Cerebral metabolic response to low blood flow:possible role of cytochrome oxidase inhibition

Albert Gjedde1,4, Peter Johannsen1, Georg E Cold2 and Leif Østergaard3,4

1Pathophysiology and Experimental Tomography Center, Aarhus University Hospital in Aarhus, Aarhus,Denmark; 2Department of Anesthesiology, Division of Neuroanesthesiology, Aarhus University Hospital inAarhus, Aarhus, Denmark; 3Department of Neuroradiology, Aarhus University Hospital in Aarhus, Aarhus,Denmark; 4Center of Functionally Integrative Neuroscience, Aarhus University, Aarhus, Denmark

The reactions of cerebral metabolism to imposed changes of cerebral blood flow (CBF) are poorlyunderstood. A common explanation of the mismatched CBF and oxygen consumption (CMRO2

)during neuronal excitation holds that blood flow rises more than oxygen consumption tocompensate for an absent oxygen reserve in brain mitochondria. The claim conversely impliesthat oxygen consumption must decline when blood flow declines. As the prevailing rate of reactionof oxygen with cytochrome c oxidase is linked to the tension of oxygen, the claim fails to explainhow oxygen consumption is maintained during moderate reductions of CBF imposed byhyperventilation (hypocapnia) or cyclooxygenase (COX) inhibition. To resolve this contradiction,we extended the previously published oxygen delivery model with a term allowing for theadjustment of the affinity of cytochrome c oxidase to a prevailing oxygen tension. The extendedmodel predicted constant oxygen consumption at moderately reduced blood flow. We determinedthe change of affinity of cytochrome c oxidase in the extended model by measuring CBF in seven,and CMRO2

in five, young healthy volunteers before and during COX inhbition with indomethacin.The average CBF declined 35%, while neither regional nor average CMRO2

changed significantly. Theadjustment of cytochrome c oxidase affinity to the declining oxygen delivery could be ascribed to ahypothetical factor with several properties in common with nitric oxide.Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196. doi:10.1038/sj.jcbfm.9600113; published online30 March 2005

Keywords: CBF; CMRO2; humans; indomethacin; PET

Introduction

Recent explanations of nonlinear flow–metabolismcoupling (Buxton and Frank, 1997; Gjedde, 1997,2002; Gjedde et al, 1999; Vafaee and Gjedde, 2000)claim that the excessively increased cerebral bloodflow (CBF) serves to prevent depletion of mitochon-drial oxygen when oxygen consumption ðCMRO2

Þrises. Conversely, the explanations imply thatit would be impossible for cytochrome oxidase to

maintain normal rates of oxygen consumption whenblood flow declines, except if increased diffusioncapacity for oxygen were to accompany the decline.In contrast, a recent study from this group revealedno effect of moderately lowered blood flow onoxygen consumption in pigs (Rasmussen et al, 2003).

The current models of nonlinear blood flowregulation rest on the premise that oxygen deliveryis regulated to maintain a minimum oxygen tensionin mitochondria. The authors’ version of the model(Gjedde, 1997; Gjedde et al, 1999; Vafaee andGjedde, 2000) accurately predicted blood flowchanges observed in states of increased oxygenconsumption (Gjedde and Marrett, 2001; Vafaeeand Gjedde, 2000, 2004) and hypothermia (Sakohand Gjedde, 2003). However, the models fail topredict the constant oxygen consumption observedin states of moderately reduced CBF. The generalprediction of the model is an obligatory decrease ofoxygen consumption with any decrease of bloodflow, but the experimental evidence accumulatedover many years clearly indicates that oxygen

Received 28 August 2004; revised 10 September 2004; accepted21 October 2004; published online 30 March 2005

Correspondence: A Gjedde, PET/CFIN, Aarhus University Hospi-tal in Aarhus 44, Norrebrogade, Aarhus 8000, Denmark.E-mail: albert@pet.auh.dk

The study was supported by the Aarhus University Research

Foundation, the NOVO Nordic Foundation, the Danish Medical

Association’s Research Foundation, MRC (Denmark) Grants

9305246, 9305427, and 9601888, and the National Science

Foundation of Denmark center-of-excellence grant to the Center

of Functionally Integrative Neuroscience at the University of

Aarhus.

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196& 2005 ISCBFM All rights reserved 0271-678X/05 $30.00

www.jcbfm.com

consumption remains approximately normal inconditions of moderately lowered blood flow(Purves, 1972; Grote et al, 1981; Kanno et al, 1988;Hirano et al, 1994; Moller et al, 2002; Rasmussenet al, 2003).

For the model to predict constant oxygen con-sumption at low CBF, it must account for the tensionof oxygen at or near the oxygen binding site ofcytochrome oxidase. We linked the mitochondrialoxygen tension to the cytochrome oxidase reactionwith a simple Michaelis–Menten expression of therelationship between oxygen flux and enzymeactivity. We then tested the properties of theextended model equation by calculating the magni-tude of the equation parameters, which wouldaccurately predict the values of oxygen consump-tion measured in the course of drug-induced low-ering of CBF. We used indomethacin, which lowersblood flow, probably by nonselectively inhibitingcyclooxygenase-1 (COX-1) and -2 (Jensen et al,1991, 1993; Dahl et al, 1996; Imberti et al, 1997;Zonta et al, 2003), although other mechanisms havealso been proposed (Wang et al, 1993; Parfenovaet al, 1995).

The COX inhibition with indomethacin estab-lished the values of key variables of a mechanismformulated to explain the maintenance of constantoxidative brain metabolism in a range of CBF ratesby means of adjustment of the affinity of mitochon-drial cytochrome c oxidase towards oxygen. Thevalues were shown to be consistent with the knownranges of these variables.

Hypothesis

The experiment tested the hypothesis that constantcerebral oxygen consumption at varying blood flowrates can be maintained by adjustment of cyto-chrome c oxidase affinity to oxygen.

Theory

A common approach underlies recent predictions ofoxygen tension profiles in brain tissue (Buxton andFrank, 1997; Gjedde, 1997). As elaborated by Gjedde(1997) and Vafaee and Gjedde (2000), the keyelement in the approach is the consideration of thelow oxygen tension in mitochondria, ‘low’ meaningquantitatively unimportant as an oxygen reserve inrelation to the oxygen tension in brain capillaries.The low tension was verified quantitatively inporcine brain by calculating the average oxygendiffusion distance in brain tissue in relation to theaverage surface area of capillary (Gjedde et al, 1999).In terms of the affinity constant of cytochromeoxidase, the tension must be consistent with theactual rate of cytochrome c oxidation and hence isnot ‘low’ in relation to the properties of the enzymethat obeys Michaelis–Menten-type kinetics.

The oxygen supply to the tissue as a whole,integrated over all tissue elements, depends closely

on the average capillary oxygen tension, which inturn relates reciprocally to the oxygen extractionfraction (Vafaee and Gjedde, 2000). The averagesaturation of hemoglobin in brain capillaries is afunction of the saturation of arterial hemoglobin andthe average extraction of oxygen from the entirelength of brain capillaries, as expressed in thefollowing relationship:

�ScapO2

¼ SartO2

1 � EO2

2

� �

where �ScapO2

is the average capillary hemoglobinsaturation, Sart

O2is the average arterial hemoglobin

saturation, and EO2is the oxygen extraction

fraction, equal to the ratio JO2=ðFCart

O2Þ between the

oxygen consumption (JO2) and the product of the

CBF (F) and the arterial oxygen concentration (CartO2

)at the arterial oxygen tension (Part

O2).

The average capillary oxygen pressure head oraverage oxygen tension in capillaries (P

capO2

) isreflected in the average saturation of capillaryhemoglobin, according to the Hill equation ofoxygen binding (test):

�PcapO2

¼ Phb50

1 � EO2=2

½Phb50=Part

O2�h þ EO2

=2

!1=h

ð1Þ

where P50hb is oxygen’s tension when hemoglobin is

half-saturated, assumed to be a constant, h is Hill’scoefficient. The omission of the ratio ðPhb

50=PartO2Þh,

which is very low at normal arterial oxygen tensionsin healthy individuals (0.015), approximately com-pensates for the omission of the physically dis-solved oxygen in the calculation of the averagecapillary oxygen tension. The omission of this termis valid only for values of EO2

that are not muchbelow the physiologic range (R.B. Buxton, personalcommunication).

