Tenecteplase (originally alteplase, switched in protocol V04) in Non-arteritic central retinal artery occlusion: Efficacy Failure | Claidex

Q: What happened in NCT04965038?

NCT04965038 was a PHASE3 study of Tenecteplase (originally alteplase, switched in protocol V04) in Non-arteritic central retinal artery occlusion sponsored by University Hospital Tuebingen. The trial was terminated or indexed as a failure claim. REVISION, the first adequately designed randomized trial of intravenous thrombolysis for non-arteritic central retinal artery occlusion, has been terminated after a prespecified interim analysis met its futility boundary. The trial randomized 127 patients to alteplase or tenecteplase versus placebo within 4.5 hours of vision loss. The result undercuts a decade of meta-analytic enthusiasm and forces the field to reconsider whether retinal ischemia behaves like cerebral ischemia at the relevant timescale.

Q: What is the Mechanism Risk Score for PLAT in Non-arteritic central retinal artery occlusion?

The Claidex Mechanism Risk Score (MRS) for PLAT in Non-arteritic central retinal artery occlusion is 42/100 (yellow band). This score quantifies the documented clinical failure burden for this target-disease pair based on post-mortems in the Claidex Failure Atlas.

Claidex Editorial Team

doi:10.5281/zenodo.20301169

Tenecteplase in central retinal artery occlusion: REVISION halts a Phase 3 thrombolysis trial for futility

NeurologyEfficacyMay 18th, 2026·6 min read·10.5281/zenodo.20301169

REVISION, the first adequately designed randomized trial of intravenous thrombolysis for non-arteritic central retinal artery occlusion, has been terminated after a prespecified interim analysis met its futility boundary. The trial randomized 127 patients to alteplase or tenecteplase versus placebo within 4.5 hours of vision loss. The result undercuts a decade of meta-analytic enthusiasm and forces the field to reconsider whether retinal ischemia behaves like cerebral ischemia at the relevant timescale.

Mechanism Risk Score

Component	Points
Phase-weighted failure burden	15.7 / 40
Archetype severity	9.8 / 25
Temporal recency	4.3 / 15
Genetic evidence deficit	9.7 / 15
Programmatic saturation	2.5 / 5

For PLAT in Non-arteritic central retinal artery occlusion, the Mechanism Risk Score is 42/100 (yellow band). The score is a failure-burden index derived from Claidex post-mortems on this target–disease pair, not a probability of approval.

MRS 42/100 (YELLOW). 1 programs across PLAT have been documented for PLAT in Non-arteritic central retinal artery occlusion: 1 Phase 3, 0 Phase 2, 0 Phase 1 — of which 1 were efficacy failures, 0 safety, 0 biomarker, and 0 operational (enrollment, sponsor, or funding). The most informative failure on file is Tenecteplase in central retinal artery occlusion: REVISION halts a Phase 3 thrombolysis trial for futility. This score quantifies the documented failure burden; the Open Targets association score of 0.35 reflects weak genetic anchoring, compounding the documented failure record. The MRS is not a prediction of future trial outcomes — it is a structured summary of the empirical record, recomputed live from the Claidex claims table, and intended to flag mechanisms where any new program must explicitly resolve each prior failure mode before pursuit is justified.

This score does not predict whether the next trial will succeed. It flags how heavy the documented mechanistic failure record is before a new program is justified.

What was tried

REVISION was a multicenter, randomized, placebo-controlled, quadruple-blinded Phase 3 trial of intravenous thrombolysis for acute non-arteritic central retinal artery occlusion (CRAO), run by University Hospital Tuebingen. Patients with sudden painless monocular vision loss within 4.5 hours of symptom onset received either intravenous thrombolytic or matching placebo on the cerebral stroke template. The protocol began with alteplase at 0.9 mg/kg and switched to tenecteplase 0.25 mg/kg at protocol amendment V04. The ClinicalTrials.gov record for NCT04965038 now lists the study as terminated with the explanation, "Stopped after prespecified interim analysis due to futility." Actual enrollment was 127.

