Stptg 4

1652

Noninvasive Determination of Age-RelatedChanges in the Human Arterial Pulse

R. Kelly, MB, FRACP, C. Hayward, MB, BSc,

A. Avolio, PhD, and M. O'Rourke, MD, FACC

Arterial pressure waves were recorded noninvasively from the carotid, radial, femoral, or allthree of these arteries of 1,005 normal subjects, aged 2-91 years, using a new transcutaneoustonometer containing a high fidelity Millar micromanometer. Waves were ensemble-averagedinto age-decade groups. Characteristic changes were noted with increasing age. In all sites,pulse amplitude increased with advancing age (carotid, 91.3%; radial 67.5%; femoral, 50.1%from first to eighth decade), diastolic decay steepened, and diastolic waves became lessprominent. In the carotid pulse, there was, in youth, a second peak on the downstroke of thewaves in late systole. After the third decade, this second peak rose with age to merge with anddominate the initial rise. In the radial pulse, a late systolic wave was also apparent, but thisoccurred later; with age, this second peak rose but not above the initial rise in early systole, even

at the eighth decade. In the femoral artery, there was a single systolic wave at all ages. Agingchanges in the arterial pulse are explicable on the basis of both an increase in arterial stiffnesswith increased pulse-wave velocity and progressively earlier wave reflection. These two factorsmay be separated and effects of the latter measured from pressure wave-contour analysis usingan "augmentation index," determined by a computer algorithm developed from invasivepressure and flow data. Changes in peak pressure in the central (carotid) artery show increasingcardiac afterload with increasing age in a normal population; this can account for the cardiachypertrophy that occurs with advancing age (even as other organs atrophy) and the predispo-sition to cardiac failure in the elderly. Identification of mechanisms responsible offers a newapproach to reduction of left ventricular afterload. (Circulation 1989;80:1652-1659)

A lthough the arterial pulse is the most funda-mental of physical signs and has been usedby clinicians for hundreds of years, it was

not until the last century that the contour of thepressure pulse was first recorded noninvasively inhumans. The sphygmograph introduced by Mareyin 18601 was sufficiently simple to be applied inclinical practice; this was done by Mahomed,2 who,utilizing the instrument, first described the clinicalentity of essential hypertension, and by Mackenzie,3who described changes in both pulse contour andrhythm observed in his general practice. This earlywork on pulse-contour recordings was soon eclipsedby the introduction of the now ubiquitous sphygmo-

From St. Vincent's Hospital (M.F.O'R.), Concord Hospital(C.H.), and University of New South Wales (A.A., M.F.O'R.),Sydney, Australia.R.K. was a Scholar of the National Health and Medical

Research Council of Australia and is now Overseas ResearchFellow with the National Heart Foundation of Australia at TheJohns Hopkins University, Baltimore, Maryland.Address for correspondence: Michael F. O'Rourke, Medical

Professorial Unit, St. Vincent's Hospital, Victoria Street, Dar-linghurst, Sydney 2010, Australia.Received October 27, 1988; revision accepted July 20, 1989.

manometer. Despite classic texts on the pressurepulse contour by Mackenzie3 and by Wiggers,4written in the early twentieth century, the sphyg-momanometer was readily embraced as being more,"scientific" because it was able to quantify bloodpressure in terms of systolic and diastolic numbers.This ascendancy of sphygmomanometric values wasalso due to an absence of theory to describe orinterpret the pulse contour, an inability to use pulserecordings usefully in clinical practice, and to prob-lems with artifacts inherent in the available mechan-ical recording systems. Thus, the use of the sphyg-mograph in describing pressure-pulse wavesdeclined, even as Einthoven's electrocardiographwas being quickly accepted for its description ofwaves of electric activity.With subsequent clinical use of the sphygmoma-

nometer, brachial arterial pressure has beendescribed in terms of the two extremes betweenwhich it fluctuates, the systolic and diastolic pres-sures, and aging changes characterized as an increasein systolic pressure with little change in diastolicpressure within the brachial artery.5 Furthermore, ithas been assumed that such change is the same in

Kelly et al Age-Related Pulse Changes 1653

APPLANATION TONOMETRY

FIGURE 1. Diagram of applanation tonometry process.

