Disruption of Frontal Parietal Communication by.14

8/13/2019 Disruption of Frontal Parietal Communication by.14

1/12

Anesthesiology, V 118 No 6 1264 June 2013

ABSTRACT

Introduction:Directional connectivity from anterior to poste-

rior brain regions (or feedback connectivity) has been shownto be inhibited by propofol and sevoflurane. In this study theauthors tested the hypothesis that ketamine would also inhibitcortical feedback connectivity in frontoparietal networks.Methods:Surgical patients (n = 30) were recruited for inductionof anesthesia with intravenous ketamine (2 mg/kg); electroen-cephalography of the frontal and parietal regions was acquired.Te authors used normalized symbolic transfer entropy, a compu-tational method based on information theory, to measure direc-tional connectivity across frontal and parietal regions. Statisticalanalysis of transfer entropy measures was performed with thepermutation test and the time-shift test to exclude false-positiveconnectivity. For comparison, the authors used normalized sym-bolic transfer entropy to reanalyze electroencephalographic datagathered from surgical patients receiving either propofol (n = 9)or sevoflurane (n = 9) for anesthetic induction.

Results: Ketamine reduced alpha power and increasedgamma power, in contrast to both propofol and sevoflurane.During administration of ketamine, feedback connectiv-ity gradually diminished and was significantly inhibited afterloss of consciousness (mean SD of baseline and anesthe-sia: 0.0074 0.003 and 0.0055 0.0027; F(5, 179) = 7.785,P< 0.0001). By contrast, feedforward connectivity was preservedduring exposure to ketamine (mean SD of baseline and anes-thesia: 0.0041 0.0015 and 0.0046 0.0018; F(5, 179) = 2.07;P= 0.072). Like ketamine, propofol and sevoflurane selectivelyinhibited feedback connectivity after anesthetic induction.Conclusions: Diverse anesthetics disrupt frontalparietal

communication, despite molecular and neurophysiologicdifferences. Analysis of directional connectivity in frontalparietal networks could provide a common metric of general

What We Already Know about This Topic

Ketamine has markedly distinct molecular and neurophysi-ologic properties compared with propofol and sevourane

Because of its unique properties, ketamine does not conform to

most general theories of anesthetic-induced unconsciousness

What This Article Tells Us That Is New

Ketamine, propofol, and sevourane all selectively impair fron-tal-to-parietal brain communication, as measured in humansurgical patients using electroencephalography and symbolictransfer entropy

Disrupted frontalparietal communication may be a commonmetric for general anesthesia and potentially a nal commonpathway by which diverse anesthetics induce unconsciousness

This article is featured in This Month in Anesthesiology.Please see this issue of ANESTHESIOLOGY, page 3A.

This art icle is accompanied by an Editorial View. Please see:Sleigh JW: The study of consciousness comes of age. ANES-THESIOLOGY2013; 118:12456.

Supplemental digital content is available for this article. DirectURL citations appear in the printed text and are available inboth the HTML and PDF versions of this article. Links to thedigital les are provided in the HTML text of this article on theJournals Web site (www.anesthesiology.org).

Copyright 2013, the American Society of Anesthesiologists, Inc. LippincottWilliams & Wilkins.Anesthesiology 2013; 118:1264-75

* Research Investigator, Department of Anesthesiology, Univer-sity of Michigan Medical School, Ann Arbor, Michigan. AssociateProfessor, Assistant Professor, Department of Anesthesiology andPain Medicine, Professor, Department of Clinical Pharmacologyand Therapeutics, Department of Anesthesiology and Pain Medi-cine, Asan Medical Center, University of Ulsan College of Medicine,Seoul, Korea. || Assistant Professor and Associate Chair for Faculty

Affairs, Department of Anesthesiology, and Faculty, NeuroscienceGraduate Program, University of Michigan Medical School.

Received from the Division of Neuroanesthesiology, Depart-

ment of Anesthesiology, University of Michigan Medical School,Ann Arbor. Submitted for publication August 18, 2012. Acceptedfor publication February 28, 2013. Supported by the National Insti-tutes of Health, Bethesda, Maryland, Grant 1RO1GM098578 (to Dr.Mashour) and the Department of Anesthesiology, University ofMichigan. Drs. Mashour and Lee hold a provisional patent (throughthe University of Michigan, Ann Arbor) on the measurement ofdirectional connectivity for anesthetic monitoring. Drs. Lee and Kucontributed equally to this article.

Address correspondence to Dr. Mashour: Department of Anes-thesiology, University of Michigan Medical School, 1H247 UH/SPC-5048, 1500 East Medical Center Drive, Ann Arbor, Michigan48109-5048. [email protected]. This article may be accessedfor personal use at no charge through the Journal Web site,www.anesthesiology.org.

Disruption of FrontalParietal Communicationby Ketamine, Propofol, and Sevoflurane

UnCheol Lee, Ph.D.,* SeungWoo Ku, M.D., Ph.D., GyuJeong Noh, M.D., Ph.D.,SeungHye Baek, M.D., Ph.D., ByungMoon Choi, M.D., Ph.D., George A. Mashour, M.D., Ph.D.||

PERIOPERATIVE MEDICINE
http://www.anesthesiology.org/mailto:[email protected]://www.anesthesiology.org/http://www.anesthesiology.org/http://www.anesthesiology.org/http://www.anesthesiology.org/mailto:[email protected]://www.anesthesiology.org/


2/12

Anesthesiology 2013; 118:1264-75 1265 Lee et al.


anesthesia and insight into the cognitive neuroscience ofanesthetic-induced unconsciousness.

KEAMINE is a phencyclidine analog that belongsto a class of anesthetics (including nitrous oxide)

that do not act viathe potentiation of -aminobutyric acid(GABA).13Studies of anesthetic-induced unconsciousnesshave consistently failed to identify common mechanisms ofboth GABAergic and non-GABAergic anesthetics, which aredistinct at both the molecular and neurophysiologic levels.Identifying a common neural correlate of anesthetic-inducedunconsciousness would be an important advance for themechanistic study of consciousness and general anesthesia,and could facilitate the development of more sophisticatedbrain monitors for surgical patients.

Information feedback from the frontal cortex to othercortical regions has been referred to as feedback or recur-

rent processing and is thought to mediate conscious expe-rience.4,5 In contrast, feedforward information flow in theposterior-to-anterior direction is thought to mediate sensoryprocessing, which can occur outside of consciousness.6,7Tere is emerging evidence for disrupted feedback and pre-served feedforward processing in patients with persistent veg-etative states,8as well as anesthetized rats.9,10We previouslydemonstrated that frontal-to-parietal feedback connectivitysignificantly exceeds feedforward connectivity in conscioushumans, and that this feedback dominance is attenuated bythe GABAergic anesthetics propofol and sevoflurane.11,12Inthe current study, we used electroencephalography and nor-

malized symbolic transfer entropy (NSE) to assess direc-tional connectivity across the frontal, parietal, and temporalregions of human surgical patients. On the basis of thisanalysis we demonstrate that drugs from three major classesof anesthetics selectively disrupt feedback communication.Tis finding supports the hypothesis that feedback connec-tivity and frontoparietal networks play an important role inconsciousness and provide evidence for a common neurobi-ology of anesthetic-induced unconsciousness.