Assuming full saturation of arterial hemoglobinand quantitatively negligible oxygen in physicalsolution in arterial plasma, the minimum oxygentension in mitochondria can now be calculated asthe tension commensurate with the actual deliveryof oxygen (test 2):

PmitO2

¼ Phb50

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2

EO2

� 1h

s !� JO2

L

� �ð2Þ

where PmitO2

is the average oxygen tension inmitochondria, and L is the oxygen diffusibility,assumed to be constant in the absence of recruit-ment. Equations (1) and (2) were combined andsolved for the CBF (F),

F ¼ JO2

2 Chb1 þ

JO2þ L Pmit

O2

L Phb50

" #h0@

1A ð3Þ

where Chb is the arterial hemoglobin concentration.The equation expresses blood flow as a function ofthe rate of oxygen consumption and the mitochon-drial oxygen tension for a given arterial oxygen

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1184

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

concentration. In this relationship, the mitochon-drial oxygen tension reflects the balance betweenthe delivery and consumption of oxygen. Thetension is given by the simple Michaelis–Menten-type kinetic expression (Gnaiger 1993; Gnaiger et al,1998):

PmitO2

¼ Pcytox50

JO2

Jmax � JO2

� �ð4Þ

where P50cytox is the apparent half-saturation tension

of the oxygen reaction with cytochrome oxidase andJmax is the maximum reaction rate. Equation (4)eliminates Pmit

O2from equation (3) and yields an

expanded expression of the relations among theprimary factors affecting oxidative metabolism:

F ¼ JO2

2 Chb1 þ 1 þ L P

cytox50

Jmax � JO2

" #JO2

L Phb50

� !h0@

1A ð5Þ

where the independent variables are F, Pcytox50 , and

Jmax, and the dependent variable is JO2, when Chb and

L are constants and arterial blood remains fullysaturated. The independent variables may be linked,of course. Equation (5) combines the Hill equationof oxygen saturation of hemoglobin (Hill, 1910,1913) and the Michaelis–Menten equation of sub-strate occupancy of an enzyme (Michaelis andMenten, 1913).

A change of a single independent variable bynecessity is reflected in the rate of oxygen consump-tion. If no change of the oxygen consumptionaccompanies a change of an independent variable,we infer a compensatory change of at least one otherindependent variable. Thus, invariant JO2

withvariable F implies compensatory change of P50

cytox orJmax or both. Previous studies of oxygen supply anddelivery suggest that the affinity of cytochrome coxidase changes inversely with the metabolicactivity of a tissue, to preserve the sensitivity ofthe reaction to changes of the maximum velocity(see Brown (2001) for a review of these findings). Wehypothesize that a similar change can explain theinvariant cerebral oxygen consumption measuredduring moderate changes of blood flow to the brain.

Predictions

Assuming JO2to be constant in equations (3)–(5), the

adjustment of cytochrome c oxidase’s affinity tooxygen yields the following specific predictions ofthe enzyme’s response to reduced blood flow:

(1) Oxygen consumption remains constant duringmoderate oligemia when the reduction of capil-lary oxygen tension is matched by decline of themitochondrial oxygen tension according toequation (3).

(2) Decline of the Michaelis constant of cytochromec oxidation according to equation (4) can main-

tain oxygen consumption constant when themitochondrial oxygen tension is low.

(3) Adjustment of the half-saturation constant ofcytochrome oxidase in response to low bloodflow implies direct or indirect regulation ofoxygen’s occupancy of cytochrome c oxidase inaccordance with equation (5) when blood flow islow.

(4) Any compensatory mechanism is limited by themaximum achievable oxygen affinity. Below thisthreshold of minimum blood flow, oxygen con-sumption must decline as dictated by equation(3) for Pmit

O2effectively equal to 0.

Subjects and methods

The study adhered to the Helsinki Declaration V. TheResearch Ethics Committee of County Aarhus approvedthe study and the National Drug Administration approvedthe use of indomethacin in healthy volunteers. Healthyvolunteers other than possibly pregnant women wererecruited by newspaper advertisement and gave writteninformed consent to the study on the basis of a detaileddescription of radiation hazards. Some variables otherthan CMRO2

before and after the indomethacin adminis-tration were reported separately (Østergaard et al, 1998).

Indomethacin was administered intravenously as abolus of 0.2 mg (kg body weight)�1 over 3 mins andthereafter as a continuous infusion of 0.2 mg/kg h (Jensenet al, 1996). No subject experienced any adverse effectfrom the indomethacin administration.

Data Acquisition and Image Analysis

The ECAT Exact HR47 PET camera (Siemens/CTI, Knox-ville, TN, USA) was operated in 3D mode. The CBF wasmeasured by brief (in less than 4 s) intravenous bolusinjection of 500 MBq [15O]water dissolved in 5 mL saline.CMRO2

measurements were performed after a single breathinhalation of 1 L of oxygen containing 500 MBq [15O]O2 asdescribed by Ohta et al (1992).

Data acquisition began at the time of injection/inhala-tion and continued for 3 mins (6 frames of 10 secs, 4frames of 15 secs, and 2 frames of 30 secs). A separate 15-min 68Ga transmission scan for attenuation correction wasperformed before each of the two sets (before and duringindomethacin administration) of emission tomography,between which seven of the nine subjects underwentmagnetic resonance (MR) and computed tomography (CT)imaging. For two subjects, all position emission tomogra-phy (PET) measurements were performed in one session,and only structural MR imaging was performed. The twoPET measurements of CBF before and during indometha-cin were placed 50 to 307 minutes apart (average229 mins). The CMRO2

measurements were placed 31 to295 mins apart (average 201 mins).

Subjects rested supine with the head fixed in a vacuumpillow. A venous catheter was inserted in the right cubitalvein for injection of the tracer and an arterial catheter wasinserted in the left radial artery for blood sampling.

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1185

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

Twelve arterial blood samples (1 mL each) were taken thefirst minute and six and three samples during the secondand third minutes of scanning, respectively. The bloodsamples were counted for 20 secs (Cobra II, Packard,Frankfurt, Germany) to provide an arterial input functionfor the calculation of the absolute flow maps. Arterialblood variables (pH, Part

O2, Chb, and HCO3

�) were measured(ABL-300, Radiometer, Copenhagen, Denmark) before andafter each emission scan.

The PET volumes (47 sections of 3.1 mm) were recon-structed by filtered backprojection using the measuredattenuation and scatter correction and filtered to aresolution of 4.6 mm full-width at half-maximum (FWHM)(Hann filter cutoff 0.5 per pixel).

For anatomical localization and coregistration, a T1-weighted, 3D MR brain image was acquired in eachsubject, covering 124 slices of 1.5 mm using a 3D-FAST-SPGR-sequence with flow compensation, (TE¼ 4 ms;TR¼ 24 ms) on a 1.0 T GE Signa MR scanner (GE MedicalSystems, Waukesha, WI, USA).

Estimates of CBF and CMRO2were determined for every

image volume element (voxel) by fitting the PET data to atwo-compartment model using nonlinear, least-squaredregression analysis as described by Ohta (Ohta et al, 1992,1996). The parametric image volumes were then smoothedwith a Gaussian filter to an isotropic resolution of 12 mmFWHM. The smoothed images were used to spatiallynormalize the image volume by automatic coregistration(Neelin et al, 1993; Collins et al, 1994) to each subject’sMR brain image (Collins et al, 1994) and into the standardTalairach coordinate system (Talairach and Tournoux,1988). The Automated Image Registration package (AIR;Woods et al, 1992, 1993) was used for spatial alignmentbetween the PET volumes.

