doi:10.1152/ajpregu.00127.2002 283:1210-1220, 2002. First published Aug 1, 2002;Am J Physiol Regulatory Integrative Comp Physiol

Riccardo Barbieri, John K. Triedman and J. Philip Saul

You might find this additional information useful...

22 articles, 16 of which you can access free at: This article cites http://ajpregu.physiology.org/cgi/content/full/283/5/R1210#BIBL

7 other HighWire hosted articles, the first 5 are: This article has been cited by

  [PDF]  [Full Text]  [Abstract]

, September 1, 2004; 126 (3): 801-807. ChestG. J. Gates, S. E. Mateika, R. C. Basner and J. H. Mateika

Continuous Positive Airway PressureBaroreflex Sensitivity in Nonapneic Snorers and Control Subjects Before and After Nasal 

[PDF]  [Full Text]  [Abstract], January 1, 2005; 288 (1): H424-H435. Am J Physiol Heart Circ Physiol

R. Barbieri, E. C. Matten, A. A. Alabi and E. N. Brown heart rate variability

A point-process model of human heartbeat intervals: new definitions of heart rate and 

[PDF]  [Full Text]  [Abstract], January 1, 2006; 100 (1): 51-59. J Appl Physiol

O-Yurvati and P. B. Raven S. Ogoh, R. M. Brothers, Q. Barnes, W. L. Eubank, M. N. Hawkins, S. Purkayastha, A.

Cardiopulmonary baroreflex is reset during dynamic exercise 

[PDF]  [Full Text]  [Abstract], July 1, 2006; 101 (1): 68-75. J Appl Physiol

O-Yurvati and P. B. Raven S. Ogoh, R. M. Brothers, Q. Barnes, W. L. Eubank, M. N. Hawkins, S. Purkayastha, A.

rest and during exerciseEffects of changes in central blood volume on carotid-vasomotor baroreflex sensitivity at 

[PDF]  [Full Text]  [Abstract], January 1, 2007; 62 (1): 86-92. J. Gerontol. A Biol. Sci. Med. Sci.

V. Novak, K. Hu, M. Vyas and L. A. Lipsitz Cardiolocomotor Coupling in Young and Elderly People

including high-resolution figures, can be found at: Updated information and services http://ajpregu.physiology.org/cgi/content/full/283/5/R1210

can be found at: and Comparative PhysiologyAmerican Journal of Physiology - Regulatory, Integrativeabout Additional material and information

http://www.the-aps.org/publications/ajpregu

This information is current as of March 12, 2008 .  

http://www.the-aps.org/.ISSN: 0363-6119, ESSN: 1522-1490. Visit our website at Physiological Society, 9650 Rockville Pike, Bethesda MD 20814-3991. Copyright © 2005 by the American Physiological Society. ranging from molecules to humans, including clinical investigations. It is published 12 times a year (monthly) by the Americanilluminate normal or abnormal regulation and integration of physiological mechanisms at all levels of biological organization,

publishes original investigations thatThe American Journal of Physiology - Regulatory, Integrative and Comparative Physiology

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

Heart rate control and mechanical cardiopulmonarycoupling to assess central volume: a systems analysis

RICCARDO BARBIERI,1,2 JOHN K. TRIEDMAN,3 AND J. PHILIP SAUL1

1Department of Pediatrics, Medical University of South Carolina, Charleston,South Carolina 29425; 2Department of Anesthesia and Critical Care, MassachusettsGeneral Hospital/Harvard Medical School, Boston 02114; and 3Department of Cardiology,Children’s Hospital and Harvard Medical School, Boston, Massachusetts 02115Received 26 February 2002; accepted in final form 27 July 2002

Barbieri, Riccardo, John K. Triedman, and J. PhilipSaul. Heart rate control and mechanical cardiopulmonarycoupling to assess central volume: a systems analysis. Am JPhysiol Regul Integr Comp Physiol 283: R1210–R1220, 2002.First published August 1, 2002; 10.1152/ajpregu.00127.2002.—Small negative changes of central volume reduce car-diac output without significant alterations of arterial bloodpressure (ABP), suggesting an adequate regulatory response.Furthermore, evidence has arisen supporting a Bainbridgereflex (tachycardia with hypervolemia) in humans. To inves-tigate these phenomena, multivariate autoregressive tech-niques were used to evaluate the beat-to-beat interactionsbetween respiration, R-R interval, and ABP at six levels ofdecreased and increased central volume. With reductions ofcentral volume below control, baroreflex and respiratory si-nus arrhythmia gains were reduced, while with increases ofvolume above control, gains increased for the first two levelsbut decreased again at the highest volume level, suggestingthe presence of a Bainbridge reflex in healthy human sub-jects. The mechanical influence of respiration on centralvenous pressure (CVP) had an unexpected shift in phase atthe point of mild central hypervolemia, with the expectednegative relation at lower volumes (inspiration lowers CVP)but a positive relation at higher volumes (inspiration raisesCVP). We conclude that multivariate techniques can quan-tify the relations between a variety of respiratory and hemo-dynamic parameters, allowing for the in vivo assessment ofcomplex cardiorespiratory interactions during manipulationsof central volume. The results identify the presence of aBainbridge reflex in humans and suggest that short-termcardiovascular control is optimized at mild hypervolemia.

baroreflex; cardiopulmonary reflex; hypovolemia; hypervol-emia; respiration; blood pressure

MODERATE CHANGES of central venous volume, such asthose that occur with changes in posture, may havedramatic effects on cardiac output without significantalterations of arterial pressure (10, 11). This simpleobservation demonstrates the potent ability of the twoprimary mechanisms responsible for short-term arte-rial pressure regulation: arterial and cardiopulmonarybaroreflexes to maintain arterial blood pressure (ABP)in a fairly narrow range during a wide variety of

hemodynamic stressors. Although it appears that thisregulation is facilitated by constant adaptation andinteraction of these two reflexes, the specific dynamicsof these processes remain a subject of considerabledebate. On the basis of the observation that smallreductions of central volume induced by low levels oflower body negative pressure (LBNP) do not lead tosignificant reductions of arterial pressure (1, 10), anumber of investigators have concluded that unloadingof cardiopulmonary receptors in the atria and ventri-cles is the primary stimulus for reflex increases insympathetic activation to muscle vascular beds, subse-quent vasoconstriction, and maintenance of arterialpressure, despite reduced cardiac output (22). How-ever, more recently, two studies have brought thishypothesis into question. Jacobsen et al. (9) first dem-onstrated that elimination of any decrease in arterialpressure with phenylephrine during LBNP markedlyreduces the muscle sympathetic response. Taylor et al.(23) then demonstrated that despite the maintenanceof arterial pressure during low levels of LBNP, systolicascending aortic area as a measure of aortic barorecep-tor input actually decreased. Both of these studiessuggest that arterial baroreflexes do play an importantrole in maintaining arterial pressure during reductionsof central volume but that the combination of cardio-pulmonary and baroreflex control is simply potentenough to prevent reductions of arterial pressure. Fur-thermore, there is considerable evidence that arterialand cardiopulmonary baroreflexes interact, such thatcardiopulmonary receptor unloading via reductions incentral volume may lead to enhancement of carotidarterial baroreflex gain (18).