Flattening ofcurved pressure containing structure allows

accurate registration of transmitted pressure, because

stresses inherent in wall of curved surface are balanced

when it is flattened.

all arteries. The effects of advancing age on thearterial pulse contour, however, have not beenprecisely defined.The aim of this study was to document age-related

changes in the arterial pulse contour, both in thecentral and peripheral arteries, using an instrumentthat would noninvasively register pressure wave

contour without artifact and with such accuracy thatcharacteristics of the contour could be clearly andquantitatively defined, much as we can now clearlydefine features of electrocardiographic waves. Suchprecisely documented changes in the normal pulsewith age will allow valid determination of the effectsof disease states on the pressure wave contour.

MethodsThe InstrumentTo document aging changes in pulse contour in

large numbers of the general population, an accu-

rate noninvasive method of measuring pressure-pulse waves was needed. The instrument devised isa pencil-shaped probe held on the skin over themaximal arterial pulsation either by hand or by a

micromanipulator. The probe incorporates a Millarmicromanometer in its tip and has the same highfidelity response as conventional Millar catheters.6Use of the probes is based on the principle ofapplanation tonometry, as is used in ocular tonom-etry for registration of intraocular pressure. Essen-tially, if one can flatten or applanate (Figure 1) thecurved surface of a pressure-containing structure,then the circumferential stresses in the wall of thestructure are balanced and the pressure registeredby the sensor is the true intra-arterial pressure.7'8Like echocardiography and other noninvasive diag-nostic techniques, applanation tonometry requiressome training but readily can be mastered. Havinglocated the point of maximal arterial pulsation, theprobe is placed over the vessel and pressed downon the artery against underlying bone. If no flat-tening is achieved, no consistent signal can beregistered. Excessive flattening produces a dis-

torted signal. Recordings are taken only when areproducible signal is obtained with high amplitudeexcursion. Large amounts of overlying tissue makeoptimal waveforms harder to obtain; the techniqueis best applicable to superficial, large arteries suchas the radial, carotid, and femoral. The sensor is astiff ceramic beam on which are mounted piezo-resistive elements forming two arms of a Wheat-stone bridge. The remaining two arms are housed inthe connector. This is attached to a standard Millarpreamplifier box that amplifies the induced voltagechange by a factor of 100. This amplified, electri-cally calibrated signal was then recorded on mag-netic tape. The accuracy of the probe has beenpreviously validated in animals and in human sub-jects with indwelling radial artery lines.9 Compari-sons were made between directly and indirectlyrecorded radial artery pressure waves in 62 humans,between directly and indirectly recorded femoralpressure waves in three dogs, and between directlyrecorded ascending aortic pressure waves and indi-rectly recorded carotid pressure waves in 17 humans.Comparison both in the time domain and by spec-tral analysis show excellent correspondence withtheoretic predictions. The tonometer provides ahigh fidelity recording of arterial pressure wavecontour under a wide variety of clinical conditionsand pulse pressures. The stringent requirementsdelineated by more recent theoretical studies8 applyto internally calibrated systems attempting to obtainaccurate noninvasive measurements of absolute sys-tolic and diastolic pressure levels. When, however,applanation tonometry is used to record pressurewave contour to supplement sphygmomanometricreadings, these exacting conditions are not neces-sary. Accurate recordings of wave contour canindeed be achieved under standard clinical condi-tions.The tonometer voltage signal registers a pressure

wave with harmonic content that does not signifi-cantly differ from that of an intra-arterially recordedwave.9 The output of the Millar preamplifier boxelectrically calibrates the signal in mm Hg (1 mV/mm Hg) in the same way as the conventional Millarunit. The use of the tonometer on an exposed vesselindeed records a waveform of amplitude virtuallyidentical to that recorded intra-arterially.9 Percuta-neous use of the probe records morphological fea-tures of the wave that are accurately reproducedover a wide range of pulse pressures9 althoughabsolute pulse pressure recorded might be lessreliable. Accordingly, the output of the instrumentis reported in this manuscript with an arbitraryamplitude scale of millivolt (mV) units.

Registration and AnalysisWaveforms were recorded on an FM TEAC

magnetic tape recorder, digitized by a 12-bit analogto digital converter with an acquisition rate of 1 kHzand entered into storage on disk by an IBM-ATcomputer. For each subject studied, six to eight

1654 Circulation Vol 80, No 6, December 1989

A1 4.

