Distinct Alterations in Oxygenation, Ion Composition and Acid-Base Balance in Cerebral Collaterals During Large-Vessel Occlusion Stroke

Purpose Disturbances of blood gas and ion homeostasis including regional hypoxia and massive sodium (Na+)/potassium (K+) shifts are a hallmark of experimental cerebral ischemia but have not been sufficiently investigated for their relevance in stroke patients. Methods We report a prospective observational study on 366 stroke patients who underwent endovascular thrombectomy (EVT) for large-vessel occlusion (LVO) of the anterior circulation (18 December 2018–31 August 2020). Intraprocedural blood gas samples (1 ml) from within cerebral collateral arteries (ischemic) and matched systemic control samples were obtained according to a prespecified protocol in 51 patients. Results We observed a significant reduction in cerebral oxygen partial pressure (−4.29%, paO2ischemic = 185.3 mm Hg vs. paO2systemic = 193.6 mm Hg; p = 0.035) and K+ concentrations (−5.49%, K+ischemic = 3.44 mmol/L vs. K+systemic = 3.64 mmol/L; p = 0.0083). The cerebral Na+:K+ ratio was significantly increased and negatively correlated with baseline tissue integrity (r = −0.32, p = 0.031). Correspondingly, cerebral Na+ concentrations were most strongly correlated with infarct progression after recanalization (r = 0.42, p = 0.0033). We found more alkaline cerebral pH values (+0.14%, pHischemic = 7.38 vs. pHsystemic = 7.37; p = 0.0019), with a time-dependent shift towards more acidotic conditions (r = −0.36, p = 0.055). Conclusion These findings suggest that stroke-induced changes in oxygen supply, ion composition and acid-base balance occur and dynamically progress within penumbral areas during human cerebral ischemia and are related to acute tissue damage. Supplementary Information The online version of this article (10.1007/s00062-023-01296-w) contains supplementary material, which is available to authorized users.


Introduction
Despite major recent advances in endovascular stroke treatment [1], large-vessel occlusion (LVO) stroke remains a leading cause of death and acquired disability worldwide [2].Hence, there is an urgent need to more clearly identify and better understand early pathophysiological events during cerebral ischemia not only in experimental settings but also in the human system [3,4].
The cessation of cerebral blood flow (CBF) during LVO stroke results in a severe reduction of oxygen and glucose supply within the downstream vascular territory [5] and is accompanied by functionally relevant alterations of intracellular and extracellular ion composition in the affected brain region [6,7].These events are part of a complex set of mechanisms in response to the ischemic stimulus which has been coined the ischemic cascade.Key parenchymal and intravascular processes that parallel and/or define tissue damage include: bioenergetic failure, excitotoxicity, intracellular calcium (Ca 2+ ) and sodium (Na + ) overload, extracellular accumulation of potassium (K + ), oxidative stress, and neuroinflammation [5][6][7][8][9][10].These processes are closely interrelated as, e.g., severely reduced CBF leads to peri-infarct depolarization which is characterized by a regional switch from blood hypoxygenation to hyperoxygenation during propagation [11,12].In addition, there is evidence that also time-dependent regional pH shifts exist which likely indicate the differential fate of penumbral subareas [13]; however, available human data on regional blood gas and electrolyte alterations supporting the crucial role of blood oxygenation, acid-base balance and ion movements in LVO stroke are scarce and partly inconsistent [14][15][16].Few groups including our own have established a sampling protocol for microcatheter-aspiration of cerebral blood samples from within the collateral circulation during LVO stroke immediately before therapeutic recanalization by endovascular thrombectomy (EVT) [14,17,18].
In this study, we now aim to retrace the aforementioned key experimental observations in a large and highly noisecontrolled cohort where microcatheter aspiration attempts and analyses of cerebral samples were performed consecutively in every eligible anterior circulation LVO stroke patient.The biological relevance of changes in blood gas and ion composition was addressed by hypothesis-based association with 1) temporal, 2) clinical functional and 3) radiological structural parameters assessed during cerebral ischemia and in the clinical course after recanalization.

Study Design
We report a prospective observational single-center study conducted between 12 December 2018 and 31 August 2020, to investigate alterations in arterial blood gas (ABG) parameters within the collateral circulation (distal to the occlusion site) of LVO stroke patients.Endovascular sampling of ischemic blood from collateral blood vessels during the early phase of infarct formation was performed by microcatheter aspiration according to a prespecified protocol during emergency EVT [18][19][20][21][22][23].Systemic arterial blood which was collected from the ipsilateral cervical internal carotid artery (ICA) under physiological flow conditions served as an intraindividual control.

