in the Right Ventricle and Pulmonary...1.5 inch NaI (T1) crystal (The Harshaw Chemical Co., Cleveland, Ohio) with a 1/8 inch cylindrical hole through its center. The pulses from the

Mixing in the Right Ventricle and Pulmonary

Artery in Man: Evaluation of Ventricular Volume

Measurements from Indicator Washout Curves

Arriuo MASERI and YALE ENSON

From the Cardiopulmonary Laboratory of the 1st (Columbia) Medical Divisionof Bellevue Hospital, and the Department of Medicine, College of Physiciansand Surgeons, Columbia University, NewYork 10016

A B S T R A C T To assess the mixing characteris-tics of the right ventricle and pulmonary artery,radioiodinated 131I serum albumin and indocya-nine green dye were injected simultaneously in16 subjects. One indicator was injected into theatrium and the other into the ventricle, or bothwere injected at different sites in the ventricle.Washout curves were obtained by rapid cathetersampling alternately just above or just below thepulmonic valve. The washout of radioisotope wasalso recorded with a precordial scintillation detec-tion probe.

Indicator washout from the ventricular inflowtract was rapid, while washout from the region ofthe ventricular apex was quite slow. Protosystolicdips in indicator concentration, noted in curvesdrawn below the pulmonic valve, suggest that theventricle emptied sequentially. Flow values com-puted from curves sampled below the valve, whencompared with reference values, suggest that a sig-nificant volume of atrial blood passed through theventricle without mixing, or mixing to only asmall extent, with the residual volume of thechamber. Peak concentration of indicator washigher below the pulmonic valve than above. Thisfinding, plus the close agreement between flowvalues computed from curves sampled above thevalve and the reference values, indicates that fur-ther mixing occurred above the valve.

Dr. Maseri's present address is Clinica Medica Uni-versita, Pisa, Italy.

Received for publication 31 July 1967 and in revisedform 27 November 1967.

Ventricular volumes computed from washoutdownslopes are systematically falsely high. Thisoverestimate appeared to be larger in normal sub-jects than in patients with low stroke volumes.Progressive mixing of blood leaving the atriumoccurs during its passage through the ventricle,pulmonic valve, and pulmonary artery and permitsaccurate estimation of flow.

INTRODUCTION

Indicator washout curves have been used exten-sively to estimate the volume of the right andleft ventricles (1-9) since the studies of Bing,Heimbecker, and Falholt (10) and of Holt (11).Although the existence of homogeneous ventricularmixing is a prerequisite condition for the validapplication of this technique, only a few descrip-tions of ventricular mixing have been reported.While the available information indicates thatmixing does not appear to be homogeneous (12-15), the significance of this observation withrespect to estimations of ventricular volume hasnot been characterized adequately. Volume maystill be estimated with validity in the absence ofperfect mixing if the inhomogeneities are randomlydistributed (16); if systematic regional differencesin mixing do occur in the ventricle, however, esti-mates of volume will be systematically in error.

Therefore, we performed the present study toexamine the conditions of mixing within the ven-tricle with particular emphasis on the possibleoccurrence of systematic regional inhomogeneities

848 The Journal of Clinical Investigation Volume 47 1968

and on the conditions of mixing beyond the semi-lunar valves (since washout curves of the indi-cator are usually recorded at this site). Theproblem was approached by studying the influenceof different injection and sampling sites on theventricular indicator washout curve. Two indi-cators (a dye and a radioactive gamma-emittingtracer) were injected simultaneously, one into theright atrium and the other into the right ventricle,or both into the ventricle at different sites. Weobtained their time-concentration curves by rapidcatheter sampling alternately just distal or justproximal to the pulmonic valve. The ventricularwashout curve of the gamma-emitting indicatorwas also recorded, at the same time, with a pre-cordial scintillation detection probe selectivelycollimated over the right ventricle.

Our findings indicate that an appreciableamount of blood entering the ventricle duringdiastole does not mix, or mixes only partially,with the residual volume and is ejected from thechamber during early systole. Blood located atthe apex of the ventricle is ejected in late systole.A considerable degree of further mixing occursbeyond the pulmonic valve.

METHODS

Patients

16 patients were selected for this study. Six had nocardiovascular abnormalities; six had chronic bronchitisand emphysema: of these, two had cor pulmonale withmoderate cardiac enlargement, three had atheroscleroticheart disease with degrees of cardiac enlargement vary-ing from slight to marked, and one had rheumatic heartdisease and had undergone mitral commissurotomy 1 yrbefore the study.

ProtocolThe subjects were studied in the nonsedated, postab-

sorptive, basal state, having received only Lugol's solu-tion in preparation for the procedure. A stethograph wasplaced around the chest to record the respiratory cycle,and the projection of the right ventricle on the anteriorchest wall was determined fluoroscopically for collimationof the scintillation detection probe. A 7 F thin-wallGoodale-Lubin catheter was inserted through a rightantecubital vein into the pulmonary artery for samplingpurposes, and a 9 F Cournand double-lumen catheter waspassed through the left arm to the right ventricular apexfor injection of the two indicators.

The injection catheter was then withdrawn about 1 cmfrom the apex so that a satisfactory pressure tracingcould be obtained from the distal lumen; in some cases it

had to be withdrawn further into the low or mid-ventriclebecause a rapid injection of saline into the distal lumenproduced a premature ventricular contraction. The ori-fice of the proximal lumen was located on the concavityof the catheter's curve and lay in the high or mid-atriumor in the ventricular inflow tract at the level of the tri-cuspid valve when the distance between the two terminalopenings was respectively 12, 8, and 4 cm. In two sub-jects two 5 F NIH catheters were used instead of thedouble-lumen. One was placed in the ventricular inflowand the other in the low or in the mid-ventricle. Thesampling catheter was next withdrawn so that its tip layjust above the pulmonic valve in 12 patients and justbelow the valve in 4.