PP

AUGMENTATION INDEX = APIPP

FIGURE 2. Graphic representation ofaugmentation indexdefined as ratio of AP/PP%. AP, pressure from shoulderto late peak; PP, pulse pressure.

consecutively recorded pulses were averaged, theiralignment being triggered at the occurrence of themaximum rise time determined by the smoothedfirst differential of the signal (dP/dt). The averagedpulse was then taken as the representative pulse forthat individual. An ensemble average of similarrepresentative waves from 30-70 other subjects inthe same age decade was obtained to form a repre-sentative wave and 2 SDs for that particular agedecade. Thus, a series of decade waves was estab-lished, each being the average of approximately 50representative waveforms from individuals of simi-lar age. Subjects with heart rates greater than 100beats/min were excluded to avoid excessive trunca-tion of the age-decade wave (to that of the subjectwith the shortest heart period) during the averagingprocess.

Carotid waveforms were further analyzed to mea-sure the "shoulder" and "peak" of the waves.Such analysis of wave contour has been donepreviously by Murgo et al0 and Takazawa"l forinvasively recorded ascending aortic waves and,also, by Fujii et a112 for noninvasively recordedwaves. We defined the augmentation index for eachwave as the ratio of height of the peak above theshoulder of the wave to the pulse pressure (Figure2). This can be done from visual inspection as hasbeen used previously.10-12 Such identification, how-ever, can be highly subjective because the shoulderis often not a clearly defined point but sometimes aless well defined plateau region on the systolicupstroke. We, therefore, chose to identify the shoul-der automatically from the time derivative of thepressure wave, as herein described.

Measurement and Justification ofAugmentation IndexFrom invasive ascending aortic pressure and flow

velocity data previously reported,'3 we have notedthere exists a correspondence between the peak offlow and the shoulder of pressure. The shoulder ofthe pressure wave is defined as the first concavityon the upstroke of the wave and separates the initialpressure rise from the late systolic peak that occursin middle aged and older subjects.10 A computeralgorithm was written to identify this shoulder in amore objective, automated way than visual inspec-tion by using time derivatives of the pressure wave.Reanalysis of the invasive data was performed toascertain the relation between the shoulder of the

10m:mHg]

10cm/sec]-_ j

FIGURE 3. Graphic plotting ofrepresentative ascendingaortic pressure (solid line above) and flow (dashed lineabove). Below is shown fourth derivative of pressure(solid line). First zero crossing offourth derivative corre-sponds to beginning of pressure wave upstroke (AA).Second zero crossing in same direction (from above tobelow) corresponds to beginning of second wave, whichis close to peak offlow (BB).

ascending aortic pressure wave, the time deriva-tives of pressure, and the peak of simultaneouslyrecorded flow. The simultaneously recorded pres-sure and flow-velocity data were digitized at 0.01-second intervals and were plotted with the first fourderivatives of pressure. When plotted against pres-sure derivatives, the timing of the shoulder wasreliably indicated by a local minimum in the firstderivative that was in the range from 0 to 50 msec ofthe peak of flow (the mean estimate from interpola-tion between sample points determined from 13patients) being 27 msec once correction had beenmade for the frequency response of the flowmeter(10 msec). To simplify the algorithm for detection ofthis point, higher-order derivatives were used toidentify the zero-crossing point equivalent to thelocal minimum of the first derivative. The first zerocrossing of the fourth derivative (in a direction fromabove to below the zero line) corresponded to thebeginning of the pressure wave upstroke (line AA,Figure 3). The second zero crossing in the samedirection corresponded to the shoulder at the begin-ning of the second wave, which constituted the latesystolic peak (line BB, Figure 3). A good correla-tion was found between the time to the second zerocrossing of the fourth derivative (x) and the timingof the peak of flow (y) in the patients studied(y=0.91+1.31x; R=0.75).Such a relation between peak flow and the shoul-

der of the pressure wave is to be expected if theshoulder of the wave indicates the pressure riseresulting from peak flow input into the vasculaturebefore the effects of wave reflection.10"13,'4 Hence,we decided to use the algorithm as described to

Kelly et al Age-Related Pulse Changes 1655

RADIAL PULSE CONTOUR CAROTID PULSE CONTOUR

I 10 mV UNITS' DECADE I10 mv UNfTS

2

3

4.