Inclusion and Exclusion Criteria
Inclusion criteria were defined as follows: (I) patients aged older than 18 years presenting with disabling first ever acute ischemic stroke which qualifies for EVT according to current guidelines and consensus recommendations [1,24,25] and (II) invasive angiographic confirmation of complete LVO of the following anterior circulation vessels: distal internal carotid artery (ICA-T), middle cerebral artery (MCA)-M1 segment or proximal MCA-M2 segment.
Patients were excluded based on the following criteria: (I) noninvasive or invasive angiographic confirmation of bilateral or multifocal vessel occlusions other than defined, (II) invasive angiographic confirmation of residual or restored antegrade CBF at the level of the EVT-qualifying lesion, (III) LVO in conjunction with either ≥ 50% cervical ICA stenosis or ICA dissection, (IV) intraprocedural percutaneous transluminal angioplasty (PTA) or stent implantation and (VI) principle deviations from the interventional protocol and sampling procedure [18][19][20][21][22][23].The flow of patient inclusion and exclusion is presented in Fig. 1.

Endovascular and Sampling Procedure
All endovascular treatments were performed by boardcertified neurointerventionalists or supervised neurointerventional fellows on a biplane angiography system (Axiom Artis Q, Siemens Healthcare, Erlangen, Germany).Endovascular access for EVT was established by means of a transfemoral approach through the common femoral artery (CFA) using the modified Seldinger technique.According to our in-house standard of practice, pressurized flush systems (0.9% saline solution with 1 IU/ml unfractionated heparin) were used to avoid thrombus formation on the inner surface of catheters.A detailed description of the standard procedure of mechanical thrombectomy has been published previously [26].Recanalization of the EVT-qualifying lesion by stent-embolus retrieval was preceded by microcatheter navigation (Neuroslider 27 or 21; Acandis, Pforzheim, Germany) into the mid-insular MCA-M2 segment.The following arterial blood samples were obtained by microcatheter aspiration: (1) distal to the occlusive lesion under persistent ischemic conditions and (2) under systemic physiological blood flow conditions at the cervical ICA level.Specifically, the first sample was obtained as follows: before device deployment and immediately after microcatheter positioning, the precise microcatheter dead space volume was aspirated with a 3 ml Luer lock syringe and discarded.Then, the target sample of 1 ml of ischemic blood from within the collateral circulation was obtained.The procedure of cerebral arterial blood sampling is displayed in supplemental Fig. 1.After termination of the therapeutic steps of EVT, systemic blood samples were obtained from cervical ICA level using the exact same technique as described for ischemic sample material.

Periinterventional collection of blood samples n=72
Exclusion: ¾ Failed to meet invasive imaging inclusion criteria, n=50

Study Population
Between 18 December, 2018, and 31 August, 2020, n = 366 consecutive patients who were treated by EVT were assessed for study eligibility according to a prespecified study protocol [18][19][20].N = 48 patients were excluded due to posterior circulation LVO and n = 11 patients due to bilateral or multifocal vessel occlusions.N = 50 patients did not qualify for inclusion upon invasive angiographic imaging either due to spontaneous recanalization or subocclusion with residual antegrade flow before EVT.Microcatheter-aspiration of ischemic blood samples was attempted in n = 257 patients with angiographically proven LVO of the following target sites: intracranial ICA-T segment, MCA-M1, and proximal MCA-M2 segment, respectively.Microcatheter-aspiration of cerebral blood samples through blood gas syringes succeeded in n = 72 patients (28%).Out of this group, n = 51 consecutive patients (20%) met all a priori defined sampling and interventional criteria of inclusion and entered data analyses.The full patient flow without missing cases is given in Fig. 1.A detailed presentation of the ABG data including temporal, clinical functional and radiological structural correlations is given in Table 1, 2, 3, 4 and 5; a graphical summary is provided in the Supplementary Information.

Patient Characteristics
The clinical, radiological, interventional, and sampling-related characteristics of the study population are presented in Table 1.

ABG Analysis in the Ischemic and Systemic Circulation
A total of n = 102 matched regional ischemic (intravascular sample from within the collateral circulation distal to the occlusion site) and systemic arterial blood samples (physiological flow conditions at cervical ICA level) were drawn from n = 51 LVO stroke patients.The data on ABG analysis of both sampling sites are presented in Table 2 and summarized in supplemental Fig. 1.

