Deep Isoflurane Anesthesia Is Associated with Alterations in Ion Homeostasis and Specific Na⁺/K⁺-ATPase Impairment in the Rat Brain

Animal procedures were approved by the animal welfare authorities in Berlin, Germany (Landesamt für Gesundheit und Soziales, G0264/14, T0096/02) and conducted following the Charité Universitätsmedizin Guidelines. Reporting complies with the Animal Research: Reporting of In Vivo Experiments Guidelines.

Coronal slices from 34 Wistar rats of both sexes (22 males and 12 females; weight, 250 to 300 g; age, 7 to 8 weeks) were prepared and maintained in interface conditions as previously described.¹⁶  The experiments in female rats were performed in response to peer review to replicate our results in both sexes. For field potential, extracellular potassium, extracellular sodium, and extracellular calcium recordings, ion-selective microelectrodes were built as previously described.¹⁶  Seizure-like events were induced by perfusion with magnesium-free artificial cerebrospinal fluid (CSF). Spreading depolarizations were induced focally by a local application of 3 M KCl.

In slice experiments, isoflurane was added to carbogen (95% O₂ and 5% CO₂, continuous gas flow, 1 l/min). TREK-1 and TASK-1/3 two-pore domain potassium channels were blocked with tetrodotoxin (1 µM), amlodipine-besylate (50 µM), and doxapram-hydrochloride (50 µM) respectively.^17–19 In vivo, 10 µM ouabain-octahydrate was solved in artificial CSF and perfused to the cortex for Na⁺/K⁺-ATPase inhibition.

The Na⁺/K⁺-ATPase activity was measured using a coupled enzyme assay as previously described.⁵  Conversion of adenosine triphosphate (ATP) to adenosine diphosphate by adenosine triphosphatases (ATPases) is coupled to NADH oxidation, which was measured spectrophotometrically (340 nm). Total Na⁺/K⁺-ATPase activity and the α2/3 fraction represented the fraction of total ATPase that was suppressible by 10 mM ouabain and 10 µM ouabain, respectively. The slices were maintained in standard conditions or exposed to 3% isoflurane for 60 min, flash-frozen in liquid nitrogen, and homogenized. In a complementary run, we added isoflurane in liquid form to the reaction mixture of untreated homogenates to achieve high isoflurane concentrations (1 and 3 vol%) and study unspecific effects. Assays and data analysis were blinded.

A total of 20 male Wistar rats (approximately 250 g, 7 to 8 weeks old) were anesthetized with isoflurane and nitrous oxide (1.5% and 70%, respectively) in an induction chamber (5 animals were excluded from the analysis; see Supplemental Digital Content, https://links.lww.com/ALN/D106). Nitrous oxide was applied only during anesthesia induction (first 3 to 5 min) to limit the isoflurane concentration before starting the electrophysiologic recordings and to achieve analgesia before local lidocaine application. After induction, anesthesia was maintained exclusively with 1 to 2% isoflurane using a mask. Pulse oximetry and body temperature (maintained at 37.0 ± 0.5°C) were monitored. After tracheotomy, mechanical ventilation was started, and end-tidal carbon dioxide was maintained at approximately 35 mmHg. Middle arterial pressure was maintained at 65 to 75 mmHg by infusing 8 to 10 ml · kg^˗1 · h^˗1 saline (for physiologic records during experiments, see supplemental tables 1 to 3, https://links.lww.com/ALN/D106). A small craniotomy in the frontal region, 2 mm from the sagittal suture and 2 mm from the coronal suture, was performed. Local anesthesia was applied (lidocaine 0.1%) before skin incisions (tracheotomy, femoral artery cannulation, and craniotomy). The craniotomy window was perfused with artificial CSF. Ion-selective microelectrodes and a stimulation electrode (in experiments concerning extracellular potassium clearance) were inserted into the frontal cortex through a small dura opening. After registration of phase 2 anesthesia (alpha or delta band, 0 to 12 Hz), isoflurane was titrated (from 1 to 1.5% up to 2.5 to 3.5%) to establish burst suppression. Complementary to extracellular corticography, a pinch test was performed at regular time points to check for the behavioral state (see also supplemental tables 1 to 3, https://links.lww.com/ALN/D106). The animals were sacrificed under deep anesthesia at the end of the experiments.

