Background

Prior clinical research demonstrated that rapid reduction in arterial carbon dioxide (PaCO₂) levels during extracorporeal membrane oxygenation (ECMO) is associated with acute brain injury (ABI), which may be due to sudden cerebral vasoconstriction and impaired cerebrovascular autoregulation (CVAR). However, the causal relationship between rapid PaCO₂ correction and its impact on ABI has not been firmly established due to the lack of high-quality evidence. We aimed to investigate whether rapid PaCO₂ correction following extracorporeal cardiopulmonary resuscitation (ECPR) causes CVAR impairment and neuronal injury in a porcine model.

Methods

In this prospective preclinical experimental study, six female pigs (mean weight: 50.75 ± 1.89 kg) were subjected to 15 min of ventricular fibrillation and were then supported by ECMO. The return of spontaneous circulation (ROSC) was attempted in animals at 20 min post-ECMO initiation. Arterial blood gas (ABG) was sampled at specific time points, while arterial blood pressure (ABP) and intracranial pressure (ICP) were continuously monitored. Sweep gas flow was set relative to each animal’s ECMO flow rate: 100% in the control group, 200% in the rapid correction group, and 25% in the slow correction group. PRx was computed as the Pearson correlation coefficient between 10-s average mean arterial pressure (MAP) and ICP values using 1-min windows updated every 30 s. Experimental phases were defined for data analysis, including baseline, fibrillation, ECMO I (0–10 min after ECMO initiation), ECMO II (10–20 min), and POST-R (post-ROSC, 20–30 min). Linear mixed-effects models were used to assess group-wise differences in ΔPRx over time. Histopathological analysis was performed to quantify neuronal injury across cortical and subcortical regions. Brain tissues were harvested and histologically analyzed for neuronal injury ischemia vulnerable regions: the midbrain, cerebellum, striatum in the basal ganglia, temporal cortex, hypothalamus, and hippocampus.

Results

In the rapid group, PaCO₂ correction caused a steep drop in PaCO₂—from 60 to approximately 30 mmHg within 5 min—and was associated with impaired CVAR. Following ECMO initiation, the rapid group exhibited a significant rise in ΔPRx, indicating impaired CVAR. Group differences in ΔPRx were significant at ECMO I (F = 8.12, p = 0.001), ECMO II (F = 6.21, p = 0.003), and POST-R (F = 13.47, p < 0.001). At ECMO II, median PRx in the rapid group was 0.50 (IQR: 0.10, 0.78), significantly higher than the control (0.11, IQR: − 0.27, 0.42) and slow (0.38, IQR: − 0.06, 0.55). Histologically, the rapid correction group exhibited significantly increased ischemic neuronal injury in ischemia-prone regions: caudate (43.1% injured neurons vs. 10.6% in control, p = 0.041), putamen (66.6% vs. 23.9%, p = 0.003), temporal cortex (34.9% vs. 8.9%, p = 0.013), and hippocampal CA-3 region (4.7% vs. 18.0%, p = 0.026). Compared to rapid correction, the slow correction group demonstrated improved gas stability (PaCO₂ decline of ~ 10 mmHg over 10 min), preserved PRx (mean PRx < 0.2), and significantly reduced neuronal injury in the putamen (p = 0.004).

Conclusion

In this experimental ECPR model, faster early PaCO₂ correction was associated with impaired CVAR (higher PRx values). Controlled CO₂ correction should be considered a key neuroprotective strategy during ECMO initiation.