When the catheters were in the desired positions theprecordial probe was collimated over the right ventricle,each lumen was preloaded with one of the indicators, andthe sampling catheter was connected to the sampling-detecting system. Indicator dilution curves were obtainedby flushing both lumens simultaneously with saline. Eachseries of curves consisted of the dye and radioisotopecurves obtained by catheter sampling and the precordialactivity recorded by the gamma detector probe. Succes-sive series of curves were obtained at 6-10-min intervalsat various combinations of injection and sampling sitesas indicated in Fig. 8.

Injection technique

The double-lumen catheters were 100 cm long; thevolume of each lumen ranged from 0.7 to 0.8 ml. TheNIH catheters were 100 cm long and had a volume of0.6 ml. For each set of injections about 100 Ac of radioio-dinated 131I serum albumin (RISA) (Abbott Labora-tories, North Chicago, Ill.) and 1.2 mg of indocyaninegreen dye (Cardio-Green, Hynson, Westcott & Dunning,Baltimore, Md.) were diluted up to 0.5 ml and preloadedin the two lumens by means of two semiautomatic cali-brated syringes. Serum albumin was added to the dyesolution in an amount of 0.1 g/mg of dye. The injectatewas delivered by flushing the lumens with 1.5 ml of saline.In 10 patients the lumens were flushed manually withtwo identical 2-ml calibrated syringes set to deliver 1.5ml. The onset and the duration of the injection was re-

corded by depressing a foot-switch; the duration variedbetween 0.5 and 0.7 sec. No attempt was made to syn-chronize the injections with any event of the cardiaccycle so that the timing of the manual injection was com-

pletely random. In five patients two automatic injectors(Sage Instruments, Inc.) were used for flushing. Theywere actuated by a variable delay circuit (Electronicsfor Medicine, Inc., White Plains, N. Y.) triggered fromthe electrocardiogram. With the catheters used, 1.5 mlwas delivered in 0.4 sec by both injectors. The delay cir-cuit was adjusted to deliver the injectate at the beginningof diastole. In three instances, however, the injection hadto be set at mid-systole because of the systematic occur-

rence of premature ventricular contractions with earlydiastolic injection. Only in one instance did an extra-

systole occur at the time of injection.

Right Ventricular and Pulmonary Artery Mixing 849

Sampling-detecting techniqueThe sampling catheter had two lateral holes within 5

mmof its open tip. In eight patients it was 80 cm longand the volume was 1.6 ml, and in the others it was 100cm long with a volume of 2 ml.

Before each pair of injections the sampling catheterwas connected in series to a flow-through scintillation de-tection probe (FTP), a densitometer, and a samplingsyringe. This system, developed earlier for the continu-ous, simultaneous detection of two indicators (17), com-prised a photomultiplier (Tracerlab DRL 2) and a 1.5 X1.5 inch NaI (T1) crystal (The Harshaw Chemical Co.,Cleveland, Ohio) with a 1/8 inch cylindrical hole throughits center. The pulses from the photomultiplier were fedto one channel of a Dual Rate Computer (Picker Nuclear,North Haven, Conn.). This instrument integrated thecounts during each 0.1-sec interval and provided an ana-logue output that was displayed and photographed on aDR8 recorder (Electronics for Medicine). The densitom-eter was a Gilford Model 103 IR. Dye time-concentra-tion curves were corrected for the time constant of thedensitometer (0.14 sec) by insertinga resistance capaci-tance combination between two dc amplifiers of the re-corder (18). We sampled blood manually into a 50-mlsyringe, at a rate of 2.5 ml/sec (timed with a metronome).The volume of blood collected in the syringe during thesampling period matched the amount calculated on thebasis of the sampling rate and duration of sampling esti-mated from the photographic record in all cases in whichit was checked.

The distortion introduced by the sampling-detectingsystem on a practically instantaneous input imposed atthe tip of the sampling catheter is rather small, and iden-tical for the two indicators (17).

We calibrated all RISA curves by drawing bloodthrough the FTP while recording the counting rate 6min after each injection; we determined the activity ofthe injectate and of the blood samples in a well counter.The dye curves were individually calibrated with thepooled sample method (19) in five patients. No calibra-tions were obtained in the others. All blood samples, to-gether with the standard, were processed and read in aBeckman DU Spectrophotometer within 90 min of theircollection.

Technique for obtaining precordial curves

The precordial scintillation detection probe consistedof a photomultiplier with a 2 X 2 inch Nal (T1) crystal(Omniprobe, Picker Nuclear) located in a lead collimatorwith a straight bore of the same diameter. The crystalsurface was recessed 20 cm from the external openingof the collimating channel. In order to further increasethe selectivity of the collimation, we divided the colli-mating channel into four equal segments by introducinga longitudinal cross, just in front of the crystal, madefrom two strips of lead 2 inches wide, 5 inches long, and1.5 mmthick. The contribution of scattered radiation tothe counting rate was reduced by interposing a 1-mmthick lead disk between the crystal and the lead cross.

The pulses from the photomultiplier were fed to the sec-ond channel of the Dual Rate Computer, the output ofwhich was recorded on the DR8 recorder.

The precordial counting rate, with this collimation,began to rise only 3-4 beats after an injection of RISAinto the pulmonary artery (when the indicator arrivedat the left heart). Hence, the contribution of extracardiacactivity to the precordial counting rate was negligibleduring the first 3-4 heart cycles after injection. Becausethe collimator was relatively wide (5 cm diameter), de-spite its selectivity, the precordial counting rate was in-fluenced in a compound fashion by changes in concen-tration at a constant ventricular volume and by changesin volume while the concentration of RI SA remainedconstant.