We used an electrophysiologic model, which describes the dynamics of the ion species sodium, potassium, and chloride and the membrane potential. Changes in extra- and intracellular ion concentrations resulted from electrodiffusive movement through voltage-dependent and -independent channels and ATP-dependent ion transport. Permeabilities for different ion species comprise basal permeabilities, summarizing passive ion transport and ion-coupled transport, gated permeabilities for potassium and sodium, describing voltage-gated channels required for action potential generation, and ligand-gated permeability for sodium describing synaptic activation. Active ion transport was modeled by the Na⁺/K⁺-ATPase, which exchanges three sodium ions for two potassium ions. Na⁺/K⁺-ATPase activity depended on extracellular potassium, extracellular sodium, and intracellular ATP. As the affinity of ATP was 1 order of magnitude below the physiologic ATP level (K_m = 300 µM, cellular ATP concentration 3 mM), the Na⁺/K⁺-ATPase would always be saturated with ATP. Passive electrodiffusion through the ion channels was calculated using the Goldman–Hodgkin–Katz equation. For mathematical equations and references, see the supplemental Materials and Methods, including supplemental tables 4 to 9 (https://links.lww.com/ALN/D106).

Data preprocessing and statistical significance testing were performed using Origin (version 6.0, OriginLab Corporation, USA), the scientific Python stack (Python version 3.8.3, Scipy version 1.7.3, Numpy version 1.21.4, Pandas version 1.1.3), and the statistical Python packages statsmodels (version 0.12.1), pingouin (version 0.5.0), and scikit-posthocs (version 0.7.0). Data visualizations were prepared using the Python plotting library Matplotlib (version 3.2.1), Matlab (version R2020B, MathWorks Inc., USA), and Origin.

All data were assessed visually regarding the underlying distributions using Q-Q plots and histograms and by testing for normality using the Shapiro–Wilk test. For null hypothesis significance testing, we conducted parametric tests in which assumptions were met; otherwise, nonparametric alternatives were employed. When two groups were compared (t tests, Mann–Whitney U test), only two-sided versions were used. When parametric assumptions were met, data from within-subject design experiments were tested using the one-way repeated-measures ANOVA test followed by pairwise t tests and the Benjamini–Hochberg procedure to control for the false discovery rate. The sphericity assumption was tested using Mauchly’s test. For the other cases, we used the nonparametric Friedman test, followed by Conover’s post hoc test of multiple comparisons using rank sums and the Benjamini–Hochberg procedure. To compare the means of three independent samples of enzyme assay data, we conducted a one-way ANOVA followed by Tukey’s honest significant difference test. For samples from a non-normally distributed population, we used the Mann–Whitney U test to compare two independent samples and the Wilcoxon signed-rank test to compare two dependent samples.

For brain slice experiments, several slices (see Supplemental Digital Content, https://links.lww.com/ALN/D106) have been obtained from the same animal. In the text, the number of experimental units is given as uppercase “N” for animals and lowercase “n” for brain slices. For consistency and unless stated otherwise, all data are given as means ± SD. Changes were stipulated to be significant for p values greater than 0.05. For a detailed description of measures to reduce subjective bias, sample size calculation, and excluded animals and slices, respectively, please refer to the supplemental Materials and Methods section in the Supplemental Digital Content (https://links.lww.com/ALN/D106).