Calculations

Stepwise calculation of the equation parameters pro-ceeded as follows:

(1) The oxygen extraction fractions (EO2) at baseline

and during indomethacin administration were calculatedfrom the measurements of oxygen consumption and bloodflow:

EO2¼ JO2

F CartO2

ð6Þ

(2) The mean capillary oxygen tensions at the controlcondition (baseline) and during indomethacin adminis-tration were calculated from equation (1):

PcapO2

¼ Phb50

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2

EO2

� 1h

sð7Þ

(3) The oxygen diffusibility (L) was assumed not tochange at any measured blood flow level, because of theabsence of capillary recruitment in human brain (Gobelet al, 1989; Kuschinsky and Paulson 1992; Vogel et al,2004). At the highest oxygen extraction fraction compa-tible with normal oxygen consumption (EO2

E0.55, see

Discussion), PmitO2

E0. At this extraction fraction, theoxygen diffusibility (L) was calculated as the valueassociated with the measured oxygen consumption ac-cording to equation (3) for Pmit

O2¼ 0:

L ¼ JO2

PcapðindomethacinÞO2

ð8Þ

(4) At the normal blood flow baseline, the mitochondrialoxygen tension was calculated according to equation (2),using the value of L calculated above:

PmitO2

¼ PcapO2

� JO2

L

� �¼ P

capðbaselineÞO2

� PcapðindomethacinÞO2

ð9Þ

(5) Also at the normal blood flow baseline, themaximum reaction velocity (Jmax) was calculated as thevelocity consistent with the known physiologic half-saturation tension of 0.5 mm Hg (Gnaiger et al, 1998),according to equation (4):

Jmax ¼ JO21 þ P

cytox50

PmitO2

" #ð10Þ

in which the Pcytox50 =Pmit

O2ratio effectively is constant when

the value of P50cytox is adjusted to the value of Pmit

O2.

Statistical Analysis

Volume of interest analysis: For comparison of CBF,CMRO2

, EO2, P

capO2

, L, and PmitO2

in ten volumes of interest(VOI) with a preponderance of gray matter in eachhemisphere were drawn manually on the average MRbrain image of the nine subjects. In the standard referencespace, the same 20 VOIs were applied to the high-resolution PET data (4.6 mm FWHM) of CBF, CMRO2

,and EO2

, acquired before and during indomethacinadministration. The average values were calculated as anaverage of the values obtained in the 20 VOIs.

Voxel analysis: The volume images acquired before andduring indomethacin administration were also comparedstatistically voxel-by-voxel after the spatial normalizationdescribed above. To compensate for interindividual varia-tion, the low-resolution PET data (12 mm FWHM) wereused for this direct statistical voxel-by-voxel analysis.Because the purpose of the voxel analysis was to detectdeviations from the average overall decrease in CBF andCMRO2

, maps were normalized to mean values of 100 forthis analysis. An initial t-statistic analysis was applied tothe difference CBF and CMRO2

, maps (rest maps sub-tracted from the indomethacin administration maps) on avoxel-by-voxel basis initially assuming a Gaussian randomfield with zero mean and a unified standard deviation(Worsley et al, 1992). The analysis showed that the s.d.was not uniform across the brain and therefore required asubsequent analysis of the local voxel-s.d. Searching graymatter (500 mL), significance corrected for multiplecomparisons (Po0.05) was reached at t44.3 for theunified-s.d. and t413 for the voxel-s.d. for 12 mm FWHMand n¼ 7 (Worsley et al, 1992, 1996).

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1186

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

Results

Subjects

Nine subjects were PET scanned before and duringindomethacin administration. Two CBF scans wereunusable because of technical problems. The CBFdata analysis included the seven remaining subjects(three women, four men of mean age 2774 (s.d.)years, range 21 to 32 years). For two subjects the firstoxygen scan, and for another two subjects the lastoxygen scan were similarly unusable because oftechnical problems. Values of CMRO2

before andduring indomethacin administration were availablein the remaining five subjects (two women, threemen aged 2774 years, range 22 to 32 years).Therefore, measurements of both variables wereavailable in only four subjects. Comparisons weremade both for the seven subjects with CBF measure-ments and the five subjects with CMRO2

measure-ments, as well as for the four subjects in whom bothmeasurements were made, and the results of thecomparisons did not differ.

The arterial blood gases did not change during orbetween scans or during the study as a whole, asshown in Table 1 (P40.5 for all comparisons).

Blood Flow

Volume of interest analysis: The average CBF of 20VOIs declined 35% from 62 to 40 mL/hg min(Po0.001), based on the VOI analysis (Table 2).Within subjects, the average reduction varied from15% to 47%. In gray matter including the deepnuclei, the average CBF declined 35% from 64.8 to42.1 mL/hg min. In white matter, CBF declined 36%from 47.5 to 30.6 mL/hg min. The average mean CBFvalues are listed in Table 2. Across subjects, the CBFreduction appeared uniform; no VOI sustained aCBF increase in any subject. The standard deviationof the percent CBF reductions among subjects in theVOIs of 20 different regions varied from 9.4 to 18.2.Across whole brain, the variation of the CBFreduction was randomly distributed, indicating notargeted indomethacin effect in any brain region. Anextreme range of reductions were observed in theleft thalamus, ranging from 4% to 54%. In no regiondid the CBF approach the conventionally accepted

ischemic threshold of 20 mL/hg min (Purves, 1972;Furlan et al, 1996; Marchal et al, 1996).

Voxel analysis: Maps of CBF before and duringindomethacin administration are shown in Figure 1.Based on the pooled s.d. of all intracranial voxels,after global normalization, the statistical analysis ofregional CBF change in each voxel revealed the sitesof relatively increased CBF in the primary visualcortices as shown in Figure 2 (lingual gyri, BA 18),as well as the left hippocampus, both thalami, andthe mid-portion of the cingulate gyri (Table 3). Thechanges in the latter regions are not seen in thesection shown in Figure 2. Based on individual-voxel s.d., after correction for multiple comparisons,the analysis revealed significant relative CBF in-creases in the mid-portion of the right cingulategyrus, right thalamus, and right lingual gyrus(Table 3). After normalization, sites of significantlydecreased CBF were seen only bilaterally in extra-cerebral tissue, anteriorly and laterally to thetemporal lobes.

Oxygen Consumption

Volume of interest analysis: The average CMRO2,

determined as the weighted average of the regionalCMRO2

in VOI (Table 2), was unaltered duringindomethacin administration with an insignifi-cant (P¼ 0.4) average decrease from (mean7s.d.)147732 to 135735 mmol/hg min among the

Table 1 Arterial pH, blood gases, and HCO3� before and during indomethacin administration (mean7s.d.)

Before indomethacin During indomethacin

Variable Before CBF After CMRO2 Before CBF After CMRO2

pH 7.4170.03 7.4170.02 7.4170.02 7.4170.01Part

CO2(mm Hg) 4175 4275 4072 4174

PartO2

(mm Hg) 109715 128748 113712 112720

HCO3� (mmol/L) 2573 2673 2572 2573

Values from blood samples after the cerebral blood flow (CBF) and before the CMRO2measurements are not shown.

Table 2 Average means of arterial hemoglobin concentration(mmol/L), cerebral blood flow (CBF) (mL/hg min), and CMRO2

(mmol/hg min), based on values from 20 predominantly graymatter VOIs before and during indomethacin administration(mean7s.d.)

Hgb CBF CMRO2

Period (n¼5) (n¼7) (n¼5)

Baseline 62710 147732Indomethacin 4074 135735Average 8.870.9 141733Change (%) �34 �8Significance* (P) 0.0006 0.19

*One-tailed paired t-test.

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1187

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

subjects. The average CMRO2change within subjects

varied between �35% and þ 15%. There were nodifferences between gray and white matter. In the20 different regions, the standard deviations of thepercent CMRO2

reductions among subjects variedfrom 7.6 to 35.2. The average subject values of CBF,CMRO2

, EO2, P

capO2

, as well as the hemoglobinconcentration, based on the VOI analysis, are listedin Table 2. The average EO2

increased 45% from 0.38to 0.55 (Table 4). The increase in EO2

was significant(Po0.04; one-tailed paired t-test).

Voxel analysis: The average CMRO2maps before and

during indomethacin administration are shown inFigure 1. The voxel-based statistical analysis re-vealed no significant changes of CMRO2

, irrespectiveof the use of absolute or normalized values, despitethe apparent decline suggested by the color codingof the figure.