Calculations

Catheter-sampling curves. The time-concentrationcurves of dye and of RISA were plotted on semilogarith-mic paper and shifted back along the time axis by a timeequal to the mean transit time between the tip of the cath-eter and the corresponding detecting device for an im-mediate, direct comparison. For a quantitative evaluationof the observed differences in peak-concentration of thewashout curves the following ratio was calculated: I/C,= DVa (where I is the injectate and DVa is the apparentdilution volume in which the injectate would be distrib-uted at a uniform concentration equal to C,). Only whenthe injectate is evenly distributed in the end-diastolic vol-ume does DVa correspond to that volume.

Flow values (Q) were calculated from these curves bythe Stewart-Hamilton formula in order to evaluate theimportance of the observed differences in area under thesimultaneous time-concentration curves.

Precordial Curves. Cardiac output was calculatedfrom the precordial curves according to the radiocardio-graphic method of Donato et al. (20-21). In each subjectthe value of cardiac output (Qref) and of stroke volume(SVref), calculated from the first radiocardiogram(RCG) obtained after injection of RISA into the atriumor into the ventricular inflow, was used for reference.

In the curves obtained after atrial systolic or protodia-stolic injection, or ventricular inflow diastolic injection,the fraction of indicator ejected during the first systoleafter injection appeared to be much larger than thatejected during the successive ones. In order to evaluatethis difference, end-diastolic volume (EDVRCG) was cal-culated according to the method of Donato et al. (20-21):EDVRcG= SVref/Ri, where R, is the average fractionof indicator ejected per beat. R, in turn is calculated as1 - K where K is the average of the ratios M3/M2 andM14/M3 (M2, M3, M4, being the average precordial count-ing rates during the second, third, and fourth cycle afterthe injection). For the first and second heart cycles, Kwas obtained as the ratio between the precordial countingrate during a 0.3-sec interval at the end of the cycle afterthe first ejection and the counting rate during the cor-responding period at the end of the cycle in which theinjection was made. EDVRcG also corresponds to the

850 A. Maseri and Y. Enson

true end-diastolic volume only in the absence of syste-matic regional differences in mixing within the ventricle.

RESULTS

Representative records of a set of curves sampledrespectively below and above the pulmonic valve,together with the simultaneous precordial curve,are illustrated in Figs. 1 and 2. The washout ofindicator from the ventricular inflow tract ap-peared to be very rapid. When RISA was injectedinto the atrium during systole or in early diastole,the precordial counting rate rose rapidly at thebeginning of the next diastole (when the indicatorentered the ventricle), reached a plateau, andpresented a very marked decrease with the firstejection (Figs. 1 and 2). The EDVRCGcalculatedfrom the fraction ejected with the first systole was26% smaller (range -21 to -33%) than theEDVRCGcalculated from the fraction ejected dur-ing the second and third beat (Fig. 3), whichindicated that under these conditions of injectionthe largest percentile fraction of the indicator leftthe ventricle with the first ejection. The pre-cordial curve showed the same behavior whenRISA was injected into the ventricular inflow

FIGURE 1 Comparison of radioiodinated "'I serum albu-min (RISA) /indocyanine green dye curves sampledproximal to the pulmonic valve. RISA was injected manu-ally into the atrium, dye into the low ventricle. From thebottom to top: timing and duration of manual injection;stethographic representation of respiratory cycle (in-spiration up); dye curves (continuous, displaced up-wards); RISA flow through scintillation detection probe(FTP) curve (discontinuous, displaced downwards);precordial RISA curve (discontinuous, displaced down-wards). The time lines are 0.1 sec apart. Of note are thedips in dye concentration which are found to occur earlyin systole during the second, third, and fourth heartcycles after the curve has been "corrected" for the meantransit time between the tip of the sampling catheter andthe densitometer. RCG, radiocardiogram.

FIGURE 2 Comparison of RISA/indocyanine green dyecurves sampled distal to the pulmonic valve. Dye was in-jected into the mid-ventricle, RISA into the atrium. Thetracings are disposed in the same fashion as in Fig. 2.Only a small amount of RISA enters the ventricle be-fore closure of the tricuspid valve (manual injection),and a small fraction is ejected into the pulmonary artery.The precordial counting rate is maximal at the beginningof diastole. The largest percentile decrease of the pre-cordial counting rate occurs during the next diastole.Note the staircase decrease of indicator concentration ob-tained by catheter sampling and the absence of earlysystolic dips.

during diastole. In similar fashion, the descendinglimb of curves obtained under these conditions bycatheter sampling proximal to the pulmonic valvewas steeper during the first beat after the peakconcentration had been attained than it was in the

300 INJECTION +20%,0 ATRIUM /

E. V INFLOW /_200 _7,-20%P200 - 00 3

100

o ,'~~~~~o

100 200 300

EDVRCG2nd and 3r EJECTION(ml)

FIGURE 3 Comparison of EDVRCGcalculated from thefractional decrease in RISA concentration during thefirst heart cycle after atrial systolic or protodiastolic(closed circles) and ventricular (V) inflow diastolic(open circles) injection, on the ordinate, with EDVRCGcalculated from the fractional decrease in RISA concen-tration during the second and third heart cycle after thesame injection, on the abscissa. The values obtained fromthe fraction ejected during the first cycle are systemati-cally lower.


z0

U-i

-JLa.

a I

ECG

V I, . . . . . I.0 2 4 6 8

Time (sac)

FIGURE 4 Comparison of time concentration curvessampled proximal to the pulmonic valve after simultane-outs manual injection of RISA into the atrium and greendye into the low ventricle. Indicator concentration, inmillimeter paper deflection, plotted logarithmically on theordinate; time in seconds on the abscissa. The largestpercentile decrease in RISA concentration occurs duringthe first heart cycle after injection. The dye curve reachesits peak only in the third cycle and displays protosystolicdips in concentration. RV, right ventricle.

lower part of the curve (Fig. 4). This phenome-non was never observed after mid, low, or apexventricular injection. The washout of indicatorfrom the apex of the ventricle, in contrast, wasparticularly slow (Figs. 4, 5, and 6).