We exposed neocortical slices to isoflurane concentrations that typically induce phase 2 (1%) and burst suppression (3%) anesthesia in rats.¹⁷  Since the concentrations of potassium and sodium are interdependent in brain tissue,^2,20  we simultaneously monitored both ions and stimulus-induced synaptic activity (fig. 1, a and b; supplemental table 10, https://links.lww.com/ALN/D106). A high isoflurane concentration of 3% significantly affected ion concentrations and neuronal excitability. Baseline extracellular potassium increased from 3.0 ± 0.0 mM to 3.9 ± 0.5 mM (P < 0.001; n = 39), extracellular sodium decreased from 153.4 ± 0.8 mM to 145.2 ± 6.0 mM (P < 0.012; n = 28), and stimulus-evoked field potential decreased from 1.2 ± 0.7 mV to 0.6 ± 0.4 mV (P < 0.001; n = 37).

Isoflurane is known to activate two-pore-domain potassium channels, resulting in an outward potassium leak and membrane hyperpolarization,¹⁴  which could potentially affect extracellular potassium. Therefore, we assessed the effects of 3% isoflurane on extracellular potassium, extracellular sodium, and extracellular calcium baseline in the presence of amlodipine (50 µM) and doxapram (50 µM). To exclude the contribution of synaptic activity, we recorded in the presence of 1 µM tetrodotoxin as well. Under these conditions, we observed extracellular potassium increase while extracellular sodium (n = 18) and extracellular calcium decreased (n = 16; all P < 0.001; fig. 1c; supplemental table 11, https://links.lww.com/ALN/D106), indicating that isoflurane effects on ion distribution might be independent of neuronal activity and two-pore-domain potassium channels. Importantly, no sex-specific differences were observed in the effects of isoflurane on interstitial ion changes (supplemental fig. 1, https://links.lww.com/ALN/D106).

To investigate whether isoflurane impaired extracellular potassium clearance, we measured its effects on seizure-like events and spreading depolarizations, which are characterized by a moderate and a near breakdown of ion balance accompanied by extracellular potassium increases from 3 to ~12 mM and from ~30 mM, respectively (fig. 2).²¹  Isoflurane significantly lowered the incidence of seizure-like events, which is in agreement with its known antiepileptic effects.²²  After seizure-like event-related extracellular potassium increases, early (T1₅₀) and late (T2₅₀) extracellular potassium clearance time constants (fig. 2, a to c; supplemental table 12, https://links.lww.com/ALN/D106) were significantly longer with stronger effects of 3% isoflurane (P = 0.001 for both T1₅₀ and T2₅₀; n = 21). Accordingly, seizure-like event duration increased during isoflurane treatment (control, 1.7 ± 0.6 s; 1%, 3.9 ± 1.2 s; and 3%, 4.9 ± 0.9 s; P < 0.001 vs. control for all; n = 21). Interestingly, these changes were accompanied by signs of increased excitability as seizure-like event-related extracellular potassium augmented (control, 2.5 ± 1.5; 1%, 3.7 ± 1.9; and 3%, 4.3 ± 2.0 mM; P < 0.001 vs. control; n = 21) and the number of spikes per seizure-like event increased (control, 13.53 ± 8.54; 1%, 29.28 ± 11.65; and 3%, 42.52 ± 10.38 spikes/seizure-like event; P < 0.001 vs. control for all; n = 21; supplemental table 12, https://links.lww.com/ALN/D106).

We next assessed isoflurane effects on extracellular potassium and extracellular sodium during spreading depolarizations (fig. 2, d and e; supplemental table 13, https://links.lww.com/ALN/D106). Isoflurane attenuated spreading depolarization-associated absolute extracellular potassium increases (only significant during 3% isoflurane; P < 0.001; n = 14). Spreading depolarization-related extracellular potassium clearance and extracellular sodium restoration time were significantly longer in the presence of 3% isoflurane. T1₅₀ doubled and T2₅₀ tripled in the presence of 3% isoflurane for both extracellular potassium (P < 0.001 for both) and extracellular sodium (P = 0.001 and < 0.001, respectively; n = 14). As shown in supplemental figure 2 (https://links.lww.com/ALN/D106), the measured changes on seizure-like events and spreading depolarization were sex-independent.

