THE DEMONSTRATIONOF PULMONARYARTERIOVENOUSSHUNTSIN NORMALHUMANSUBJECTS, AND THEIR INCREASE IN

CERTAIN DISEASE STATES*

By ANTHONYD. JOSEt AND WILLIAM R. MILNOR

(From the Department of Medicine, Johns Hopkins University and Hospital,Baltimore, Md.)

(Submitted for publication April 28, 1959; accepted July 9, 1959)

The evidence for existence of arteriovenousshunts in the normal human lung is strong butnecessarily indirect. Simultaneous measurementsof oxygen tension in alveolar air and in arterialblood consistently show a degree of venous ad-mixture in both normal and abnormal subjects (1),but no information has been obtained about thepath taken by this unventilated blood. The pas-sage of small glass or plastic beads through thepulmonary circulation has demonstrated potentialpathways as large as 200ju in diameter (2, 3), butthis technique involves an artificial distention ofthe vascular bed and gives no information aboutthe extent to which such pathways may functionphysiologically.

The present investigation was based on thehypothesis that the decreased venous inflow to theright atrium and reduced blood flow through thepulmonary vessels during the Valsalva maneuver,as well as the rebound effects after release of themaneuver, might alter the flow through suchshunts, and that this alteration could be detectedby continuous measurement of the systemic arterialoxygen saturation. No distinction would neces-sarily be made by this technique between truearteriovenous anastomoses, with ventilation-per-fusion ratio of zero, and poorly ventilated alveolarcapillaries, with ventilation-perfusion ratio belowthe average for the whole capillary bed.

Wehave studied nine normal subjects, and 16patients with cardiac or pulmonary disease; re-sults in 12 additional patients with intracardiacshunts have been included for comparison.

METHODS

Ear oximetry proved unsatisfactory in recordingchanges in saturation during the Valsalva maneuver, ow-

* This investigation was supported in part by a grantfrom the American Heart Association.

t Present address: School of Medicine, Vanderbilt Uni-versity, Nashville, Tennessee.

ing to the large variations in blood content of the earcaused by the maneuver. The Colson Model 103 cuvettedensitometer 1 was therefore used, recording the opticaldensity of arterial blood at 620 m,u wave length. Asshown by Sabiston, Khouri and Gregg (4), in this por-tion of the spectrum the optical density of blood is alinear function of the per cent oxygen saturation of hemo-globin, provided no change in either hematocrit or ve-locity of blood flow through the cuvette occurs.

Patients were examined resting in the supine position.The cuvette densitometer was connected by approximately20 cm. of polyethylene tubing (internal diameter = 1.14mm.) to a needle inserted into the brachial or femoralartery. Arterial blood was drawn through the cuvetteby a motor-driven syringe at a constant rate, usually 0.3ml. per second. The output of the densitometer was re-corded on one chanel of a Sanborn "Poly-Viso" or an

"Electronics for Medicine" recorder. The sensitivity ofthe record was adjusted so that a 1 per cent change inoxygen saturation would produce a deflection of 1 to 2mm., using the neutral density wedge of the Colson cu-vette as a guide.

Calibration of the record for each patient was calcu-lated after completion of the test, when his arterial he-matocrit had been determined. Since the change in opti-cal density of blood for a given change in oxygen satura-tion depends on hemoglobin concentration, a calibrationchart could be prepared to give the sensitivity of our in-strument at different hematocrit levels. The error intro-duced by using hematocrit as an indication of hemoglobinconcentration (i.e., assuming the same mean corpuscularhemoglobin concentration for all patients) is negligiblefor this purpose.

The calibration chart shown in Figure 1 was con-structed from data obtained by in vitro measurements onfour different human blood samples of widely differenthematocrit values. In each sample the oxygen satura-tion was varied by aliquots of high and low saturationand the linear relation between optical density and oxygensaturation (4) determined. The slope of this line foreach sample is plotted on the ordinate of Figure 1, andits hematocrit on the abscissa. Using this chart and acalibrating signal of known optical density on each rec-ord, the recorded deflections could be converted intochanges in oxygen saturation. Figure 2 shows one of theoriginal records, with the calculated sensitivity scale at-

1 The Colson Company, Elyria, Ohio.

1915

ANTHONYD. JOSE AND WILLIAM R. MILNOR

06

regular changes in saturation synchronous with respira-tion and of magnitude 0.4 to 0.5 per cent on either sideof the mean. Saturation was recorded in six patientsduring simple breath-holding for 12 seconds; in four nosaturation change was found, while in two there was afall in saturation, of 0.6 per cent in one and of 1.1 percent in the other. Three rapid deep inspirations in thesame six patients caused no change in saturation infour, a rise of 1.2 per cent in one and a rise of 1.8 percent in one. In no case was there any difficulty in distin-guishing the effect of the Valsalva maneuver on arterialsaturation from the effects of simple respiration. Thiswas equally true in the patients with intracardiac shunts,although those with "balanced" shunts frequently showedrespiratory fluctuations in saturation as large as 2 percent on either side of the mean.