The site of sampling markedly affected the peakconcentration of the washout curves obtained bycatheter sampling, but had no appreciable influ-ence on the washout slope after the first two orthree beats. During this initial period, curvessampled beyond the pulmonic valve showed amore rounded peak. The descending limb showedthe same rate of decrease, however, although at alower concentration. Regardless of the site of in-jection, the peak concentrations of the curvessampled distal to the pulmonic valve were muchlower than those of the curves sampled proximalto it within a few minutes, the injectate and thesite of injection being unchanged. In five subjectsin whom curves were obtained in succession fromthe pulmonary artery and the ventricular outflow,the values of DVa calculated from the curvessampled below the pulmonic valve were signifi-cantly smaller (-48%o average, range -25 to

Time (sec)

FIGURE 5 Comparison of time concentration curvessampled proximal to the pulmonic valve. In this caseRISA was injected automatically in protodiastole into thelow ventricle and the dye into the high atrium. The ordi-nate and abscissa are the same as in Fig. 4. The closedcircles on the RISA curve represent the average deflec-tion per heart cycle. The descending limbs of both curvesare practically rectilinear after the third heart cycle, butthe slope of the atrial curve is appreciably steeper. TheRISA curve displays protosystolic dips in concentrationon this occasion.

-80%) than those calculated from the curvessampled above the valve (Fig. 7).

The area under the washout curves sampledproximal to the pulmonic valve after ventricularinjection was systematically larger than that of the

(mn

z0

w-Jw0

Time (eec)

FIGURE 6 Comparison of time concentration curvessampled distal to the pulmonic valve. The ordinate andabscissa are the same as in Fig. 4. RISA was injectedinto the ventricular inflow, dye into the apex. The wash-out of indicator from the apex is significantly slowerthan the washout of indicator from the inflow. The in-flow washout becomes slower only after the fifth heartcycle. PA, pulmonary artery.


400

Effi 300

:,,, 200

100

+20%INJECTION //

V ATRIUM /0 INFLOW / '-20%OMID / //20A LOW d/ -

-* APEX X/ ,'

/ // .-

/ 4I

I I I I

100 200 300 400

PASDVa (ml)

FIGURE 7 Comparison of apparent dilution volume (DVa)obtained from curves sampled beyond the pulmonic valve(PA.), on the abscissa, with DVa obtained with ventricu-lar sampling (RV.), on the ordinate. The former valuesare much larger than those obtained in the same subjectwith ventricular sampling.

curves sampled beyond the valve. The significanceof these differences in area can be better appre-ciated by comparing the flow values calculatedfrom these curves by the Stewart-Hamilton for-mula with those obtained with the referencemethod (Fig. 8). While atrial injection with pul-monary artery sampling yielded cardiac output

values very close to those obtained with the refer-ence technique, Es = 6%o (standard error of esti-mate) (Fig. 8 A), ventricular injection with ven-tricular outflow sampling yielded lower valuesin all instances (- 53% on the average, range- 20 to - 757%). This underestimate was smallin patients with enlarged hearts and low refer-ence cardiac output, but was quite large in pa-tients whose heart size and reference flow valueswere normal. In all cases the discrepancy wasgreater after apex or low ventricular injection(Fig. 8 D). When the sampling site was abovethe pulmonic valve, however, ventricular injectiongave cardiac output values significantly lower thanthe reference method in only 4 instances out of 21(Fig. 8 C). In three of these curves the indicatorwas injected into the apex and in one into themid-ventricle. Atrial injection with ventricularoutflow sampling yielded cardiac output values inagreement with those of the reference method in6 curves out of 7 (Fig. 8B).

The morphology of the descending limb ofcurves obtained by catheter sampling was influ-enced both by the site of sampling and by that of

ATRIAL INJECTIONA

(liters/ 10 +10%min) ,',-O

8 /// //

bPA 64 -

2 /,'

O2 4 6 8 106REFERENCE(liters/min)

+10%(li:ters/ 0 / /min.) 0-,

8 ,'-1%p

6PA 6 i*/4

2 ;'

2 4 6 8 10

( REFERENCE(liters/min)

liters/Imin.)

6RV

8

6

4

2

B+10%

,/,s~~~~~~'h

// /'

/0

2 4 6 8 10bREFERENCE(liters/min)

D

+10%(liters/ 10 /(meins)/l

8 . ,// /

6RV 6-

4-

2 P,8 ,

IL IL -I

2 4 6 8 10

6REFERENCE(liters/min)

FIGURE 8 Comparison of flow values obtainedby catheter sampling of the indicator, on the or-dinate, with those obtained by the reference tech-nique, on the abscissa. Injection sites: atrial-O= high, 0 = mid, A = low; ventricular-E= inflow, 0 = mid, A = low, K = apex; opensymbols = RISA; closed symbols = dye. A,Atrial injection and sampling in the pulmo-nary artery. Agreement is good. B, Atrial in-jection and ventricular sampling. Agreement isgood in six of seven cases. C, Ventricular in-jection and pulmonary artery sampling. Thecatheter sampled curves are lower than the ref-erence in only four instances: one case of mid-ventricular injection, and all three cases of api-cal injection. D, Ventricular injection and ven-tricular sampling. The values calculated fromcatheter sampled curves are markedly lowerthan the reference. The underestimate increasesconsiderably with the level of flow. Apex andlow ventricular injection appear to yield thelowest values.

Right Ventricular and Pulmonary Artery Mixing

VENTRICULAR INJECTIONC

853

injection. The curves sampled proximal to the pul-monic valve showed a "staircase" decrease witheach diastole, more evident at slow heart rates.In most curves the diastolic "staircase" decreasewas systematically preceded by a "dip" in earlysystole. This pattern was observed with both indi-cators (Figs. 1, 4, and 5). It was more evidentin the 6 curves drawn after injection into the apex,but was also apparent in 5 of 6 curves afterinjection into the mid or low ventricle, in 2 of 4curves after injection into the ventricular inflow,and in 3 of 11 curves after atrial injection. Sam-pling beyond the pulmonic valve also resulted ina "staircase" decrease of indicator concentrationwith each systole, more evident at slow heartrates (Fig. 2). However, at this site no proto-systolic "dip" was noted.