Association Between Ischemic Blood Gas Parameters and the Duration of Stroke
We used Spearman's rank correlation coefficient to assess associations between ischemic ABG parameters and the time from stroke onset to ischemic sampling.The median time from stroke onset to distal sampling was 274 min (IQR 205-369 min).Regional ischemic pH values were found to be negatively associated with the duration of stroke, but the correlation was just above the threshold for statistical significance (r = -0.36;p = 0.055).There was also a borderline statistical insignificance for the association of local HCO3 -std (r = -0.34;p = 0.074) and BE (B) (r = -0.34;p = 0.074) with occlusion duration.All other ischemic ABG parameters showed no correlation with the time from stroke onset to ischemic sampling (|r| < 0.3; Table 3 and supplemental Fig. 1). K

Association of Ischemic ABG Parameters with Clinical Stroke Severity Before and After Vessel Recanalization
Spearman's rank correlation coefficient was used to investigate associations between the ischemic ABG parameters and clinical stroke severity at baseline and at 48 h after recanalization.These data are given in Table 4 and supplemental Fig. 1.Blood glucose levels within the collateral circulation were positively correlated with the NIHSS score at 48 h after recanalization (r = 0.37; p = 0.015); however, none of the other ischemic ABG parameters showed any association with either clinical stroke severity on admission or at 48 h.

Association of Ischemic ABG Parameters with Infarct Extent Before and After Vessel Recanalization
Results of correlation analysis between regional ischemic ABG parameters and infarct extent before and after EVT are given in Table 5 and supplemental Fig. 1.Univariate analysis revealed significant associations between preserved tissue integrity (ASPECTS) at admission and intravascular Na + :K + ratios (r = -0.32;p = 0.031), K + concentrations (r = 0.3; p = 0.041), and Cl -concentrations (r = -0.35;p = 0.018) distal to the occlusion site.No other associations between baseline infarct extent and ischemic ABG parameters were seen.Analysis of dichotomized data (more extensive infarcts, ASPECTS ≤ 7 vs. minor infarcts, ASPECTS ≥ 8; poor vs. moderate to good collaterals) did not show a difference between any of the ischemic ABG parameters (Supplementary Information).The Na + concentrations within collateral blood vessels were found to be negatively correlated with postinterventional infarct extent 24-48 h after recanalization (r = -0.34,95% CI: -0.58--0.05;p = 0.02).There were no other correlations between the ischemic ABG parameters and tissue integrity on follow-up imaging.
We further analyzed the association between imaging defined preinterventional to postinterventional worsening of stroke ( ASPECTS) and regional ischemic ABG parameters.Infarct extent remained constant in n = 20 (39.2%) patients, whereas dynamic infarct progression equivalent to a median ASPECTS of 1 (IQR 1-2) was observed in n = 31 (60.8%)patients.Progressive infarction was positively correlated with the Na + concentrations within the occluded vascular territory (r = 0.42; p = 0.0033).No other associations were found with respect to preinterventional to postinterventional infarct progression.