As extracellular potassium distribution and clearance are importantly controlled by the Na⁺/K⁺-ATPase in neuronal tissue,¹  we measured Na⁺/K⁺-ATPase activity in brain slices that were exposed to isoflurane (fig. 3, a and b). Compared to control slices, tissue exposure to 3% isoflurane for 60 min resulted in an approximately 25% decrease in the activity of the α2/3 fraction (i.e., the astrocytic or neuronal isoforms) of the Na⁺/K⁺-ATPase (control, 2.1 (1.5, 2.7) µmol P/mg · h; treated, 1.5 (1.0, 2.1) µmol P/mg h; n = 23 and 24 slices, respectively; P = 0.011). By contrast, 3% isoflurane did not significantly affect the ubiquitous α1 fraction or other ATPases, which are involved in the translocation of magnesium and Ca²⁺.²³  To test for indirect effects of isoflurane on membrane-bound enzymes, such as mediated by modifications of the lipid bilayer,²⁴  we measured Na⁺/K⁺-ATPase activity in homogenates exposed to concentrations that far exceed its clinical application by adding isoflurane in liquid form to the reaction mixture (fig. 3, c and d). Using this approach, the α2/3 activity fraction reduction increased to approximately 50%, while the effect on the housekeeping isoform (α1) remained comparably small. Of note, at this high concentration, the assay revealed unspecific effects of isoflurane on other ATPases (fig. 3d, right), demonstrating its capacity to detect unspecific isoflurane action when present.

To investigate isoflurane effects in vivo, we next determined extracellular potassium clearance (first group) and baseline extracellular potassium (second group) in the frontal cortex of Wistar rats during phase 2 and burst suppression anesthesia. Compared to phase 2, the early (T1₅₀) and late (T2₅₀) extracellular potassium clearance time constant after cortical stimulation almost doubled during burst suppression (N = 4; P = 0.001 for both; fig. 4, a and b; supplemental table 14, https://links.lww.com/ALN/D106). We assessed extracellular potassium baseline while deepening isoflurane anesthesia from phase 2 into burst suppression, which showed a biphasic response with an initial dip followed by a sustained increase. Accordingly, extracellular potassium rose from 2.9 ± 0.1 mM during phase 2 to 3.5 ± 0.6 mM 50 min after burst suppression induction and further increased to 4.1 ± 0.7 mM after 90 min (N = 6; supplemental table 15, https://links.lww.com/ALN/D106). During burst suppression establishment, burst amplitude and burst-associated extracellular potassium rises increased, while the associated extracellular potassium clearance was impaired as the steepness of extracellular potassium decay slopes, after single bursts, decreased from ˗55.5 ± 14.6 to ˗33.5 ± 14.3 µM.s^-1 (P < 0.001; N = 6; supplemental table 16, https://links.lww.com/ALN/D106).

We performed computational modeling for theoretical verification and assessment of functional consequences of partial inhibition of the Na⁺/K⁺-ATPase by isoflurane. Finally, we studied changes in cortical activity during partial inhibition of the Na⁺/K⁺-ATPase in vivo (fig. 5).