5

6

7

a

0 1 50 300 450 600 750

TIME(msec)

FIGURE 4. Contours showing averaged radial wavesfrom 420 subjects. Radial waveforms are displayed aboveeach other from infants in the first decade (1), throughchildren, to adults up to the eighth decade (8). Each pulseis average of 40-70 individual pulses. *Amplitude isexpressed in uncalibrated voltage (mV) units althoughthe Millar preamplifier box electrically calibrates theoutput signal in mm Hg.

detect the contribution of wave reflection to chang-ing carotid pulse contour. The method used is auto-matic and excludes subjective bias. It identifies apoint on the pressure wave very close to peak flow inthe ascending aorta, and is justifiable on the theoret-ical basis of wave travel and reflection in the aorta.

SubjectsVolunteers studied were from the out-of-hospital

community-based population considered to repre-sent the normal population. Volunteers were Cau-casians from both urban and rural areas who were2-91 years old. They were screened on the basis ofcardiovascular history and examination; those withvalvular heart disease or chronically treated cardio-vascular disease were excluded. Subjects found tohave mild hypertension (diastolic blood pressure>95 and <105 mm Hg), 4.8% of the sample, wereincluded in analysis as part of the spectrum in thegeneral population. Radial pressure waveforms wereobtained from 420 subjects (207 men and 213 wom-en), of whom 38 were smokers; carotid waveformswere from a further 407 subjects (181 men and 226women), of whom 82 were smokers; and femoralwaveforms were from 178 subjects of similar age(110 men and 68 women), of whom five weresmokers. Of all the subjects studied, 28 were onmonotherapy for hypertension at the time of study.

Statistical AnalysisStatistical analysis was performed using a two-tail

unpaired Student's t test with a significant differencebeing at the level ofp equaling less than 0.05.

0 1 50 300 450 600 7S0

TIME(msec)

FIGURE 5. Contours showing averaged carotid waves

from 407 subjects. Carotid pulse contourfrom first dec-ade (1) to eighth decade (8), each pulse being average of40-70 individual pulses. *Amplitude is expressed in mVunits.

ResultsThe radial pulse contour recorded in 420 subjects

showed characteristic changes with increasing age

(Figure 4). The radial pulse contour in childrenshows multiple prominent fluctuations. With advanc-ing age, these become less distinct and the systolicpeaks progressively broader, although the maxi-mum still usually occurs in early systole. In con-trast, the carotid wave recorded in 407 subjectsshows not so much a broadening but the emergenceof the late systolic peak, which determines thesystolic pressure level (Figure 5). With increasingaging, the principal change in the carotid pulse is aprogressive rise in the second systolic peak that,after the third decade, merges with and dominatesthe initial rise. The femoral waveforms (Figure 6)also show a progressive rise in the systolic wave

and loss of any diastolic wave.The amplitude (pulse pressure) of the carotid

wave increases by 91.3% from the first to eighthdecade compared with a 67.5% increase in theradial pulse and a 50.1% increase in the femoralpulse. This greater increase is due to change in twoparts of the carotid wave. After the third decade oflife, a late systolic peak becomes dominant, whichadds to the initial pressure rise and so defines theshoulder on the upstroke of the wave. The total 91%increase in carotid pulse pressure from the first tothe eighth decade is due to an increase in both therise to this initial shoulder (by an average 20 mVunits or 53% of the increase) and the height of thelate peak above the shoulder (averaging 18 mVunits) (Table 1). This change in carotid pulse con-tour was quantified by the augmentation index

DECADE

2

3

4

5

6

7

8

1656 Circulation Vol 80, No 6, December 1989

FEMORAL PULSE CONTOUR

DECADE

2

3

4

5

6

8

0 1 50 300 450 600 750

TIME(msec)

FIGURE 6. Contours showing averaged waves from 178subjects. Femoral pulse contour from first decade (1) toeighth decade (8). *Amplitude is expressed in mV units.

measured by the algorithm developed using inva-sive aortic pressure and flow data.When this algorithm was applied to the carotid

wave age-decade data, it was found that althoughthe timing of the shoulder was within 14 msec for alleight decades (range, 102-116 msec, after the onsetof systole), the height of the shoulder and the latepeak both showed a substantial increase with age.(Table 1). The height of the shoulder increased from39 to 59 mV units above diastolic pressure from thefirst to the eighth decade, whereas the late peakincreased from less than 1 to 19 mV units above theshoulder. The corresponding augmentation indexincreased from 1.6% to 24.1%.Changes in pressure-pulse contour are due to

changes in the systemic vasculature itself or todiffering flow input into the vasculature from theheart. Previous studies have shown that at restthere is little change in stroke volume or ejectionfraction with increasing age.15 Aortic flow-velocityprofile does not alter in contour with increasing age,