The changes in optical density of blood observed dur-, ing and

be duerhang,e.

HEMATOCRITFIG. 1. RESULTS OBTAINED BY THE IN VITRO STUDY

OF FOURHUMANBLOOD SAMPLES, SHOWINGTHE RELA-TIONSHIP BETWEEN HEMATOCRIT AND DENSITOMETERCALIBRATION

See text.

tached. To facilitate comparison, the records in Figure3 have been replotted at uniform sensitivity.

A sample of arterial blood was taken immediatelybefore each record and the control arterial oxygensaturation measured in a photometric oximeter (5)which was calibrated frequently with simultaneousVan Slyke determinations. Blood withdrawal throughthe cuvette was then begun, and after establishinga baseline on the record, the patient was asked toperform a previously rehearsed Valsalva maneuver,

blowing through a mouth tube into a closed air chamber.A mercury manometer visible to the patient measured thepressure thus produced and the patient endeavored tokeep this pressure at 40 mm. Hg for 12 to 14 seconds.In most cases, special care was taken to avoid a deep in-spiration before the maneuver; when this was neglected,a deflection related to this inspiration frequently appearedin the record of arterial saturation, preceding that relatedto the maneuver itself, (Patients E.J. and M.G., Figure 3).

In each case the mouth pressure developed during thetest was recorded by a strain gauge and amplifier to givean approximate indication of the changes in intra-thoracic pressure (6) which constitute the primary effectof the Valsalva maneuver.

Pulmonary arterial pressure was recorded in three pa-

tients in whom the study was performed during rightheart catheterization.

Baseline stability in the group of normal subjects was

such that no measurable saturation changes occurred dur-ing quiet respiration. In the group with heart or lungdisease, those whose condition was primarily cardiacshowed an equally stable baseline; three patients, in eachof whom the primary pathology was in the lungs, showed

A2RT. O2SAT.

PA

after the Valsalva maneuver could theoreticallyto acute changes in hematocrit rather thanin oxygen saturation, although hematocrit

902 LjfAA

IUU! -I

FIG. 2. ORIGINAL RECORD, SHOWING CHANGES IN

BRACHIAL ARTERYOXYGENSATURATION (ART. 02 SAT.),PULMONARYARTERY PRESSURE (PPA, MM. HG) AND

MOUTHPRESSURE(Pm, MM. HG.) DURING AND AFTER A

VALSALVA MANEUVERIN PATIENT J.PR.Effective pulmonary vascular perfusion pressure, esti-

mated as the difference between mean pulmonary arterypressure and mouth pressure, falls from a control levelof 24 mm. Hg to 6 mm. Hg within a few seconds after theonset of the Valsalva maneuver, maintains this level dur-ing the maneuver and rebounds temporarily to 38 mm.

Hg after release. Increased arterial saturation followsthe decreased perfusion pressure during the maneuver,

and decreased saturation below control levels follows therebound elevation of perfusion pressure. The time lagof approximately five seconds between pressure and satu-ration changes presumably represents transit time of bloodfrom pulmonary capillaries to peripheral artery.

20*

15-

10

#wj

.4

0

c O

qz_=

a >-_ l_

C -zZ Wl<a

, -J

< pun X

5a02 04

1916

1%1%

10

".. 10-

"....o _ -;

Iw-

PULMONARYARTERIOVENOUSSHUNTSIN MAN

',

..8

0

o 0

o o oo 0b o

Iiij

0

Id

ow o

(D (I r

0Ii

S 0 0

I; o oo o

I

z

n

- P

0Z

c'J U)

En

o cn

rr Z

oz>4

E-44

o M

rzN

9O

O t¢V\j2

b z

U0

O C

M 00

z ¢O0 Z

CDL;

1917

d2

-izSI za ig 1.-hq is I 8

all

ANTHONYD. JOSE AND WILLIAM R. MILNOR

changes of the order of 10 per cent would be neces-sary. To rule out this possibility the studies were re-peated in four patients using light of 800 mAwave length,so that the record would not be affected by changes inoxygen saturation, but would reflect changes in hemato-crit with a sensitivity approximately two-thirds thatat 620 m,. The records of these four patients at 620 m,uall showed significant changes during the Valsalva ma-neuver, larger than normal in two. At 800 mis, however,three of the records were completely flat, while the fourthshowed a very small deflection opposite in direction tothat seen at 620 m,u.