In silico, gradual Na⁺/K⁺-ATPase inhibition generated a extracellular potassium increase, leading to a new steady-state consistent with previous observations using ouabain in vitro and in vivo.²⁵  An increase in extracellular potassium from 3 to 4 mM (as observed experimentally) occurred at an approximately 35% Na⁺/K⁺-ATPase inhibition (fig. 5a). Notably, at approximately 60% inhibition, a steep rise in extracellular potassium was predicted, indicating a breakdown of ion homeostasis. We further simulated the effects of Na⁺/K⁺-ATPase inhibition on the neuronal membrane potential and extracellular potassium clearance during stimulus-induced network activation and spreading depolarization simulations (fig. 5, b and c). As observed in vivo (fig. 4d), Na⁺/K⁺-ATPase inhibition (35%) impaired extracellular potassium clearance after simulated cortical stimulation. Subsequently, spreading depolarizations were simulated by increasing extracellular potassium to approximately 35 mM locally while increasing sodium permeability to 20 × 10^˗9 m/s for 20 s. Mass conservation was achieved by removing extracellular potassium over a time period of 20 s. This mimicked the contribution of neighboring neurons to potassium clearance during spreading depolarization, which depolarizes due to increasing extracellular potassium and excitatory neurotransmitter release. Excess of extracellular potassium is cleared by an increase in Na⁺/K⁺-ATPase activity, spread through the astrocyte network, and transported to the bloodstream. As shown in figure 5c, partial inhibition of the Na⁺/K⁺-ATPase activity prolongs extracellular potassium clearance, as well as neuron depolarization.

Changes in transmembrane potassium gradients alter the resting membrane potential, which directly affects neuronal excitability and network activity.¹  We modeled the effects of Na⁺/K⁺-ATPase inhibition on neuronal excitability by generating periodic excitatory inputs (increase in Na⁺ permeability to 20 × 10^˗9 m/s) and by decreasing Na⁺/K⁺-ATPase activity by 10% and 35% (fig. 5d). While at 10% inhibition, only subthreshold membrane depolarization and small changes in extracellular potassium were elicited, at 35% inhibition of the Na⁺/K⁺-ATPase, the excitatory input generated neuronal spiking, leading to local extracellular potassium accumulation, which in turn decreased the excitatory threshold.

To assess isoflurane effects on excitability experimentally, we applied artificial CSF-containing ouabain (10 µM) to the cortex during phase 2 anesthesia. In these in vivo recordings, cortical excitability increased and displayed periods of bursting, resulting in a significant increase in the oscillatory power range (P = 0.02; N = 5; fig. 5e).

We observed that isoflurane, at concentrations that induce burst suppression anesthesia, substantially slowed extracellular potassium clearance and increased baseline extracellular potassium in the rat brain. A specific decrease of the α2/3 fraction of Na⁺/K⁺-ATPase activity, as well as the observed changes in ion homeostasis, suggested sodium pump impairment as a contributing mechanism.

We found a specific decrease of the α2/3 fraction activity by isoflurane in tissue treated with concentrations relevant to deep anesthesia (fig. 3, a and b). Importantly, brain homogenates exposed to very high concentrations of isoflurane display lower activity in other subunits, but the effects on the α2/3 fraction remained the main source of Na⁺/K⁺-ATPase impairment (fig. 3, d and e). While Na⁺/K⁺-ATPase inhibition by isoflurane has been reported for cultured alveolar type 2 cells,²⁶  to our knowledge, this is the first report of specific Na⁺/K⁺-ATPase inhibition in the mammalian brain. The observed isoform specificity suggests direct, specific interaction with distinct α subunits rather than unspecific effects based on interactions of the membrane and lipophilic portions of the transmembrane proteins. The resulting dose-dependent effect of isoflurane on extracellular potassium clearance favors the conclusion that the astrocytic (α2) isoform, which is arguably the most important for removing extracellular potassium in the wake of neuronal activity, might be targeted and impaired by isoflurane.²⁷  Unlike α1, α2 and α3 serve more specialized cellular roles as they are functionally coupled to secondary active transporters.²⁸  The α2 isoform colocalizes with the sodium/calcium exchanger, which is a key mechanism to maintain physiologic cytosolic Ca²⁺ concentration. A reduction in α2 isoform activity leads to increased intracellular Ca²⁺ and amplification of intracellular Ca²⁺ signaling.²⁹  Thus, our experimental results on extracellular calcium in slices treated with isoflurane (fig. 1c) are in good agreement with a sodium/calcium exchanger dysfunction secondary to Na⁺/K⁺-ATPase inhibition. Furthermore, the α2 isoform provides the local steep ion gradients that drive glutamate reuptake transporters crucially involved in the maintenance of extracellular glutamate concentration.³⁰  In this line, α2 isoform dysfunction compromises glutamate reuptake and potassium clearance, promoting spreading depolarization.³¹  Isoflurane has also been shown to inhibit the astrocytic Kir4.1 channel, which mediates spatial potassium buffering after neuronal activation and could contribute to the observed changes in ion dynamics.³² 