TABLE 1. Carotid Waveform Analysis

Height ofHeight of peak Augmen-

Time to shoulder above tationAge shoulder above foot shoulder index(yr) (msec) (mV units) (mV units) (%)1-10 116 39.4 0.6 1.6

11-20 116 50.1 1.6 3.021-30 110 51.3 2.7 4.931-40 110 46.5 3.7 7.441-50 102 42.6 8.4 16.551-60 102 43.0 11.1 20.561-70 102 47.1 14.9 24.171+ 106 59.0 18.7 24.1

TABLE 2. Ascending Aortic Blood Flow Velocity Profile

Age Age<30 yr >60 yr p

Average peak flowvelocity (cm/sec) 94+10 67+13 0.01

Average acceleration(cm/sec2) 921±+ 113 650+280 0.05

Mean ejection time(msec) 290+30 293+38 NS

Values are mean+one SD.

although rate of flow acceleration and peak flowvelocity are reported to decline with increasingage.16 We sought to verify that our study populationconcurred with these results by studying a sampleof subjects using Doppler flow techniques. Fromthe original cohort, a group of 12 older (>60 years)and 13 younger (<30 years) subjects were studied.Parameters compared were peak flow velocity (cm/sec), acceleration of systolic upstroke, and ejectiontime. This group spanned a wider range of age(19-80 years) than the data previously reported byNichols et al.14 Ultrasonography was done with aToshiba Sonolayer SSH 65 A phased array ultra-sonograph and a 2.5-MHz pulsed-wave Dopplerprobe. Recordings were made with subjects in asupine, rested position. Recordings of ascendingaortic flow were taken from the suprasternal notch(where sampling was done near the inner curvatureat the aortic root), the right parasternal, and theapical positions. Signals analyzed were those thatgave the best delineated velocity profile at any ofthe three positions.As has been found by other workers, ejection

time remained constant with increasing age, althoughpeak flow velocity and acceleration did decline withincreasing age (Table 2). As will be discussed, thechanges in pressure wave contour with increasingage are explicable mainly on the basis of changes inthe systemic vasculature with increasing age becausethere were few changes in flow input into thesystem as detected by ascending Doppler flow mea-surements (Figure 7).

DiscussionThese characteristic age-decade pulse waveforms

represent the largest sample of noninvasive record-ings of the pulse yet made in human subjects. Theaccuracy of the recordings is dependent on both theprinciple of applanation tonometry, which is basedon well established theory,8 and also the use of highfidelity Millar micromanometers incorporated intothe tip of the probes. The technique has been shownto be accurate in registering intra-arterial pressurewaves as compared in the time domain and byspectral analysis.9 Previous noninvasive pulserecordings have used wave tracings obtained bydisplacement of a volume of air or fluid contained ina capsule fixed over the arterial pulsation.'7,18 Suchvolume displacement techniques are limited by

Kelly et al Age-Related Pulse Changes 1657

100

U

c-

40 mVT 80 mV

wL UNITS1 UNITS'a. ' .

YOUNG OLDFIGURE 7. Upper: Ascending aortic blood flow velocityprofile recorded by Doppler ultrasonography in 19-year-old male (young) and 72-year-old male (old). Lower:Carotidpressurepulse contour recorded by arterial tonom-etry in same individuals. With increasing age, there ismarked increase in pulse pressure, mostly due to latesystolic peak, with concomitant lesser reduction in peakflow velocity. *Amplitude is expressed in mV units.

