Incidence,hemodynamic,and electrical characteristics of spreading depolarization in a swine model are affected by local but not by intravenous application of magnesium

Animal preparation

The experiments were approved by the ethics commission Regierungspräsidium Karlsruhe, Karsruhe, Germany (Nr. G-167/14) according to the 2010/63/EU guidelines and following the ARRIVE guidelines for reporting in vivo experiments. Three groups of five German Landrace swine (mean weight of 29.3 kg, 3–4 months of age) were craniotomized and monitored for over 6 h, to test local and i.v. Mg²⁺ vs. controls. The animal handling protocol was described previously ¹¹ and in the Supplementary material. An extensive craniotomy and dura mater excision was performed to expose both hemispheres. The translucent border of a 6-contact platinum ECoG strip was placed on the cortex surface for ECoG monitoring. The ECoG was recorded in one of the two hemispheres from a subdural strip. IOS settings and recording were performed as described before.^12,13

SD induction and structure of the experiment

SDs were induced with a small drop (1–3 μl) of KCl (1 mol/l), which was placed directly onto the cortical surface in a central visible area, using a dull needle from an insulin syringe. The brain cortex was stimulated hourly in both hemispheres. The experiment consisted in two hourly stimulations before therapy or vehicle, followed by three hourly stimulations (Figure 1(a)). Comparison of SD from the first stimulation with the SD from the third simulation, and the SD from the second stimulation with the SD from the fourth stimulation was performed. They presented the same recovery time after washout and placement of the paraffin pool. The change of medium affects the temperature and ion concentrations and it can interfere with IOS variables, as seen in previous experiments, ¹² but the washout is better to maintain an adequate visualization of the SDs free of blood and detritus. OPEN IN VIEWER

Experimental groups

Stimulations were followed by a 1-h observation period; if more than one SD appeared in the 1-h period of observation (clusters), only the first SD was considered for analysis. All of the animals had two hourly stimulations before the first medium change occurred. One hour after the second stimulation, the medium covering the brain was changed to either NaCl 0.9% (in control and Mg²⁺ i.v. groups) or NaCl 0.9% with 20 mmol/l=8.12 mEq/l MgSO₄ (in the local Mg²⁺ group). Simultaneously, an i.v. application of either 40 ml MgSO₄ 10% (in the Mg²⁺ i.v. group) or NaCl 0.9% (in the control and local Mg²⁺ group) was performed, according to the corresponding group. After 10 min of brain exposure, solutions were washed out and a pool of paraffin covered the brain again. After stabilization for 10 min, the brain was again stimulated (third, fourth, and fifth stimulations).

Data analysis

IOS and ECoG recordings were analyzed for SD events, according to the methods described previously^14,15 using the LabChart7 software. In IOS, for each experiment, a number of six regions of interest (ROIs) of about 10×10 pixels (approximately 0.06–0.2 mm²) in size were selected and distributed along the hemispheres. Details on how ROIs were distributed and used for analysis are given in the Supplementary material. After obtaining a baseline intensity value in IOS, the total amplitude of the intensity change, the minimum of the intensity change (equal to the peak hyperemia due to increased hemoglobin concentration), and the maximum of the intensity change (equal to peak oligemia due to decreased hemoglobin tissue concentration), which are expressed as changes relative to the baseline in percantage, were compared to the wave components of the SDs without therapy. Measurements of the time to reach peak hyperemia, the duration of oligemia, the duration of the total hemodynamic response, and the number of ROIs reached per SD, were performed.

The speed was measured separately. Two dedicated ROIs separated by a distance of 6 mm were chosen in a direction perpendicular to the SD expansion, in order to calculate the time to expand the given distance.