Changes in flow rate of blood through the cuvette, ifthey were to occur during the maneuver, would also

cause changes in optical density and lead to a similarfalse interpretation of the record. An increase in velocityof blood through the cuvette causes a diminution of therecorded optical density, owing probably to incompleteturbulence of blood in the cuvette at the rates of bloodflow employed. This sensitivity to flow is equal at both620 m, and 800 m,u wave lengths; the absence of changein records taken at 800 mA therefore indicates that nochanges in flow rate took place. Moreover, the syringewithdrawal unit was designed especially to avoid the oc-currence of such flow changes, and with experience, itsoperator can readily detect any irregularity in its func-tion. Changes in blood pressure in the cuvette causeno change in the record of optical density.

TABLE I*

Arterial 02 saturation

Rise Fallduring after

Patient Age Diagnosis At rest maneuver maneuver

Group 1

NormalNormalNormalNormalNormalNormalNormalNormalNormal

Group 2CHF, (Isch.)CHF, MI, AlMSMS, PHMS (mild)Previous LVF (Isch.)Cirrhosis of liverPrevious LVF (Isch.)Previous LVF (Isch.)Previous LVF (Isch.)PH, cause uncertainPH, lung fibrosisPH, lung diseasePH, cause uncertainPulmonary emboliPrevious cor pulmonale

Group 3ASD, Left-to-ASD rightASD shuntASDVSDASD, PHASD, Ebstein's diseaseVSD, PSVSD, PHVSD, PHVSD, PHPDA, PH

97 2.596 1.799 0.899 0.896 0.996 0.798 None96 1.095 1.1

94 None85 None95 1.898 1.298 1.997 1.497 2.792 2.198 2.194 6.092 6.485 7.381 6.992 5.494 3.993 0.5

95 2.495 1.699 0.899 0.998 1.292 5.570 0.490 None86 0.987 2.794 3.696 3.5

A. C.D. B.J.W.L. D.E. J.J. C.C. M.D. L.A. B.

B. B.M. V.A. Br.G. H.J. Pr.B. Br.L. E.F. M.R. F.M. L.H. G.H. A.A. Cu.C. Y.D. H.J. P.

M. T.R. S.M. G.G. W.M. B.E. W.I. N.D. G.J.S.J. R.J. B.J. M.

183240285843311832

47582736324556614846584827415436

251819263448383322562824

None0.50.71.71.34.50.90.92.6

NoneNone

0.21.01.50.4

None0.5

NoneNone

2.0None

0.5None

3.37.2

None1.54.40.90.47.94.7

10.77.88.2

16.613.3

* Key to abbreviations: CHF = congestive heart failure; Isch. = coronary artery disease; MI = mitral insuffi-ciency; Al = aortic insufficiency; MS= mitral stenosis; PH = pulmonary hypertension; ASD = atrial septal defect;VSD = ventricular septal defect; PS = pulmonic stenosis; PDA = patent ductus arteriosus.

1918

PULMONARYARTERIOVENOUSSHUNTSIN MAN

PATIENT SELECTION

Group 1 comprised two healthy members of the medi-cal staff (C.M. and A.B.) and seven hospital patients whohad no cardiovascular disease.

Group 2 patients all had obvious disease of the heartor lungs or both. Patients with severe emphysema were

not included. The individual diagnoses are listed inTable I. The stated diagnosis was confirmed by autopsyin Patients M.L., A.Cu. and L.E., and at mitral valvotomyin A.Br. and G.H.; none of these five patients had a pa-

tent foramen ovale. Patients H.G., H.A., D.H. and J.Pr.were studied by cardiac catheterization and dye dilutiontechniques, and H.A. by angiocardiography; in none was

there any evidence of an intracardiac shunt. The diag-nosis in the remaining seven patients was made clinically.In J.P. the diagnosis of "Pickwickian syndrome" had beenmade one year prior to this study, after a complete hos-pital investigation. By strict dietary treatment, he hadsince lost over 100 pounds in weight and made a com-

plete clinical recovery.

Group 3 included five patients with large left-to-rightintracardiac shunts, and seven in whom the shunt was bi-directional (Table I).

RESULTS

The changes recorded in all the patients studiedhave been listed in Table I.