movement artifact, by errors due to the large size ofthe capsule relative to the size of the artery, byinability to separate carotid and venous pulsations,by need for a significant "hold-down" force foroptimal signal registration, and by a high dampingcoefficient causing attenuation of high frequencycharacteristics. Nevertheless, these techniques havegiven valuable information and extended the earlierwork of Marey,' Mahomed,2 and Mackenzie.3 Incontrast to such capsules that rely on magnificationof vessel wall displacement, tonometry is not depen-dent on volume changes. It requires minimal hold-down force and is much less subject to artifactrelated to placement over the pulse because themicromanometer in the tip of the probe is verysmall (0.5 x 1.0 mm, o.d.) compared to the diameterof the artery. Tonometry, thus, registers the arterialpressure pulse much as the clinician feels it bypalpation.The changes in the femoral pulse with age are

characterized by a progressive increase in the sizeof the systolic peak with a resultant increase inpulse pressure, whereas diastolic fluctuations dimin-ish. There are no secondary systolic fluctuations inthe femoral pulse. In contrast, the carotid and radialwaves show an increase in pulse amplitude withage, as well as important changes in the contour ofsecondary systolic fluctuations. The averaging pro-

cess used to obtain age-decade waveforms resultedin obvious smoothing of high frequency compo-nents of the wave such as the incisura. Never-theless, there is still a clear change in pressure wavecontour with increasing age. There is a progressiverise in a late systolic fluctuation in both so that itbecomes the late peak of pressure in the carotidartery after the third decade, and that becomesequal to the initial pressure peak in the radial pulseat the eighth decade. The result is a greater increasein carotid pulse pressure with increasing age than ineither of the peripheral pulses because, in thecarotid, the late peak is added to the initial pressurerise. The rise in this late systolic fluctuation in boththe radial and carotid pulses is explicably due tofaster return of wave reflection from the lowerbody. Such altered timing of wave reflection hasbeen discussed at length elsewhere and is largelyresponsible for the suboptimal coupling of the vas-culature to left ventricular function that is known tooccur with increasing age.14As aging occurs, there are dramatic changes in

the vasculature that increase cardiac afterload.14,19Both nonpulsatile (peripheral resistance) and pulsa-tile (characteristic impedance and wave reflections)components of arterial load increase with increasingage. The age-related increase in left ventricularpressure load is primarily due to increased aorticstiffness (characteristic impedance) that increasesaortic and left ventricular pressure at its earlysystolic peak or shoulder and secondarily due toaugmented early wave reflection that raises pres-sure from the shoulder to the late systolic peak.10"4The results in Table 1 show these two effects on thecarotid pressure pulse as they increase from the firstcompared with the eighth decade. In the interveningyears, however, the increase in pulse pressure withincreasing age is primarily due to an increase inwave reflection (height of the late systolic peak).Although the 20-mV-units increase in shoulder heightacross the full age spectrum is consistent with theknown increase in aortic stiffness of large arterieswith increasing age, its constancy in middle agemay be explained by the concomitant decrease inpeak velocity that occurs with increasing age.'6Although, in middle years the progressive increasein arterial stiffness is balanced by a gradual decreasein peak flow from the heart, with the passage oftime, the degenerative vascular changes are greaterand, by the seventh and eighth decades, there is anincreased characteristic impedance and consequentincreased height of the shoulder above diastolicpressure. This study shows that, with increasingage, increased aortic stiffness and increased earlywave reflection contribute approximately equally toventricular load as assessed by the central pressurepulse (20 versus 18 mV units).The increase in the late systolic pressure peak

with increasing age is substantial and represents anincrease of 25% of pulse pressure in the carotidartery between the ages of 30 and 60 years (Figure

1658 Circulation Vol 80, No 6, December 1989

55 65 75+

index for carotid

8). This may be compared with previous data ofMurgo et al10 and Takazawa"l for invasive data fromthe ascending aorta. These two studies show a 40%increase in pulse pressure between the ages of 30and 60 years (Figure 9). Comparison of Figures 8and 9 shows that although the carotid pressurepulse tends to underestimate the magnitude of age-related changes, the carotid pulse detected by appla-nation tonometry is a good noninvasive guide toevents in the central aorta.The relevance of these pressure changes to left

ventricular afterload is apparent when one consid-ers known changes in left ventricular structure andfunction with increasing age. The increase in thelate systolic peak in the carotid pulse occurs fromthe third decade onward, which is the same time atwhich the age-related increase in left ventricularmass commences.20,21 This increase in mass withincreasing age is accompanied by changes in leftventricular function.22 Furthermore, recent studieshave shown that the increased ventricular load andsuboptimal ventricular-vascular coupling repre-sented by these pressure-pulse changes are, at least

AUGMENTATION INDEX - ASCENDING AORTA80

60

J 40 r

20zWccW' 0-

-20v TWZAWA

-40*0 10 20 30 40 50 60 70 80

AGE

FIGURE 9. Scatterplot ofaugmentation indexfor ascend-ing aorta as function of age; from data published byMurgo et a110 (n) and Takazawa" (A).

