AN EXPERIMENTALSTUDYOF THE IMMEDIATE HEMODY-NAMIC ADJUSTMENTSTO ACUTEARTERIOVENOUS

FISTULAE OF VARIOUS SIZES'

BY CHARLESW. FRANK, HSUEH-HWAWANG,2 JACQUESLAMMERANT,3ROBERTMILLER, ANDRENEWEGRIA

(From the Department of Medicine, Columbia University College of Physicians and Surgeons,and the Presbyterian Hospital, New York, N. Y.)

(Submitted for publication November 17, 1954; accepted January 26, 1955)

In the past thirty years, numerous observationshave been made on the hemodynamic effects ofarteriovenous fistulae. Arteriovenous fistulae ofvariable size and location have been studied in pa-tients as well as in animals. Such studies haveconsisted a) in observing the effect of suddenlyopening and closing a fistula in acute experiments,b) in studying the effects of suddenly closing andopening a fistula in an animal or a patient whohad had such a fistula for a relatively long time,and c) in recording the effect of complete surgi-cal eradication of a long-standing fistula. Whilesome agreement has been reached on many points,there persists some controversy about most ofthem (1). It is now generally agreed that anacute as well as a chronic arteriovenous fistulaincreases the cardiac output (1-8). However,Lewis and Drury (9) came to the conclusion thatthe cardiac output is not increased unless thearteriovenous shunt is very large. Van Loo andHeringman (10) concluded that acute arteriove-nous fistulae increase the cardiac output althoughthe flow of blood in certain vascular areas outsidethe fistula circuit is decreased. The venous pres-sure is definitely increased near the site of thearteriovenous communication, but most investiga-tors have found that the increment of venous pres-sure decreases rapidly as the pressure is meas-ured further down the venous bed toward theright atrium, and in the venae cavae and rightatrium the pressure is elevated little or not at all

I This work was made possible by a grant-in-aid fromthe NewYork Heart Association, the Sidney A. LegendreGift, and the Charles A. Frueauff Gift. It was presentedat the Annual Meeting of the American PhysiologicalSociety in New York City in 1952.

2Fellow of the New York Heart Association.8 Fulbright Fellow, from the Department of Medicine,

University of Louvain Medical School, Louvain, Bel-gium.

(3, 4, 11). However, one group of workers (12)has reported more striking increases of right atrialpressure in dogs with an arteriovenous fistula.The blood volume seems to be increased by anarteriovenous fistula of a certain size and dura-tion (3, 13) although some observers (4) sug-gest that this increase in blood volume occursonly in the presence of an incipient or frankcardiac failure.

Since the magnitude of the effects of an arterio-venous fistula would seem a priori to depend uponthe volume of blood shunted through the ab-normal pathway, it is rather surprising that in nostudy has any attempt been made to measure si-multaneously the rate of flow through the fistulaand the other functions of interest such as cardiacoutput, arterial and venous pressures. The pres-ent study was undertaken in an attempt to de-termine the acute cardiovascular adjustments toarteriovenous fistulae of different sizes, by re-cording simultaneously and continuously the flowof blood through the fistula, the flow of blood intothe systemic circulation exclusive of the arterio-venous fistula circuit, the mean arterial blood pres-sure and the mean central venous pressure.

METHODS

Fifteen dogs weighing between 13 and 28 kilogramswere anesthetized by the intravenous infusion of 100 mil-ligrams of chloralose per kilogram of body weight. Thechest was opened by a midsternal incision, and respira-tion was maintained by a positive pressure respirator.Segments of the brachiocephalic trunk, subclavian ar-tery, common carotid arteries, upper part of the descend-ing aorta, as well as the femoral veins were exposedand dissected. Ten milligrams of heparin per kilogramof body weight were then administered intravenously andrepeated injections of one-half that amount were givenevery half hour. Cannulation of the arteries mentionedabove was then performed in a manner previously de-scribed (14) and as pictured by the diagram of Figure 1.