Other than the impairment of the activity-dependent extracellular potassium clearance, the source of baseline extracellular potassium (and decrease of extracellular sodium and extracellular calcium) rises might include other mechanisms than Na⁺/K⁺-ATPase impairment. Two-pore-domain background potassium channel inhibition, activation of ATP-dependent potassium channels, and decrease in extracellular space might contribute to the observed alterations in ion distribution.^14,15,33  Although complete pharmacologic inhibition of TREK and TASK is difficult, the effects in ion distribution under isoflurane were well replicated in slices treated with amlodipine and doxapram, suggesting that Na⁺/K⁺-ATPase inhibition might contribute to ion redistribution (fig. 1c). Moreover, the observed redistributions in extracellular sodium and extracellular calcium in our experiments would not necessarily be expected in the case of an isolated two-pore-domain potassium channel activation. Thus, isoflurane-induced Na⁺/K⁺-ATPase impairment might act synergistically with two-pore-domain potassium channel opening, activation of ATP-dependent potassium channels, and/or decrease of the interstitial space to generate the observed changes in ion distribution. Concerning anions, burst suppression has been associated with increasing extracellular chloride,⁹  which is in line with the observed decrease in extracellular cations.

A biophysical model reproduced the main observations of the experimental part of this study, thereby supporting the plausibility of our mechanistic assumptions (fig. 5). Partial Na⁺/K⁺-ATPase inhibition by approximately 35% had similar effects on both extracellular potassium baseline and activity-dependent potassium clearance as 3% isoflurane. Additionally, Na⁺/K⁺-ATPase inhibition increased excitability, converting subthreshold membrane depolarizations to spiking episodes, which resembled a burst-suppression pattern (fig. 5e). Accordingly, seizure-like event duration, extracellular potassium increases, and the number of spikes per event increased under 3% isoflurane in slices. In vivo, partial Na⁺/K⁺-ATPase inhibition induced bursting activity in the cortex (figs. 2b and 5e).

During burst-suppression anesthesia, profound synaptic inactivation and cortical hyperexcitability coincide.⁶  Isoflurane suppresses neuronal activity by activation of γ-aminobutyric acid type A receptors, inhibition of presynaptic vesicle exocytosis, and inhibition of N-methyl-d-aspartate receptors.^34,35  On the other hand, the exact mechanisms underlying cortical hyperexcitability and the generation of bursts are still under debate. Interestingly, neocortical application of 10 µM ouabain to achieve a partial Na⁺/K⁺-ATPase inhibition (i.e., of the α2/3 fraction) in vivo induced burst activity during phase 2 anesthesia (fig. 5e). This finding is particularly intriguing since it suggests that isoflurane-induced Na⁺/K⁺-ATPase inhibition contributes to the generation of cortical hyperexcitability during deep anesthesia. Impairment of extracellular potassium clearance, glutamate reuptake decrease, and abolished long-lasting afterhyperpolarization³⁶  secondary to Na⁺/K⁺-ATPase inhibition represent potential mechanisms.