Materials and Methods

In brief, 30 surgical patients were anesthetized with ketamine

(2 mg/kg intravenous infusion) while recording eight-channelelectroencephalography of frontal, parietal, and temporalregions. NSE is a model-free, nonlinear measure of the causalrelationship between two signals, and can be used to inferdirected functional connectivity.1317NSE between frontaland parietal or frontal and temporal regions was calculated forpatients in the conscious state and after induction of anesthe-sia with ketamine. We also reanalyzed electroencephalographicdata from surgical patients anesthetized with the intravenousanesthetic propofol (n = 9) or the inhaled anesthetic sevoflu-rane (n = 9),12in order to compare directly the spectral changesand NSE with that of ketamine. For the purpose of statistical

analysis, the electroencephalogram time series during baseline

consciousness and after anesthetic exposure was divided intothree equal epochs for each state, resulting in six substates(baselineB1, B2, B3; anesthesiaA1, A2, A3).

Ketamine Experiments

Participants. Te study was approved by the InstitutionalReview Board of Asan Medical Center (Seoul, South Korea),

and written informed consent was obtained in all cases; elec-troencephalographic data were gathered at Asan Medical Cen-ter and analyzed at the University of Michigan Medical School(Ann Arbor, MI). Investigators from both sites participated inthe design of the study. Patients scheduled for elective stom-ach, colorectal, thyroid, or breast surgery (n = 30, male/female= 15 /15, American Society Anesthesiologists Physical StatusI or II, aged 2264 yr) were enrolled in this study. Exclusioncriteria included previous cardiovascular disease (includinghypertension), a previous brain surgery, a history of drug oralcohol dependence, known neurological or psychiatric disor-ders, or current use of psychotropic medications.Anesthetic Procedures. Patients received no sedatives orother medications before induction of anesthesia, and weremonitored with electrocardiography, pulse oximetry, end-tidal carbon dioxide concentration, and noninvasive bloodpressure measurement. Ketamine (2 mg/kg diluted in 10 mlof 0.9% normal saline) was infused over 2 min (Baxter infu-sion pump AS40A; Baxter Healthcare Corporation, Deer-field, IL). We assessed noninvasive blood pressure every 30 sduring ketamine infusion, and administered 510 mg labet-alol if systolic blood pressure was increased 30% higher thanbaseline blood pressure. ime to loss of consciousness was

determined by checking every 10 s for the loss of response toverbal command (squeeze your right hand twice). Electro-encephalographic and electromyographic data were acquireduntil 5 min after loss of consciousness. At the end of the study,patients received an effect-site target concentration (2.5 g/ml) of propofol in combination with an effect-site target con-centration (5 ng/ml) of remifentanil.

Propofol and Sevoflurane Experiments

Participants.Tese data were originally gathered for a pre-vious study of the frontalparietal system, although NSEwas not applied for the original analysis. Eighteen patientsscheduled for elective abdominal or breast surgery (n = 18,male/female = 8/10, American Society AnesthesiologistsPhysical Status I and II, aged 2966 yr) were enrolled in thestudy. Propofol (n = 9) or sevoflurane (n = 9) was admin-istered to induce general anesthesia while eight-channelelectroencephalogram was recorded. Propofol (Diprivan;AstraZeneca, London, United Kingdom), was initiallyadministered with a target-controlled infusion of 2.0 g/ml and was increased at a rate 1.0 g/ml per 20 s until lossof consciousness. Sevoflurane (Sevorane, Abbott, IL), wasinitially administered as 2 vol% and increased at a rate of2 vol% per 20 s (up to 8%) until loss of consciousness; see

study by Ku et al.12for experimental details.


3/12


Neural Correlate of General Anesthesia

Electroencephalogram and Physiologic Data Acquisition

for the Three Anesthetics

In all experiments, the electroencephalogram was recordedat eight monopolar channels in the frontal, parietal, andtemporal regions (Fp1, Fp2, F3, F4, 3, 4, P3, and P4

referenced by A2, which followed the international 1020system for electrode placement) by a WEEG-32 (LXE3232-RF; Laxtha Inc., Daejeon, Korea) with a sampling frequencyof 256 Hz. Te electrode impedance was kept below 5 K.Band-pass filtering with the fifth-order Butterworth filterwas applied to electroencephalographic data (forward andbackward), correcting the phase shifting after band-passfiltering (butterworth.m and filtfilt.m in Matlab SignalProcessing oolbox; MathWorks, Natick, MA). Electromy-ography was concurrently recorded at four bipolar channels(bilateral frontalis and temporalis muscles) by a QEMG-4(Laxtha Inc.) with a sampling frequency of 1,024 Hz.

Analysis of Directional Connectivity

ransfer entropy (E) offers a nonlinear and model-free esti-mation of directed functional connectivity based on informa-tion theory, quantifying the degree of dependence of Y onX or vice versa.16,17Te E can be defined as the amountof mutual information between the past of X (XP) and thefuture of Y (YF) when the past of Y (YP) is already known,i.e.,

EX YF P P F P F P P I Y X Y H Y Y H Y X Y = ( )= ( ); ( , )

(1)

where H(YF/YP) is the entropy of the process YFconditionalon its past.

Te distributions of XP, YP, and YF can be writtenexplicitly as

E log 2X YF P P

F P P P

P P F P P Y Y X

P Y Y X P Y

P Y X P Y Y ( )

( )( )

= ,, , ( )

, ( , )

(2)

I Y X Y I Y Y F P P F P X Y; , ; = ( )+ E (3)

Equation (3) shows that E represents the amount of infor-mation provided by the additional knowledge of the past ofXwith respect to the future of Y.

One disadvantage of E is the subjective decision forthe bin size in the probability calculation in Equation (2).o avoid this problem, symbolic E (SE) can be used. InSE, each vector for YF, XP, and YP in Equation (2) is asymbolized vector point. For instance, a vector Y

t consists

of the ranks of its components Y y y y t m=[ ]1 2, , , , wherey yj t m j= - ( - )1 is replaced with the rank in ascending order,y mj 1 2, , ,[ ] for j m= 1 2, , , . Here m is the embed-ding dimension and is the time delay. SE is defined in

the same way as Equation (2), but replacing the embedded

vector points with the symbolized vector points. In com-parison with E, SE is a more robust and computationallyefficient method.16,18A schematic illustrating the principlesof E is presented in figure 1.

Bias and Normalization of STEo remove the bias of SE for a given electroencepha-lographic data set, the shuffled-data method was used.19Te shuffled data retain the same signal characteris-tics as the original signal does, but the causal relationis completely eliminated. Tis shuffling process wasapplied only to the source signal (X), leaving the tar-get signal (Y) intact. Te SE with the shuffled sourcesignal ( ), ( ) ( , )X H Y Y H Y X YP X Y

F P F P P Shuff

ShuffledShuffSE = ,

estimates the bias caused by the signal characteristics ofthe source signal (X). Te unbiased SE was normalized asfollowing,

NSESE SEShuffled

X YX Y X Y

F PH Y Y

=

( | )[ , ]0 1 (4)

NSE is normalized SE (dimensionless), in which the biasof SE is subtracted from the original SE and then dividedby the entropy within the target signal, H (YF/Y P). Intui-tively, NSE represents the fraction of information in thetarget signal Y, which is not explained by its own past andexplained by the past of the source signalX.19

Finally, the asymmetry between NSEX Y andNSEY X was defined as following,

DFX YX Y Y X

X Y Y X

=+

NSE NSENSE

NSE[ , ]1 1 (5)

Terefore, if DFX Y has a positive value, the connectivityfrom X to Y is dominant, and vice versa for a negative value.Te feedback and feedforward connections in the frontopa-rietal network were evaluated with NSEf p and NSEp f over the 48 subjects (30 ketamine, 9 propofol, 9 sevoflu-rane). Te average NSEf p and NSEp f were calcu-lated over the eight pairs of electroencephalogram channelsbetween the frontal and parietal regions for each subject;

NSE NSEf p

f pi ji j

n n

n n

f p

( )=

=

1

1i

,

,

, wheren

f = 4 andn

p= 2. Te asymmetry of information flow between the two

brain regions was defined as DFf p (Equation 5) for eachsubject.