722

IMMEDIATE ADJUSTMENTSTO ACUTE ARTERIOVENOUSFISTULAE

The tube with the tunnel clamp T leads into the centralend of a femoral vein. When clamp T is screwed tight,the blood expelled by the left ventricle leaves the aorticarch via the brachiocephalic and subclavian arteries, andafter passing through rotameter A, flows into the de-scending aorta and common carotid arteries. This flowwill be referred to as the body flow, and was measuredand recorded continuously by rotameter A. When thefistula to the femoral vein was opened by unscrewingtunnel clamp T, the flow of blood through the fistulawas measured and recorded continuously by rotameterB. Thus, when the fistula was patent, the cardiac outputwas equal to the sum of the two recorded flow rates.Each rotameter was supplemented with a shunt S toallow the recording of the zero flow without interruptionof the circulation. When blood flows were being meas-ured, the shunts were occluded by clamps. The meanarterial blood pressure and the mean venous pressurewere recorded by damped Gregg optical manometers ofsuitable sensitivity. The venous pressure was recordedfrom the point of entrance of the superior vena cava intothe right atrium. Clamp T was unscrewed so as to openthe fistula progressively and reach the desired level offistula flow over a period of from one-half to one min-ute; from the time of the unscrewing of Clamp T, thefistula was maintained open for five minutes or longer,then it was suddenly occluded. Continuous photographicrecords of pressures and flows were obtained during acontrol period, during the time the fistula was patent, andduring a recovery period which followed the occlusionof the fistula. An example of the type of record obtainedis illustrated in Figure 2. To facilitate the presentation

FIST. Ccl

BPMMINg

BcEm

FIG. 1. DIAGRAM OF THIE TECHNIQUE USED TO PRO-DUCE ARTERIOVENOUSFISTULAE AND TO MEASUREANDRECORDCONTINUOUSLYAND SEPARATELY THE FLOW OFBLOOD THROUGHTHE FISTULA (FISTULA FLOW) ANDTHE FLOw OF BLOODTO THE REST OF THE BODY (BODYFLOW)-EXPLANATION IN TEXT

V p MM**Hpo

FIG. 2. CONTINUOUSRECORDOBTAINED DURINGTHE PATENCYOF AN ARTERIOVENOUSFISTULA ANDAFTER ITS SUDDENOCCLUSION

At the beginning of the record, from top to bottom, fistula flow, mean arterial bloodpressure, body flow and venous pressure. At the arrow, sudden occlusion of the ar-teriovenous fistula. Scale for the flows in cc. per minute; scales for the arterial bloodpressure and the venous pressure in mm. of mercury and water, respectively.

1 2

"01"W

DOO.

58 50

723

C. W. FRANK, HSUEH-HWAWANG, J. LAMMERANT,R. MILLER, AND R. WEGRIA

140V. P MMH2O 0

a.p. MHg 50

BEATS IIIMIN. 170 X

T*P.R. 0MMH1so _I

1400

1200

1000

FLOW C C 800MI N.

600

400

200

0.2 I 2 3 4 5

+ TIME IN MINUTES

FIG. 3. PLOT OF AN EXPERIMENTIN WHICHAN ARTERIOVENOUSFISTULAACCOMMODATINGAPPROXIMATELY27 PER CENT OF THE CONTROLCARDIACOUTPUTWASPRODUCED

From top to bottom, mean arterial blood pressure in mm. of mercury;

venous pressure in mm. of water; heart rate per minute; body total periph-eral resistance in arbitrary units; cardiac output, body flow and fistula flowin cc. per minute. At the first and second arrow, opening and closing offistula, respectively. Time in minutes.

and analysis of the data the results of each experimentwere plotted on graph paper as indicated in the succeedingfigures. The resistance to the body flow or body totalperipheral resistance (TPR) was calculated in arbitraryunits by dividing the mean arterial blood pressure ex-

pressed in mm. of mercury by the body flow expressedin liters per minute.

RESULTS

Fifty fistulae of various size were studied in 15dogs. Examples of the hemodynamic adjustmentsto small, medium and large fistulae are illustratedby Figures 3, 4, and 5.