Flow-metabolism couple: The model-derived para-meters were calculated according to equations (6)–(10) for the four subjects in whom both blood flowand oxygen consumption measurements were com-pleted. The calculations yielded values for theaverage capillary oxygen tension P

capO2

, the oxygen

Figure 1 Transaxial sections of average cerebral blood flow(CBF) (top) and CMRO2

(bottom) maps, before (left) and after(right) indomethacin administration. Scaling of maps is thesame before and during indomethacin. Sections are spatiallysynchronized at Talairach coordinates (x, y, z) �5, �35,10 mm.

Figure 2 Section (z¼�7 mm) of 3D matrix of change ofcerebral blood flow (CBF) associated with indomethacinadministration, based on pooled s.d. of all intracranial voxelsafter global normalization of CBF value of each voxel. Statisticalanalysis of CBF change in each voxel revealed significantlyincreased CBF shown for primary visual cortex (left and rightlingual gyri, BA 18), as well as hippocampus, both thalami, andmid-portion of cingulate gyrus (Table 3).

Table 3 Cerebral sites with locally increased normalized cerebral blood flow (CBF) with the t-statistic calculated with both the globalunified s.d. and the local voxel s.d.

Global s.d. analysis Local voxel-s.d. analysis

Location x,y,z (mm) BA t x,y,z (mm) BA t

Left lingual gyrus �7, �81, �3 18 5.8Right lingual gyrus 7, �80, �8 18 5.5 16, �93, �9 17 8.8Left hippocampus �28, �18, �14 — 4.9Right thalamus 11, �18, 8 — 4.6 13, �23,0 — 11.8Left thalamus �16, �25, 12 — 9.3Cingulate gyrus 0, �19, 36 24 4.6 3, �31,36 31 13.0

Talairach coordinates of the peak t-value: x: right (+)/left (–) to midline; y: anterior (+)/posterior (–) to anterior commisure; z: cranial (+)/caudal (–) to theAC–PC plane. BA Brodmann area according to the Talairach atlas; t-statistic from the t-statistics.

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1188

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

diffusibility L, the mitochondrial oxygen tensionPmit

O2at the normal baseline, and the maximum

reaction rate of cytochrome oxidase Jmax. The resultsare listed in Table 4. The oxygen extraction fractionof 0.55 was deemed to approximate the generallyreported threshold of 0.56 to 0.59 (see Discussion).With a Michaelis constant of 0.5 mmHg, the Pmit

O2/

P50cytox ratio at baseline was 17, signifying cytochrome

oxidation very close to saturation.Equation (5) predicts the blood flow rates that are

sufficient to support a constant rate of oxygenconsumption, given a constant ratio between P50

cytox

and PmitO2

when the values of the experimentallyestimated parameters Jmax and L listed in Table 4 areinserted into the equation. The predicted relation-ship between blood flow and the variable P50

cytox of

Eq. (5) is shown in Figure 3 for the constant rate ofoxygen consumption of 141 mmol/hg min measuredin the 10 subjects. The lower limit of P50

cytox was fixedat 0.05 mm Hg, as determined by Chance (1976) forcytochrome oxidase in vitro.

In the predominantly gray matter volumes, bloodflow rates below 45 mL/hg min failed to sustain thenormal rate of oxygen consumption and predicted adeclining ratio of Pmit

O2to P50

cytox, reaching 2 at 20 mL/hg min and 1 at 10 mL/hg min, at which the oxygenextraction fraction reached unity. Given the para-meter values listed in Table 4, Figure 4 shows theregression of Eq. (5) to the measured average rate ofoxygen consumption of 141 mmol/hg min, measuredin the 10 subjects for blood flow rates ranging from10 to 95 mL/hg min.

Table 4 Measured or calculated oxygen delivery and metabolism variables

Mean7s.e.m. (n¼ 4) (One-tailed paired t) P

Variable Unit Baseline Indomethacin

CBF mL/hg min 6775 4071 0.01CMRO2 mmol/hg min 150719 138720 0.27EO2 ratio 0.3870.05 0.5570.07 0.044

PcapO2

mm Hg 4473 3773 0.033L mmol/hg min mm Hg 3.270.5Pmit

O2mm Hg 8.573.0

Jmax mmol/hg min 159720

Figure 3 Relationship between P50cytox and CBF (mL/hg min)

calculated from equation (5) and parameter estimates listedin Table 4 for a constant oxygen consumption rate of 141mmol/hg min. Magnitude of P50

cytox was assumed not to declinebelow 0.05 mm Hg. Abscissa: CBF (mL/hg min). Ordinate:P50

cytox (mm Hg).

Figure 4 Extended nonlinear flow-metabolism couple accordingto equation (5) calculated from equation parameters listed inTable 4 and determined from experimentally observed means ofCBF (mL/hg min) and CMRO2

(mmol/hg min) before and duringindomethacin administration, using relation between CBF andP50

cytox shown in Figure 3. Abscissa: CBF (mL/hg min). Ordinate:CMRO2

(mmol/hg min).

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1189

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

Figure 5 shows the effect of changing the value ofJmax listed in Table 4. Changes of Jmax had no effect onoxygen consumption at blood flow rates below theblood flow threshold of 45 mL/hg min. Above thisthreshold, the effect depended on the magnitudes ofboth blood flow and Jmax, that is, the higher theblood flow, the greater the effect of a change of Jmax.The graph made the important prediction that Jmax

values below normal may reverse the relationshipbetween blood flow and oxygen consumption,paradoxically causing oxygen consumption to riseat low blood flow rates.

Discussion

The normal baseline values of CBF were higher thanconventionally reported for whole brain. We ascribethe higher values to the voxel mapping and thepredominance of gray matter in the selected corticalregions. In voxel mapping, average regional CBF andCMRO2

values are computed from values of indivi-dual voxels, for which the exponential regressionmay yield lower signal-to-noise ratios.

Effect of Indomethacin on cerebral blood flow

The study confirmed that the nonselective COXinhibitor indomethacin potently reduced CBF, prob-ably by inhibiting prostaglandin synthesis andconstricting precapillary resistance vessels (Castel-lano et al, 1998; Rasmussen et al, 2003). The averagereduction of CBF was 35%. Similar reductions werefound in previous human studies using indirectmethods (Wennmalm et al, 1981; Hemler et al, 1990;

Jensen et al, 1993, 1996). The findings are alsosimilar to the result of the only previous tomo-graphic study showing 28% to 40% reduction ofCBF in baboons, depending on the region (Schu-mann et al, 1996).

The effect of indomethacin administration on CBFwas measured by the 133xenon method in adults(Amano and Meyer, 1981; Jensen et al, 1996) and bythe Doppler flow method in children (Austin et al,1992; Hemler et al, 1990; Nitter et al, 1995) andanimals (Nilsson et al, 1995). By itself, xenon causesCBF globally to increase, but significant changes areseen only at high concentrations such as those usedto obtain Xe-CT-images.

Both the VOI and voxel-based analyses indicatedthat no single region was differentially vulnerable toCBF reduction during the indomethacin adminis-tration. When the standard deviation of CBF variesacross the spatial extent of the brain, assumptions ofthe standard brain mapping methods are no longerfulfilled (Worsley et al, 1992), and the pooled s.d.cannot be used to assess significance. In these cases,the statistical analysis of the data must use thenormalized CBF values and the s.d. of each voxel.The correction for multiple comparisons furtherincreased the significance threshold of the t-statistic.Nevertheless, the analysis revealed a significantrelative increase of the normalized CBF, and thus asmaller absolute decrease in the cingulate cortexand less significantly also in the visual cortex andthe thalamus, suggesting that these regions were lessaffected by the indomethacin-induced CBF changes.