DISCUSSIONMixing in the ventricle and in the pulmonaryartery

Our data confirm the existence of systematicregional differences in blood turnover rate in theventricle with striking uniformity. They indicate,as well, that the ventricle empties in a sequentialfashion and that further mixing occurs in thepulmonary artery beyond the pulmonic valve.

A. REGIONALDIFFERENCESIN BLOODTURNOVERRATE IN THEVENTRICLE

The observation that the largest percentile frac-tion of indicator injected into the ventricularinflow during diastole is ejected from the ventriclewith the first systole (Fig. 3) indicates that theblood in this region has a rapid turnover rate.Since the same behavior is observed after eitheratrial systolic or early diastolic injection (Figs.1-4) one can infer that an appreciable portion ofatrial blood does not mix evenly with the residualvolume during diastole, remains predominantly inthe inflow tract, and is ejected with the firstsystole.

At the other extreme, the uniform observationof a slower washout of indicator from the apex,consistent with the findings of Scarpa-Smith (14)and of Carleton, Bowyer, and Graettinger (15)in the left ventricle, indicates that the turnoverrate of blood at the ventricular apex is slow.Unless the bulk of the indicator is injected intothis region, however, its turnover rate does not

appear to determine the slope of the ventricularwashout curve until very late in its course, ifever. The descending limbs of the washout curveswere reasonably rectilinear on the semilogarithmicplot, after the first few beats, and their slopes didnot differ significantly after atrial, ventricularinflow, or mid-ventricular injection. These find-ings are reasonable if one considers that dur-ing the beat after injection the indicator in theventricle is in the residual volume and (unlessit were specifically injected in a peripheral partsuch as the apex) it would behave approximatelyin the same way regardless of the site of injection.On the other hand this linearity can not representevidence of homogeneous ventricular mixing. Infact, regardless of the site of injection, the concen-tration of indicator in the regions close to theinflow tract must be very close to zero after thefirst ejection, and the slope of the ventricularwashout curve must then be determined only bychanges in concentration of the indicator in thoseregions with slower turnover rates. The observedslope of our curves therefore repesents the wash-out of indicator from a region that has a bloodturnover intermediate between that of the apexand of the inflow tract.

Some quantitative estimate of the amount ofatrial blood which passes through the ventriclewithout mixing, or mixing only partially, withthe residual volume can be derived from the analy-sis of the washout curves sampled in the ventricleafter ventricular injection. When sampling is per-formed at a constant rate in the presence ofpulsatile flow, the concentration of indicator hasto be constant during each cycle to permit a cor-rect estimate of flow by the indicator dilutionprinciple (22). While this condition may be metin the pulmonary artery where the concentrationmay be considered constant, as a first approxima-tion, during each heart cycle, it is not met in theventricle. It can easily be seen, however, that, inthe presence of homogeneous mixing, ventricularsampling would yield a small area and a falselyhigh value of cardiac output when the indicatoris injected into the ventricle at the end of diastoleor during systole, but never a large area and a

falsely low value of cardiac output even afterprotodiastolic injection (Fig. 9). The large under-estimate of cardiac output observed in thosecurves sampled proximal to the pulmonic valve


Lf IL10-

C

5 - t Ra-I-

o tinjection

I.L..JECG ECG

rsme Tinjection

A.....-. ...

Time

FIGURE 9 A comparison of hypothetical time concentra-tion curves of indicator recorded in the pulmonary arteryand in the ventricle after an instantaneous ventricular in-jection marked by the arrow. It is assumed that mixingin the ventricle is homogeneous, and that no further mix-ing occurs in the pulmonary artery. Injection is at theend of diastole on the left, and at the beginning of diastoleon the right. The area under the curve recorded in thepulmonary artery (enclosed by the continuous line) isindependent of the timing of the injection, while the areaunder the curve recorded in the ventricle (shaded) is not.This latter area can be smaller than that under the curverecorded in the pulmonary artery but never significantlylarger.

after ventricular injection (Fig. 8 D) indicatesthat a large amount of blood containing little orno indicator passes systematically through theventricular outlet during a fraction of systole.Therefore, the amount of blood passing throughthe ventricle without mixing, or mixing only par-tially, with the residual volume is quite large.Since the underestimate was much larger in sub-jects with higher flow and normal hearts than inpatients with enlarged hearts and low flow, theamount of atrial blood traversing the ventriclewithout mixing seems to be directly related to theflow and inversely related to the ventricular vol-ume. This does not imply that mixing is morehomogeneous in the cavity as a whole in largeventricles with small stroke volumes, but onlyindicates that under these circumstances the indi-cator is better mixed with blood at the ventricularoutlet. On the other hand, as would be predictedfrom the behavior of atrial blood postulated above,flow values calculated from curves sampled in theventricle after atrial injection are not systemati-cally lower than those obtained with the referencemethod (Fig. 8 C).