Figure 3 is the plot of an experiment in whichwas produced an arteriovenous fistula accommo-

dating approximately 300 cc. per minute, whichamounted to approximately 27 per cent of the con-

trol cardiac output. During the control period,two minutes of which are shown in Figure 3, themean arterial blood pressure oscillated between150 and 154 mm. of mercury, the cardiac output

ranged between 1120 and 1170 cc. per minute, and

the central venous pressure varied from 45 to 46mm. of water. The body total peripheral re-

sistance ranged between 130 and 135 units. Im-mediately upon opening the fistula, the cardiacoutput increased and thirty seconds after the be-ginning of the opening of the fistula, the fistulaflow amounted to 320 cc. per minute, the bodyflow had decreased slightly to 1080 cc. per minute,and the cardiac output was 320 plus 1080, or 1400cc. per minute. During the five-minute periodthat the fistula was patent, the mean arterial bloodpressure remained within its control range, andin view of the slight decrease of the body flow itis apparent that some increase in the peripheralresistance to body flow was present, the TPRreaching a peak value of 144 units. A slight in-crease in central venous pressure was recordedduring this fistula. After five minutes the fistulawas suddenly closed. The body flow immediatelyincreased and reached a maximal value of 1280cc. per minute within six seconds, then returnedto its control level of 1140 cc. per minute within

6

724

I I

MSFrftIft -!!

7

I MMEDIATE ADJUSTMENTSTO ACUTE ARTERIOVENOUSFISTULAE

5.P. mmHe.

V.P.

ATMH20

SETS/IMIN.

150-140

6050

180170

t R.f! MM Hg 130L/mI.. 120

a erfn

1400

.0AA

bOO1000-

FLOW Sc C 800MIN.

600

400

200

z

i0

0

310

b.

IL

A 'I ! VI r̂ I 11I I I I I 1 IJ I 1 .1il -I --::1:1

-2 -I I 2 3 4 5 6 7XTIKE IIMKUNUTES.

FIG. 4. PLOT OF AN EXPERIMENTIN WHICHAN ARTEROVENOUSFISTULAACCOMMODATINGAPPROXiMATELY57 PER CENT OF THE CONTROLCADIACOUTPUTWASPRODUCED

Same legend as in Figure 3.

NORMALSALINE

H20~00ca/OOc=2~ ~ ~ ~f.R

130

JR P.

*11-.. .. .... .. MM:No.] ; r;~. 1l -_~~~~. 11 | + _ E_<tTI30

oo-A_. V. P.;R_><__<_>_ ,R -B0_. _ _ ___.-._____I --: u "20too~ ~ ~ X1 A 4I x lll10 MMP.600- -.. ___ i _ 1 LII : 170 ""

Boo.. . .1t. 1 !SQ iI0|l TLh

-2 -1 1 a 3 4 1 7 9 10

TIME IN MINUTESFIG. 5. PLOT OF AN EXPERMENTIN WHICHAN ARTERIOVENOUSFISTULA Accommo-

DATING 100 To 104 PER CENTOF THE CONTROLCARDIAC OUTPUTWASPRODUCEDFrom top to bottom, mean arterial blood pressure in mm. of mercury; venous pres-

sure in mm. of water; heait rate per minute; body total peripheral resistance in arbi-trary units; cardiac output, body flow and fistula flow in cc. per minute. At firstand fourth arrows, opening and closing of fistula, respectively. Between second andthird arrows, intravenous infusion of 100 cc. of isotonic sodium chloride solution.Time in minutes.

725

C. W. FRANK, HSUEH-HWAWANG, J. LAMMERANT,R. MILLER, AND R. WAGRIA

15 seconds of the time of the occlusion of thefistula. The mean arterial blood pressure also ex-hibited a transient increase of 5 mm. of mercuryduring this 15-second period immediately follow-ing the closing of the fistula. A distinct fall of thecentral venous pressure was recorded after theclosure of the fistula. No significant changes inheart rate were noted. In summary, it is ap-parent that this fistula, accommodating a bloodflow equivalent to around 30 per cent of thecontrol cardiac output, caused an increase incardiac output almost equal to the amount of bloodflowing through the fistula so that only a slightdecrease in body flow occurred. A fall in arterialblood pressure was prevented by vasoconstrictionof the systemic circulation exclusive of the fistulacircuit, as manifested by a slight increase in thebody total peripheral resistance.