Effect of Indomethacin on CMRO2

The present study revealed no effect of the loweredCBF on CMRO2

in humans. Previous animal studiesof the effects of indomethacin on CMRO2

gaveconflicting results of no change (Dahlgren et al,1981; Pickard and MacKenzie, 1973) or reduction(Nowicki et al, 1987). The only tomographic studyon living animals (Schumann et al, 1996), found nochange of the global CMRO2

but interestingly (seebelow) a small and significant increase of CMRO2

inthalamus and pons. In adult humans, the effect ofindomethacin on CMRO2

was assessed only by thearterio-venous difference (Wennmalm et al, 1981;Jensen et al, 1991; Dahl et al, 1996; Bundgaard et al,1996) and no significant change was observed in anyof these studies. However, a study of preterm infantswith near-infrared-spectroscopy of cytochrome oxi-dase indicated reduced oxygenation (Benders et al,1995).

Neither the average nor the regional CMRO2

changed significantly as a result of indomethacinadministration in this group of young healthysubjects. We found a nonsignificant average declineof CMRO2

in both gray and white matter ofapproximately 10%, but with very large variabilityamong subjects and VOIs. The present study did not

Figure 5 Extended nonlinear flow–metabolism couple accord-ing to equation (5) with parameters listed in Table 4 as functionof F and P50

cytox relationship shown in Figure 3, for each of threemagnitudes of maximum cytochrome c oxidation rates (Jmax)listed in graph. Abscissa: CBF (mL/hg min). Ordinate: CMRO2

(mmol/hg min).

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1190

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

confirm any significant local changes, regardless ofwhether absolute or normalized CMRO2

values werecompared. Although the CMRO2

analysis could becompleted in only five of the eight subjects studiedin the present study, significant average change ofCMRO2

can be ruled out with a high degree ofcertainty.

It is well known that the average oxygen con-sumption of brain tissue is maintained at the normallevel during moderate declines of blood flow (e.g.,Purves, 1972; Grote et al, 1981; Kanno et al, 1988;Hirano et al, 1994; Moller et al, 2002; Rasmussenet al, 2003). The threshold of this homeostaticmechanism is reached when EO2

exceeds 0.56 to0.59 (Nemoto et al, 2002). The current models ofconstantly maintained mitochondrial oxygen re-serves do not explain this ability of cytochromeoxidase to maintain normal oxygen consumptionwhen the oxygen tension is reduced by diminisheddelivery. To explain this ability, the authors incor-porated a simple Michaelis–Menten expression intotheir model of flow–metabolism coupling and testedthe response to low blood flow by means ofindomethacin infusion. The analysis revealed aphysiological mitochondrial tension of 8.5 mm Hg,on assumption of constant oxygen diffusion condi-tions, including absent recruitment of capillaries.The absence is well supported for physiologicconditions, but it is unclear whether it extends tononphysiologic states below the threshold of flow-limitation of oxygen consumption (Gobel et al, 1989,Kuschinsky and Paulson, 1992; Vogel et al, 2004).

Nonlinear Flow–Metabolism Coupling

A key factor in the present extension of thenonlinear flow–metabolism coupling is the cyto-chrome oxidase affinity. At low blood flow rates,this factor controls the sensitivity of oxygen con-sumption to depletion of mitochondrial oxygen atthe prevailing cytochrome oxidase activity, as notedin previous reports of variable cytochrome oxidaseaffinity (Brown and Cooper 1994; Brown 1995;Gnaiger et al, 1995, 1998; Brown, 2001).

The theory of nonlinear flow–metabolism cou-pling predicts that incommensurate changes ofblood flow must accompany changes of oxygenconsumption, but it fails to predict the changes ofmetabolism when blood flow changes are imposedby external factors. The present extension of theauthors’ model shows that oxygen consumptionremains normal only if the affinity of cytochromeoxidase is adjusted to the prevailing oxygen tensionwhen the maximum rate of cytochrome c oxidation(Jmax) is constant. The model also shows how andwhen the changes of oxygen consumption arelimited by the magnitude of blood flow. In theextended model, increase of blood flow appears tobe a necessary but not sufficient condition forincrease of metabolism (Vafaee and Gjedde, 2004).

The model makes two important predictions that arethe results of the primary but separate role of flowadjustment:

First, provided the flow is maintained above thethreshold identified in the present study, oxidativemetabolism may, in principle, vary independently,within the limits set by the reaction rate ofcytochrome oxidase. This means that transientchanges of oxygen consumption may exceed theimmediate changes of blood flow that follow priorflow adjustment, as recently observed in studies byPET (Vafaee and Gjedde, 2004) and near-infraredspectroscopy (Chance and Nioka, 2004).

Second, the model makes the unexpected predic-tion that the relationship between oxygen consump-tion and blood flow may be inverted at blood flowrates above the threshold if the maximum rate ofcytochrome c oxidation is low. The reason for thisparadox is cytochrome oxidase’s affinity towardsoxygen, which rises more than the oxygen fallswhen blood flow approaches the limit of sufficientdelivery. This prediction has been verified indepen-dently in reports of elevated CMRO2

in brain tissueunder ischemic stress in patients with an oxygenextraction at or slightly below the limit of sufficientdelivery (Nariai et al, 2001, Nemoto et al, 2002,2004).

Single Common Factor?

The question of the agent responsible for the adjust-ment of cytochrome oxidase affinity in response to achange of blood flow cannot be answered withcertainty because many factors play a potential role.These factors were discussed in details by Gnaigeret al (1998), who predicted several of the propertiesof nonlinear flow–metabolism coupling evaluatedin the present study. Among the factors that maycontribute to the decline of affinity in response toelevated mitochondrial oxygen tension, cited byGnaiger et al (1998), is the access of oxygen to thecytochrome oxidase binding site. Jointly, thesefactors must possess at least two properties toaccount for the measured effect.

First, the factor(s) competively or noncompeti-tively must modulate the cytochrome oxidase reac-tion. The simplest such property would be theability to slow the rate-controlling reaction offerrocytochrome c (‘c2þ ’) with oxidized cytochromec oxidase, according to the equation presented byChance (1996):

JO2 ¼ k3a3½c2þ�1 þ k3½c2þ�=k1Pmit

O2

ð11Þ

where k3 and a3 are the association constant (turn-over number) and quantity of cytochrome aa3 orcytochrome c oxidase, [c2þ ] is the concentration offerrocytochrome c, and k1 is the association constantof the reaction between oxygen and ferricytochromec. The apparent Michaelis–Menten kinetic constants

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1191

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

of the steady-state reaction between oxygen andreduced cytochrome c oxidase include the half-saturation oxygen tension:

PcytoxM ¼ k3½c2þ�=k1 ð12Þ

and the maximum flux

Jmax ¼ k3a3½c2þ� ð13Þwhere the turnover rate, equal to the product of k3

and [c2þ ], controls the reaction rate when theoxygen level remains constant. In the simplest case,a common inhibitory factor, relative to its inhibitoryconstant (ratio wi), would reduce the flux byreducing the turnover according to the expansionof equations (4) and (10) for cooperative (oruncompetitive rather than noncompetitive) inhibi-tion (Pearce et al, 2003),

JO2 ¼JmaxPmit

O2

PcytoxM þ ð1 þ wiÞPmit

O2

ð14Þ

where wi is the weighted average sum of potentialfactors relative to their inhibitory constants, Ci/Ki.Equation (14) and Figures 4 and 5 show that it ispossible to proceed from magnitudes of commonfactors to blood flow rates on the one hand and ratesof oxygen consumption on the other. This equationpredicts the necessary concentration of an NO-likesubstance from the following relationship:

Pcytox50 ¼ P

cytoxM þ wiP

mitO2

ð15Þwhich yields a predicted NO concentration of

Ci ¼ KiP

cytox50 � P

cytoxM

PmitO2

!¼ 2:4nmol=L ð16Þ

when the inhibitory constant is 46 nmol/L at themitochondrial oxygen tension calculated in thisstudy (Moncada and Erusalimsky, 2002).

Second, the common factor must be sensitive tothe rate of CBF in a manner which maintains therelationship of negligible presence at low blood flowand significant presence at higher blood flow rates.To satisfy this mechanism, the common factor mustbe released continuously in relation to the magni-tude of CBF. Thus, physiological blood flow changescan be said to prepare the cytochrome oxidaseaffinity for changes of oxidative metabolism, whilethe common factor maintains the oxygen consump-tion at its normal level, until or unless a change ofmetabolism is enabled by increase or decrease of themaximum enzyme reaction rate (Jmax).