The primary endpoint was functional recovery at 30 days, defined by best-corrected visual acuity. Eligibility required baseline best-corrected visual acuity worse than LogMAR 1.3, retinal pathology confirmed by fundoscopy and OCT, and exclusion of arteritic causes. The design assumed an effect on the order of the meta-analytic estimate, roughly 29% recovery in the thrombolysis arm versus 11% in the conservative arm (Poli et al., International Journal of Stroke 2024).

REVISION was conceived in the wake of two influential individual-participant meta-analyses suggesting that intravenous tissue plasminogen activator within 4.5 hours produces visual improvement in 39 to 74 percent of CRAO patients (Shahjouei et al., International Journal of Stroke 2024; Xie et al., Stroke 2025). It was supposed to be the confirmatory trial.

The biological hypothesis

CRAO is a stroke of the eye. A clot or embolus, usually from the carotid bifurcation or the heart, lodges in the central retinal artery and starves the inner retina of oxygen. The reasoning behind extrapolating cerebral thrombolysis to CRAO is mechanical. The thrombus is small, often platelet-fibrin in composition, and the central retinal artery is a 100-micron caliber vessel with no collateral arterial supply distal to the cribrosa. Plasmin generated by recombinant tissue plasminogen activator should cleave fibrin, restore flow, and rescue ischemic but salvageable retina before infarction completes (Hayreh and Zimmerman, Ophthalmology 2005).

The clinical evidence base for this hypothesis was indirect but consistent. The EAGLE study, the only prior randomized controlled trial in CRAO, compared local intra-arterial fibrinolysis to conservative standard therapy in 84 patients and was stopped early when adverse events accumulated in the interventional arm without an efficacy difference (Schumacher et al., Ophthalmology 2010). After EAGLE, the field shifted to intravenous administration on the same template that had transformed acute ischemic stroke care.

The 2025 AGILE meta-analysis tightened the case. Across 783 patients, IV thrombolysis within 4.5 hours produced recovery from severe vision loss in 28.8% versus 11.1% with conservative care, an odds ratio of 3.32 (95% CI 1.24 to 8.92). Monte Carlo simulation published with that paper concluded that a confirmatory randomized trial would need approximately 95 participants per arm to achieve 80% power. REVISION was designed and largely enrolled before that benchmark was published.

What actually happened

The Data and Safety Monitoring Committee stopped REVISION at the prespecified interim analysis. With 127 patients randomized, the conditional power to reach a positive primary endpoint at full enrollment crossed the futility boundary. Granular per-arm results are not yet public, but the futility decision implies that the observed difference in 30-day functional recovery between thrombolysis and placebo was too small to recover statistical significance at full accrual. REVISION was funded by the German Federal Ministry of Education and Research and Boehringer Ingelheim, which supplied tenecteplase.

REVISION primary endpoint context. Left, Wilson 95% confidence intervals for visual recovery across the EAGLE trial (intra-arterial), the Shahjouei and AGILE individual-participant meta-analyses (intravenous), and the REVISION interim. Right, a power curve for IV thrombolysis versus conservative care assuming 28.8% versus 11.1% recovery, marking REVISION's roughly 64 patients per arm and the AGILE-derived 95 per arm needed for 80% power. Data: ClinicalTrials.gov NCT04965038, Schumacher et al. 2010, Shahjouei et al. 2024, Xie et al. 2025.

Safety did not drive the decision. Tenecteplase in the cerebral stroke literature carries documented risks: cerebral haemorrhage (84 serious FAERS reports), intracranial haemorrhage (72), and angioedema (59) across 774 total serious reports for the molecule. None of these signals was reported as a stopping rule driver. The trial stopped on efficacy.