Figure 4 is the plot of an experiment performedon the same dog as that used in the experiment ofFigure 3. In this experiment an arteriovenousfistula accommodating a flow of 600 cc. per min-

160 -

150 -

140

0-1300 120

0

4 100cjw 90'41

80Ui.

70-

to 60-

50

Z 40-

30-

20

10

* .P.at controtl± 54m Hg

O.B.RcControl I

II

OuSteedy infusion unning(see figure 7)

Iafter

be f ore

;oT I

~~~~0

~~~~0

~~~0

*0

o e 2030 50 070 80 00110 120 130FISTULA FLOW

expressed as % of control cardiac output

FIG. 6. COMPOSITEPLOT OF THE PHENOMENAOBSERVEDIN THE 50 FISTULAE STUDIED

The percentile increase in cardiac output produced ineach fistula is plotted along the ordinate; the correspond-ing values of the fistula flow expressed in terms of percent of the control cardiac output are plotted along theabscissa.

ute, or 57 per cent of the control cardiac outputwas produced. Although the increment in cardiacoutput (500 cc. per minute) was greater than withthe smaller fistula, the increase in cardiac outputwas insufficient to compensate fully for the fistulaflow, and the body flow decreased by 100 cc. perminute, a 9 per cent decrease. Body vasoconstric-tion, manifested by an increase in body TPR from126 to a peak of 145 units, was effective in main-taining the mean arterial pressure within its con-trol range. An increase of 5 mm. of water in thecentral venous pressure was present during thepatency of the fistula. Upon closing the fistula,the mean arterial blood pressure temporarily rose10 mm. of mercury above previous and controllevels, the body flow rose and remained elevatedabout 15 seconds, after which the body flow, i.e.,the cardiac output, returned to control value.The venous pressure promptly fell to its controllevel. The heart rate showed no striking changethroughout this experiment except for a temporaryslowing of five beats per minute immediately af-ter the occlusion of the fistula.

Figure 5 is the plot of the response of the samedog to a fistula accommodating a blood flow ofup to 1100 cc. per minute, i.e., approximately 100per cent of the control cardiac output. Althoughthe cardiac output reached a peak of 2000 cc. perminute, this increase was clearly inadequate, forthe body flow decreased by 200 to 250 cc. per min-ute. Despite the body vasoconstriction revealedby the rise in body TPR, the arterial blood pres-sure fell 10 mm. of mercury. The central venouspressure rose 9 mm. of water. During the sixthminute of the period during which the fistula waskept open, 100 cc. of isotonic sodium chloride solu-tion were administered to the dog via the intactfemoral vein. This infusion produced a dramaticincrease in cardiac output to a peak of 2840 cc.per minute, an increase of 970 cc. per minute aboveits value just prior to the infusion. However,most of this increment of cardiac output, 850 cc.of the 970 cc., was absorbed by the body circula-tion and only 120 cc. went through the fistula cir-cuit. As the body flow rose, there was a markeddecrease in the body TPR. Despite this fall inbody TPR, the increase in body flow was suchthat the mean arterial blood pressure rose aboveits control level. After the end of the infusion,

726

IMMEDIATE ADJUSTMENTSTO ACUTE ARTERIOVENOUSFISTULAE

the fistula was kept open for another two minutes,during which the effects of the infusion subsidedsomewhat: cardiac output and body flow as wellas venous and arterial blood pressures decreased,whereas the body TPRrose. Upon closing of thefistula the arterial blood pressure rose, thenpromptly stabilized at its control level; venouspressure and cardiac output fell but remainedslightly above control levels, probably because ofthe persistent effect of the infusion, and the TPRremained below its control value presumably forthe same reason.

Figure 6 is a composite plot of the results ob-served in the 50 arteriovenous fistulae studied.Along the abscissa is plotted the fistula flow ex-pressed as percentage of the control cardiac out-put. On the ordinate is plotted the correspondingincrease in cardiac output expressed in similarterms. To facilitate the discussion, all fistulaehave been divided into four groups, delineated bythe vertical interrupted lines. The basis for thisdivision of all fistulae into four groups will becomeapparent in the course of the discussion.