Nitric Oxide

As a substance known to cause vasodilatation, raiseCBF, and competitively inhibit oxygen binding tocytochrome oxidase in vitro and tentatively alsoin vivo, nitric oxide (NO) is an attractive candidatefor a role as common factor. Goadsby et al (1992)first demonstrated that NO links changes in CBF

and metabolism, at least for spreading depression-elicited increases in metabolic activity, and theresults implied that NO may have a more generalrole in flow–metabolism coupling (Moncada andErusalimsky, 2002). Brown and Cooper (1994),Cleeter et al (1994), and Schweizer and Richter(1994) first reported that cytochrome c oxidation canbe inhibited by NO. Brown (1995) proposed that NOmay be the physiologic regulator of the affinity ofmitochondrial respiration for oxygen, enablingmitochondria to act as sensors of oxygen over theentire physiologic range. Torres et al (1995) foundthat cytochrome c oxidase rapidly binds NO whenthe enzyme enters the turnover phase. The rapidonset of inhibition suggested that nanomolar con-centrations of NO are effective under physiologicconditions of low oxygen tension: The concentra-tions of NO measured in many biological systemsappear to be similar to those sufficient to inhibitcytochrome oxidase and mitochondrial respiration.Conversely, inhibition of NO synthesis stimulatedrespiration in a number of systems.

Thus, nanomolar concentrations of NO immedi-ately, specifically and reversibly inhibit cytochromeoxidase in cooperative competition with oxygen invivo. Animal models display respiratory inhibitionby NO endogenously produced by constitutiveisoforms of NO synthase (NOS), which may belargely mediated by NO inhibition of cytochromeoxidase, suggesting that the inhibition of cyto-chrome oxidase by NO may be involved in thephysiologic and/or pathologic regulation of respira-tion, and the affinity of cytochrome oxidase foroxygen in vivo (Brown, 2001).

Neurons release NO in response to activation ofreceptors of the excitatory amino acid N-methyl-D-aspartate (NMDA). Faraci and Breese (1993) foundthat activation of the receptors for NMDA producesdilatation of the cerebral microcirculation that ismediated by NO. This and other findings suggestthat NO generated in response to activation of theNMDA receptors in vivo is neuronally derived.Thus, NO may mediate increases in local bloodflow during increases of neuronal activity inresponse to excitatory amino acids.

Another body of evidence in the literature isconsistent with tonic NO release by capillaryendothelium and by nNOS within neurons (Reutenset al, 1997; White et al, 1999). Tonic release wouldmatch the baseline level of blood flow to theinhibition of cytochrome oxidase to the degreenecessary to maintain constant cerebral oxidativemetabolism, unless the effect of the inhibition isovercome by increase of Jmax. Buerk et al (2003)showed that NO acutely is elevated when CBF risesin vivo. Thus, when appropriately stimulated, theconstitutive forms of NOS present in endothelium(eNOS), neurons (nNOS), and inducible NOS(iNOS) in glial cells, appear to produce additionalNO, resulting in transient inhibition of cytochromeoxidase in surrounding cells.

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1192

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

It has been suggested that changes of tissue pHaffect the reduction of nitrite to NO in the absence ofoxygen (Zweier et al, 1999), potentially influencingboth the rate of blood flow and the inhibition ofcytochrome oxidase. This phenomenon in turn mayexplain the effects of hypo- and hypercapnia onblood flow and the absent effect on oxygen metabo-lism observed in most studies. Thus, findings byPuscas et al (2000) indicate that NO- as well asprostaglandin-induced carbonic anhydrase inhibi-tion is involved in the vasodilation produced byhypercapnia. However, findings by Heinert et al(1999) suggest that the NO- and indomethacin-sensitive pathways involved in the hypercapnicresponse are distinct, although they interact syner-gistically. The interaction is also suggested byBeasley et al (1998), who concluded that the eNOSresponse to global ischemia involves an indometha-cin-sensitive mediation.

An additional finding supports the candidature ofNO for at least a part role as common factor: NOgeneration appears to be saturable at higher oxygenlevels. Although, the P50 of NO synthase for oxygenis at least an order of magnitude higher than that ofcytochrome oxidase (in the absence of NO; Brown,2001), saturation of NO synthase eventually rendersthe enzyme unable to produce more NO to furtherinhibit cytochrome oxidase at the highest oxygenlevels.

Safety of Indomethacin

We studied only young healthy volunteers. Inpatients with compromised CBF, further reductioncould affect the CMRO2

. Abnormally low CBF is acommon finding in patients with severe traumaticbrain injury (Glasgow Coma Scale less than 8; Cold,1989; Marion et al, 1991; Obrist et al, 1979; Over-gaard et al, 1981) and hypoxia may be the singlemost important secondary factor affecting the out-come in these patients (Miller, 1985). Therefore,indomethacin should be used with care in the acutemanagement of traumatic brain injury, especially inconditions of a high intracranial pressure. The use ofindomethacin should be cautioned in older patientswith reduced CBF because it may provoke furtherreductions of CBF. Specifically, the extraction ofoxygen should not be allowed to rise above 0.56 to0.59, or the venous hemoglobin saturation shouldnot be allowed to decrease below 38% to 40%.

Conclusions

This study confirms an average decrease of CBF of35% and no significant reduction of CMRO2

afterindomethacin administration in young healthyawake humans. The study supports the extendedmodel of diffusion-limited oxygen delivery to brain.The maintenance of normal oxygen consumption

during the flow reduction was ascribed to adjust-ment of cytochrome oxidase affinity by a mechanisminvolving a hypothetical common factor, which hasseveral properties in common with NO.

References

Amano T, Meyer JS (1981) Prostaglandins and human CBFcontrol: effect of aspirin and in-domethacin. J CerebBlood Flow Metab 1:403–4

Austin NC, Pairaudeau PW, Hames TK, Hall MA (1992)Regional cerebral blood flow velocity changes afterindomethacin infusion in preterm infants. Arch DisChild 67:851–4

Beasley TC, Bari F, Thore C, Thrikawala N, Louis T, BusijaD (1998) Cerebral ischemia/reperfusion increases endo-thelial nitric oxide synthase levels by an indomethacin-sensitive mechanism. J Cereb Blood Flow Metab18:88–96

Benders MJ, Dorrepaal CA, van de Bor M, van Bel F (1995)Acute effects of indomethacin on cerebral hemody-namics and oxygenation. Biol Neonate 68:91–9

Brown GC (1995) Nitric oxide regulates mitochondrialrespiration and cell functions by inhibiting cytochromeoxidase. FEBS Lett 369:136–9

Brown GC (2001) Regulation of mitochondrial respirationby nitric oxide inhibition of cytochrome c oxidase.Biochim Biophys Acta 1504:46–57

Brown GC, Cooper CE (1994) Nanomolar concentrations ofNO reversibly inhibit synaptosomal respiration bycompeting with oxygen at cytochrome oxidase. FEBSLett 356:295–8

Buerk DG, Ances BM, Greenberg JH, Detre JA (2003)Temporal dynamics of brain tissue nitric oxide duringfunctional forepaw stimulation in rats. Neuroimage18:1–9

Bundgaard H, Jensen K, Cold GE, Bergholt B, FrederiksenR, Pless S (1996) Effects of perioper-ative indomethacinon intracranial pressure, cerebral blood flow, andcerebral metabolism in patients subjected to cranio-tomy for cerebral tumors. J Neurosurg Anesthesiol8:273–9

Buxton RB, Frank LR (1997) A model for the couplingbetween cerebral blood flow and oxygen metabolismduring neural stimulation. J Cereb Blood Flow Metab17:64–72

Castellano AE, Micieli G, Bellantonio P, Buzzi MG,Marcheselli S, Pompeo F, Rossi F, Nappi G (1998)Indomethacin increases the effect of isosorbide di-nitrate on cerebral hemodynamic in migraine patients:pathogenetic and therapeutic implications. Cephalalgia18:622–30

Chance B (1976) Pyrimidine nucleotide as an indicator ofthe oxygen requirements for energy-linked functions ofmitochondria. Circ Res 38(suppl 1):I-31–8