Na⁺/K⁺-ATPase impairment by isoflurane and effects on ion distributions might also contribute to the suppression of cortical activity during deep anesthesia. The observed extracellular potassium baseline increase, combined with a longer extracellular potassium decay time, might extend the suppression period between bursts since neuronal repolarization and sodium channel activation are delayed. The significant drop of extracellular calcium from 1.6 to 1.2 mM, measured in brain slices (fig. 1c), suggests a substantial accumulation of intracellular Ca²⁺ secondary to a Na⁺/K⁺-ATPase dysfunction. Indeed, extracellular Ca²⁺ depletion has been proposed to generate the quasiperiodic burst suppression pattern by impairing synaptic transmission.^7,9 

While we observed impairment in extracellular potassium clearance both ex vivo and in vivo, changes concerning extracellular potassium accumulation started much later in leaving animals than in brain slices (50 min vs. 7 min; figs. 1, a and c, and 4, c and d). Differences in the underlying mechanisms controlling extracellular potassium between in vivo and ex vivo could account for the observed discrepancies: (1) loss or dysfunction of astrocytes and less extracellular space tortuosity after slicing,³⁷  (2) absence of blood circulation and therefore impairment of the neurovascular unit with reduced capacity to remove potassium from brain to blood,³⁸  and (3) differences in isoflurane pharmacokinetics in vivo due to transport and distribution between more compartments.³⁹  Furthermore, isoflurane-induced inhibition of mitochondrial complex I and consequent decrease in ATP availability⁴⁰  could also impair Na⁺/K⁺-ATPase. However, such a decrease in ATP availability would severely compromise tissue viability (not observed in our experiments) and might occur at very high enzyme inhibition.¹⁷ 

While inhalational anesthetics are considered safe regarding organ toxicity, the potential neurotoxicity of general anesthesia remains a controversial topic. We observed similar effects of isoflurane exposure on ion homeostasis in acute brain slices obtained from both male and female rats. This is particularly relevant in light of the reported sex-specific sensitivity to general anesthesia and increases the generalizability of the experimental findings of this study.^41,42  Deeper anesthetic levels, such as during burst suppression anesthesia, have been associated with postoperative neurologic complications and poorer clinical outcomes in intensive care patients.^11,12  It is conceivable that Na⁺/K⁺-ATPase impairment contributes to neurotoxicity during long-lasting deep anesthesia while impaired removal of extracellular glutamate and slowing of the sodium/calcium exchanger could result in excessive cytosolic Ca²⁺ accumulation, thereby activating cytotoxic cascades. However, these adverse effects could also be partially compensated by the inhibitory effects of isoflurane on N-methyl-d-aspartate receptors³⁴  and activity-dependent reduction in neuronal energy demand.¹⁷  Certainly, further research will be necessary to better understand the relevance of our findings.

The authors dedicate this work to Uwe Heinemann, M.D., PhD., (deceased, Charité - Universitätsmedizin Berlin, Berlin, Germany) an unforgettable mentor. They thank Jörg R.P. Geiger, Ph.D. (Charité - Universitätsmedizin Berlin, Berlin, Germany) and Dietmar Schmitz, M.D. (Charité - Universitätsmedizin Berlin, Berlin, Germany) for their support and Christoph Barner, M.D. (Charité - Universitätsmedizin Berlin, Berlin, Germany), Jérémie Sibille, Ph.D. (Charité - Universitätsmedizin Berlin, Berlin, Germany), and Iwona Wallach, Ph.D. (Deutsches Herzzentrum der Charité, Berlin, Germany) for critical reading of the manuscript.

Supported by Deutsche Forschungsgemeinschaft grants 408355133 (to Dr. Liotta and Dr. Berndt), DR 323/5-1 (to Dr. Dreier), and DR 323/10-1 (to Dr. Dreier) and by Bundesministerium für Bildung und Forschung grant 0101EW2004 (to Dr. Dreier). Supported in part by the Sonnenfeld Stiftung (to Dr. Kovács) and the Clinician Scientist Program, Berlin Institute of Health (Berlin, Germany; to Dr. Liotta).

The authors declare no competing interests.

An extended description of the authors’ methods, references, analysis, and results can be found in the Supplemental Digital Content (https://links.lww.com/ALN/D106).