As can be seen in Figure 6, it is clear that thedegree of increase in cardiac output is a functionof the size of the fistula. In six of the seven fis-tulae in which the fistula flow was equal to orsmaller than 19 per cent of the control cardiac out-put (Group I), the increase in cardiac output wasequal to the fistula flow, so that the body flow re-mained at its control value, as did the arterialblood pressure and body TPR. The largest fistulain which the increase in cardiac output was equalto the fistula flow accommodated a flow equal to27 per cent of the control cardiac output. In allthe fistulae larger than 27 per cent (Groups IIIand IV) and in 11 of the 14 fistulae of Group II,i.e., fistulae equal to 19 to 27 per cent of the con-trol cardiac output, the increment in cardiac out-put was less than the amount of the fistula flowso that the body flow decreased. Body vaso-constriction as indicated by an increase in thebody TPR, was present in all these experimentsin which the body flow decreased. Although inall these fistulae the body flow decreased, the meanarterial blood pressure was maintained in 7 of

V..MH0 H.R.V.P. MMH206o 5 9 t l5O BEATS

60 15~~~~~~~~~~~~~~~0SEmATS

CC 200

200 -10 TPRV

1800

1600-

-2 -, 0 9 2 3 4 5 6 7 9* TIME IN MINUTES +

FIG. 7. PLOT OF AN EXPERIMENTIN WHICH A DOGWASGIVEN INTRAVENOUSLYAPPROXIMATELY100 CC. OF ISOTONIC SODIUM CHLORIDE SOLUTION OVERA PERIOD OF10 MINUTES, BEGINNING 1 MINUTE: BEFORETIME 0, THE TIME AT WHICH THE AR-TEROVENOUSFISTULA WASOPENEDPROGRESSIVELYOVER A PERIOD OF 4 MINUTES

From top to bottom, mean arterial blood pressure in mm. of mercury, venous pres-sure in mm. of water; heart rate per minute; body total peripheral resistance in arbi-trary units; cardiac output, body flow and fistula flow in cc. per minute. At first andlsecond arrow, progressive opening and sudden closing of arteriovenous fistula, re-spectively. Time in minutes.

727

C. W. FRANK, HSUEH-HWAWANG, J. LAMMERANT,R. MILLER, AND R. WEGRIA

the 11 fistulae of Group II (experiments repre-sented by solid dots in Figure 6) and in 5 of the13 experiments of Group III. Group IV com-prises 15 large fistulae during which both bodyflow and mean arterial blood pressure decreased.In seven experiments of Group IV, an infusionof 100 cc. of isotonic sodium chloride solution wasadministered over a one-minute period during thepatency of the fistula. During each infusion thecardiac output increased and body flow as well asmean arterial blood pressure rose to or even abovecontrol levels. As mentioned in the discussion ofFigure 5, most of the increase in cardiac outputwas accommodated by the body flow circuit.The increase in cardiac output during these in-fusions is indicated in Figure 6 by the arrows.The large solid dot of Figure 6 was yielded bythe experiment pictured in Figure 7. In this ex-periment, the dog was given intravenously a con-tinuous infusion of isotonic sodium chloride solu-tion at a constant rate. The total amount of sa-line solution administered in 10 minutes, from 1minute before the opening of the fistula to ap-proximately 3 minutes after the closing of thefistula, was approximately 100 cc. In this ex-periment a fistula finally accommodating 77 percent of the control cardiac output was attainedand the cardiac output increased by the full amountof the fistula flow. The central venous pressurerose no more than 6 mm. of water during thisfistula. As is apparent from Figure 6, this fistulais almost three times as large as any other fistulafound to be fully compensated with respect tobody flow without benefit of infusion.

The changes in heart rate, as indicated by thefour examples shown in Figures 3, 4, 5, and 7,were slight and variable. As the fistulae weregradually opened, no significant changes werenoted. Upon sudden closing of the fistula, atemporary slowing of the heart rate was observedin half of the experiments. The magnitude ofthis change varied from 5 to 22 beats per minute.

The changes in central venous pressure wereslight, but a definite increase was noted in 21 ofthe 22 fistulae in which the venous pressure wasrecorded continuously. Table I summarizes theincreases seen in central venous pressure with thedifferent sizes fistulae.

DISCUSSION

An increase in cardiac output was observed inevery fistula studied and the magnitude of the in-crease appeared to be a function of the size of thefistula. Inasmuch as the heart rate did not changeconsistently or markedly with the progressiveopening of the fistula, it might be concluded thatthe increase in cardiac output was due mainly toan increase in systolic output.