in part, reversible by the use of nitroglycerin, whichreduces the reflected wave component.13,23-24These characteristic age-related changes in the

pulse waveform have established a profile of normal-aging changes that will allow the effects of diseasestates to be more precisely defined. The increase insystolic and pulse pressure, particularly in the cen-tral carotid pulse, have important implications forthe aging population. Increase in peak systolicpressure and increased pulsation around the meanpressure level result in increased cyclic stresses onthe structure of arterial walls and account for thepropensity of large arteries to rupture or developatheromatous occlusive disease with increasingage.25 The findings in this study concur with previ-ous epidemiologic data, showing the importance ofsystolic pressure as a risk factor for cardiovascularmorbidity and mortality with advancing age.26'27 Inaddition, these findings provide important answersto the question posed by the most recent of thesestudies,28 regarding the determinants of systolicpressure levels in both central and peripheralarteries.

Furthermore, the fact that the radial pulse showsdifferent but complementary changes to those in thecarotid pulse has important implications for diag-nostic approaches and treatment. It is clear that theincrease in radial (and femoral) systolic pressuresunderestimate the age-related rise in central systolicpressure because it does not include the character-istic rise in the late systolic peak seen in the centralarteries of mature adults. Hence, in subjects aged40-70 years, measurements of sphygmomanometricpressure might substantially underestimate the effectof vasoactive drugs on ventricular afterload becausea reduction in the late carotid pressure peak mightoccur with little or no change in peripheral systolicpressure. This is because, in the peripheral pulse,the reflected wave is on the downstroke of systolerather than at its peak and, although reduced byvasoactive drugs, it has little effect on peak systolicpressure. On the other hand, further investigationis warranted to see if a predictive index of centralsystolic pressure changes might be obtained fromanalysis of the late systolic fluctuation in theperipheral (radial) pressure pulse waveform.The determination of age-related changes in the

arterial pulse by high fidelity applanation tonome-try, thus, provides important supplementary infor-mation to that obtained by use of the sphygmoma-nometer. The use of this modern-day sphygmographenhances new investigations of the ill effects ofaging and disease states on cardiovascular functionand extends the earlier observations of Mahomed,2Mackenzie,3 and Freis et al.17

AcknowledgmentsWe acknowledge J. Ganis, BSc, and J. Daley,

RN, for assistance in data collection and prepara-tion of this manuscript.

AUGMENTATION INDEX - CAROTID ARTERY

60

40 -

20 -

-

zW0cc0.

5 1 5 25 35 45

AGE

FIGURE 8. Graph of augmentationartery as function of age.

Kelly et al Age-Related Pulse Changes 1659

The arterial tonometer was developed with theassistance of Mr. Huntley Millar, Houston, and Dr.D. Winter, South-West Research Institute, SanAntonio, Texas.

References1. Marey EJ: Recherches sur le Pouls au Moyen d'un Nouvel

Apareil Enregistreur-le Sphygmographe. Paris, E. Thunotet Cie, 1860

2. Mahomed F: The aetiology of Bright's disease and theprealbuminuric stage. Med Chir Trans 1874;57:197-228

3. Mackenzie J: The Study of the Pulse. Edinburgh, McMillan,1902, pp 8-28, 67-82

4. Wiggers CJ: The Pressure Pulses in the CardiovascularSystem. London, Longmans, Green and Co, 1928, pp 65-90

5. Roberts J (ed): Blood Pressure Levels of Persons 6-74Years. United States, 1971-1974

6. Murgo JP, Giolma JP, Altobelli SA: Signal acquisition andprocessing for human haemodynamic research. Proc IEEE1977;65:696-702

7. Mackay RS, Marg E, Oechsli R: Automatic tonometer withexact theory: Various biological applications. Science 1960;131: 1688-1689

8. Drzewiecki GM, Melbin J, Noordergraaf A: Deformationalforces in arterial tonometry. IEEE Front Eng Comput HealthCare 1984;28:642-645

9. Kelly R, Hayward C, Ganis J, Daley J, Avolio A, O'RourkeM: Non-invasive registration of the arterial pressure pulsewaveform using high-fidelity applanation tonometry. J VascMed Biol 1989;3:142-149

