Temporal dissociation between cerebral blood flow and brain tissue oxygenation during CPR: observations from a porcine model

Article information

Korean J Anesthesiol. 2026;.kja.251104

Publication date (electronic) : 2026 February 2

doi : https://doi.org/10.4097/kja.251104

Michael G. Silverman^1,*, Ki Tae Jung^2,*, Stefan A. Carp^3,*, Bryce Carr³, Ailis C. Muldoon³, Bonsung Koo⁴, Dibbyan Mazumder³, Ekaterina Creed¹, Kichang Lee^,1

¹Cardiology Division, Massachusetts General Hospital, Boston, MA, USA

²Department of Anesthesiology and Pain Medicine, College of Medicine and Medical School, Chosun University, Gwangju, Korea

³Department of Radiology, Massachusetts General Hospital, Charlestown, MA, USA

⁴Department of Anesthesiology and Pain Medicine, Bucheon Hospital, Soonchunhyang University College of Medicine, Bucheon, Korea

Corresponding author: Kichang Lee, Ph.D. Cardiology Division, Massachusetts General Hospital, 55 Fruit Street, Edwards #325, Boston, MA 02114, USA Tel: +1-617-726-1706 Email: kichang.lee@mgh.harvard.edu

*Michael G. Silverman, Ki Tae Jung, and Stefan A. Carp have contributed equally to this work as co-first authors.

Received 2025 December 3; Revised 2025 December 28; Accepted 2026 February 1.

Dear Editor,

Cerebral injury remains a major determinant of survival and long-term neurologic outcome after cardiac arrest [1]. Although systemic parameters such as mean arterial pressure (MAP) and end-tidal CO₂ (EtCO₂) are routinely used as surrogates of perfusion during resuscitation, these metrics may inadequately reflect cerebral oxygen delivery, and both experimental and clinical data suggest a potential disconnect between macro-circulatory indices and actual cerebral tissue oxygenation.

Experimental data indicate that postarrest patients may exhibit normal MAP and EtCO₂ despite persistent cerebral hypoxia, suggesting that systemic hemodynamic targets may not reliably represent the underlying cerebral microcirculation and oxygenation status [2]. One porcine study, for example, reported that approximately 70% of animals demonstrated low brain tissue oxygen tension (PbtO₂ < 20 mmHg) within the first hour after return of spontaneous circulation (ROSC) despite preserved systemic hemodynamics [3].

To further investigate this dissociation between systemic hemodynamics and cerebral oxygenation, we performed a brief observational analysis measuring cerebral blood flow (CBF) and PbtO₂ simultaneously during cardiopulmonary resuscitation (CPR) in a porcine ventricular fibrillation (VF) cardiac arrest model.

This study was approved by the Institutional Animal Care and Use Committee of Massachusetts General Hospital (Approval No. 2019N000244). Thirteen healthy Yorkshire swine were mechanically ventilated to maintain normocapnic conditions (tidal volume: 7–10 ml/kg; respiratory rate: 10–15 breaths/min; end-tidal CO₂: 35–45 mmHg; FiO₂: 0.35–0.4; I:E ratio: 1:2). VF was electrically induced via a 5-F pacing catheter advanced into the right ventricle and connected to a programmable cardiac stimulator (Model 5326, Medtronic). After VF was confirmed, ventilation was discontinued, and VF was left untreated for 12±4 minutes to simulate a no-flow interval. Defibrillation (X-series, ZOLL Medical) was then performed, followed by manual chest compressions at a rate of 100–120 compressions per minute in accordance with current American Heart Association resuscitation guidelines. If ROSC was not achieved, additional cycles of defibrillation and CPR were performed. During resuscitation, ventilation was resumed using the same ventilator settings as at baseline; however, no epinephrine was administered. Cerebral hemodynamics were monitored through two burr holes created 10 mm lateral to the midline. Cortical CBF was measured using a laser Doppler (LD) flowmetry needle probe (PeriFlux System 5000, Perimed AB), inserted into the brain parenchyma. Intracranial pressure (ICP), PbtO₂, and brain temperature were measured using the Licox^® Brain Tissue Oxygen Monitoring system (Integra LifeSciences). A combined oxygen and temperature probe (CC1P1, Integra LifeSciences) and a compatible Camino ICP probe (1104B, Integra LifeSciences) were inserted approximately 20 mm into the cerebral parenchyma. Signals were digitized at 400 Hz using a multichannel data acquisition system (PowerLab^® 16/30, ADInstruments). Physiological variables—including CBF, PbtO₂, ICP, arterial pressure, electrocardiogram, EtCO₂, and temperature—were continuously recorded throughout baseline, untreated VF, CPR, and the early ROSC period.