B. SEQUENTIAL VENTRICULAREMPTYING

The finding of protosystolic dips in concentra-tion on the downslope of curves sampled proximal

to the pulmonic valve suggests that the ventricleempties in a sequential fashion: atrial blood, whichpasses through the ventricle without mixing, isejected systematically in early systole; blood fromthe apex is ejected in late systole. That these dipsrepresent real changes in concentration is demon-strated by their constancy and by their presenceeither in both curves or only in one, depending onthe site of injection. These changes in concentra-tion, which occur in a very short time, must berelatively large not to be completely obscured bythe catheter-sampling technique. This phenomenonwas first noted by Swan and Beck (13) in 16%of the washout curves recorded by sampling ata very high speed above the aortic valve, afterventricular injection, in dogs. These authors sug-gested that the "dips" were due to a very lowconcentration of indicator in the blood ejected atthe beginning of systole. According to our findingsit appears likely that it is the blood in the inflowtract, where the concentration of the indicatordecreases very rapidly, which is ejected in earlysystole. Further, the fact that the "dips" weremore evident after injection into the apex indicatesthat blood from the apical region is ejected in latesystole. The absence, in some cases, of dips afteratrial or tricuspid injection is probably due to thefact that the changes in concentration were toosmall or took place too rapidly to be detected byour catheter-sampling system. The fact that Swanand Beck (13) noted the dips in only 16%o oftheir curves is likely due to the distance of theirsampling catheter from the valve (further mixingat this site preventing detection of the phenome-non). The absence of dips in our pulmonary arterycurves is probably ascribable to the same cause,as well as to the characteristics of our catheter-sampling system.

The extremely low values of DVa, in two in-stances even lower than the stroke volume, whichwere obtained after low ventricular and apicalinjection (despite the fact that the peak concen-tration was reached only two or three beats afterthe injection), are consistent with the concept thatthe blood from the apex and low ventricle isejected in the last part of systole. It would be hardto account for the finding of these very high peakconcentrations two or three beats after the injec-tion without assuming sequential regional empty-ing of the ventricle.

Right Ventricular and Pulmonary Artery Mixing

-I

855

C. FURTHER MIXING ABOVE THE PULMON1CVALVE

The large increase in DVa which occurred whenthe point of sampling was moved from the ventricleto the pulmonary artery (Fig. 7) indicates thata significant admixture of the stroke volume withblood already present in the pulmonary arterytakes place. This conclusion is supported by thecloser agreement between flow values obtainedby catheter sampling after ventricular injectionand the reference values which occur under thesame circumstances (Fig. 8 C). Since the des-cending limb of curves sampled distal to the valveshowed the same slope, after the first two or threebeats, as the curves sampled in the ventricle, thevolume of blood above the valve that is involvedin mixing must be small in comparison with itsturnover rate, which, in turn, is evidently far morerapid than that of the ventricular regions thatdetermine the slope of the washout curve.

These findings are consistent with the observa-tion of Freis and Heath (23) and of Swan,Knoop, and David (24), that a considerable de-gree of mixing occurs beyond the aortic valve.However, according to the findings of Freis andHeath (23), Swan and Beck (13), and Rhode,Kines, and Holt (25), as well as the presentstudy, mixing is also not homogeneous above thesemilunar valves. The significant underestimationof flow observed in the three curves sampled inthe pulmonary artery after apex injection mayindicate that blood with a high concentration ofindicator coming from the apical region, ejectedin late systole, does not mix completely with theblood ejected in the first part of systole and re-mains just above the valve until the next ejection.

The agreement with the reference method ob-served in curves sampled from the pulmonaryartery after atrial injection, similar to that foundin previous comparisons between the radiocardio-graphic method and systemic arterial indicatordilution methods (26, 27) as well as the Fickprinciple (4), is consistent with previous reportsby Fritts et al. (28, 29), Fox and Wood (30),and Rochester, Durand, Parker, Fritts, and Har-vey (31). These authors also found a close agree-ment between flow values calculated from curvessampled in the pulmonary artery after atrialinjection and those obtained with a reference tech-

nique. In agreement with independent evidence(32) these observations suggest that an indicatorinjected into the right atrium is sufficiently mixedwith blood, by the time it reaches the pulmonaryartery, to yield a correct estimate of cardiac out-put. They also show that flow can be estimated byindicator dilution methods with reasonable accu-racy, even in the absence of a localized perfectmixing site between the injection and samplingpoints, provided that the indicator attains a suf-ficiently homogeneous dispersion in blood by thetime it reaches the sampling point (33).

A description of the sequence of events whichnormally occurs during the heart cycle may facili-tate evaluation of the results of this study. Atrialblood entering the ventricle at the opening of thetricuspid valve has a greater opportunity and alonger time available for mixing with the post-systolic residual blood than does blood enteringduring atrial systole. Blood in the central portionsof the ventricle has a greater opportunity for mix-ing with atrial blood than does blood at the apexof the chamber. Furthermore, considering therespective positions of the tricuspid and of thepulmonic valves, it seems likely that blood enteringthe ventricle in late diastole is distributed adjacentto the tricuspid and below the pulmonic valve.With the onset of ventricular systole the bloodbelow the pulmonic valve and in the middle of theventricle will be ejected first while a portion ofthe blood in the low ventricle and in the apex maybe ejected only in late systole. The blood ejectedin the first part of systole, at higher velocity, ismore likely to mix with blood already presentabove the valve -than that ejected at low velocityin late systole. Since it would seem reasonable toexpect that blood behaves in a similar way in bothventricles and above both semilunar valves, ourfindings in the right ventricle and in the pul-monary artery may also be valid for the leftventricle and the aorta.

Evaluation of ventricular volume measure-ments by dilution methods

Ventricular end-diastolic washout volume(EDVW) is currently calculated from the strokevolume and the fraction of indicator ejected perbeat after a sudden atrial or ventricular injection(1-11). The major assumption of this method isthat this fraction equals the fraction of the end-


diastolic volume ejected. While studies on in vitromodels substantiate the validity of the mathemati-cal analogue for calculation of volume and of theinjection and detecting techniques used (34),proof of the validity of volume measurements invivo is lacking. Rolett, Sherman, and Gorlin (35)found good agreement between left ventricularEDVWmeasured by thermodilution in three dogsand the end-diastolic volume measured at the samepressure after the animals had been sacrificed.However, because of differences in the myocardialdistensibility of living and dead animals, this ob-servation is of doubtful significance. Indeed, com-parisons of this method with angiocardiographicassessments of ventricular volume indicate thatvolumes obtained with the indicator washout tech-nique are uniformly and significantly larger (15,36, 37).