Statistics

Descriptive statistics were calculated for all outcome variables of interest. Continuous variables were assessed for normality, using histograms and Kolmogorov–Smirnov tests. In order to test for differences in the means, medians and number of events, the (paired) Student’s t-test, the Wilcoxon-signed rank test, the Kruskal–Wallis rank sum test (ANOVA on ranks), and the one-way analysis of variance (ANOVA) were used, where appropriate. Descriptive statistics and test group differences were obtained, using the statistical analysis software SPSS 20.0. P values <0.05 were considered statistically significant.

Physiological parameters and SDs before therapy

Physiological parameters showed no significant difference when comparing the mean value one hour before intervention to the mean value 1 h after intervention (paired Student’s t-test and Wilcoxon-signed rank test accordingly, Supplementary Table 1).

Before therapy administration (i.e. during the first and second stimulations), SDs could be evoked through KCl in both groups, according to the established stimulation times. SDs were viewed mostly in IOS as biphasic and/or triphasic hemodynamic changes (Figure 1(b)).

Effect of local Mg²⁺ on SD

A profound inhibitory effect on SD induction was observed following localized exposures to Mg²⁺ (Figure 1(b) and (c)). Ten minutes after exposure to localized Mg²⁺, SDs were not able to be elicited. SDs were completely blocked in IOS for the third stimulation. Nevertheless, we detected an electrical signal in the ECoG, in the electrode adjacent to the stimulation, which did not spread. Susceptibility to SD progressively recovered over the subsequent 2 h, such that by the fourth stimulation there were SDs in 5 of 10 hemispheres, an expansion occurred in 43% of the predefined ROIs and in the half of the cases SDs appeared to have reached 90% of the ECoG channels. After the fifth stimulation, SDs were detected in 8 of 10 hemispheres and presented an expansion in 60% of the ROIs. In 80% of the cases, SDs for this stimulation expanded to 100% of the ECoG channels (see supplementary video online).

IOS characteristics for the fourth SD of the local groups showed a significant decrease in total amplitude when compared to the second SD (−22.37%, SE 7.6, p = 0.023, ANOVA with Bonferroni correction) for both the control and (−1%, SE 7.6, p = 0.027) the i.v. therapy groups. Peak hyperemia and peak oligemia showed no significant difference between groups (Figure 2(a)). The time to reach peak hyperemia showed no significant differences (ANOVA on ranks) between groups. The duration of the oligemia of the second SD was 10:04 (mm:ss), SE 00:08 and it was significantly reduced after local Mg²⁺ application compared to the control group (−01:34, SE 00:34, p = 0.036). The duration of the oligemia after local therapy (fourth simulation) also showed a significant difference compared to the SD from the second stimulation (01:51, SE 00:15, p = 0.019) (Figure 2(b)). The total duration of the hemodynamic response of the second SD was 12:14, SE 00:09 and it was significantly reduced after local Mg²⁺ application in the fourth simulation SD, compared to the control (−01:44, SE 00:35, p = 0.021) (Figure 2(b)) and compared to the SD from the second stimulation (−02:03, SE 00:30, p = 0.016). The mean speed was 4.1, SE 0.5 mm/min. The third and fourth SD showed no significant difference with the first and second SD, respectively (ANOVA on ranks). For the analysis across the groups, we found a significant decrease in speed in the SDs from the fifth stimulation (−1.7 mm/min p = 0.039) between the local Mg²⁺ and control group (Figure 2(c)). OPEN IN VIEWER

In ECoG, the SDs from the third stimulation were between local Mg²⁺and control groups. Given that the stimulation place was close to a strip electrode (4 mm), an electrical signal could be measured in all cases, at least in one electrode, although the expansion was blocked in coherence with the hemodynamic signal. When comparing the electrical characteristics of the depolarization wave, a difference was seen for local for amplitude (−32.5%, p = 0.023) but not for duration (−2.5 min, p = 0.89, ANOVA), (Figure 2(d)).

Effect of i.v. Mg²⁺ on SD

No effect was found on SD induction and expansion during i.v. Mg²⁺ infusion. In IOS, SDs properties were found to be unaffected after i.v. Mg²⁺ administration when compared with the control group. Similarly, in EcoG, no difference in amplitude was found after the intravenous administration (Figure 2(a) to (d)).