All thirteen pigs achieved ROSC, and a consistent and notable physiological pattern emerged across the animals. As shown in Fig. 1A, CBF (LD) demonstrated a sharp decline immediately after VF induction, reaching near-zero flow within seconds. In contrast, PbtO₂ declined at a significantly slower rate and typically remained above 20–25 mmHg for over a minute despite the complete cessation of perfusion. When CPR was initiated, CBF (LD) rose from the VF nadir, with mean values increasing by approximately 60% of baseline despite large inter-animal variability (Fig. 1B). In contrast, PbtO₂ demonstrated negligible change, remaining near baseline with only minimal fluctuations despite these marked increases in CBF. This divergence indicates that the perfusion generated during CPR did not translate into a proportional improvement in PbtO₂. Following ROSC, CBF (LD) exhibited a rapid hyperemic increase, reaching 50% of its baseline within less than one minute. In contrast, PbtO₂ recovered only after a substantial delay, often requiring several minutes to reach even 50% of baseline (Fig. 1C). The temporal pattern parallels the deoxygenation observed at VF onset. This discrepancy between perfusion and tissue oxygen tension suggests that restoration of microscopic and macroscopic CBF does not immediately produce a corresponding recovery of tissue-level oxygen diffusion.

Fig. 1.

Cerebral perfusion and oxygenation dynamics during VF, CPR, and early ROSC. (A) Representative hemodynamic data from a single animal across experimental stages. (B) Percentage change in CBF (LD) and PbtO₂ during CPR, illustrating the marked increase in CBF in contrast to the minimal change in PbtO₂. (C) Time required to reach 50% reduction (during VF) and 50% recovery (after ROSC) of baseline CBF (LD) and PbtO₂, demonstrating the temporal dissociation between cerebral perfusion and cerebral oxygenation. SBP: systolic blood pressure, DBP: diastolic blood pressure, MAP: mean arterial pressure, HR: heart rate, EtCO₂: end-tidal carbon dioxide, CBF: cerebral blood flow measured by laser Doppler (LD), PbtO₂: brain tissue oxygen tension, ICP: intracranial pressure, VF: ventricular fibrillation, CPR: cardiopulmonary resuscitation, ROSC: return of spontaneous circulation.

Several physiological mechanisms may account for this delay. Post-ischemic microvascular dysfunction—including endothelial swelling, capillary obstruction, and impaired capillary recruitment—can limit tissue oxygenation even when microscopic and macroscopic blood flow improves [1,2,4]. In addition, altered cerebral metabolic demand during arrest and reperfusion may further uncouple the relationship between CBF and PbtO₂ [2,5]. The early hyperemic response after ROSC likely reflects transient vasoparalysis or impaired cerebrovascular autoregulation, rather than true restoration of microcirculatory oxygen delivery [1,2]. These considerations align with prior findings demonstrating poor correlation between systemic hemodynamics and direct measurements of brain oxygenation in the immediate postarrest period [3].

The implications of this physiologic dissociation are clinically important. Reliance on systemic markers such as MAP or EtCO₂ may create a false sense of adequacy by inaccurately reflecting cerebral perfusion and oxygenation during CPR and early ROSC. Our observations indicate that even when CBF improves promptly with chest compressions and rebounds rapidly following ROSC, PbtO₂ may remain depressed for several minutes, potentially extending cerebral hypoxic exposure during a vulnerable recovery window. These results challenge the assumption that normalization of systemic hemodynamics equates to adequate cerebral oxygen delivery and underscore the need for monitoring strategies that more directly interrogate cerebral physiology.

Although limited by small sample size, the use of healthy animals, and the short post-ROSC observation period, this brief analysis demonstrated an important physiological phenomenon with potential implications for CPR quality assessment and early post-arrest care. Real-time cerebral monitoring, such as PbtO₂, near-infrared spectroscopy, or advanced perfusion metrics (i.e., CBF and cerebral perfusion pressure [CPP; MAP-ICP]), may help identify occult cerebral hypoxia that would otherwise remain undetected using systemic parameters alone. Future studies are needed to clarify whether perfusion-directed resuscitation or individualized oxygen titration could mitigate such mismatch and improve neurologic outcomes.

In conclusion, the differential temporal recovery of CBF and PbtO₂ observed in this porcine VF model illustrates that systemic hemodynamics alone are insufficient to assess cerebral oxygenation during CPR and early ROSC. Approaches that directly evaluate cerebral physiology may help identify unrecognized cerebral hypoxia and guide more targeted resuscitative interventions.