Heart rate control has been a less emphasized, butpotentially important, component of the reflex responseto changes in central volume. As with arterial pressure,low levels of LBNP produce minimal if any change inmean heart rate (10), an observation previously used tosupport the argument that arterial baroreflexes playedno role in the response. However, a number of investiga-tors, beginning with Bainbridge (3, 27), have demon-

Address for reprint requests and other correspondence: J. P. Saul,165 Ashley Ave., PO Box 250915, Charleston, SC 29425 (E-mail:saulp@musc.edu).

The costs of publication of this article were defrayed in part by thepayment of page charges. The article must therefore be herebymarked ‘‘advertisement’’ in accordance with 18 U.S.C. Section 1734solely to indicate this fact.

Am J Physiol Regul Integr Comp Physiol 283: R1210–R1220, 2002.First published August 1, 2002; 10.1152/ajpregu.00127.2002.

0363-6119/02 $5.00 Copyright © 2002 the American Physiological Society http://www.ajpregu.orgR1210

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

strated reflex tachycardia with volume loading and reflexbradycardia with volume reduction in dogs, changes op-posite those expected with an arterial baroreflex re-sponse. Such observations introduce the possibility thatthe lack of a change in mean heart rate during nonhypo-tensive LBNP is due to offsetting reflex effects on heartrate via the cardiopulmonary and arterial baroreflexes.Several recent findings support such a hypothesis. First,although the presence of the Bainbridge reflex has beencontroversial in humans (3), one report has clearly dem-onstrated reflex tachycardia with volume loading in hu-mans (17). Furthermore, Triedman et al. (25) found thatmild hypovolemia from a 10% blood donation signifi-cantly changed arterial baroreflex control of heart ratewithout changing either mean heart rate or arterial pres-sure, consistent with the findings of others during lowlevels of LBNP (8).

In this study, both unloading and loading of cardio-pulmonary receptors have been employed to assess theinfluence of central volume on short-term control ofheart rate and arterial pressure. Because of their abil-ity to identify interactions between cardiovascularvariables, autoregressive (AR) multivariate techniques(4) are used to investigate the relations between respi-ration, R-R interval, and ABP.

METHODS

Experimental protocol. The study protocol was approved bythe institutional human studies committee, and written in-formed consent was obtained from all participants. A total of13 healthy human subjects between the ages of 20 and 25 yr(median 23 yr), weight range 57–86 kg (median 65.5 kg),underwent the protocol.

Surface electrocardiogram (ECG), instantaneous lung vol-ume [respiration (Resp)], central venous pressure (CVP), andABP were monitored during 8-min segments in a variety ofconditions. Instantaneous lung volume was recorded with aRespitrace two-belt impedance plethysmograph (Noninva-sive Monitoring Systems, Ardsdale, NY) and calibrated byhaving subjects inflate and deflate an 800-ml bag. A centralvenous line was inserted via the left median cubital or ce-phalic vein, positioned in the superior vena cava, and flushedcontinuously with heparinized saline. CVP was transducedwith a Statham strain-gauge transducer, which was cali-brated with a water manometer and positioned at the level ofthe right atrium. ABP was measured with a noninvasivefinger photoplethysmograph (Finapres, Ohmeda, Englewood,CO) applied to the right third digit. Data were recorded afterthe subjects were comfortable and cardiovascular variableswere observed to be stable. Room temperature was main-tained at 24°C.

Respiratory intervals were set for each subject with anauditory cue (beep). Subjects were accustomed to these cuesfor several minutes at a constant respiratory interval of 4 s(0.25 Hz, 15 breaths/min). For each experimental period of 8min, the cues occurred with intervals between 1 and 15 s,randomly chosen from a Poisson distribution, with a meaninterval of 4 s, or 15 breaths/min. The random-intervalbreathing technique broadens the frequency spectrum of therespiratory signal in a manner approximating filtered whitenoise. The technique and its use in physiological frequency-domain studies have been described previously (7). LBNPwas applied by use of a Plexiglas chamber that accommo-dated the hips and lower extremities of the subject. This

device allowed for rapid application of negative pressuresdown to �50 cmH2O. Cardiopulmonary receptors were firstdeactivated with small, incremental applications of LBNP(�5, �15, �30 mmHg), then control conditions were re-stored, and finally cardiopulmonary receptors were activatedby passive leg raising (20° and 60°) and a 500-ml infusion ofnormal saline (NS). Data were acquired for a minimum of 8min of random breathing in each condition.

Data acquisition and analysis. Signals were recorded on aHewlett-Packard 3968 eight-track FM magnetic tape re-corder. Simultaneous analog-to-digital conversion was per-formed at 360 Hz using specialized software on an 80386-based computer. Data were transferred to a workstation(SPARC 2, Sun Microsystems, Mountain View, CA) for off-line analysis. R-wave positions were detected digitally, andR-R intervals were converted into a smoothed instantaneousR-R interval time series constructed at 3.0 Hz by use of analgorithm described previously (6). CVP and Resp signalswere digitally filtered at 1.5 Hz and decimated to 3.0 Hz. TheABP signal was also filtered at 1.5 Hz and decimated to 3.0Hz to provide a mean ABP signal. Systolic (SBP) and dia-stolic blood pressure (DBP) values were detected from theunfiltered arterial pressure signal, and a third-order spliningand filtering algorithm was used to construct 3-Hz timeseries of SBP and DBP. The result was synchronized 3.0-Hztime series of all the signals of interest. All signals werevisually inspected for artifact. Time series of �6 min (341 s,or 1,024 points) of R-R interval, Resp, CVP, SBP, and DBPwere selected from each 8-min study period.