Appropriate Embedding Parameters in Baseline

and Anesthetized States

During ketamine anesthesia, we found that NSEf p andNSEp f have multiscale properties, showing distinctinformation transfer between frontal-parietal regions onshort- and long-term scales. Tis may be associated with

simultaneous increases of gamma and delta powers (relatively


4/12



short- and long-term dynamics). Terefore, informationtransmission of a single time scale would not be able torepresent the multiscale connectivity of ketamine anesthesia.In this study, we found that the maximum information transferbetween frontal and parietal regions provides a consistentconnectivity feature of ketamine, propofol, and sevoflurane.Tree embedding parameters, embedding dimension (d

E),

time delay (), and prediction time (), are needed forNSE. Te parameter set that provides the maximuminformation transfer (NSE) from the source signal to thetarget signal was selected as the primary connectivity for agiven electroencephalographic data set, instead of applyinga conventional embedding method. By investigating theNSE in the broad parameter space of d

E(from 2 to 10) and

(from 1 to 30), we fixed the embedding dimension (dE)

at three, which is the smallest dimension providing a similarNSE, and found the time delay () producing maximumNSE. In this parameter space, a vector point could coverfrom 11.7 ms (with = 1 and d

E= 3) to 351 ms maximally

(with = 30 and dE= 3). If a parameter set for maximum

information transfer was determined in one direction, weused the same parameters for the opposite direction. akingthe maximum NSE as the primary connectivity for a givenelectroencephalographic data set, it can be concluded thatall other processes are nonparametric without subjectivedecisions for embedding parameters. Te prediction time wasdetermined with the time lag (1100, 3.9390 ms) resultingin maximum cross-correlation, assuming the time lag as the

interaction delay between the source and target signals.

Statistical Analysis of Six Substates

For the purpose of statistical evaluation, we segmented10-min long electroencephalographic data into sixsubstates: three substates for baseline (B1, B2, and B3)and three substates for the period of anesthetic exposure(A1, A2, and A3), with each substate being a 100-slong electroencephalographic epoch. Te feedback orfeedforward connectivity of each substate was defined asthe mean value of connectivity over 10 small windows(10-s long). For each small window, the average NSE,

NSEf p , and NSEp f , between four frontal and twoparietal electroencephalogram channels was calculated.Te small window size of 10 s may satisfy pseudostationaryconditions for NSE calculation, and the mean value over10 small windows may reflect the connectivity of a substate.Terefore, the data format of the ketamine group is 30

subjects and 6 substates for each connection. Te feedbackand feedforward connectivity, NSEf p and NSEp f ,were compared across the six substates; the significancewas assessed by a repeated measure ANOVA and apost hocanalysis using ukey multicomparison test. Te mean SDfor each connection is presented in Supplemental DigitalContent 1 (http://links.lww.com/ALN/A923 ) and theresults of thepost hoctest are presented in the Results. TeDAgostino-Pearson omnibus normality test and Freedmannonparametric test were conducted before the ANOVAtest. A Pvalue less than 0.01 was considered significant.GraphPad Prism Version 5.01 (GraphPad Software Inc.,

San Diego, CA) was used for the tests.

Fig. 1.Schematic illustration of transfer entropy. Symbolic transfer entropy measures the causal influence of source signalXon target signal Y, and is based on information theory. The information transfer from signalXto Yis measured by the differ-

ence of two mutual information values, I [YF; XP, YP]and I [YF; YP], whereXP, YP, and YFare, respectively, the past of source

and target signals and the future of the target signal. The difference corresponds to information transferred from the past

of source signalXPto the future of the target signal YFand not from the past of the target signal itself. The average overall

vector points measures the information transferred from the source signal to the target signal. The vector points are symbol-

ized with the rank of their components: e.g.,a vector point (30,78,51) is symbolized to (1,3,2) with the rank of components

in ascending order.
http://links.lww.com/ALN/A923http://links.lww.com/ALN/A923


5/12



Statistical Analysis for False-positive Connectivity:

Permutation Test and Time-shift Test

Despite efforts to remove bias from NSE measurements,there is still a potential for false-positive causality. Wetherefore applied the permutation and time-shift tests to

evaluate, respectively, the significance and the linear mix-ing effect on the connectivity. At each substate (B1, B2,B3, A1, A2, and A3), 10 trial data sets were generated bycontinuously dividing 100-s long electroencephalographicepochs into 10-s long epochs. Te permutation and time-shift tests were then applied to the 10 trial data sets, derivedfrom eight-channel electroencephalogram. Te test statistics(NSE NSEf p p f and ) were calculated for eight pairsof electroencephalographic channels.

Te permutation test is a nonparametric statistical signifi-cance measure that is used to assess whether the test statisticsof two groups are interchangeable.13,14Te null hypothesis is

that the means of the test statistics (NSE) of original andrandomized electroencephalographic data are interchange-able. Te null hypothesis was considered to be rejected withPvalue less than 0.01. Te significance test for connectivitywas applied to each pair of electroencephalographic channelswith 10 trials and the corresponding 10 shuffled data sets.

With the same data format, the time-shift test was applied toevaluate possible false-positive connectivity due to an instanta-neous linear mixing effect.13,14In the test, because the future ofthe source signal has no causal relationship with the present of

the target signal, if NSTE(t) < NSTE(t), t = t + , where tisthe future time and is an interaction time delay between twosignals, then the two signals have an instantaneous linear mixingwith time delay . Tis also holds for the shifted signal with thepast of source signal (t= t ). As with the permutation test,

the time-shift test was applied to the original and time-shifted(i.e.,shuffled) electroencephalographic data. Te significance ofthe mean difference between the two groups was evaluated withthe permutation test (nonparametric method, P< 0.01).

