Section 3
Results and follow-up data
All values for subject MU-01 are compared against the pooled flash-reperfusion control arm (n = 6), in which identical cooling was applied but oxygen was restored in a single step. Follow-up is ongoing at 214 days as of 12 August 2026.
Quantitative electroencephalography was computed from an eight-channel montage (Fp1, Fp2, F3, F4, C3, C4, O1, O2, referenced to linked mastoids) sampled at 512 Hz, band-pass filtered from 0.5 to 45 Hz, notch filtered at 60 Hz, and segmented into four-second epochs with automated rejection of epochs exceeding 100 µV. Spectral power was estimated by Welch periodogram with a Hann window and 50 percent overlap, then expressed as a percentage of the mean power recorded across three pre-arrest sessions. Analysts were blinded to arm allocation during epoch scoring.
Figure 3
Cortical EEG spectral power recovery, % of pre-arrest baseline
Table 3
Serum neurofilament light chain, pg·mL⁻¹
| Time | MU-01 | Control mean |
|---|---|---|
| 0 h | 18 | 21 |
| 6 h | 142 | 1,910 |
| 24 h | 388 | 6,240 |
| 72 h | 211 | 8,905 |
| 168 h | 62 | 7,410 |
Peak NfL in MU-01 was 4.4 % of the control peak and returned to within 3.4× baseline by day 7, consistent with transient axonal stress rather than structural loss.
Table 4
Behavioural assessment, day 9
| Measure | Day 9 | Baseline | Call |
|---|---|---|---|
| Modified Glasgow canine neurological scale | 18 / 18 | 18 / 18 | No deficit |
| Handler voice discrimination (4-way) | 94 % | 97 % | Within noise |
| Obstacle course completion speed | 88 % | 100 % | Mild slowing |
| Gait symmetry (pressure plate) | 96 % | 100 % | Within noise |
| Startle habituation trials to criterion | 4 | 4 | Unchanged |
| Appetitive motivation index | 1.02 | 1.00 | Unchanged |
Section 3.4
Systemic and imaging outcomes
Organ-level recovery was tracked because whole-body washout exposes every tissue bed, not only the brain, to 41 minutes of oxygen-free perfusion.
Renal and hepatic function
Serum creatinine peaked at 2.1 mg·dL⁻¹ at 24 hours with a corresponding rise in neutrophil gelatinase-associated lipocalin, consistent with stage 1 acute kidney injury by IRIS criteria, and resolved without renal replacement therapy by day 5. Alanine aminotransferase peaked at 318 U·L⁻¹ at 12 hours and normalised by day 8. Neither organ showed histological evidence of cortical necrosis on ultrasound-guided biopsy at day 30.
Cardiac structure and rhythm
Echocardiography at day 14 showed a left ventricular ejection fraction of 54 percent with mild regional hypokinesis of the right ventricular free wall attributable to the underlying cardiomyopathy rather than to the arrest. An implanted cardioverter-defibrillator recorded four non-sustained ventricular tachycardia episodes across 214 days, none requiring shock therapy. Cardiac magnetic resonance at day 90 quantified late gadolinium enhancement at 6 percent of myocardial mass.
Neuroimaging
Diffusion-weighted magnetic resonance imaging at 72 hours showed no restricted-diffusion lesions in the hippocampus, thalamus, cerebellar Purkinje layer or cortical watershed territories, the four regions where hypoxic-ischaemic injury reliably appears first. Apparent diffusion coefficient values across all sampled regions of interest fell within one standard deviation of the age-matched control cohort. Volumetric analysis at day 180 detected no hippocampal atrophy against the pre-arrest scan of record.
Sleep architecture and long-term monitoring
Ambulatory 24-hour electroencephalography at day 60 and day 180 showed preserved slow-wave and rapid-eye-movement proportions, with a mild reduction in sleep spindle density at day 60 that had normalised by day 180. No interictal epileptiform discharges were recorded at any time point, which is notable because post-arrest myoclonus and subclinical status epilepticus are among the most common findings in survivors of prolonged normothermic arrest.
Section 3.5
Statistical treatment
With a single treated subject, conventional frequentist comparison is inappropriate. Primary inference was therefore pre-specified as a Bayesian single-case analysis: observed MU-01 values were evaluated against a posterior predictive distribution constructed from the six flash-reperfusion controls and from eleven historical normothermic arrests in the institute veterinary registry, using weakly informative half-normal priors on residual variance. The posterior probability that MU-01 day-6 EEG spectral power exceeded the control distribution was greater than 0.999, with a Bayes factor of 214 in favour of a treatment effect on the primary endpoint.
Tissue-control histology, which contributed n = 34 independent samples across time points, was analysed by mixed-effects regression with sample nested within animal and a false discovery rate correction across the seven pre-registered markers. All analysis code, raw telemetry and the pre-registration timestamp are deposited with dataset MU-EEG-001 so that the distinction between pre-specified and exploratory results can be independently audited.
Section 4
Limitations
This is a single subject. No claim of generalisability is made, and the result should be read as a proof of biological possibility rather than as evidence of clinical efficacy.
The arrest was witnessed, monitored and occurred in a theatre with a pre-positioned perfusion team. Time from arrest to washout was 9 minutes; no out-of-hospital system currently achieves this. The interval, not the chemistry, may prove to be the binding constraint on translation.
Behavioural preservation was assessed against a 14-month pre-arrest record for this animal, which is unusually complete but was not collected for this purpose. Subtle changes in olfactory discrimination and in sleep architecture cannot be excluded and are under continued measurement.
The control arm received identical cooling but single-step reoxygenation, which isolates the reoxygenation schedule as a variable but does not isolate the individual contribution of each perfusate constituent. A four-arm factorial design separating KI-118, the hydrogen sulfide donor and the calpain antagonist is required before any component can be described as necessary rather than merely present. Sample sizes for that design are specified in the supplementary statistical appendix.
Species differences are material. Canine cerebral white matter proportion, collateral circulation and hypothermic tolerance all differ from human values, and the subject was a small-breed animal with a favourable surface-area-to-mass ratio for rapid cooling. Extrapolation of the 41-minute figure to human physiology is not supported by the present data and is not claimed.
Finally, the certification of death met veterinary clinical standards. It does not settle the philosophical question the result raises: whether a state that can be reversed 41 minutes later should have been called death at all.