Changes in SD characteristics after local Mg²⁺

In our experiments, local Mg²⁺ not only blocked the hemodynamic and electric response, but also decreased the amplitude of the hemodynamic and electric change, the total duration and the duration of the oligemic phase of the SDs. It is not known how detrimental SDs with reduced amplitude and duration in IOS and ECoG are for the neurons. SDs with less amplitude might be less harmful per se, but SDs appearing as clusters tend to reduce their amplitude also when they get closer. On the other hand, speed has a modest sensitivity and specificity in predicting a change in SD susceptibility, with positive and negative values of 84% and 56%, respectively, in a susceptibility animal model. ¹⁷ Higher SD susceptibility should correlate with clusters in patients. Clusters of SDs have been correlated with worse brain metabolism and outcome. ¹⁸ We speculate that the main neuroprotective effect would be the reduction of SD frequency and the effectiveness of the therapy could be indirectly estimated with the reduction of the expansion, speed, and other SD characteristics in our model.

Action mechanism of Mg²⁺

One of the Mg²⁺ action mechanisms is through an NMDA receptor that is activated by a voltage-dependent process. There is a voltage nonlinear dependence of the NMDA conductance in the presence of extracellular Mg²⁺ at least from 1 μl to 10 mM, as seen in whole cell recordings. ¹⁹ NMDA activation depends on the Mg²⁺ removal of the receptor. At high Mg²⁺ concentrations, an NMDA receptor can be inactivated, reducing the flow of Na⁺, K⁺ and particularly the Ca²⁺ influx, keeping the neuron in a stable state of excitation and avoiding cellular damage during SD.

We consider that, under our experimental conditions, Mg²⁺ has an effect on the vascular response, independent from the effect on neuronal depolarization and glutamate release. It is well known that Mg²⁺ also alters the Na⁺, K⁺, and Ca²⁺ currents of the brain vessels via direct effects on vascular muscle and indirect effects on endothelial cells. ²⁰

Translation from basic research to clinical studies

Theoretically, in a clinical study, neuroprotection to the noxious SD effects could be achieved when SDs have been inhibited or the harmful hemodynamic components have been reduced. SDs were not monitored in IMAGES and MASH-2 trials but it is possible that the i.v. dose given in those studies would not have had an effect on SDs. This corresponds to basic studies, where, for example, the high intravenous MgSO₄ dose of 720 µmol/kg had no effect on the infarct volume in rats.^21,22 Intravenous Mg²⁺ reaches very low concentrations in cerebrospinal fluid when it is given over a long period of time. For example, inducing an hypermagnesemia (2.1–2.5 mmol/l) in brain-injured patients could increase the amount of Mg²⁺ in cerebrospinal fluid only by 15% (1.43 ± 0.13 mmol/l). ²³

This study supports the concept that Mg²⁺ may still benefit patients with SAH and stroke if it were applied topically, either intracisternally or intrathecally, as also supported by rat experiments.^4,5,7 A recent monocentric study randomized 70 patients presenting with aneurysmal SAH (WFNS Grades 1–4, Fisher Grades 2–3). One group received standard treatment with cisternal irrigation therapy and the other with continuous infusion of MgSO₄ in addition. No significant differences in outcome were observed after three months between the two groups, ²⁴ but a concentration of 1.2 mmol/l would be too low to suppress SDs in many animal models and brain slices.^17,25 In our study, we applied Mg²⁺ at significantly higher concentrations (≈20 mmol/l).

Study limitations

This study has some limitations that should be considered. One of them is the disadvantage presented by IOS in which the visualization of SDs is not only selective for cerebral hemoglobin in a qualitative manner, but also for other not well known intracellular changes related to the mitochondrial redox status seen in histological preparations without hemoglobin.^26,27