Bivariate AR spectral estimation. A bivariate AR model oforder p can be described by the following equation

X�n� � � �k�1

p

A�k�X�n � k� � W�k�

X�n� � �x1�n�x2�n�� A�k� � �a11�k� a12�k�

a21�k� a22�k��W�n� � �w1�n�

w2�n��(1)

where x1(n) and x2(n) are the output signals, w1(n) and w2(n)are white noises, and aij(k) are the AR coefficients. The pmatrices of coefficients A(k) were calculated by solving theextended Yule-Walker equations (13). Information about thewhiteness of the input noise for each identification, and aboutthe optimum orders, according to the Akaike criteria (2),allows for determination of the best order to choose for aproper identification of the spectral characteristics of the twosignals and their interaction. Each transfer block describedin Fig. 1 can then be identified by properly choosing theoutput signals x1(n) and x2(n) among Resp, R-R, SBP or DBP,and CVP. Relations between the output biological signals canbe attributed to specific physiological mechanisms, indicatedby the transfer boxes in Fig. 1. Each transfer function, rep-resented by the boxes, can be identified using bivariate ARspectral estimation.

Parameter identification. Mean values for each experimen-tal condition were calculated. The variability of each signalwas calculated as the integrated power spectral density from0 to 0.5 Hz. For each signal, the spectral parameters wereextracted using a division into a low-frequency (LF) interval(0.04–0.15 Hz) and a high-frequency (HF) interval (0.15–0.5Hz) (19). LF and HF coherent powers between each pair ofsignals were computed using the cross-spectrum normalizedby the two autospectral power densities. For the baroreflexrelations, SBP was used to identify the feedback response,

R1211HR CONTROL AND CARDIOPULMONARY COUPLING

AJP-Regul Integr Comp Physiol • VOL 283 • NOVEMBER 2002 • www.ajpregu.org

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

similarly to most other studies of the heart rate baroreflex(16, 26), while DBP was used to identify the feedforwardresponse because of the clear effects of a change in R-Rinterval on DBP. Gain and phase values were also extractedat the two frequencies where the coherence between eachpair of signals reached its maximum inside each of thefrequency bands (LF and HF). Indexes were considered reli-able for coherence values over 0.5. Data and some of theanalyses from a single subject during one control epoch areshown in Fig. 2. The time domain signals (Fig. 2, Aa–Ae)clearly demonstrate the random interval nature of thebreathing pattern and its irregular effects on the variabilityof each of the other signals displayed. The autospectra of R-R,SBP, and Resp are shown along with their AR components inFig. 2, Ba–Bc. Despite the random nature of the respiratorycues, preferential peaks can be seen in the respiratory spec-trum at about 0.08, 0.15, and 0.25 Hz, which are reflected inthe spectra of both R-R and SBP. The coherence, gain, andphase spectra for the transfer relations between Resp andR-R (respiratory sinus arrhythmia), and SBP and R-R(baroreflex feedback), are shown in Fig. 2, Ca and Cb, respec-tively. Of note, the coherence spectra demonstrate peaks atthe same frequencies as the autospectra in Fig. 2B, presum-ably related to the stronger respiratory signal input at thosefrequencies. All signals are highlighted at the points wherecoherence is �0.5, where the transfer relations were used toyield reliable estimates of gain and phase in the LF and HFregions. Phase values are defined between �360 and 0 de-grees to evidence the physiological delay between each pair ofvariables, which will be addressed in RESULTS and DISCUSSION.

Statistics. The parameters identified by the multivariateprocedure were averaged among the 13 subjects for each ofthe seven epochs. Parameters identified for the two controlepochs were averaged together to give a single value for eachindex. The statistical analysis was performed using a t-test

(paired 2 sample for the means). All epochs were consideredfor the statistical test: the results for the control epoch werecompared first with each of the three volume reduction(LBNP) epochs, and then with each of the three volumeloading epochs. A P value of �0.05 was considered signifi-cant. Because the average data revealed what appeared to bea maximum in many of the gain parameters during volumeloading, statistical comparison was also performed betweenthe last two volume loading epochs, where the trend of someof the parameters appeared to change.

RESULTS

Most of the identified parameters are summarized inTables 1 and 2, while some are shown separately inFigs. 3–7, as presented below. The x-axis of each graphrepresents the seven experimental epochs. LBNP ep-ochs are shown on the left of the control epoch, andvolume loading epochs on the right. The values areconnected with solid lines, to highlight any trends dueto the induced physiological changes. Each epoch valuehas vertical bars, which represent the SE of the mean.Asterisks mark the values statistically different fromthe combined control epoch, while the statistical differ-ence between the last two volume loading epochs ismarked with diamonds over the connecting line be-tween the two points.

Mean values (Table 1, Fig. 3). The changes of CVPgenerally reflect the effects of the volume unloadingand loading procedures. As expected, CVP decreaseswith each increase of LBNP, while it increases consis-tently with the 20° leg raise and the NS infusion, butminimally with the 60° leg raising (Fig. 3A). Mean R-Rinterval decreases (tachycardia) with decreases in CVPduring LBNP, with significant changes by �15 mmHgLBNP. Similarly, there is initially a bradycardia (in-creased mean R-R) with volume loading at 20° and 60°leg raising; however, a tachycardia (reduced R-R) wasobserved at the highest level of volume induced withthe NS infusion (Fig. 3B). Although the changes werenot significantly different from control, R-R was signif-icantly lower after NS than at 60° leg raise, when onlythose two epochs were considered. The observed rela-tive tachycardia with volume loading is consistent witha Bainbridge reflex in this study (3, 17).

Mean arterial pressure, represented by ABP, wasconstant during LBNP, supporting the notion thatmild unloading of cardiopulmonary receptors leads tominimal changes in arterial pressure (Fig. 3C). How-ever, during volume loading, ABP values increasedsignificantly with each change in loading, potentiallyinvoking the dynamics of the arterial baroreflex loop(Fig. 3C).

Variance (Fig. 4). Although there were few significantchanges in either total variance or variance in the twofrequency bands LF and HF for any of the parametersmeasured, the variance for a number of parameters dem-onstrated trends that are consistent with changes in boththe parameter means (Fig. 3) and transfer gains (Figs. 5and 6). The total LF and HF variances of the respiratorysignal were nearly constant for all the epochs, confirmingthe efficiency of the random breathing control technique

Fig. 1. Simplified model of cardiovascular short-term regulation.Relations between the output biological signals can be attributed tospecific physiological mechanisms, indicated by proper abbreviationsin each box. Mech, mechanical influence through muscle activity;RSA, respiratory sinus arrhythmia; -Baro, baroreceptors gain;-FF, feedforward mechanical gain through vasculature; Cardiopul-monary, cardiopulmonary receptor activity. Each transfer function,represented by the boxes, can be identified using multivariate au-toregressive (AR) spectral estimation.