Results

Distinct Spectral Changes Induced by Ketamine, Propofol,

and Sevoflurane

Te relative power of the electroencephalogram spectrumis the percentage of power contained in a frequency bandrelative to the total spectrum (0.135 Hz). We removedfrequency bands above 35 Hz to avoid muscle artifact. Te

relative powers of , , and bands increased after injec-tion of ketamine (across 3 substates of baseline consciousnessand 3 substates of ketamine anesthesia: F(5, 179) = 3.54,P= 0.0047 for ; F(5, 179) = 11.28, P< 0.0001 for ; andF(5, 179) = 68.86; P< 0.0001 for ), whereas the relativepowers of and bands decreased (across 6 substates: F(5,179) = 52.25, P< 0.0001 for ; F(5, 179) = 14.38, P< 0.0001for ). Te simultaneous increase of the relative powers forboth slow waves (and ) and fast waves () was unique tothe ketamine power spectrum (fig. 2, first column). Propofol

0 2 4 6 8 10

15

20

25

30

35

Propofol

0 2 4 6 8 10

10

15

20

0 2 4 6 8 10

10

15

20

0 2 4 6 8 10

20

30

40

0 2 4 6 8 10

10

15

20

Time(minutes)

0 2 4 6 8 10

162024283236

Sevoflurane

0 2 4 6 8 10

12

16

20

24

28

0 2 4 6 8 10

8

12

16

20

0 2 4 6 8 10

20

30

40

0 2 4 6 8 10

8

12

16

20

Time (mintues)

0 2 4 6 8 10

20

25

30

B1 B2 B3 A1 A2 A3

Delta

Ketamine

0 2 4 6 8 10

22

24

26

28

30

Theta

0 2 4 6 8 10

10

15

20

Alpha

0 2 4 6 8 10

26

28

30

32

34

Beta

0 2 4 6 8 10

12

14

16

18

20

Time (minutes)

A

B

C

D

E

Gamma

B1 B2 B3 A1 A2 A3 B1 B2 B3 A1 A2 A3

Fig. 2.The relative power spectrum for ketamine, propofol, and sevoflurane. The relative powers for each frequency band, (A) (0.14 Hz), (B) (48 Hz), (C) (813 Hz), (D) (1325 Hz), and (E) (2535 Hz), are presented. Error bardenotes the standarderror for each 10-s electroencephalogram epoch over 30 subjects. Theblueshade indicates induction of anesthesia (from 5 to

7 min). The time spans for propofol and sevoflurane anesthesia were rescaled to match with the time of ketamine. Redcolor

indicates the increase of relative power after anesthetic-induced unconsciousness, whereas bluecolor indicates the decrease

of relative power. The relative power of and demonstrated different responses among the three anesthetics. Six substates(B1, B2, B3 in baseline state and A1, A2, and A3 in anesthesia) are denoted.


6/12



and sevoflurane (fig. 2, second and third columns) induceda spectral pattern distinct from that of ketamine anesthe-sia: there was an increase of , , and powers (F(5, 53)> 3.3, P< 0.05 for propofol; F(5, 53) > 4.4, P< 0.001 forsevoflurane) and decrease of and powers (F(5, 53) > 3.5,

P< 0.01 for propofol; F(5, 53) > 11, P< 0.001 for sevoflu-rane). Te difference among the three anesthetics was mostsalient in (reduced by ketamine, increased by propofoland sevoflurane) and (increased by ketamine, reduced bypropofol and sevoflurane). Tus, the effects of GABAer-gic and non-GABAergic anesthetics were associated withdistinct electroencephalographic spectra. In terms of theregional difference of relative powers for the five frequencybands, the F3, F4 and P3, P4 channels showed a signifi-cant power difference. See table 1 in Supplemental DigitalContent 1 (http://links.lww.com/ALN/A923), which showsregional differences in relative power after administration ofketamine.

o examine whether the increased power after admin-istration of ketamine was due to electromyographic artifact,we investigated Pearson correlation coefficients betweenelectromyography and electroencephalography in the fre-quency bands (2535 Hz). Te waves for the electromyo-gram and electroencephalogram are presented in figure 3,A and B; a scatter plot of both signals during anesthesia ispresented in figure 3C. Te mean and SD of Pearson corre-lation coefficient was 0.0106 0.05, which was calculatedonly with the pairs of channels with Pvalue less than 0.01.No correlation was found between the two signals.

Effects of Ketamine on Directional ConnectivityDistinctive patterns of connectivity were demonstratedacross different brain regions (fig. 4, AF). Figure 4A demon-strates feedback dominance between frontalparietal regionsduring the baseline conscious state, which is consistent with

our previous findings.11,12 During administration of ket-amine, feedback connectivity was gradually reduced and sig-nificantly inhibited after loss of consciousness (F(5, 179) =7.785, P< 0.0001 for feedback connectivity; P< 0.001 forA2 and B1, B2, B3). Mean time to loss of consciousness was

89.4 18.4 s. All but two patients were unresponsive at theend of the 2-min infusion and all were within 30 s thereafter.Te reduced feedback connectivity began a notable but sta-tistically insignificant rebound (P> 0.01, not significant forB2, B3, and A3), which may reflect the waning effect of thesingle drug bolus. By contrast, the feedforward connectivitywas preserved for the entire experimental period (F(5, 179) =2.07, P= 0.072; P> 0.01, not significant for any substates).

Changes in the two directions of connectivity gave riseto a significant change in the asymmetry of feedback andfeedforward connectivity after anesthesia (F(5, 179) = 23.88,P < 0.0001). Te asymmetry of information flow in

the frontalparietal network was largely reduced duringketamine injection, which induced a balanced informationflow during the first 1 min after loss of consciousness(P < 0.001 between B1, B2, B3 and A1, A2, A3). Tisbalanced bidirectional connectivity began to break downfrom the substate A3 (P < 0.01, for A2 and A3). Oninvestigating the frontal and temporal regions (fig. 4B), wealso found feedback dominance in the baseline consciousstate and selective inhibition of feedback connectivity afterketamine injection (F(5, 179) = 6.59 and P< 0.0001 forfeedback, but F(5, 179) = 2.42, P= 0.038 for feedforward;A2 is most significant). Figure 4E demonstrates that

ketamine was associated with a reduction in the asymmetry,but the feedback connectivity still maintained a relativelygreater information flow. Tus, the inhibition of feedbackconnectivity in the frontaltemporal network was not asrobust as that of the frontalparietal network. Te tendency

4

2

0

2

4

Gamma waves of EEG

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 24

3

2

1

0

1

2x 1015

Gamma waves of EMG

Time (second)

A

B

C

505

4

3

2

1

0

1

2

3

4

normalized EMG

normalize

dEEG

Fig. 3.The relationship of waves between electroencephalography and electromyography. (A)Wave of 2-s long electroen-cephalogram epoch. (B)Wave of 2-s long electromyogram epoch. (C)Scatter plot of two waves of 5-min long electroenceph-alogram and electromyogram during ketamine anesthesia. No correlations were identified using Pearson correlation coefficient.

EEG = electroencephalogram; EMG = electromyogram.
http://links.lww.com/ALN/A923http://links.lww.com/ALN/A923


7/12



of rebound of feedback and feedforward connectivity in thefrontaltemporal network in substate A3 was still observed.Ketamine had no effect on horizontal connectivity

across the hemispheres (F(5, 179) = 1.76, P = 0.124 forleft to right; F(5, 179) = 2.25, P= 0.052 for right to left),despite the disruption of the rostralcaudal connectivity inthe brain. Figure 4, C and F shows that the informationflow between left and right hemispheres was balanced andno asymmetry was identified, even in the baseline wakingstate. Te connectivity among eight electroencephalographicchannels in six substates is presented in figures 1 and 2 ofSupplemental Digital Content 1 (http://links.lww.com/ALN/A923). Te variance of individual connectivity duringketamine anesthesia is presented in figure 3 of SupplementalDigital Content 1 (http://links.lww.com/ALN/A923 ).