The initial alteration of the circulation uponopening an arteriovenous fistula consists in theopening of a path of less resistance through whichblood passes directly from the arterial into thevenous part of the vascular bed. The opening ofthat path of less resistance decreases the resistanceto the outflow of blood from the left ventricle,thus producing an increase in the emptying ofthe left ventricle. In the case of a small fistula,the increase of the venous return is such that thecardiac output is increased by the full amount ofthe fistula flow, the body flow remains unchanged,and the mean arterial blood pressure is promptlyrestored to its control level through the increasein cardiac output. It is rather doubtful whether,even in a small fistula, the more complete empty-ing of the left ventricle ever is the only mecha-nism through which the increase in venous returnoccurs and the cardiac output is increased suffi-ciently to be augmented by the full amount ofthe fistula flow. Other mechanisms, myocardialas well as peripheral, which may play a role inthe acute adjustment to the establishment of anarteriovenous fistula have to be considered, es-pecially in large fistulae. Indeed, upon openingthe fistula the inflow of arterial blood into the veinscauses a considerable increase in the local venouspressure ( 11 ), the veins become distended and thevolume of blood in the venous system must be as-sumed to increase until a new equilibrium isreached. At this point a) the amount of bloodleaving the veins per unit of time to enter theright heart is equal to that which enters the ve-nous system coming from the body and the fistula,b) the amount of venous return is larger thanbefore the fistula was opened, and c) the venoussystem most probably contains more blood thanbefore the opening of the fistula. The rate of ve-nous return in the fistula circuit is determined bythe pressure gradient between the veins and the

728

IMMEDIATE ADJUSTMENTSTO ACUTE ARTERIOVENOUSFISTULAE

heart and obviously the head of pressure in thissystem is limited for a fistula of a given size bythe distensibility of the veins and the volume ofblood available to fill them.'

Thus it is suggested that, with the opening ofa small fistula, there is an increase in the amountof blood flowing into the venous bed which ac-commodates a greater volume of blood, then thepressure gradient rises so that the flow of bloodinto the venous bed is equal to the flow of bloodinto the right heart. At this new equilibrium, thecontrol cardiac output is increased by the fullamount of the fistula flow so that body flow andarterial blood pressure remain at control levels.Such a fistula has been characterized as being fullycompensated. However, a redistribution of bloodmust have occurred, leading to an increase of theblood contained in the venous part of the vascu-lar bed. The source of the blood accounting forthe increase in the venous blood volume is notclear. Since there was no evidence of constrictionof the arterial tree, at least in the case of a smallfistula, it is probable that the increment in the ve-nous blood volume comes not from the arterialtree itself but from blood reservoirs. With largerfistulae a progressively larger volume of blood ispooled in the venous system, and eventually afistula flow rate is reached at which the pressuregradient is insufficient to ensure an increase in thevenous return to the right heart equal to theamount of the fistula flow. Under such circum-stances the increase in cardiac output will be lessthan the amount of the fistula flow and the bodyflow will fall below control value.

As indicated by experiments similar to thatplotted in Figure 7, when the total volume of in-travascular blood is increased by the continuousinfusion of an isotonic solution of sodium chlo-ride, a much larger fistula is tolerated withoutdecrease in body flow. Similarly, as pictured inFigures 5 and 6, when a dog with a fistula largeenough to produce a decrease in body flow re-ceives an intravenous infusion of isotonic sodium