Discussion
To the best of our knowledge, no dedicated animal and only few preliminary human reports exist on regional disturbances of blood gas and ion homeostasis measured directly within the cerebral collateral circulation during acute ischemic stroke [14][15][16]32].This represents an observation gap to understand stroke pathophysiology, because pioneering experimental observations have highlighted the relevance of low blood flow and metabolic failure including tissue ion shifts particularly during the initial phase of infarct formation; however, without providing data on the cerebral vascular compartment [5,6].Recently, we have developed a method which enables the noise-controlled extraction and analysis of cerebral blood samples which were obtained from collateral arteries under persistent ischemic conditions [18,19].This approach proved to be highly consistent across different research designs [18][19][20][21][22][23]33] and was now used to investigate the clinical significance of cerebral blood gas and electrolyte alterations in human LVO stroke.Our main findings are the following: human LVO stroke results in a (1) regional reduction in oxygen partial pressure (-4.29%; p = 0.035) and (2) potassium concentrations (-5.49%; p = 0.0081) within the collateral circulation.
The relative reduction in oxygen partial pressure distal to the occlusion site is consistent with previous observations which, however, found oxygen partial pressure both within (paO2ischemic = 73.90mm Hg vs. paO2systemic = 78.90mm Hg) and well above (paO2ischemic = 213.98mm Hg vs. paO2systemic = 251.43mm Hg) the limits for arterial blood [14,15].Substantial elevations in paO2, as were observed in our study, are explained by preinterventional and peri-interventional high-flow oxygenation which is known to lead to an up to fourfold increase in paO2 [34].In our study, disruption of antegrade cerebral blood flow led to regional normocapnia and negative base excess.This condition plausibly reflects an incipient metabolic (nonrespiratory) acidosis [35].At this time of observation, full compensation, as reported previously [15,16], is unlikely as counterregulation is not accomplished within 12 h [36,37].Interestingly, ABG analysis revealed both a fine but significant shift to more alkaline overall pH values within collateral blood vessels and a time-dependent decrease in cerebral pH values during occlusive ischemia.As more pronounced acidosis, reflecting impaired CO2 removal during no/low-flow conditions or inadequate oxidative phosphorylation, is to be expected in or in close vicinity to the infarct core, it can be inferred that cerebral samples were extracted from penumbral regions where an infarct milieu is gradually developing [13,[38][39][40].This interpretation is supported by the favorable imaging profile of the study population which is characterized by predominantly small baseline infarcts (i.e., median ASPECTS of 9 and penumbral imaging-based patient selection).Importantly, the initial ASPECTS in this study is numerically identical to that of pooled patient data from EVT trials in intermediate time windows (HERMES metaanalysis: 9) [25], and one point higher compared to that of both pooled patient data from EVT trials in extended time windows (AURORA meta-analysis: 8) and large prospec-K .Hence, our observations may also apply to these populations and are likely not driven by a selection bias due to overrepresentation of patients with small ischemic lesions.Normal neuroelectric activity and water content of the brain require the careful orchestration and proper distribution of intracellular and extracellular ions.The observed relative hypokalemia and increase in Na + /K + ratios support the notion of significant ion movements within ischemic brain regions which are characterized by net K + losses and/or Na + gains [6,7].The literature suggests that astrocytes may form a functional syncytium for extracellular and intravascular potassium clearance as a means to control neuronal excitability in viable tissue [7].Correspondingly, the lack of massive sodium release supports the conclusion that largescale cell death may not have occurred in EVT patients at the time of cerebral sample acquisition [6,43].This is again consistent with the fact that more than one third of patients presenting with unknown or extended time of ischemia were selected for recanalization based on the absence of extensive infarction on imaging at presentation.Accordingly, there was no temporal association between cerebral ion composition and the duration of stroke.Finally, preclinical data from others indicate that there is a near-perfect linear correlation between sodium and potassium ion shifts and changes in brain water content in the lesioned hemisphere early before the disruption of the blood-brain barrier (r = 0.992; p < 0.001) [43].Given that brain water content is inversely correlated with X-ray attenuation, the results imply that the radiological measures of ASPECTS and ASPECTS decline reflect the extent and dynamics of ionic brain edema [29,43,44].
The major strength of this study is its prospective consecutive design including control of a large set of baseline clinical, radiological, interventional, and analytical variables.Still, this study remains observational and could be carried out only at a single-center limiting causal inference and generalizability.Furthermore, additional information regarding cerebral sodium content and pH based on, e.g., sodium and chemical exchange saturation transfer (CEST) magnetic resonance imaging (MRI) is not available [45,46]; however, such a type of extended and methodically demanding MRI set-up would be highly time-consuming and this would represent a relevant time conflict with significant delay before EVT [24,45,46].
In conclusion, subtle but distinct disturbances of cerebral oxygen supply, ion composition and acid-base balance occur and dynamically progress during the phase of occlusive ischemia in stroke patients and are related to acute tissue damage.Combined, these data additionally advocate to restore physiologic cerebral circulatory conditions in LVO stroke, notwithstanding that promising new concepts of add-on cerebroprotection are emerging.

Fig. 1
Fig. 1 Flow chart of patient inclusion and exclusion.CT computed tomography, ICA internal carotid artery, LVO large-vessel occlusion, PTA percutaneous transluminal angioplasty

Table 2
ABG parameters during LVO stroke by the site of sampling

Table 3
Correlation analysis between ischemic ABG parameters and duration of stroke All parameters were tested by Spearman's rank correlation coefficient (rs) and are given with 95% confidence interval (CI).A two-sided P value of < 0.05 was used to determine statistical significance.BE (B) base deviation in plasma, Cl -chloride ion concentration, COHb carboxyhemoglobin, ctCO2 total content of CO2, HCO3 -std standard bicarbonate, Glu glucose, Hct hematocrit, HHb deoxyhemoglobin, iCa 2+ ionized calcium ion concentration, K + potassium ion concentration, MetHb methemoglobin, Na + sodium ion concentration, Na + :K + ratio sodium-to-potassium ratio, O2Hb fractional oxyhemoglobin, paCO2 arterial partial pressure of carbon dioxide, paO2 arterial partial pressure of oxygen, sO2 oxygen saturation of hemoglobin, tHb total hemoglobin

Table 4
Correlation analysis of ischemic ABG parameters with baseline and follow-up NIHSS All parameters were tested by Spearman's rank correlation coefficient (rs) and are given with 95% confidence interval (CI).A two-sided P value of < 0.05 was used to determine statistical significance.BE (B) base deviation in plasma, Cl -chloride ion concentration, COHb carboxyhemoglobin, ctCO2 total content of CO2, HCO3 -std standard bicarbonate, Glu glucose, Hct hematocrit, HHb deoxyhemoglobin, iCa 2+ ionized calcium ion concentration, K + potassium ion concentration, MetHb methemoglobin, Na + sodium ion concentration, Na + :K + ratio sodium-to-potassium ratio, NIHSS National Institutes of Health Stroke Scale, O2Hb fractional oxyhemoglobin, paCO2 arterial partial pressure of carbon dioxide, paO2 arterial partial pressure of oxygen, sO2 oxygen saturation of hemoglobin, tHb total hemoglobin