Chance B (1996) Spectrophotometric and kinetic studiesof flavoproteins in tissues, cell suspensions, mitochon-dria and their fragments. In: Flavins and flavoproteinsSlater EC, (ed), New York: Elsevier, 496–510

Chance B, Nioka S (2004) Oxygen extraction as a fastmetabolic signal from tissue [abstract]. Int Soc OxygenTransport Tissue, Bari, Italy, August 20–26

Cleeter MW, Cooper JM, Darley-Usmar VM, Moncada S,Schapira AH (1994) Reversible inhibition of cyto-chrome c oxidase, the terminal enzyme of the

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1193

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

mitochondrial respiratory chain, by nitric oxide.Implications for neurodegenerative diseases. FEBS Lett345:50–4

Cold GE (1989) Does acute hyperventilation provokecerebral oligaemia in comatose patients after acutehead injury? Acta Neurochir (Wien) 96:100–6

Collins DL, Neelin P, Peters TM, Evans AC (1994)Automatic 3D intersubject registration of MR volu-metric data in standardized Talairach space. J ComputAssist Tomgr 18:192–205

Dahl B, Bergholt B, Cold GE, Astrup J, Mosdal B, Jensen K,Kjaersgaard JO (1996) CO2 and indomethacin vaso-reactivity in patients with head injury. Acta Neurochir(Wien) 138:265–73

Dahlgren N, Nilsson B, Sakabe T, Siesjo BK (1981) Theeffect of indomethacin on cerebral blood flow andoxygen consumption in the rat at normaland increasedcarbon dioxide tensions. Acta Physiol Scand 111:475–85

Faraci FM, Breese KR (1993) Nitric oxide mediatesvasodilatation in response to activation of N-methyl-D-aspartate receptors in brain. Circ Res 72:476–80

Furlan M, Marchal G, Viader F, Derlon JM, Baron JC(1996) Spontaneous neurological recovery after strokeand the fate of the ischemic penumbra. Ann Neurol40:216–26

Gjedde A (1997) The relation between brain function andcerebral blood flow and metabolism. In: Cerebrovascu-lar disease (Batjer HH, ed) Philadelphia: Lippincott-Raven, 23–40

Gjedde A (2002) Cerebral blood flow change in arterialhypoxemia is consistent with negligible oxygen tensionin brain mitochondria. NeuroImage 17:1876–81

Gjedde A, Marrett S (2001) Glycolysis in neurons, notastrocytes, delays oxidative metabolism of humanvisual cortex during sustained checkerboard stimula-tion in vivo. J Cereb Blood Flow Metab 21:1384–92

Gjedde A, Poulsen PH, Østergaard L (1999) On theoxygenation of hemoglobin in the human brain. AdvExp Biol Med 471:67–81

Gnaiger E (1993) Homeostatic and microxic regulation ofrespiration in transitions to anaerobic metabolism. In:The Vertebrate Gas Transport Cascade: Adaptations toEnvironment and Mode of Life (Bicudo JEPW, ed) BocaRaton: CRC Press, 358–70

Gnaiger E, Lassnig B, Kuznetsov A, Rieger G, Margreiter R(1998) Mitochondrial oxygen affinity, respiratory fluxcontrol and excess capacity of cytochrome c oxidase.J Exp Biol 201:1129–39

Gnaiger E, Steinlechner-Maran R, Mendez G, Eberl T,Margreiter R (1995) Control of mitochondrial andcellular respiration by oxygen. J Bioenerg Biomembr27:583–96

Goadsby PJ, Kaube H, Hoskin KL (1992) Nitric oxidesynthesis couples cerebral blood flow and metabolism.Brain Res 595:167–70

Gobel U, Klein B, Schrock H, Kuschinsky W (1989) Lack ofcapillary recruitment in the brains of awake rats duringhypercapnia. J Cereb Blood Flow Metab 9:491–9

Grote J, Zimmer K, Schubert R (1981) Effects of severearterial hypocapnia on regional blood flow regulation,tissue PO2 and metabolism in the brain cortex of cats.Pflugers Arch 391:195–9

Heinert G, Nye PC, Paterson DJ (1999) Nitric oxide andprostaglandin pathways interact in the regulation ofhypercapnic cerebral vasodilatation. Acta PhysiolScand 166:183–93

Hemler RJ, Hoogeveen JH, Kraaier V, Van Huffelen AC,Wieneke GH, Hijman R, Glerum JH (1990) A pharma-cological model of cerebral ischemia. The effects ofindomethacin on cerebral blood flow velocity, quanti-tative EEG and cognitive functions. Methods Find ExpClin Pharmacol 12:641–3

Hill AV (1910) The possible effect of aggregation of themolecules of haemoglobin on its dissociation curve.J Physiol 40, iv–vii

Hill AV (1913) The combination of haemoglobin withoxygen and with carbon monoxide. Biochem J 7:471–80

Hirano T, Minematsu K, Hasegawa Y, Tanaka Y, HayashidaK, Yamaguchi T (1994) Aceta-zolamide reactivity on123I-IMP single photon emission computed tomogra-phy in patients with major cerebral artery occlusivedisease: correlation with positron emission tomographyparameters. J Cereb Blood Flow Metab 14:763–70

Imberti R, Bellinzona G, Ilardi M, Bruzzone P, Pricca P(1997) The use of indomethacin to treat acute rises ofintracranial pressure and improve global cerebralperfusion in a child with head trauma. Acta Anaes-thesiol Scand 41:536–40

Jensen K, Freundlich M, Bunemann L, Therkelsen K,Hansen H, Cold GE (1993) The effect of indomethacinupon cerebral blood flow in healthy volunteers. Theinfluence of moderate hypoxia and hypercapnia. ActaNeurochir (Wien) 124:114–9

Jensen K, Kjaergaard S, Malte E, Bunemann L, TherkelsenK, Knudsen F (1996) Effect of graduated intravenousand standard rectal doses of indomethacin on cerebralblood flow in healthy volunteers. J Neurosurg Anesthe-siol 8:111–6

Jensen K, Ohrstrom J, Cold GE, Astrup J (1991) The effectsof indomethacin on intracranial pressure, cerebralblood flow and cerebral metabolism in patients withsevere head injury and intracranial hypertension. ActaNeurochir (Wien) 108:116–21

Kanno I, Uemura K, Higano S, Murakami M, Iida H, MiuraS, Shishido F, Inugami A, Sayama I (1988) Oxygenextraction fraction at maximally vasodilated tissue inthe ischemic brain estimated fro the rgional CO2

resonsiveness measured by positron emission tomo-graphy. J Cereb Blood Flow Metab 8:227–35

Kuschinsky W, Paulson OB (1992) Capillary circulation inthe brain. Cerebrovasc Brain Metab Rev 4:261–86

Marchal G, Beaudouin V, Rioux P, de la Sayette V, Le DozeF, Viader F, Derlon JM, Baron JC (1996) Prolongedpersistence of substantial volumes of potentially viablebrain tissue after stroke: a correlative PET-CT studywith voxel-based data analysis. Stroke 27:599–606

Marion DW, Darby J, Yonas H (1991) Acute regionalcerebral blood flow changes caused by severe headinjuries. J Neurosurg 74:407–14

Michaelis L, Menten ML (1913) Zur Kinetik der Invertin-wirkung. Biochem Z 49:333–69

Miller JD (1985) Head injury and brain ischaemia—implications for therapy. Br J Anaesth 57:120–30

Moller K, Strauss GI, Qvist J, Fonsmark L, Knudsen GM,Larsen FS, Krabbe KS, Skinhoj P, Ped-ersen BK (2002)Cerebral blood flow and oxidative metabolism duringhuman endotoxemia. J Cereb Blood Flow Metab22:1262–70

Moncada S, Erusalimsky JD (2002) Does nitric oxidemodulate mitochondrial energy generation and apop-tosis? Nat Rev Mol Cell Biol 3:214–20

Nariai T, Mataushima Y, Imae S, Ohta Y, Ishii K, Senda M,Ohno K (2001) Variable hemodynamic stages in