R1212 HR CONTROL AND CARDIOPULMONARY COUPLING

AJP-Regul Integr Comp Physiol • VOL 283 • NOVEMBER 2002 • www.ajpregu.org

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

in maintaining a constant respiratory stimulus over avariety of conditions and in both of the relevant frequencybands (Fig. 4A). The variance of CVP increased duringvolume unloading and decreased during volume loading,

changes that appear most obvious in the HF band, andreached significance at maximum volume loading (Fig.4B). R-R interval total variance changes had a similarpattern to that observed for R-R mean, with a maximum

Fig. 2. Aa–Ae: the 5 cardiovascular vari-ables considered for the experimentalprotocol. Example from a recording of 1control epoch: respiration (Resp; Aa);central venous pressure (CVP; Ab); R-Rinterval (R-R; Ac); systolic blood pres-sure (SBP; Ad); diastolic blood pressure(DBP; Ae). Ba–Bc: autospectra of the R-R(Ba), SBP (Bb), and Resp signals (Bc)obtained by applying the bivariate anal-ysis for R-R-SBP and R-R-Resp. Exam-ple from a recording of 1 control epoch.Low (LF)- and high-frequency (HF) spec-tral components are also shown. Ca andCb: coherence, gain, and phase betweenrespiration and R-R interval (Resp-R-R;Ca) and between SBP and R-R interval(SBP-R-R; Cb). Example from a record-ing of 1 control epoch. Areas in grayindicate frequency ranges where coher-ence is �0.5.

R1213HR CONTROL AND CARDIOPULMONARY COUPLING

AJP-Regul Integr Comp Physiol • VOL 283 • NOVEMBER 2002 • www.ajpregu.org

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

peak at 60° leg raising, which seemed to be due com-pletely to changes in the HF band (Fig. 4C). Furthermore,as with R-R mean, there was a significant reduction inR-R variance between 60° leg raising and maximumvolume loading. SBP and DBP total and HF variance(Fig. 4D, DBP not shown) were nearly constant duringvolume unloading; however, LF variance appeared toincrease slightly. Alternatively, total and LF variance forSBP decreased with volume loading, while HF varianceremained nearly constant. Of note, although in generalthe changes in total R-R variance are in the oppositedirection of those in total SBP variance, that was not thecase for the last volume loading epoch, indicating that thedecrease in R-R variance is not merely attributable to achange in SBP variability.

Coherence. Coherences were calculated for all thepossible interactions between R-R, SBP (or DBP),Resp, and CVP. In Table 2, values are shown in the

standard format for each signal pair (LBNP values onthe left, and volume loading values on the right of thecontrol results). For the effects of respiration on R-Rinterval [respiratory sinus arrhythmia (RSA)] andCVP (Resp-R-R and Resp-CVP), coherence values weregenerally higher in the HF than in the LF range.However, for those involving the baroreflex relations(SBP-R-R and R-R-DBP), LF coherence was generallyhigher than HF. The highest coherence values werefound for the mechanical effect of respiration on cen-tral filling pressure (Resp-CVP), where coherence wasclose to 1 (mean 0.93) for all epochs in the HF rangeand averaged 0.78 in the LF range. These values reflectthe close relation between Resp and CVP shown for thetime domain signals in Fig. 2, Aa and Ab. High coher-ence values were also found for the baroreflex relations(SBP-R-R and R-R-DBP) and for the RSA (Resp-R-R).Importantly, for both the baroreflex and RSA at both

Table 1. CVP, R-R interval, SBP and DBP averaged among all subjects for each experimental epoch

MeanValues �30 �15 �5 Control Leg� Leg�� 500 ml

CVP, mmHg 0.56�1.1* 1.58�0.7* 2.78�0.6* 4.55�0.8 5.61�1.2 5.53�1.2 7.57�1.4†R-R, ms 840�30* 884�32* 921�31 948�25 993�26* 997�29* 927�25†SBP, mmHg 104�1.9 106�1.4 106�1.7 107�1.3 111�1.1* 113�1.2* 115�1.4*DBP, mmHg 61�2.6 61�1.9 59�1.9 59�2.8 62�2.6* 66�2.9* 69�3.5*

Values are means � SE. �30, �15, And �5 are �30, �15, and �5 mmHg lower body negative pressure, respectively; leg� and leg�� are20° and 60° passive leg raising, respectively; 500 ml is 500-ml infusion of normal saline. CVP, central venous pressure; R-R, R-R interval;SBP and DBP, systolic and diastolic pressure, respectively. *Values significantly different from combined control epoch (Control).†Significantly different from leg�� volume loading epoch.

Table 2. Summary of gain and coherence identified by bivariate AR estimation from each pair of signals forboth the LF and HF ranges, averaged among all subjects for each experimental epoch

BivariateParameters �30 �15 �5 Control Leg� Leg�� 500 ml

Resp-R-R, ms/lRSA gain LF 103�18* 117�15* 126�23 174�30 188�33 171�29 144�37RSA gain HF 73�11* 96�14* 128�22 139�27 169�33 188�41 137�35†Coherence LF 0.53�0.06 0.56�0.06 0.53�0.08 0.65�0.06 0.65�0.06 0.72�0.04 0.66�0.04Coherence HF 0.71�0.04 0.82�0.03 0.82�0.02 0.79�0.03 0.75�0.07 0.79�0.04 0.79�0.04

Resp-SBP, mmHg/lMech gain LF 8.5�1.6 9.0�2.1 9.3�2.9 7.3�1.2 6.6�1.2 5.8�1.1 5.5�1.6Mech gain HF 4.3�0.7 3.6�0.5 3.5�0.6 3.6�0.5 3.5�0.6 3.4�0.6 3.4�0.9Coherence LF 0.41�0.05 0.56�0.06 0.49�0.07 0.59�0.05 0.62�0.07 0.59�0.08 0.53�0.08Coherence HF 0.46�0.04 0.54�0.04 0.52�0.04 0.55�0.04 0.62�0.02 0.59�0.03 0.52�0.07

Resp-CVP, mmHg/lMech gain LF 4.8�0.7 6.3�1.8* 6.2�2.8 3.2�0.9 2.3�0.7 2.4�0.6 2.3�0.6*Mech gain HF 9.4�2.2* 9.3�2.2* 6.8�1.8 5.3�1.2 3.2�0.8 3.4�0.8 3.3�0.8Coherence LF 0.88�0.05* 0.86�0.06 0.88�0.04 0.77�0.05 0.70�0.09 0.67�0.07 0.69�0.07Coherence HF 0.94�0.03 0.92�0.05 0.96�0.02 0.92�0.02 0.94�0.01 0.90�0.03 0.90�0.04