The relative constancy of the fraction of indi-cator ejected per beat, as indicated by a rectilineardownslope of the washout curve on semilogarith-mic paper, and its reproducibility, have been con-sidered the strongest indication of adequate ven-tricular mixing (1, 3-9, 11, 16, 35, 38). Ourfindings do not support this interpretation sincethe downslopes of simultaneous washout curvesobtained by catheter sampling were usually recti-linear on semilogarithmic paper after two or threebeats, despite the fact that their slopes weremarkedly different (Fig. 5). Furthermore, thesecurves were reproducible and their slopes com-pared well with those obtained over the precor-dium at the same time. It seems reasonable toconclude that these washout curves are an expres-sion of the turnover rate of the blood in the regionscontaining the indicator and that their rectilineardownslope on semilogarithmic paper indicates onlythat the turnover rate of blood in these regions isrelatively constant in successive beats.

Information concerning the volume of the re-gions that determine the downslope of the wash-out curve after the first or the first few beats maybe obtained only from a knowledge of the flowthrough them. This flow must be only a fractionof the stroke volume to account for the remarkablylow values of cardiac output obtained from thecurves sampled below the pulmonic valve afterventricular injection, and a significant fraction ofthe stroke volume must pass through the ventriclewithout mixing, or mixing only partially, with the

residual volume. Since EDVWis calculated fromthe downslope of the washout curve after the first,or the first few beats, its value will obviouslybe significantly larger than the true end-diastolicvolume regardless of the indicator used (be it dye,isotope, or temperature), regardless of the pointof injection, and regardless of where or how thewashout curve was recorded.

On the basis of these considerations, the indi-cator-washout method of estimating ventricularvolume can only set an upper limit to the valueof the end-diastolic volume. The overestimate mustbe rather large in normal subjects whereas it maybe of modest proportions in patients with an en-larged heart and low cardiac output. The appli-cation of this method to the study of ventricularresponses to various stimuli can only provideinformation concerning the direction of changesin volume if the pattern of blood transfer acrossthe ventricle remains unchanged. Recent findingssuggest that this condition may be met in theabsence of marked changes in heart rate or in theejected fraction (15). Reported alterations inEDVWin response to physiological manipulations(39-41) should be evaluated with these considera-tions in mind. Certainly studies of ventricularfunction derived from the measurement of EDVWappear to be unjustified at present.

ACKNOWLEDGMENTSThis investigation was supported by research grantsHE-05741, HTS-5443-05, and HE-02001-11 from the Na-tional Heart Institute, National Institutes of Health,U. S. Public Health Service.

Dr. Maseri is a Research Trainee (HTS-5443-05) ofthe National Heart Institute. Dr. Enson is a recipient ofan Investigatorship of the Health Research Council ofthe City of New York under contract No. 1-176.

REFERENCES

1. Holt, J. P., and J. Allensworth. 1957. Estimation ofthe residual volume of the right ventricle of the dog'sheart. Circulation Res. 5: 323.

2. Freis, E. D., G. L. Rivara, and B. L. Gilmore. 1960.Estimation of residual and end-diastolic volumes ofthe right ventricle of men' without heart disease,using the dye-dilution method. Am. Heart J. 60: 898.

3. Folse, R., and E. Braunwald. 1962. Determination offraction of left ventricular volume ejected per beatand of ventricular end-diastolic and residual volumes.Experimental and clinical observations with a pre-cordial dilution technic. Circulation. 25: 674.

4. Lewis, M. L., C. Giuntini, L. Donato, R. M. Harvey,and A. Cournand. 1962. Quantitative radiocardiog-


raphy. III. Results and validation of theory andmethod. Circulation. 26: 189.

5. Rapaport, E., B. D. Wiegand, and J. D. Bristow.1962. Estimation of left ventricular residual volumein the dog by a thermodilution method. CirculationRes. 11: 803.

6. Bristow, J. D., R. L. Crislip, C. Farrehi, W. E.Harris, R. P. Lewis, D. W. Sutherland, and H. E.Griswold. 1964. Left ventricular volume measure-ments in man by thermodilution. J. Clin. Invest. 43:1015.

7. Gorlin, R., E. L. Rolett, P. M. Yurchak, and W. C.Elliott. 1964. Left ventricular volume in man mea-sured by thermodilution. J. Clin. Invest. 43: 1203.

8. Rapaport, E., M. Wong, R. E. Ferguson, P. Bern-stein, and B. D. Wiegand. 1965. Right ventricularvolumes in patients with and without heart failure.Circulation. 31: 531.

9. Wilcken, D. E. L. 1965. The measurement of the end-diastolic and end-systolic, or residual, volumes of theleft ventricle in man, using a dye-dilution method.Clin. Sci. 28: 131.

10. Bing, R. J., R. Heimbecker, and W. Falholt. 1951.An estimation of the residual volume of blood in theright ventricle of normal and diseased human heartsin vivo. Am. Heart J. 42: 483.

11. Holt, J. P. 1956. Estimation of the residual volumeof the ventricle of the dog's heart by two indicatordilution technics. Circulation Res. 4: 187.

12. Irisawa, H., M. F. Wilson, and R. F. Rushmer. 1960.Left ventricle as a mixing chamber. Circulation Res.8:183.

13. Swan, H. J. C., and W. Beck. 1960. Ventricular non-mixing as a source of error in the estimation of ven-tricular volume by the indicator-dilution technic.Circulation Res. 8: 989.

14. Scarpa-Smith, C. 1960. Mixing of indicator withblood in the left ventricle. Federation Proc. 19: 117.(Abstr.)

15. Carleton, R. A., A. F. Bowyer, and J. S. Graettinger.1966. Overestimate of left ventricular volume by theindicator dilution technique. Circulation Res. 18: 248.