'Two other factors, the size of the aperture betweenthe arterial and the venous circuit, and the possibility ofthe development of turbulence at the site of the fistula,have to be considered. The first one is dealt withthroughout the manuscript; the second one is not dis-cussed because it is not susceptible of quantitation withthe data at hand.

chloride solution, the cardiac output may increaseby the full amount of the fistula flow or more,and the body flow return to or above control level,although the fistula flow itself is affected relativelylittle by the infusion. It is therefore most prob-able that the essential factor which limits thedegree of increase in cardiac output in the pres-ence of a large acute arteriovenous fistula is theinadequacy of the venous return and not the car-diac reserve. Furthermore, the critical factorsetting the limit to the rate of the venous returnis apparently the total available volume of blood.Therefore, the increase in blood volume which de-velops in chronic arteriovenous fistulae may be in-terpreted as a beneficial physiological adjustmentand not necessarily as an indication of cardiacfailure.

The cardiac output increased by the full amountof the fistula flow in 6 out of 7 experiments inwhich the fistula accommodated up to 19 per centof the control cardiac output (Group I). In 11out of the 14 fistulae accommodating a flow equalto from 19 to 27 per cent of the control cardiacoutput (Group II) and in all fistulae accommo-dating a flow amounting to more than 27 per centof the control cardiac output (Groups III and IV),the cardiac output did not increase by the fullamount of the fistula flow and the body flow de-creased. The total peripheral resistance rose inall those experiments in which the increase incardiac output was inadequate and, as a result ofthis vasoconstriction, the mean arterial blood pres-sure was maintained at control level in 7 out ofthe 11 experiments of Group II and in 5 of the13 experiments of Group III. However, despitevasoconstriction, the arterial blood pressure fellin all the experiments in which the fistula accom-modated more than 60 per cent of the controlcardiac output.

Emphasis has been placed upon the fact that whenthe fistula is open, the central venous pressure iselevated little or not at all in the presence of a con-siderable increase of the stroke volume (15). Itmust be remembered indeed that experiments onthe heart-lung preparation indicate that the rightauricular pressure has proven to be an insensi-tive indicator of the rate of venous return, the ex-tent of ventricular filling and the volume of sys-tolic discharge (16). However, as can be seenin Table I, definite, though sometimes slight, in-

729

C. W. FRANK, HSUEH-HWAWANG, J. LAMMERANT,R. MILLER, AND R. WEGRIA

TABLE I

Increases in mean central venous pressure during thedifferent arteriovenous fistulae

NumberOf

experi-ments

Fistula flow(expressed in

per cent ofcontrolcardiacoutput)

Central venous pressure(mm. of water)

Mean Range ofincrease increase

8 20-30 3.7 0-77 30-60 5.6 3-107 60-130 7.7 6-9

creases in the central venous pressure were re-

corded during the period that the fistula was pa-

tent in 21 out of 22 experiments. It is conceiv-able that in clinical or experimental studies inwhich the thorax remained intact and continuouspressure recording was not obtained, changes ofthis magnitude could remain undetected becauseof the magnitude of the fluctuations of venous

pressure under the influence of respiration. Fig-ure 2 demonstrates the rapidity with which thecentral venous pressure falls upon closure of thefistula. An abrupt reduction of 8 mm. of wateroccurred within two seconds. It would seem thatthis abrupt fall of the central venous pressure isdue to the sudden decrease of the rate of venous

return. The sudden increase in the resistanceto the left ventricular outflow and the reduction inventricular filling which both occur upon occludingthe fistula circuit, would appear to be an adequateexplanation for the sudden decrease in strokevolume.

The changes in heart rate which may be ob-served with the opening and closing of arterio-venous fistulae do not seem to be the essentialdeterminant of the changes in cardiac output, in-asmuch as similar responses in output are seen inexperiments on animals or humans in which thechanges in heart rate are prevented by anesthesiaor atropine (5, 8).

The data presented may also help the under-standing of some of the circulatory adjustments to