SBP-R-R, ms/mmHg-Gain LF 11.4�1.5* 17.1�2.4 16.7�2.9 21.7�4.1 26.3�3.7 31.0�5.1* 27.6�5.8-Gain HF 10.1�1.6* 14.8�2.5* 18.4�3.1 23.8�4.1 32.3�5.4 31.2�6.7 22.4�4.6Coherence LF 0.62�0.06 0.74�0.03 0.72�0.05 0.70�0.07 0.78�0.04 0.76�0.05 0.64�0.07Coherence HF 0.60�0.06 0.60�0.06 0.64�0.05 0.68�0.04 0.73�0.04 0.71�0.04 0.61�0.07

R-R-DBP, mmHg/ms-Gain LF 0.071�0.008* 0.069�0.009* 0.06�0.013 0.044�0.006 0.047�0.005 0.042�0.008 0.035�0.005*-Gain HF 0.037�0.005* 0.031�0.005* 0.025�0.003* 0.021�0.003 0.022�0.001 0.022�0.003 0.027�0.005Coherence LF 0.62�0.07 0.75�0.03 0.72�0.06 0.70�0.07 0.78�0.04 0.76�0.05 0.63�0.07Coherence HF 0.60�0.06 0.60�0.06 0.64�0.05 0.68�0.04 0.74�0.04 0.71�0.04 0.62�0.07

Values are means � SE; coherence low-frequency (LF) and coherence high-frequency (HF) values are unitless. AR, autoregressive; Resp,respiration; Mech, mechanical: -gain, -baroreceptor gain; -gain, -feedforward gain. *Values statistically different from combined controlepoch. †Statistically different from leg�� volume loading epoch.

R1214 HR CONTROL AND CARDIOPULMONARY COUPLING

AJP-Regul Integr Comp Physiol • VOL 283 • NOVEMBER 2002 • www.ajpregu.org

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

LF and HF, the average coherence for all epochs wasabove the 0.5 threshold, indicating that significantrelations existed between the variables and validatingthe transfer relations. This observation also highlightsthe ability of the techniques used, in particular randominterval breathing, to identify parameters such as RSAin both their traditional frequency ranges and nontra-ditional ones (HF vs. LF for RSA). Finally, the mechan-ical effect of respiration on SBP (Resp-SBP) had lowcoherence near 0.5 at both LF and HF, indicating thatthere are multiple influences on SBP other than respi-ration.

Respiratory effects (Table 2, Fig. 5). For demonstra-tion purposes, and because the coherence was highestfor HF, values for the RSA gain and phase are shownonly for the HF band. First, as confirmation of a correctidentification of the RSA, average HF phase values forall epochs were just above �180°, indicating that in thefrequency range of normal respiratory activity, therewas the expected decrease in R-R interval (increase ofheart rate) with inspiration, consistent with previousobservations on neurally mediated RSA (12, 21). Fur-thermore, the gain changes with volume manipulationhad a nearly identical pattern to that seen for bothmean R-R and R-R variance, with a peak at the mildhypervolemia stage of 60° leg raise. As with R-R vari-ance, the changes were not always significantly differ-ent from control; however, a significant decrease of

gain did occur between 60° leg raise and the NS infu-sion (P � 0.02). The mechanical influence of respirationon SBP was virtually constant, regardless of volumestate, with the possible exception of a slight increase ofgain at �30 mmHg LBNP (Table 2). The negative 90°phase at all epochs is consistent with that found pre-viously (20) and appears to reflect the fact that themechanical influence of respiration on intrathoracicpressure is related to the rate of change of lung volume(derivative), or flow (28).

Baroreflex (Table 2, Fig. 6). Again for demonstrationpurposes, only a single frequency range is shown, butfor the baroreflex relations, the more traditional LFband is used (5). As noted for the RSA identification,the phase values obtained for the baroreflex relationsseem reasonable for the nature of the physiologicalinteractions between heart rate and arterial pressure.The feedback path from SBP to R-R (-Baro in Fig. 6)has a phase just below 0°, indicating that R-R intervalincreases (heart rate decreases) when SBP increaseswith a very short delay, all consistent with a vagallymediated response. Alternatively, the feedforwardpath from R-R to DBP (-FF) has a phase around 90°,consistent with a 180° shift at 0 Hz (decreased R-Rincreases DBP) and a 2- to 3-s time delay for propagat-ing a change in heart rate into a change in arterialpressure. Importantly, the -baroreflex feedback gainrevealed the same characteristic trend shown by theR-R monovariate parameters and by the RSA gain,with the peak for baroreflex control of heart rate lo-cated at the same point of mild hypervolemia from the60° leg raise as seen in the other parameters. Thefeedforward gain increased significantly at the twohighest levels of hypovolemia (P � 0.01) and decreasedsignificantly at the highest level of hypervolemia (P �0.01).

Resp-CVP. The mechanical influence of respirationon central filling pressures had an unusual andsomewhat unexpected finding. The gain increasedslightly during the hypovolemia induced by LBNPand decreased about the same amount during vol-ume loading. However, the phase underwent a dra-matic nearly 180° shift at the first level of hypervol-emia and remained significantly different fromcontrol at all levels of volume loading (Fig. 7). Theseobservations may best be visualized by examiningthe time domain respiratory and CVP signals fromone subject (Fig. 8). CVP clearly decreases with in-spiration at �30 and �5 mmHg LBNP, has a some-what biphasic response at control, and increaseswith inspiration during central volume expansion,beginning with the 20° leg raise.

DISCUSSION

There are three important new findings in thisstudy. First, and most important, the data demon-strate that short-term cardiovascular control (RSA andbaroreflex feedback) appears to be optimized at mildhypervolemia, suggesting that even awake supine nor-movolemic conditions can represent a state of very mild

Fig. 3. Mean values of CVP (A), R-R interval (B), and arterial bloodpressure (ABP; C): averages for the group. Epochs: �5, �15, and �30are �5, �15, and �30 mmHg lower body negative pressure (LBNP),respectively; C, control; leg� and leg�� are 20° and 60° passive legraising, respectively; 500 ml, 500-ml infusion of normal saline. Barsindicate SEs on the averaged values. *Epochs that significantlydiffer (P � 0.05) from the control epoch. }Significant differencebetween the leg�� and the 500-ml epoch.