Multiscale Feedback and Feedforward Connectivity

Te simultaneous increase of slow and fast electroencephalo-gram waves (and bands) were associated with multiscaleinformation transmission structures during ketamine anes-thesia (fig. 5, AF). In particular, the rebound of feedback/feedforward connectivity was identified in short time scales(= 1, 2, and 3 with d

E=3; 11, 23, and 35 ms in fig. 5, A

and D) during ketamine anesthesia, but did not appear atlonger time scales ( = 10, 15, and 20 with d

E

= 3; 117,175, and 234 ms in fig. 5, B and E). However, despite the

reactivation of connectivity, the reduced asymmetry betweenthe feedback and feedforward connectivity was maintained(F(5, 179) > 7, P< 0.001; B1, B2 and B3 > A1, A2, andA3 for the asymmetries of all in fig. 5, C and F). Distinctfrom the baseline conscious state, the rebound feedback con-nectivity was accompanied by increasing feedforward con-nectivity, which maintained the reduced asymmetry duringketamine anesthesia. Tus, although activity of feedback/feedforward connections was variable across time scales, thereduced asymmetry was consistently associated with the lossof consciousness. Furthermore, the feedback dominance inthe baseline state and the selective inhibition of feedbackconnectivity during anesthesia were found irrespective oftime scales.

Robustness of NSTE Measures during Ketamine

Anesthesia

Spurious interpretation of correlation or causality canoccur when measuring connectivity. As such, the time-shift and permutation tests13,14 were applied to all pairsof electroencephalographic channels between frontal andparietal regions in order to test the significance and assess thefalse-positive connections generated by NSE analysis. oapply these tests, 10 trials (10-s long electroencephalogramepochs) and 10 corresponding shuffled data sets weregenerated from each substate for each subject receiving

0 1 2 3 4 5 6 7 8 9 10

3

4

5

6

7

8

9

10

x 103 3

FB&

FF

0 1 2 3 4 5 6 7 8 9 10

3

4

5

6

7

8

9

10

3

4

5

6

7

8

9

10

x 103

x 10

0 1 2 3 4 5 6 7 8 9 10

feedback

feedforward

feedback

feedforward

feedback

feedforward

0 1 2 3 4 5 6 7 8 9 10

0.1

0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Time(minute)

0 1 2 3 4 5 6 7 8 9 10

0.1

0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Time(minute)

Frontal & Parietal

KetKet

Ket

Ket

Ket

Frontal & Temporal Left & RightCBA

FED

Ket

0 1 2 3 4 5 6 7 8 9 10

0.1

0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Time(minute)

As

ymmetry

Fig. 4. The feedback (FB) and feedforward (FF) connectivity measured by normalized symbolic transfer entropy between

(A) frontal and parietal regions, (B) frontal and temporal regions, and (C) left and right hemispheres. The FB/FF asymmetry is

shown between (D) frontal and parietal regions, (E) frontal and temporal regions, and (F) left and right hemispheres. The positive

value of asymmetry indicates that the FB connectivity is dominant over the FF connectivity. The mean and standard error over 30

subjects are denoted. The Ket inblueshade indicates the ketamine injection for 2 min. The mean time to loss of consciousness

occurred at approximately 6.5 min on the x axis.
http://links.lww.com/ALN/A923http://links.lww.com/ALN/A923http://links.lww.com/ALN/A923http://links.lww.com/ALN/A923http://links.lww.com/ALN/A923http://links.lww.com/ALN/A923


8/12



ketamine. Te rejection rates of the null hypothesis forthe two statistical tests are summarized in table 1. If thenull hypotheses of both tests are rejected, the strength ofconnectivity is deemed robust and reliably deviates fromrandom connectivity generated from potential linear mixing.Te percentage of connections passed for both tests wasevaluated over all possible pairs of the subjects (2,880 pairs:2 directions 6 substates 8 channel pairs for frontal andparietal regions 30 subjects). Te feedback connections inthe baseline state had a higher null hypothesis rejection ratefor both tests (30% among all 720 feedback connections forB1, B2, and B3), indicating that the feedback connections

in the baseline state were relatively less susceptible to thepotential risk of linear mixing. Te feedforward connections

in the baseline state had a rejection rate of 16%; the feedbackand feedforward connections during ketamine anesthesiahad rejection rates, respectively, of 23 and 17% on average.Comparing the directional connections, the feedforwardconnections had relatively lower null hypothesis rejectionrates for both tests, which may be due to relatively smallerNSE values in the feedforward direction. Regarding thedependence on state, the baseline conscious state had ahigher rejection rate than the anesthetized state.

Analysis of only the connections that survived both thepermutation and time-shift tests for each substate demon-strated the same patterns of the feedback/feedforward con-

nectivity. Te selectively inhibited feedback connectionand preserved feedforward connection during ketamine

00 1 2 3 4 5 6 7 8 9 10

2.5

3

3.5

4

4.5

5

5.5

6

6.5

7

FB&F

F

Feedback

Feedforward

Feedback

Feedforward

Feedback

Feedforward

1 2 3 4 5 6 7 8 9 10

1

2

3

4

5

6

7

0 1 2 3 4 5 6 7 8 9 101

2

3

4

5

6

7

8

Asymm

etry

00 1 2 3 4 5 6 7 8 9 10

Time(minute)Time(minute) Time(minute)

Asy

mmetry

1 2 3 4 5 6 7 8 9 10

0.4

0.2

0

0.2

0.4

0.6

0.8

0 1 2 3 4 5 6 7 8 9 100.2

0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.1

0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

x 103

x 103

x 103

3

Ket

Ket

KetKet

KetKet

Short term scale Long term scale All scalesCBA

FED

Fig. 5.Distinct connectivity response of the frontalparietal network at different scales. The mean feedback/feedforward (FB/

FF) connectivity over 30 subjects is demonstrated for (A) a short time scale (= 1), (B) a longer time scale (= 20), and (C) vari-ous time scales (= 1, 3, 5, 10, 15, 20). The shorter time scales (= 1, 3, and 5) are denoted with filled squaresand the longertime scales (= 10, 15, and 20) are denoted with empty circles. The corresponding asymmetries of FB and FF connectivity arepresented for (D) = 1, (E) = 20, and (F) various delay times (= 1, 3, 5, 10, 15, 20). The ket denotes the period of ketamineadministration. Error barreflects standard error over 30 subjects.

Table 1. The Null Hypothesis Rejection Rates (%) for the Permutation and Time-shift Test

n = 240 B1 B2 B3 A1 A2 A3

Permutation (%) FB 75 69 67 77 84 83FF 27 21 18 19 40 40

Permutation and time shift (%) FB 31 28 28 27 22 19

FF 18 16 13 15 18 18

Note the dependence on state and direction of connectivity: the FB connections in the baseline conscious state have a higher nullhypothesis rejection rate. The number of connections for each substate is 240 (8 pairs of electroencephalogram channels 30 subjects).

A1, A2, A3 = three Anesthetized states; B1, B2, B3 = three Baseline conscious states; FB = feedback; FF = feedforward.


9/12



anesthesia, as well as the significantly reduced asymme-tries of feedback/feedforward connectivity, were confirmed(F(5,179) = 9.4, P< 0.0001, A2/A3 < A1 and baseline statesfor feedback connections; F(5, 179) = 0.54, P= 0.74 acrosssix substates for feedforward connections; F(5, 179) = 8.13,P< 0.0001 for feedback/feedforward asymmetry).

Comparison of Feedforward/Feedback Connectivity

during Ketamine-, Propofol-, and Sevoflurane-induced

UnconsciousnessFigure 6, AF shows feedback/feedforward connectivity andfeedback/feedforward asymmetry in frontalparietal net-works during baseline consciousness and unconsciousnessinduced by ketamine, propofol, and sevoflurane. Becausepropofol and sevoflurane had longer electroencephalogramepochs for the induction of anesthesia compared with ket-amine (from start of drug administration to loss of con-sciousness: 4.09 1 min for propofol; 3.8 0.73 min forsevoflurane), we resampled the feedback and feedforwardconnections measured by NSE with the same data length.For convenience of comparison, we rescaled the time axis.