10. Murgo JP, Westerhof N, Giolma JP, Altobelli SA: Aorticinput impedance in normal man: Relationship to pressurewaveforms. Circulation 1980;62:105-116

11. Takazawa K: A clinical study of the second component ofleft ventricular systolic pressure. J Tokyo Med Coll 1987;45:256-270

12. Fujii M, Yaginuma T, Takazawa K, Noda T, Komatsu H,Katsui T, Watabiki H, Kawada Y, Hosoda S: Non-invasivedetection for reflection wave in the arterial system, inAutomedica vol. 9. New York, Gordon & Breach, 1987, p 49

13. Yaginuma T, Avolio AP, O'Rourke MF, Nichols WW,Morgan JP, Roy P, Baron D, Branson J, Feneley M: Effectsof glyceryl trinitrate on peripheral arteries alters left ventric-ular hydraulic load in man. Cardiovasc Res 1986;20:153-160

14. Nichols W, O'Rourke M, Avolio A, Yaginuma T, Murgo J,Pepine CJ, Conti CR: Effects of age on ventricular-vascularcoupling. Am J Cardiol 1985;55:1179-1184

15. Weisfeldt M, Lakatta EG, Gerstenblith G: Aging and cardiacdisease, in Braunwald E (ed): A Textbook of CardiovascularMedicine. Philadelphia, Saunders, 1988, pp 1650-1662

16. Gardin JM, Burn CS, Childs WJ, Henry WL: Evaluation ofblood flow velocity in the ascending aorta and main pulmo-nary artery of normal subjects by Doppler echocardiogra-phy. Am Heart J 1984;107:310-319

17. Freis ED, Heath WC, Luchsinger PC, Snell RE: Change inthe carotid pulse which occur with age and hypertension.Am Heart J 1966;71:757-765

18. Craige E, Smith D: Heart sounds: Phonocardiographs;carotid, apex and jugular venous pulse tracing; and systolictime intervals, in Braunwald E (ed): A Textbook of Cardio-vascular Medicine. Philadelphia, Saunders, 1988, pp 41-64

19. Yin FCP: Aging and vascular impedance, in Yin FCP (ed):Ventricular/Vascular Coupling. New York, Springer-Verlag, 1988, pp 115-137

20. Linzbach AJ, Akuomoa-Boateng E: Alternsveranderungendes menschlichen Herzens. 1. Das Herzgewicht in Alter.Klin Wochenschr 1973;51:156-163

21. Gerstenblith G, Fredericksen J, Yin FCP, Fortuin NJ,Lakatta EG, Weisfeldt ML: Echocardiographic assessmentof a normal adult aging population. Circulation 1977;56:273-278

22. Nichols WW, O'Rourke MF, Avolio AP, Yaginuma T,Murgo JP, Pepine CJ, Conti CR: Age-related changes in leftventricular/arterial coupling, in Yin FCP (ed): Ventricular!Vascular Coupling. New York, Springer-Verlag, 1988, pp79-114

23. Latson TW, Hunter WC, Katoh N, Sagawa K: Effect ofnitroglycerin on aortic impedance, diameter and pulse wavevelocity. Circ Res 1988;62:884-890

24. Fitchett D, Simkus G, Beaudiy J, Marpole D: Reflectedpressure waves in the ascending aorta: Effect of glyceryltrinitrate. Cardiovasc Res 1988;22:494-500

25. O'Rourke MF: Hypertension, in O'Rourke MF (ed): ArterialFunction in Health and Disease. Edinburgh, Churchill Liv-ingstone, 1982, pp 185-195, 210-224

26. Kannel WB, Castelli WP, McNamara PM, McKee PA,Feinlab M: Role of blood pressure in the development ofcongestive heart failure. N Engl J Med 1972;287:781-787

27. Kannel WB, Wolf PA, McGee DL, Dawber TR, McNamaraP, Castelli WP: Systolic blood pressure, arterial rigidity andrisk of stroke. The Framingham Study. JAMA 1981;245:1225-1229

28. Rutan GH, Kuller LH, M Neaton JD, Wentworth DN,McDonald RH, McFate-Smith W: Mortality associated withdiastolic hypertension and isolated systolic hypertensionamong men screened for Multiple Risk Factor InterventionTrial. Circulation 1988;77:504-514

KEY WORDS * applanation tonometry * arteriosclerosis .

wave reflection