Failure mechanism, best guess

Three explanations fit the data. The first is selection bias in the meta-analyses. The Shahjouei and AGILE pools draw from 72 and 35 studies respectively, almost all observational and retrospective. Spontaneous improvement in CRAO is well documented, particularly in patients with a cilioretinal artery that perfuses the macula, and selection into a treated cohort can mask that natural history.

The second is biology. The retinal nerve fiber layer tolerates only about 100 minutes of complete ischemia before irreversible loss in primate models (Hayreh, Progress in Retinal and Eye Research 2011). Most CRAO patients reach medical care after that window, so even successful recanalization within 4.5 hours rescues cells that are already dead. The cerebral stroke template assumes a penumbra at hours, not minutes.

The third is statistical. REVISION enrolled about 64 patients per arm. At the AGILE effect size of 28.8% versus 11.1%, that sample delivers approximately 56% power, well short of 80%. If the true effect in REVISION's population was even modestly smaller, perhaps 22% versus 14%, the trial was destined to cross a futility line at the first protocol-mandated look.

What this means for similar programs

The downstream program closest to REVISION is the THEIA trial in France, which uses tenecteplase in CRAO on a similar template. The TEMPURA trial in Canada and other IV thrombolysis protocols now under development in Germany should re-examine their assumed effect size against REVISION's interim and consider tighter time windows. Neuroprotective and reperfusion-adjacent strategies such as hyperbaric oxygen retain modest clinical support and may deserve a higher prior than the field has assigned them (Mac Grory et al., Stroke 2021).

How to prevent this next time

REVISION is a textbook case of a confirmatory trial under-powered for the effect size that motivated it. With 127 patients randomized 1:1, assuming the AGILE pooled estimate of 28.8% versus 11.1% recovery and a two-sided alpha of 0.05, the study delivered approximately 56% power. Achieving 80% power against that effect requires roughly 95 patients per arm, exactly the number Monte Carlo simulation in the AGILE paper produced. REVISION is the rare Phase 3 that was correctly designed against an unsourced effect estimate and incorrectly designed against the best published estimate available at planning time. A Bayesian predictive probability calculation at the interim, $P P (s u ccess) = \int_{θ} P (f ina l s u ccess ∣ θ, in t er im d a t a) \cdot π (θ ∣ in t er im d a t a) d θ$ , against a flat prior centered on the AGILE estimate would yield a posterior predictive probability below the typical 0.10 futility threshold, consistent with what the DSMC observed.

The deeper fix is biomarker enrichment by time. CRAO enrollment in REVISION extended to 4.5 hours, the cerebral stroke window. Retinal cell death in primate ischemia models is essentially complete by 240 minutes. A trial that restricted enrollment to symptom onset within 90 minutes would shrink the eligible pool by roughly a factor of three but raise the prior probability of benefit substantially. Combined with stratification by cilioretinal artery presence on OCT-angiography, present in approximately 32% of patients and protective against complete macular infarction, the enriched fraction approaches a population in which odds ratios for benefit are preclinically two to three times higher than in the unrestricted cohort.

The single most actionable change is straightforward. Future CRAO thrombolysis trials should restrict to symptom onset within 90 minutes and stratify by cilioretinal artery status before randomization rather than power against an unrestricted 4.5-hour cerebral stroke window.

Open questions

What is the per-arm recovery rate in REVISION when the full dataset is released, and how does it compare to the AGILE benchmark of 28.8% versus 11.1%?
Does a subset analysis of patients treated within 90 minutes show any signal that survives multiplicity correction?
Is tenecteplase pharmacokinetically equivalent to alteplase in the small-vessel retinal compartment, given the protocol switch midway through the trial?
What rate of symptomatic intracranial haemorrhage and angioedema was observed in REVISION, and how does it compare to the cerebral stroke baseline of 2 to 7 percent?
Should the AGILE individual-participant meta-analytic estimate be revised downward when REVISION results enter the synthesis?