in cardiac arrhythmia, with no approved mechanism-annotated CaMKII inhibitor in ChEMBL for this target (ChEMBL v34). Programs aiming at calcium-handling kinases in rare inherited arrhythmias should assume that human genetic support sits on the channel, not the kinase, and design accordingly. The lesson generalizes to any rare-disease asset where the genetics implicate a substrate and the drug targets an upstream modifier. Build the smallest possible study around a target-engagement biomarker, not around a safety catalogue, so the trial retains value independent of corporate timing.

Open questions

Did CRD-4730 achieve measurable CaMKII inhibition at tolerated doses in the 7 enrolled participants, and will Cardurion release any pharmacodynamic data? Is the program paused or fully discontinued? Would a substrate-anchored endpoint, such as phospho-RyR2, have changed the internal calculus that led to termination? Does CaMKII inhibition add benefit over optimized beta-blockade and flecainide in CPVT1, or is the addressable residual risk too small to power economically?

Sources

1. • ClinicalTrials.gov, NCT06005428 (study design, masking, enrollment 7, primary outcomes, termination reason, primary completion 2025-05-31). - Cardurion Pharmaceuticals, CaMKII program description (mechanistic rationale). - Open Targets Platform v26: CAMK2D (ENSG00000145349) association scores for CPVT1 0.0485, CPVT 0.0747, CPVT4 0.0518, animal_model datatype. - openFDA FAERS, queried 2026-05-31 (1 report for CRD-4730; insufficient for disproportionality). - ChEMBL v34: no mechanism-annotated CaMKII delta (CHEMBL2801) inhibitor drugs. - Kim et al., Cardiovascular Research, 2016,- Reactive oxygen species-mediated CaMKII activation in arrhythmia,- Ryanodine-receptor CPVT mechanism review, 2025,. Available from: https://clinicaltrials.gov/study/NCT06005428.