R1215HR CONTROL AND CARDIOPULMONARY COUPLING

AJP-Regul Integr Comp Physiol • VOL 283 • NOVEMBER 2002 • www.ajpregu.org

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

cardiovascular stress to normal human subjects. Sec-ond, both the time and frequency domain data stronglysupport the presence of a Bainbridge reflex (hypervol-emia-induced tachycardia) at moderately elevated lev-els of central volume, presumably present, as Bain-bridge suggested (3), to reduce cardiac preload undervolume loading conditions. Third, an unexpected effectof volume loading was identified, in which the normaleffect of respiration on CVP is reversed, serving as anindicator of intrathoracic volume status. The resultsfrom this study also confirm prior observations thatmild to moderate levels of hypovolemia do not causesignificant reductions in arterial pressure (10), ex-plained in part in this study by a mild tachycardia andincreased feedforward gain from heart rate to arterialpressure.

Optimization of heart rate control at mild hypervol-emia. Multiple studies have demonstrated that mild hy-povolemia induced by low levels of LBNP does not lead tosignificant reductions of arterial pressure (1, 9, 10);however, all of the precise mechanisms underlying thisobservation have never been fully elucidated. Someinvestigators have concluded that the unloading ofcardiopulmonary receptors in the atria and ventriclesleads to a reflex increase in sympathetic outflow tomuscle vascular beds, subsequent vasoconstriction,and maintenance of arterial pressure, despite reducedcardiac output (22). However, other studies have

brought this hypothesis into question, suggesting thatarterial baroreflex mechanisms play the largest role(9). This study has used a different approach to quan-tify the cardiovascular response to mild to moderatelevels of both hypo- and hypervolemia, employing thefrequency response to small fluctuations of heart rateand arterial pressure induced by random respiratoryactivity to assess the response of the various controlsystems involved. Our findings demonstrate that in-deed the gain and phase characteristics of multiplecontrol systems do change with shifts in central volumeand are consistent with some previous findings demon-strating a change in vagal heart rate control during lowlevels of LBNP (8) or nonhypotensive hemorrhage (25).

In this study, the two systems most responsible forcontrolling short-term fluctuations of heart rate, thearterial baroreflex and the RSA, both appeared to beoptimized (point of maximum gain) in a state of mildhypervolemia. These findings are supported by a peakin R-R variance and a maximum for the mean R-Rinterval (lowest heart rate) at the same point, suggest-ing that indeed the cardiovascular system is in a con-dition of minimized stress. The fact that both the heartrate arterial baroreflex and RSA gains continually de-cline with progressive hypovolemia while arterial pres-sure is maintained suggests that neither of these con-trol elements is fully responsible for the maintenanceof pressure. Because central volume was not directly

Fig. 4. Total variance (left), LF variance (middle), and HF variance (right) of Resp (A), CVP (B), R-R interval (C),and SBP (D): averages for the group. Bars indicate SEs on the averaged values. *Epochs that significantly differ(P � 0.05) from the control epoch. }Significant difference between the leg�� and the 500-ml epoch.

R1216 HR CONTROL AND CARDIOPULMONARY COUPLING

AJP-Regul Integr Comp Physiol • VOL 283 • NOVEMBER 2002 • www.ajpregu.org

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

perturbed on a beat-to-beat basis using a techniquesuch as random LBNP (24), and muscle sympatheticoutflow was not directly measured, the data from thisstudy cannot be used to comment directly on eitherarterial baroreflex or cardiopulmonary reflex control ofperipheral arterial resistance as the mechanism. How-ever, we did observe increases in the feedforward gainfrom heart rate to arterial pressure (R-R-DBP), the“other half ” of the arterial baroreflex. These changes

indicate that in the volume-reduced state, the samechange in heart rate is able to produce a larger changein arterial pressure and supports the notion that theelevation in mean heart rate helps prevent a fall inarterial pressure. The mechanism by which this feed-forward gain might be increased might be related tothe reduction in thoracic aortic size previously docu-mented by Taylor et al. (23) during similar volumereductions. With arterial capacitance reduced, the

Fig. 5. Gain and phase calculated forthe group in HF for the influence ofrespiration on R-R interval (bottomright, RSA) and for the influence ofrespiration on SBP (top left, Mech).*Epochs that significantly differ (P �0.05) from the control epoch. }Signifi-cant difference between the leg�� andthe 500-ml epoch.

Fig. 6. Gain and phase calculated for the groupin LF for the baroreflex loop, respectively, forthe influence of SBP on R-R (bottom, -Baro)and for the influence of R-R on DBP (top, -FF).*Epochs that significantly differ (P � 0.05)from the control epoch.

R1217HR CONTROL AND CARDIOPULMONARY COUPLING

AJP-Regul Integr Comp Physiol • VOL 283 • NOVEMBER 2002 • www.ajpregu.org

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

same change in cardiac output will lead to a largerchange in pressure. Although we report on the feedfor-ward gain to DBP, the findings were similar for SBPwith a slightly longer phase lag. Regardless of whichblood pressure parameter is used, the data from thisstudy clearly address more directly the role of heartrate control and its optimization at mild hypervolemia

than the precise mechanism of blood pressure mainte-nance during hypovolemia. Furthermore, it should benoted that the magnitude of the feedforward effectappears to be species and circumstance dependent,with a smaller effect noted by some in species such asrabbits (14), but a clear effect noted by others in thesame species (26).

Bainbridge reflex in humans. The phenomenon ofhypervolemia-induced tachycardia was first describedby Bainbridge in 1915 (3) after observing heart rateelevations in anesthetized dogs during very large rapidinfusions of volume to the left atrium, which markedlyincreased left atrial pressure. Bainbridge hypothesizedthat the reflex served the purpose of increasing cardiacoutput in response to increased venous return to theheart, reducing the elevation of cardiac preload. Al-though Bainbridge himself did not describe a bradycar-dia component to the reflex during lowering of atrialpressure, the basis of the Bainbridge reflex in animalswas eventually described as the full cardiopulmonaryreflex (15), parallel activation and deactivation of bothsympathetic and parasympathetic efferent activitywith hypo- and hypervolemia, respectively. Thus, al-

Fig. 7. Gain and phase calculated for the group in LF for theinfluence of Resp on CVP. *Epochs that significantly differ (P � 0.05)from the control epoch.

Fig. 8. Comparison between Resp (A) andCVP (B) during volume reduction (�30 and�5 mmHg) and volume loading (leg� and500 ml): 150-s segments extracted from re-cordings of a single complete experimentalprocedure.

R1218 HR CONTROL AND CARDIOPULMONARY COUPLING

AJP-Regul Integr Comp Physiol • VOL 283 • NOVEMBER 2002 • www.ajpregu.org

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

though the cardiac sympathetic component of the car-diopulmonary reflex will generally be in the same di-rection as the cardiac component of the arterialbaroreflex response (increased sympathetic outflowduring hypovolemia rather than reduced), the cardiacparasympathetic or vagal component will generally bein the opposite direction (enhanced vagal outflow dur-ing hypovolemia). Given all of these complexities, it isnot surprising that there has been a failure to consis-tently observe a clear Bainbridge reflex in humans. Infact, on the hypovolemia side, the heart rate responseto mild reductions in central venous volume in humanshas varied across the many studies in the literaturefrom mild tachycardia as we observed in this study, tono change in heart rate to mild bradycardia (9, 18).