circumstances other than those resulting fromarteriovenous fistulae. Indeed, these observationsmay be taken to indicate that when in acute situa-tions, vasodilatation occurs in an important vas-

cular area, the degree of increase in venous returnattainable without the action of compensatorymechanisms may well be quite limited.

SUMMARY

Some of the immediate circulatory adjustmentsto the opening and closing of arteriovenous fistu-lae of different sizes have been studied in theanesthetized dog. An increase in stroke volumeand cardiac output was observed in all experi-ments. With the smaller fistulae, i.e., fistulae ac-commodating less than 20 per cent of the controlcardiac output, the cardiac output increased bythe full amount of the fistula flow. Greater in-creases in cardiac output occurred with largerfistulae, but the amount of the increase in outputwas less than the amount of the fistula flow whichresulted in a decrease of the blood flow to thesystemic capillary bed. Vasoconstriction super-vened, which prevented a fall in the mean arterialblood pressure in some of the fistulae accommo-dating up to 60 per cent of the control cardiacoutput, but with larger fistulae the arterial bloodpressure fell despite vasoconstriction.

It is suggested that the essential factor whichlimits the degree of increase in cardiac output inthe presence of a large fistula is the inadequacy ofthe venous return, and under the conditions ofthese experiments, the limit to the rate of venousreturn seems to be the volume of blood available.

In the light of these findings, the increase inblood volume noted in chronic experimental andclinical arteriovenous fistulae may be consideredas a compensatory mechanism, and not necessarilyas an indication of the development of cardiacfailure.

REFERENCES

1. McGuire, J., Hauenstein, V., Stevens, C. D., andSharretts, K. C., Effects of arteriovenous fistulaeon the heart and circulation in Blood, Heart andCirculation, F. R. Moulton, ed., publication of theA.A.A.Sc., No. 13, The Science Press, Washington,1940.

2. Harrison, T. R., Dock, W., and Holman, E., Ex-perimental studies in arteriovenous fistulae: car-diac output. Heart, 1924, 11, 337.

3. Holman, E., Arteriovenous Aneurysm; AbnormalCommunications Between the Arterial and VenousCirculations. New York, The Macmillan Co.,1937.

4. Reid, M. R., and McGuire, J., Arteriovenous aneu-rysms. Ann. Surg., 1938, 108, 643.

5. Elkin, D. C., and Warren, J. V., Arteriovenous fistu-las: their effect on the circulation. J.A.M.A.,1947, 134, 1524.

730

IMMEDIATE ADJUSTMENTSTO ACUTE ARTERIOVENOUSFISTULAE

6. Cohen, S. M., Edholm, 0. G., Howarth, S., Mc-Michael, J., and Sharpey-Schafer, E P., Cardiacoutput and peripheral bloodflow in arteriovenousaneurysm. Clin. Sc., 1948, 7, 35.

7. Warren, J. V., Nickerson, J. L., and Elkdn, D. C.,The cardiac output in patients with arteriovenousfistulas. . Clin. Invest., 1951, 30, 210.

8. Nickerson, 3. L., Elkin, D. C., and Warren, J. V.,The effect of temporary occlusion of arteriove-nous fistulas on heart rate, stroke volume, andcardiac output. J. Clin. Invest., 1951, 30, 215.

9. Lewis, T., and Drury, A. N., Observations relating toarteriovenous aneurism. Heart, 1923, 10, 301.

10. Van Loo, A., and Heringman, E. C., Circulatorychanges in the dog produced by acute arteriove-nous fistula. Am. J. Physiol., 1949, 158, 103.

11. Nickerson, J. L., Cooper, F. W., Jr., Robertson, R.,and Warren, J. V., Arterial, atrial and venous

pressure changes in the presence of an arteriove-nous fistula. Am. J. Physiol., 1951, 167, 426.

12. He E. C., Davis, H. A., and Rives, J. D.,Effect of acute A-V fistula on circulation time andauricular pressure in dogs. Proc. Soc. Exper.Biol. & Med., 1945, 60, 371.

13. Warren, J. V., Elkin, D. C., and Nickerson, J. L,

The blood volume in patients with arteriovenousfistulas. J. Clin. Invest., 1951, 30, 220.

14. Wegria, R., Frank, C. W., Misrahy, G. A., Sioussat,R. S., McCormack, G. H., and Sommer, L. S., Anew technic for the continuous recording of thecardiac output. Proc. Soc. Exper. Biol. & Med.,1950, 74, 551.

15. Stead, E. A., Jr., and Warren, J. V., Cardiac outputin man; an analysis of the mechanisms varyingthe cardiac output based on recent clinical studies.Arch. Int. Med., 1947, 80, 237.

16. Patterson, S. W., and Starling, E. H., On the me-chanical factors which determine the output ofthe ventricles. J. Physiol., 1914, 48, 357.

731