Te epoch lengths for induction and anesthesia were set as

5 and 8 min, respectively, and lengths were matched usingMatlab function (resample.m). Tus, the x-axis in figure 6,B, C, E, and F does not represent the true timeline for eachpatient. Te mean and standard error of NSE for the base-line state and the mean and standard error for the resam-pled NSE for induction and anesthesia over all subjects(9 subjects for propofol and 9 subjects for sevoflurane) arepresented as continuous in figure 6, B, C, E, and F.

Te dominant feedback connectivity in the baseline state

and the selective inhibition of feedback connectivity afterinduction was clearly demonstrated across all three anesthet-ics (F(5, 179) = 9.18, P< 0.0001 for propofol and F(5, 179)= 4.11, P= 0.0042 for sevoflurane 6 six substates). In con-trast, feedforward connectivity was preserved irrespective ofstate, a consistent finding across all three anesthetics (F(5,179) = 0.179, P= 0.967 for propofol and F(5, 179) = 0.7,P= 0.625 for sevoflurane across 6 substates). As a result, thereduction of feedback dominance and feedback/feedforwardasymmetry (F(5, 179) = 4.97, P= 0.0012 for propofol andF(5, 179) = 5.3, P= 0.0008 for sevoflurane across six sub-states) in the frontalparietal network was a common neu-

ral correlate of anesthetic-induced unconsciousness across

0 1 2 3 4 5 6 7 8 9 10

3

4

5

6

7

8

9

10

x 103

x 103

x 103

FB&F

F

0 2 4 6 8 10 12 14

2

3

4

5

6

7

8

0 2 4 6 8 10 12 14

2

3

4

5

6

7

8

9

10

11

0 1 2 3 4 5 6 7 8 9 10

0.1

0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Time(minute) Time(minute) Time(minute)

Asymmetry

0 2 4 6 8 10 12 14

2

1

0

1

2

3

4

5

0 2 4 6 8 10 12 142

1

0

1

2

3

4

5

6

7

SevofluranePropofolKetamineA B C

D E F

Feedback

Feedforward

Feedback

Feedforward

Feedback

Feedforward

B1 B2 B3 A1 A2 A3 B1 B2 B3 A1 A2 A3 B1 B2 B3 A1 A2 A3

Fig. 6. A common neural correlate of anesthetic-induced unconsciousness. The inhibition of asymmetry between the feed-

back (FB) and feedforward (FF) connectivity is a common feature found across three heterogeneous anesthetics. The FB (red)/

FF (blue) connections (AC) and their asymmetry (DF) in the frontalparietal network are shown for (A and D) ketamine (n = 30),

(Band E) propofol (n = 9), and (Cand F) sevoflurane (n = 9). The means and standard errors are denoted at each window. Anesthetic

induction is highlighted withblue shading. Six substates (B1, B2, B3 in baseline state and A1, A2, and A3 for anesthesia) are denoted.

Note that the timeline of propofol and sevoflurane induction has been rescaled for the purpose of comparison with ketamine induction.

Higher variation in propofol and sevoflurane groups may relate to fewer patients (n = 9 for each anesthetic) compared to the ketaminegroup (n = 30).


10/12



ketamine, propofol, and sevoflurane, which represent threemolecularly distinct classes of anesthetics. Te mean SD offeedback and feedforward connections for six substates arepresented for ketamine, propofol, and sevoflurane in table 2of Supplemental Digital Content 1 (http://links.lww.com/ALN/A923).

Discussion

Tis is the first study to assess changes in directional con-nectivity during ketamine anesthesia in humans and alsothe first to provide evidence for a common correlate of both

non-GABAergic (ketamine) and GABAergic (propofol,sevoflurane) anesthetics that is based in the neurobiology ofconsciousness. Despite the distinct spectral changes of ket-amine compared with propofol and sevoflurane, all anesthet-ics inhibited feedback connectivity, preserved feedforwardconnectivity, and thus inhibited the feedback dominanceobserved during the baseline conscious state. Tese findingshave implications for the measure and mechanism of generalanesthesia, as well as the role and mechanism of feedbackconnectivity in consciousness. It is important to note thatthis study assessed only connected or external conscious-ness (i.e.,consciousness of environmental stimuli), which is

thought to be mediated by lateral frontoparietal networks,rather than disconnected or internal consciousness (e.g.,dream states), which is thought to be mediated by moremedial networks. Unresponsiveness to an unambiguous ver-bal command was used as a surrogate for anesthetic-inducedloss of external consciousness.

General anesthetics can be roughly organized into threegroups.20Group 1 includes primarily GABA

Areceptor ago-

nists such as propofol, etomidate, and barbiturates, whichinduce unconsciousness but do not effectively suppress move-ment. Group 2 includes ketamine, nitrous oxide, and xenon,which are not thought to act primarily viaGABA

Areceptors,

but rather through N-methyl-D-aspartate or other receptortypes. Tese drugs have a potent analgesic effect, but are

relatively weak hypnotics and immobilizers. Finally, group 3drugs include volatile anesthetics, such as sevoflurane, desflu-rane and isoflurane, which are potent hypnotics, immobiliz-ers, and amnesics, and have diverse molecular effects such asGABA

Aand glycine receptor agonism, two-pore potassium

channel activation, and excitatory neurotransmitter recep-tor antagonism. As can be seen in table 2, ketamine (group2) does not share the same molecular, neuroanatomic, orneurophysiologic properties of propofol (group 1) or sevo-flurane (group 3). Ketamine is thought to act via glutama-tergicN-methyl-D-aspartate1receptors or HCN1 channels3;unlike virtually all other inhaled and intravenous anesthet-

ics, ketamine does not metabolically inhibit the thalamus.21

Furthermore, ketamine does not activate the sleep-associatedventrolateral preoptic nucleus as other Group 1 and Group 2anesthetics do, but rather activates numerous brainstem anddiencephalic arousal centers.22Tis arousal mechanism maycontribute to the increase in high-frequency and desynchro-nized waveforms observed on the electroencephalogram.23Due to their unique neurophysiologic profile, the effectsof ketamine and nitrous oxide are not accurately reflectedin indices generated by commercially available brain moni-tors,24which generally have algorithms that rely on a slow-ing electroencephalogram frequency. Terefore, the effects

of ketamine and other non-GABAergic drugs have eludedgeneralized frameworks of anesthetic mechanism (table 2)and are not reliably detected in the clinical setting by con-ventional monitoring technology.

Te current study suggests that measurement of direc-tional connectivity in the frontoparietal or frontotemporalregion could reflect the effects of all major groups of anes-thetics. Because the analytic measure of NSE is rooted ininformation theoryand because the frontoparietal networkis an important site of information convergenceour find-ings are consistent with the integrated information theoryof consciousness. We predict that measures of informationintegration, such as the complex phi (),25would also bereduced by diverse anesthetics. However, as a real-time

Table 2. Select Characteristics of Three Major Classes of General Anesthetics

Explanatory LevelGroup 1

(e.g.,Propofol)Group 2

(e.g.,Ketamine)Group 3

(e.g.,Sevoflurane)

Molecular Major GABA receptor agonist? Yes No Yes

Neuroanatomic target Depression of the thalamus? Yes No YesSystems neuroscience VLPO activation? Yes No Yes

Neurophysiology Increased alpha power? Yes No Yes

Information theory Inhibition of cortical feedbackconnectivity?