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1194

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

moyamoya disease classified among 60 patients withpositron emission tomography [abstract]. J Cereb BloodFlow Metab 21(Suppl 1):S347

Neelin P, Crossman J, Hawkes DJ, Ma Y, Evans AC(1993) Validation of an MRI/PET landmark registra-tion method using 3D simulated PET images andpoint simulation. Comput Med Imaging Graph 17:351–6

Nemoto EM, Yonas H, Kuwabara H, Pindzola RR,Sashin D, Meltzer C, Price JC, Chang Y (2002)Detection of stage II compromised cerebrovascularreserve by xenon-CT cerebral blood flow with azetazol-amide and oxygen extraction fraction by positronemission tomography. In: Brain imaging using PET(Senda M, Kimura Y, Herscovitch P, eds) San Diego:Elsevier, 259–67

Nemoto EM, Yonas H, Kuwabara H, Pindzola RR, SashinD, Meltzer CC, Price JC, Chang Y, Johnson DW (2004)Identification of hemodynamic compromise by cerebro-vascular reserve and oxygen extraction fraction inocclusive vascular disease. J Cereb Blood Flow Metab24:1081–9

Nilsson F, Bjorkman S, Rosen I, Messeter K, Nordstrom CH(1995) Cerebral vasoconstriction by indomethacin inintracranial hypertension. An experimental investiga-tion in pigs. Anesthe-siology 83:1283–92

Nitter WH, Johnsen LF, Eriksen M (1995) Acute effects ofindomethacin on cerebral blood flow in man. Pharma-cology 51:48–55

Nowicki JP, Jourdain D, MacKenzie ET (1987) NADHfluorescence in vivo: changes in cerebral oxidativemetabolism and perfusion induced by pentobarbital,indomethacin, and salicylate. J Cereb Blood Flow Metab7:280–8

Obrist WD, Gennarelli TA, Segawa H, Dolinskas CA,Langfitt TW (1979) Relation of cerebral blood flow toneurological status and outcome in head-injuredpatients. J Neurosurg 51:292–300

Ohta S, Meyer E, Fujita H, Reutens DC, Evans AC, GjeddeA (1996) Cerebral [15O]water clearance in humansdetermined by PET: I. Theory and normal values.J Cereb Blood Flow Metab 16:765–80

Ohta S, Meyer E, Thompson CJ, Gjedde A (1992) Oxygenconsumption of the living human brain measured aftera single inhalation of positron emitting oxygen. J CerebBlood Flow Metab 12:179–92

Østergaard L, Johannsen P, Host-Poulsen P, Vestergaard-Poulsen P, Asboe H, Gee AD, Hansen S.B, Cold GE,Gjedde A, Gyldensted C (1998) Cerebral blood flowmeasurements by dynamic MRI bolus tracking: com-parison with [15O]H2O PET in humans. J Cereb BloodFlow Metab 18:935–40

Overgaard J, Mosdal C, Tweed WA (1981) Cerebralcirculation after head injury. Part 3: Does reducedregional cerebral blood flow determine recovery ofbrain function after blunt head injury? J Neurosurg55:63–74

Parfenova H, Zuckerman S, Leffler CW (1995) Inhibitoryeffect of indomethacin on prostacyclin receptor-mediated cerebral vascular responses. Am J Physiol268:H1884–90

Pearce L, Bominaar L, Bruce C, Petersen J (2003) Reversalof cyanide inhibition of cytochrome c oxidase bythe auxiliary substrate nitric oxide. J Biol Chem 278:52139–45

Pickard JD, MacKenzie ET (1973) Inhibition of prostaglan-din synthesis and the response of baboon cerebral

circulation to carbon dioxide. Nat New Biol 245:187–8

Purves MJ (1972) The physiology of the cerebral circula-tion. University Press: Cambridge

Puscas I, Coltau M, Domuta G, Baican M, Puscas C, PascaR (2000) Carbonic anhydrase I inhibition by nitricoxide: implications for mediation of the hypercapnia-induced vasodilator response. Clin Exp PharmacolPhysiol 27:95–9

Rasmussen M, Poulsen PH, Treiber A, Delahaye S, TankisiA, Cold GE, Therkelsen K, Gjedde A, Astrup J (2003) Noinfluence of the endothelin receptor antagonist bosen-tan on basal and indomethacin-induced reduction ofcerebral blood flow in pigs. Acta Anaesthesiol Scand47:200–7

Reutens DC, McHugh MD, Toussaint PJ, Evans AC,Gjedde A, Meyer E, Stewart DJ (1997) l-Arginineinfusion increases basal but not activated cerebralblood flow in humans. J Cereb Blood Flow Metab17:309–15

Sakoh M, Gjedde A (2003) Neuroprotection in hypother-mia linked to redistribution of oxygen in brain. Am JPhysiol Heart Circ Physiol 285:H17–25

Schumann P, Touzani O, Young AR, Verard L, Morello R,MacKenzie ET (1996) Effects of indomethacin oncerebral blood flow and oxygen metabolism: a positronemission tomographic investigation in the anaesthe-tized baboon. Neurosci Lett 220:137–41

Schweizer M, Richter C (1994) NO potently and reversiblydeenergizes mitochondria at low oxygen tension.Biochem Biophys Res Commun 204:169–75

Talairach J, Tournoux P (1988) Co-planar stereotaxic atlasof the human brain. Stuttgart: Thieme Verlag

Torres J, Darley-Usmar V, Wilson MT (1995) Inhibition ofcytochrome c oxidase in turnover by nitric oxide:mechanism and implications for control of respiration.Biochem J 312:169–73

Vafaee MS, Gjedde A (2000) Model of blood–brain transferof oxygen explains nonlinear flow-metabolism cou-pling during stimulation of visual cortex. J Cereb BloodFlow Metab 20:747–54

Vafaee MS, Gjedde A (2004) Spatially dissociated flow-metabolism coupling in brain activation. NeuroImage21:507–15

Vogel J, Gehrig M, Kuschinsky W, Marti HH (2004)Massive inborn angiogenesis in the brain scarcelyraises cerebral blood flow. J Cereb Blood Flow Metab24:849–59

Wang Q, Paulson OB, Lassen NA (1993) Indomethacinabolishes cerebral blood flow increase in responseto acetazolamide-induced extracellular acidosis: amechanism for its effect on hypercapnia? J Cereb BloodFlow Metab 13:724–7

Wennmalm A, Erikson S, Wahren J (1981) Effectof indomethacin on basal and carbon dioxidestimulated cerebral blood flow in man. Clin Physiol 1:227–34

White RP, Hindley C, Bloomfield PM, Cunningham VJ,Vallance P, Brooks DJ, Markus HS (1999) The effect ofthe nitric oxide synthase inhibitor L-NMMA on basalCBF and vasoneuronal coupling in man: a PET study.J Cereb Blood Flow Metab 19:673–8

Woods RP, Cherry SR, Mazziotta JC (1992) Rapid auto-mated algorithm for aligning and reslicing PET images.J Comput Assist Tomogr 16:620–33

Woods RP, Mazziotta JC, Cherry SR (1993) Automatedimage registration. Ann Nucl Med 7:S70–1

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1195

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196

Worsley KJ, Evans AC, Marrett S, Neelin P (1992) A three-dimensional statistical analysis for CBF activationstudies in human brain. J Cereb Blood Flow Metab12:900–18

Worsley KJ, Marrett S, Neelin P, Vandal AC, Friston KJ,Evans AC (1996) A unified statistical approach fordetermining significant signals in images of cerebralactivation. Hum Brain Map 4:58–73

Zonta M, Angulo MC, Gobbo S, Rosengarten B,Hossmann KA, Pozzan T, Carmignoto G (2003)Neuron-to-astrocyte signaling is central to the dynamiccontrol of brain microcirculation. Nat Neurosci 6:43–50

Zweier JL, Samouilov A, Kuppusamy P (1999) Non-enzymatic nitric oxide synthesis in biological systems.Biochim Biophys Acta 1411:250–62

Low blood flow and cytochrome oxidase inhibition in brainA Gjedde et al

1196

Journal of Cerebral Blood Flow & Metabolism (2005) 25, 1183–1196