There have been relatively few studies of the heartrate response to hypervolemia in humans; however,recently Pawelczyk and Levine (17) found that tachy-cardia could indeed be induced with hypervolemic ex-pansion in humans. These data are in agreement withour findings on mean R-R interval, but in addition, thedata from this study go further to identify reductions inheart rate arterial baroreflex and RSA gain that par-allel the reduction in R-R interval (tachycardia). Fur-thermore, we observed a significant drop in HF powerfor R-R interval, indicating a reduction in vagal mod-ulation of heart rate, consistent with Bainbridge’s orig-inal hypothesis (3). These changes and the observationthat HF power for R-R interval declines during LBNP-induced hypovolemia suggest that in humans, vagallymediated arterial baroreflex control of heart rate dom-inates during most mild central volume changes, con-sistent with other findings (9). However, with signifi-cant increases in volume, “Bainbridge” takes over byreducing mean and beat-to-beat vagal activity.

Respiratory modulation of CVP: phase changes withmild hypervolemia. Perhaps the most unique and cer-tainly the most unexpected observation of this study isthe dramatic nonlinear shift in the phase relation be-tween respiratory activity and CVP with even themildest level of hypervolemia (see Figs. 7 and 8 atleg�). Virtually every physiology text describes thewell-known reduction in intrapleural pressure duringinspiration (28), which subsequently reduces intravas-cular pressure within the thorax and increases venousreturn during inspiration, leading to increased rightventricular output and a widening of the split in thesecond heart sound. The data from this study areconsistent with such a scenario during control condi-tions and hypovolemia (Fig. 8, control and �5 and �30mmHg) but demonstrate the opposite effects duringhypervolemia, elevation of CVP during inspiration(Fig. 8, leg� and 500 ml), presumably reducing anyinspiratory-related increases in venous return. Al-though the methodology used in the study does notallow us to specifically address the underlying mecha-nism of this phenomenon, there are a few clues in thedata. Because there was no notable change in CVPbetween the first two levels of hypervolemia (Fig. 3,leg� and leg��), one can speculate that most of theblood pooled in the leg veins was emptied into the

splanchnic and thoracic veins during the first leg ele-vation, representing a relatively large shift in volume.After that point, with the splanchnic bed full and withprimarily diaphragmatic breathing in the supine posi-tion, inspiration probably led to compression of theabdominal veins, which with the legs raised forcedblood into the thorax. Apparently, this effect overcameany reductions in pressure due to decreased intratho-racic pressure. The precise effect on total venous re-turn to the heart is less clear, because blood in thesuperior systemic thoracic veins would actually beforced out of the thorax by the elevated pressure,rather than drawn into the thorax by negative pres-sure, but more blood would be forced into the thoraxthrough the inferior thoracic systemic veins by positivepressure from the abdomen. Regardless of the mecha-nism underlying it, this sudden shift in phase betweenrespiration and CVP represents a sensitive marker ofcentral volume status, which may have clinical utilityin the identification of optimum volume status.

Clinical utility of the findings. The three primaryobservations in this study can all be combined to derivea clinical algorithm for determining optimal volumestatus in critically ill patients. The simplest metric inthis study to utilize in a clinical setting is the respira-tory-CVP phase shift, which can be observed withoutany special tools on a standard cardiac monitor. Apatient whose volume status is unclear could be vol-ume loaded until the patient is at or near the point ofa 0°, in-sync, relation, with periodic checks by slightelevations of the lower extremities. Because R-R inter-val is always available and arterial pressure usuallyavailable, a more sophisticated monitoring systemcould also continuously measure R-R interval, R-Rinterval HF power, RSA gain, and arterial baroreflexgain, as previously described (4). Optimized clinicalstatus will theoretically be present at the maximumpoint for each of these variables, which occurs verynear the Resp-CVP phase shift. Finally, any sign ofactivation of the Bainbridge reflex, as indicated by adecrease in one of these variables during volume load-ing, should be taken as a sign of volume overload.

Clearly, the above scenarios make a number of as-sumptions that would have to be tested in clinical trialsbefore their routine use. Perhaps the most importantassumption is that the abnormal heart and vascula-ture will respond similarly to that in the subjectspresented here who have presumably normal physiol-ogy. Nonetheless, the possibility of having such sensi-tive clinical volume measures is intriguing.

Limitations. As noted previously, the data in thisstudy cannot be used to fully explain the mechanismsby which mean arterial pressure is maintained duringmild hypovolemia. In particular, without measurementof efferent sympathetic activity or systemic vascularresistance, the full role of the arterial and cardiopul-monary reflexes cannot be elucidated. Despite the clearobservation of the phase shift between Resp and CVP,neither the precise mechanism underlying it nor theprecise effects on venous return can be identified fromthe data collected here. Furthermore, despite using a

R1219HR CONTROL AND CARDIOPULMONARY COUPLING

AJP-Regul Integr Comp Physiol • VOL 283 • NOVEMBER 2002 • www.ajpregu.org

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from

causal technique that attempts to separate out barore-flex effects from other effects on heart rate, randominterval breathing introduces RSA in the LF rangewhere the baroreflex normally predominates, poten-tially confounding the identification of the baroreflex.Finally, because no clinical studies have been per-formed to test the abilities of the described parametersto measure volume status, at this time the clinicalutility of the findings is purely theoretical.

Conclusions. Systems analysis techniques can pro-vide unique insights into cardiovascular regulation. Inthis study, the relations between respiration, arterialpressure, and CVP were used to investigate the phys-iological effects of mild to moderate reductions andelevations of central volume. Furthermore, randominterval breathing was used to broaden the frequencycontent of the relations being studied. The data dem-onstrated first that heart rate control appeared to bemaximized in a state of mild central hypervolemia, to apoint where increases in volume lead to reductions inheart rate control gain, and tachycardia. These find-ings also confirmed the presence of a Bainbridge reflexin humans. Finally, an unexpected nonlinear phaseshift in the mechanical coupling between respirationand CVP was identified during mild hypervolemia,near the point of maximum heart rate control. To-gether, the heart rate control and respiratory-CVPfindings may potentially serve as sensitive markers ofcentral volume status in critically ill patients.