Yes Yes Yes

Group 1 anesthetics include primarily GABAA

agonists such as propofol, etomidate, and thiopental. These drugs tend to be stronghypnotics, but weak immobilizers and analgesics. Group 2 anesthetics include non-GABAergic drugs (such as ketamine, nitrous oxide)that may antagonize the N-methyl-D-aspartate glutamatergic receptor. These drugs tend to be strong analgesics, but weak hypnoticsand immobilizers. Groups 3 anesthetics have a mixed profile of GABA

Aagonism, two-pore potassium channel agonism, and excitatory

neurotransmitter antagonism. These drugssuch as sevoflurane, isoflurane, and desfluraneare strong hypnotics and immobilizers.Inhibition of cortical feedback connectivity is potentially a common mechanism of anesthetic-induced unconsciousness across all threegroups. VLPO contains neurons that are active during sleep.

GABA = -aminobutyric acid; VLPO = ventrolateral preoptic nucleus.
http://links.lww.com/ALN/A923http://links.lww.com/ALN/A923http://links.lww.com/ALN/A923http://links.lww.com/ALN/A923


11/12



measure of integrated information may require high-densityelectroencephalography, a perturbational approach such astranscranial magnetic stimulation, and significant compu-tational burden for accurate calculation, all of which arecurrently impractical for a routine clinical care setting. Our

past and current studies demonstrate proof of principle thatfeedback and feedforward valuesas well as the asymmetrybetween themcan be determined with relatively few chan-nels in human surgical patients, although real-time analysishas not yet been attempted.

A recent study using high-density electroencephalogra-phy also found selective inhibition of feedback connectivityin frontoparietal networks after propofol-induced uncon-sciousness and, using dynamic causal modeling, suggestedthat this finding reflects a corticocortical interaction ratherthan the cortical signature of an underlying thalamocorti-cal event.26Another recent study using positron-emissiontomography suggested a unique importance of frontopari-etal connectivity in the recovery of consciousness after anes-thesia.27Tus, it is likely that the measurement of feedbackand feedforward connectivity demonstrated in the currentstudy reflects a corticocortical interaction and the importantrole of frontalparietal communication in consciousnessand anesthesia.25,28 Tese findings support the hypothesisthat the disruption of topdown communication from thefrontal cortex is a common mediator of general anestheticeffects,5 and provide a framework for future mechanisticstudies of anesthetic-induced unconsciousness that link dis-tinct molecular and neurophysiologic actions to a commoninformation-processing endpoint. For example, the thala-

mocortical hypersynchrony and increased power of wavescaused by propofol29,30likely block the flexible corticocorti-cal communication required for consciousness. However, wedemonstrate that ketamineunlike propofoldecreases power in association with anesthetic-induced unconscious-ness (although synchrony was not measured directly). Tus,there is likely an alternative neurophysiologic pathway thatnonetheless results in the inhibition of frontalparietalcommunication.

Feedback or recurrent processing, both within sensorymodalities and across association cortices, has been pro-posed to be critical for consciousness.47Although focused

mainly on the frontal, parietal, and temporal regions usinglow-resolution scalp electroencephalography, our past stud-ies11,12and the current study demonstrate, by two distinctanalytical methods, that consciousness of the environmentis associated with a dominant cortical feedback originatingin the frontal region. Tese empirical findings are consistentwith a recent neural mass model based on diffusion ten-sor imaging of structural connectivity, in which phase lagindex measures revealed a dominant flow of informationfrom the frontal cortex to an information-processing hubof the posterior parietal cortex.31On the molecular level, ithas been suggested that feedback processing is mediated by

N-methyl-D-aspartate receptor activity whereas feedforward

connectivity is mediated by -amino-3-hydroxy-5-methyl-isoxazolepropionic acid receptors,4 which hasrecently been confirmed in a study of the macaque visualsystem.32Te current study further supports this hypothesisbecause ketaminewhich has glutamatergic N-methyl-D-as-

partate-receptor antagonist effectsselectively inhibits feed-back connectivity while preserving feedforward connectivity.Tis investigation is limited by the spatial resolution of an

eight-channel electroencephalogram and thus our findingscannot be extrapolated beyond the regions analyzed. How-ever, obtaining such findings in surgical patients with rela-tively few channels in a real-world clinical setting enhancesthe translational potential of this study. As noted, a recentstudy of propofol using high-density electroencephalogra-phy is consistent with our finding and suggests that it reflectsa direct corticocortical interaction; the stringent statisticaltests applied in the current study also indicate that the effectsof ketamine on feedback connectivity are robust. Nonethe-less, further study with high-density electroencephalographyand stepwise titration of multiple anesthetics will be impor-tant confirmation. Further confirmation of increased fron-toparietal connectivity and communication during recoveryfrom general anesthesia will also be required.12,27Te currentstudy did not address recovery because ketamine was usedonly for induction in these surgical cases.

It should be noted that two recent studies assessingdirectional connectivity with Granger causality did not finda decrease of connectivity after propofol-induced uncon-sciousness.33,34Tese results, however, are inconsistent withthe known effects of general anesthetics. General anesthesia

is associated with a metabolic deactivation of frontoparietalnetworks35and a decrease in frontoparietal functional con-nectivity28it is therefore unlikely that feedback connectiv-ity would be maintained or increased under these conditions.However, we must acknowledge that different analytictechniques and protocols may yield different findings. Forinstance, variations in electroencephalogram reference, com-mands, eye opening, and time scale of analytic techniquesmay influence connectivity results. Furthermore, like otherconnectivity measures, NSE analysis has numerous limita-tions that weaken the assertion of a truly causal interactionbetween brain regions. NSE does not work well in higher

dimensions (more than 5) because the number of possiblesymbols increases by a factorial order with dimension. Tus,NSE requires a data set long enough for proper estima-tion of conditional probability; the short stationary epochin the electroencephalogram conflicts with this requirement.Te NSE measure itself is not sensitive to the effects of alatent third source, which can generate spurious causality.Te high dimensionality, nonstationarity, and third-sourceproblems limit many current analyses of connectivity; con-siderably more work must be done to establish definitivelythat a particular direction of connectivity or communicationis critical to the mechanism of consciousness and anesthesia.

However, the consistent results across multiple anesthetics,


12/12

A th i l 2013 118 1264 75 1275 L t l


multiple analytic methods, and multiple studies encouragefurther research to test the hypothesis that impaired commu-nication in the frontalparietal network reflects a commonfinal pathway of anesthetic mechanism.

In conclusion, we demonstrate that anesthetics from

three major drug classes disrupt frontalparietal communi-cation in a characteristic and consistent way, as measuredby NSE in the feedback and feedforward directions. Tesedata provide the foundation for a common metric and com-mon neurobiology of anesthetic-induced unconsciousness.