16. Holt, J. P. 1966. Indicator-dilution methods: indi-cators, inj ection, sampling and mixing problems inmeasurement of ventricular volume. Am. J. Cardiol.18: 208.

17. Maseri, A., and Y. Enson. 1967. A technique for con-tinuous simultaneous detection of two indicators inthe circulation. J. Appl. Physiol. 22: 153.

18. Gomez, D. M., M. Demeester, P. R. Steinmetz, J.Lowenstein, B. P. Sammons, D. S. Baldwin, andH. Chasis. 1965. Functional blood volume and distri-bution of specific blood flow in the kidney of man.J. Appl. Physiol. 20: 703.

19. McNeely, W. F., and M. A. Gravallese, Jr. 1954.Measurement of cardiac output by dye dilution tech-nique: use of an 'integrated' sample collection incalibration of the photometric instrument. J. Appl.Physiol. 7: 55.

20. Donato, L., C. Giuntini, M. L. Lewis, J. Durand,D. F. Rochester, R. M. Harvey, and A. Cournand.1962. Quantitative radiocardiography. I. Theoreticalconsiderations. Circulation. 26: 174.

21. Donato, L., D. F. Rochester, M. L. Lewis, J. Durand,J. 0. Parker, and R. M. Harvey. 1962. Quantitativeradiocardiography. II. Technique and analysis ofcurves. Circulation. 26: 183.

22. Stow, R. W. 1954. Systematic errors in flow deter-minations by the Fick method. Minn. Med. 37: 30.

23. Freis, E. D., and W. C. Heath. 1964. Hydrodynamicsof aortic blood flow. Circulation Res. 14: 105.

24. Swan, H. J. C., T. J. Knopp, and P. R. David. 1965.Effect of aortic mixing on determination of ventricularvolumes by washout. Physiologist. 8: 284. (Abstr.)

25. Rhode, E. A., H. Kines, and J. P. Holt. 1963. Studieson the accuracy of ventricular volume measurementsby the indicator-dilution technique. Physiologist. 6:263. (Abstr.)

26. Giuntini, C., A. Maseri, and R. Bianchi. 1966. Pul-monary vascular distensibility and lung compliance asmodified by dextran infusion and subsequent atropineinjection in normal subjects. J. Clin. Invest. 45: 1770.

27. Mariani, M., A. Maseri, and C. Giuntini. Com-parison between cardiac output measurement by radio-cardiography and arterial dilution method. J. Biol.Med. Nucl. In press.

28. Fritts, H. W., Jr., P. Harris, C. A. Chidsey, III,R. H. Clauss, and A. Cournand. 1957. Validation of amethod for measuring the output of the right ventriclein man by inscription of dye-dilution curves fromthe pulmonary artery. J. Appl. Physiol. 11: 362.

29. Fritts, H. W., Jr., P. Harris, C. A. Chidsey, III,R. H. Clauss, and A. Cournand. 1961. Estimation offlow through bronchial pulmonary vascular anasto-moses with use of T-1824 dye. Circulation. 23: 390.

30. Fox, I. J., and E. H. Wood. 1957. Applications ofdilution curves recorded from the right side of theheart or venous circulation with the aid of a newindicator dye. Proc. Staff Meeting Mayo Clinic. 32:541.

31. Rochester, D. F., J. Durand, J. 0. Parker, H. W.Fritts, Jr., and R. M. Harvey. 1961. Estimation ofright ventricular output in man using radioactivekrypton (Kr'). J. Clin. Invest. 40: 643.

32. Hamlin, R. L., W. P. Marsland, C. R. Smith, andL. A. Sapirstein. 1962. Fractional distribution of rightventricular output in the lungs of dogs. CirculationRes. 10: 763.

33. Andres, R., K. L. Zierler, H. M. Anderson, W. N.Stainsby, G. Cader, A. S. Ghrayyib, and J. L. Lilien-thal, Jr. 1954. Measurement of blood flow and volumein the forearm of man: with notes on the theory ofindicator-dilution and on production of turbulence,hemolysis and vasodilation by intra-vascular injection.J. Clin. Invest. 33: 482.

34. Salgado, C. R., and P. M. Galletti. 1964. In vitroevaluation of the thermodilution technique for the


measurement of end-diastolic ventricular volume.Federation Proc. 23: 302. (Abstr.)

35. Rolett, E. L., H. Sherman, and R. Gorlin. 1964.Measurement of left ventricular volume by thermo-dilution: an appraisal of technical errors. J. Appl.Physiol. 19: 1164.

36. Bartle, S. H., and M. E. Sanmarco. 1966. Measure-ment of left ventricular volume by biplane angiocardi-ography and indicator washout techniques: A com-parison in the canine heart. Circulation Res. 19: 295.

37. Hallermann, F. J., G. C. Rastelli, and H. J. C.Swan. 1963. Comparison of left ventricular volumesby dye dilution and angiographic methods in the dog.Am. J. Physiol. 204: 466.

38. Rapaport, E. 1966. Usefulness and limitations ofthermal washout techniques in ventricular volumemeasurement. Am. J. Cardiol. 18: 226.

39. Holt, J. P. 1957. Effect of plethora and hemorrhageon left ventricular volume and pressure. CirculationRes. 5: 273.

40. Bristow, J. D., R. E. Ferguson, F. Mintz, and E.Rapaport. 1963. Thermodilution studies of ventricularvolume changes due to isoproterenol and bleeding.J. Appl. Physiol. 18: 129.

41. Maseri, A., R. Bianchi, C. Giusti, P. Toni, and L.Donato. 1965. Early effects of digitalis on centralhemodynamics in normal subjects. Am. J. Cardiol.15: 162.


Documents

in the Right Ventricle and Pulmonary...1.5 inch NaI (T1) crystal (The Harshaw Chemical Co., Cleveland, Ohio) with a 1/8 inch cylindrical hole through its center. The pulses from the