This work was supported by a grant from the Whitaker Founda-tion, Mechanicsburg, Pennsylvania, a Milton Fund Research Grantfrom Harvard Medical School, and by National Institutes of HealthGrant R01-HL-62385.

REFERENCES

1. Abboud FM. Integration of reflex responses in the control ofblood pressure and vascular resistance. Am J Cardiol 44: 903–911, 1979.

2. Akaike H. A new look at the statistical model identification.IEEE Trans Autom Control AC-19: 716–723, 1974.

3. Bainbridge FA. The influence of venous filling upon the rate ofthe heart. J Physiol 50: 65–84, 1915.

4. Barbieri R, Waldmann RA, DiVirgilio V, Triedman JK,Bianchi AM, Cerutti S, and Saul JP. Continuous quantifica-tion of baroreflex and respiratory control of heart rate by use ofbivariate autoregressive techniques. Ann Noninvasive Electro-cardiol 1: 264–277, 1996.

5. Baselli G, Cerutti S, Civardi S, Malliani A, and Pagani M.Cardiovascular variability signals: towards the identification of aclosed-loop model of the neural control mechanisms. IEEE TransBiomed Eng 35: 1033–1046, 1988.

6. Berger RD, Akselrod S, Gordon D, and Cohen RJ. Anefficient algorithm for spectral analysis of heart rate variability.IEEE Trans Biomed Eng BME-33: 900–904, 1986.

7. Berger RD, Saul JP, and Cohen RJ. Transfer function anal-ysis of autonomic regulation. I. The canine atrial rate response.Am J Physiol Heart Circ Physiol 256: H142–H152, 1989.

8. Floras JS, Butler GC, Ando SI, Brooks SC, Pollard MJ, andPicton P. Differential sympathetic nerve and heart rate spectraleffects of nonhypotensive lower body negative pressure. Am JPhysiol Regul Integr Comp Physiol 281: R468–R475, 2001.

9. Jacobsen TN, Morgan BJ, Scherrer U, Vissing SF, LangeRA, Johnson N, Ring WS, Rahko PS, Hanson P, and Victor

RG. Relative contributions of cardiopulmonary and sinoaorticbaroreflexes in causing sympathetic activation in the humanskeletal muscle circulation during orthostatic stress. Circ Res 73:367–378, 1993.

10. Johnson JM, Rowell LB, Niederberger M, and EismanMM. Human splanchnic and forearm vasoconstrictor responsesto reductions of right atrial and aortic pressures. Circ Res 34:515–524, 1974.

11. Jong-de Vos V, Wieling W, Johannes JM, Harms MP, KuisW, and Wesseling KH. Incidence and hemodynamic character-istics of near-fainting in healthy 6- to 16-year old subjects. J AmColl Cardiol 25: 1615–1621, 1995.

12. Katona PG, Poitras JW, Barnett GO, and Terry BS. Cardiacvagal efferent activity and heart period in the carotid sinusreflex. Am J Physiol 218: 1030–1037, 1970.

13. Kay SM. Modern Spectral Estimation. Englewood Cliffs, NJ:Prentice-Hall, 1988.

14. Liu HK, Guild SJ, Ringwood JV, Barrett CJ, Leonard BL,Nguang SK, Navakatikyan MA, and Malpas SC. Dynamicbaroreflex regulation of blood pressure control: influence of theheart vs. peripheral resistance. Am J Physiol Regul Integr CompPhysiol 283: R533–R542, 2002.

15. Mark AL and Mancia G. Cardiopulmonary baroreflexes inhumans. In: Handbook of Physiology. The Cardiovascular Sys-tem. Peripheral Circulation and Organ Blood Flow. Bethesda,MD: Am. Physiol. Soc., 1983, sect. 2, vol. III, pt. 2, chapt. 21, p.795–813.

16. Parati G, DiRienzo M, Bertinieri G, Pomidossi G, CasadeiR, Groppelli A, Pedotti A, Zanchetti A, and Mancia G.Evaluation of the baroreceptor-heart rate reflex by 24-hour in-tra-arterial blood pressure monitoring in humans. Hypertension12: 214–222, 1988.

17. Pawelczyk JA and Levine BD. Cardiovascular responses torapid volume infusion: the human Bainbridge reflex. Circulation92: I-656, 1995.

18. Pawelczyk JA and Raven PB. Reductions in central venouspressure improve carotid baroreflex responses in conscious men.Am J Physiol Heart Circ Physiol 257: H1389–H1395, 1989.

19. Saul JP. Beat-to-beat variations of heart rate reflect modulationof cardiac autonomic outflow. News Physiol Sci 5: 32–37, 1990.

20. Saul JP, Berger RD, Albrecht P, Stein SP, Chen MH, andCohen RJ. Transfer function analysis of the circulation: uniqueinsights into cardiovascular regulation. Am J Physiol Heart CircPhysiol 261: H1231–H1245, 1991.

21. Saul JP, Berger RD, Chen MH, and Cohen RJ. Transferfunction analysis of autonomic regulation. II. Respiratory sinusarrhythmia. Am J Physiol Heart Circ Physiol 256: H153–H161,1989.

22. Sundlof G and Wallin BG. Effect of lower body negativepressure on human muscle nerve sympathetic activity. J Physiol278: 525–532, 1978.

23. Taylor JA, Halliwill JR, Brown TE, Hayano J, and Eck-berg DL. Non-hypotensive hypovolaemia reduces ascendingaortic dimensions in humans. J Physiol 483: 289–298, 1995.

24. Triedman JK and Saul JP. Blood pressure modulation bycentral venous pressure and respiration. Buffering effects of theheart rate reflexes. Circulation 89: 169–179, 1994.

25. Triedman JK, Cohen RJ, and Saul JP. Mild hypovolemicstress alters autonomic modulation of heart rate. Hypertension21: 236–247, 1993.

26. Van Vliet BN, Belforti F, and Montani JP. Baroreflex stabi-lization of the double (pressure-rate) product at 0.05 Hz inconscious rabbits. Am J Physiol Regul Integr Comp Physiol 282:R1746–R1753, 2002.

27. Vatner SF, Boettcher DH, Heyndrickx GR, and McRitchieRJ. Reduced baroreflex sensitivity with volume loading in con-scious dogs. Circ Res 37: 236–242, 1975.

28. West JB. Mechanics of breathing. In: Respiratory Physiology,edited by West JB. Baltimore, MD: Williams & Wilkins, 1979, p.86–113.

R1220 HR CONTROL AND CARDIOPULMONARY COUPLING

AJP-Regul Integr Comp Physiol • VOL 283 • NOVEMBER 2002 • www.ajpregu.org

on March 12, 2008

ajpregu.physiology.orgD

ownloaded from