References 1. Yamamura T, Harada K, Okamura A, Kemmotsu O: Is the site

of action of ketamine anesthesia the N-methyl-D-aspartatereceptor? Anesthesiology 1990; 72:70410

2. Jevtovi-TodoroviV, TodoroviSM, Mennerick S, Powell S,Dikranian K, Benshoff N, Zorumski CF, Olney JW: Nitrousoxide (laughing gas) is an NMDA antagonist, neuroprotec-tant and neurotoxin. Nat Med 1998; 4:4603

3. Chen X, Shu S, Bayliss DA: HCN1 channel subunits are amolecular substrate for hypnotic actions of ketamine. JNeurosci 2009; 29:6009

4. Dehaene S, Changeux JP: Experimental and theoreticalapproaches to conscious processing. Neuron 2011; 70:20027

5. Changeux JP: Conscious processing: Implications for generalanesthesia. Curr Opin Anaesthesiol 2012; 25:397404

6. Lamme VA, Supr H, Spekreijse H: Feedforward, horizon-tal, and feedback processing in the visual cortex. Curr OpinNeurobiol 1998; 8:52935

7. Lamme VA, Roelfsema PR: The distinct modes of visionoffered by feedforward and recurrent processing. TrendsNeurosci 2000; 23:5719

8. Boly M, Garrido MI, Gosseries O, Bruno MA, Boveroux P,Schnakers C, Massimini M, Litvak V, Laureys S, Friston K:

Preserved feedforward but impaired top-down processes inthe vegetative state. Science 2011; 332:85862

9. Imas OA, Ropella KM, Ward BD, Wood JD, Hudetz AG:Volatile anesthetics disrupt frontal-posterior recurrent infor-mation transfer at gamma frequencies in rat. Neurosci Lett2005; 387:14550

10. Imas OA, Ropella KM, Wood JD, Hudetz AG: Isourane dis-rupts anterio-posterior phase synchronization of ash-inducedeld potentials in the rat. Neurosci Lett 2006; 402:21621

11. Lee U, Kim S, Noh GJ, Choi BM, Hwang E, Mashour GA:The directionality and functional organization of frontopa-rietal connectivity during consciousness and anesthesia inhumans. Conscious Cogn 2009; 18:106978

12. Ku SW, Lee U, Noh GJ, Jun IG, Mashour GA: Preferentialinhibition of frontal-to-parietal feedback connectivity is aneurophysiologic correlate of general anesthesia in surgical

patients. PLoS ONE 2011; 6:e2515513. Vicente R, Wibral M, Lindner M, Pipa G: Transfer entropyA

model-free measure of effective connectivity for the neuro-sciences. J Comput Neurosci 2011; 30:4567

14. Lindner M, Vicente R, Priesemann V, Wibral M: TRENTOOL: AMatlab open source toolbox to analyse information ow in timeseries data with transfer entropy. BMC Neurosci 2011; 12:119

15. Martini M, Kranz TA, Wagner T, Lehnertz K: Inferring direc-tional interactions from transient signals with symbolic trans-fer entropy. Phys Rev E Stat Nonlin Soft Matter Phys 2011;83(1 Pt 1):011919

16. Staniek M, Lehnertz K: Symbolic transfer entropy. Phys RevLett 2008; 100:158101

17. Schreiber T: Measuring information transfer. Phys Rev Lett2000; 85:4614

18. Staniek M, Lehnertz K: Symbolic transfer entropy: Inferringdirectionality in biosignals. Biomed Tech (Berl) 2009; 54:3238

19. Gourvitch B, Eggermont JJ: Evaluating information trans-fer between auditory cortical neurons. J Neurophysiol 2007;97:253343

20. Forman SA, Chin VA: General anesthetics and molecularmechanisms of unconsciousness. Int Anesthesiol Clin 2008;

46:435321. Lngsj JW, Maksimow A, Salmi E, Kaisti K, Aalto S, Oikonen

V, Hinkka S, Aantaa R, Sipil H, Viljanen T, Parkkola R,Scheinin H: S-ketamine anesthesia increases cerebral

blood ow in excess of the metabolic needs in humans.ANESTHESIOLOGY2005; 103:25868

22. Lu J, Nelson LE, Franks N, Maze M, Chamberlin NL, SaperCB: Role of endogenous sleep-wake and analgesic systemsin anesthesia. J Comp Neurol 2008; 508:64862

23. Schwartz MS, Virden S, Scott DF: Effects of ketamine on theelectroencephalograph. Anaesthesia 1974; 29:13540

24. Hirota K, Kubota T, Ishihara H, Matsuki A: The effects ofnitrous oxide and ketamine on the bispectral index and 95%spectral edge frequency during propofol-fentanyl anaesthe-sia. Eur J Anaesthesiol 1999; 16:77983

25. Alkire MT, Hudetz AG, Tononi G: Consciousness and anes-thesia. Science 2008; 322:87680

26. Boly M, Moran R, Murphy M, Boveroux P, Bruno MA,Noirhomme Q, Ledoux D, Bonhomme V, Brichant JF, TononiG, Laureys S, Friston K: Connectivity changes underlyingspectral EEG changes during propofol-induced loss of con-sciousness. J Neurosci 2012; 32:708290

27. Lngsj JW, Alkire MT, Kaskinoro K, Hayama H, MaksimowA, Kaisti KK, Aalto S, Aantaa R, Jskelinen SK, RevonsuoA, Scheinin H: Returning from oblivion: Imaging the neuralcore of consciousness. J Neurosci 2012; 32:493543

28. Boveroux P, Vanhaudenhuyse A, Bruno MA, Noirhomme Q,Lauwick S, Luxen A, Degueldre C, Plenevaux A, SchnakersC, Phillips C, Brichant JF, Bonhomme V, Maquet P, GreiciusMD, Laureys S, Boly M: Breakdown of within- and between-

network resting state functional magnetic resonance imagingconnectivity during propofol-induced loss of consciousness.ANESTHESIOLOGY2010; 113:103853

29. Ching S, Cimenser A, Purdon PL, Brown EN, Kopell NJ:Thalamocortical model for a propofol-induced alpha-rhythmassociated with loss of consciousness. Proc Natl Acad SciUSA 2010; 107:2266570

30. Supp GG, Siegel M, Hipp JF, Engel AK: Cortical hypersyn-chrony predicts breakdown of sensory processing duringloss of consciousness. Curr Biol 2011; 21:198893

31. Stam CJ, van Straaten EC: Go with the ow: Use of a directedphase lag index (dPLI) to characterize patterns of phase rela-tions in a large-scale model of brain dynamics. Neuroimage2012; 62:141528

32. Self MW, Kooijmans RN, Supr H, Lamme VA, Roelfsema PR:

Different glutamate receptors convey feedforward and recur-rent processing in macaque V1. Proc Natl Acad Sci USA 2012;109:110316

33. Barrett AB, Murphy M, Bruno MA, Noirhomme Q, Boly M,Laureys S, Seth AK: Granger causality analysis of steady-stateelectroencephalographic signals during propofol-inducedanaesthesia. PLoS ONE 2012; 7:e29072

34. Nicolaou N, Hourris S, Alexandrou P, Georgiou J: EEG-basedautomatic classication of awake versusanesthetized statein general anesthesia using Granger causality. PLoS ONE2012; 7:e33869

35. Kaisti KK, Metshonkala L, Ters M, Oikonen V, Aalto S,Jskelinen S, Hinkka S, Scheinin H: Effects of surgicallevels of propofol and sevourane anesthesia on cerebral

blood ow in healthy subjects studied with positron emis-sion tomography. ANESTHESIOLOGY2002; 96:135870

Disruption of Frontal Parietal Communication by.14

Documents

Transcript of Disruption of Frontal Parietal Communication by.14