Group 1

A measurable change in saturation was recordedduring or after the maneuver in every normalsubject. The pattern of change was closely simi-lar in each case; a rise in saturation occurred dur-ing the maneuver in eight of the nine subjectsand a fall in saturation followed release of themaneuver in eight of the nine; both changes were

seen in seven patients. The saturation rise aver-

aged 1.2 per cent, varying from 0 to 2.5 per cent;

the saturation fall averaged 1.5 per cent, varyingfrom 0 to 4.5 per cent. The three records ofnormal subjects shown on the left of Figure 3 il-lustrate the greatest (J.C.) and least (C.M.)changes observed in normal subjects.

Group 2

The two patients who were in congestive heartfailure at the time of study showed no changewhatever in saturation during or after release ofthe maneuver; they were the only patients studiedin whom no change was observed. In one ofthem (B.B.), a right atrial pressure record was

made during the maneuver and showed the char-acteristic "square-wave" pattern found in pa-tients with congestive heart failure (7).

In the other 14 patients of this group, thechanges seen were of the same pattern as those innormal subjects, with a rise in saturation duringthe maneuver in all 14 and a fall in saturationafter release in 12, reaching to below the restinglevel in nine. In seven of these 14 patients, themagnitude of the saturation changes was withinthe range seen in normal subjects, but in the re-maining seven, changes of increased magnitudewere recorded, rises of from 3.9 to 7.3 per centoccurring during the maneuver in six and a fallof 7.2 per cent following release in one. Therecords of three such patients, showing increasedsaturation changes, are illustrated in the center ofFigure 3.

Group 3

The direction of saturation changes was againsimilar. Three of the four patients with pre-dominant left-to-right shunt at atrial level showedchanges differing in no way from those seen innormal subjects (e.g., Patient R. S. in Figure 3);the fourth (M.G. in Figure 3), differed not inthe magnitude but in the abruptness of the fallin saturation following release. The one patientwith left-to-right shunt at ventricular level showedchanges identical to those in normal subjects.

The seven patients in whom the shunt was"balanced" showed changes in the same direction,but of generally greater magnitude, as illustratedby the record of J.M. in Figure 3. In this group,respiratory fluctuations of plus or minus 2 percent were common during quiet respiration andincreased markedly to plus or minus 4 to 7 percent on hyperventilation.

DISCUSSION

Patients without intracardiac shunts

The arterial saturation changes recorded in the23 patients with no intracardiac shunt (Groups 1and 2) presumably arose in the lungs. Func-tional patency of the foramen ovale might permita small right-to-left flow of blood after release ofthe Valsalva maneuver, as does an atrial septaldefect (8), but this would be expected in less

1919

ANTHONYD. JOSE AND WILLIAM R. MILNOR

than 10 per cent of the adult population (9) andwould not in any case explain the observed risein saturation during the maneuver. Although asmall amount of coronary venous blood is re-turned to the left side of the heart through theThebesian vessels, this flow is probably too smallto be of significance in the present results.

The intrapulmonary mechanisms which mightaccount for our results may be divided into:changes in alveolar oxygen tension (alveolar PO2),in diffusion capacity, in velocity of capillary bloodflow or in distribution of blood flow within thepulmonary vascular bed. Our evidence and thatof other investigators tend to exclude the first threeof these factors.

Although we did not measure oxygen tensions,the calculated effect of the initial 40 mm. Hg risein intrathoracic pressure is to raise a normalalveolar PO2 (104 mm.) by 5 to 6 mm. at most;as the maneuver is held for 12 to 14 seconds, thePO2 gradually falls, reaching close to the controlvalue; when intrathoracic pressure is released,there is an immediate fall followed by a small risecaused by a brief period of hyperventilation. Thewhole range of alveolar PO2 changes in normallyventilated alveoli probably does not exceed 5 to 6mm., and this could not account for any measur-able saturation changes in arterial blood. Inpoorly ventilated alveoli (e.g., PO2 50 mm.), therise in PO2 due to increased intrathoracic pressurewould be proportionally less (at most 3 mm.) and,except with gross ventilation defect and markedarterial blood desaturation, this could not causesignificant arterial saturation changes.

The Valsalva maneuver is unlikely to changegas distribution where this is already abnormaland we see no reason to suppose that the alveolarPO2 in such an area would be elevated by anymechanism other than that discussed above. Evenif it were to occur, such a change would be smallerthan that caused by several deep inspirations, andas has been described, these never caused satura-tion changes comparable in magnitude to thoseaccompanying the Valsalva maneuver.

An alteration in the diffusion properties of thealveolar membrane sufficient to account for theobserved saturation changes seems improbable, be-cause of the rapidity, the brief duration and thereversibility of the changes.

The absence of saturation changes in the twopatients studied during congestive heart failuresuggests that the changes are dependent upon thehemodynamic effects of the Valsalva maneuver,which are not obtained in such patients (7), ratherthan the ventilatory effects.

Simple changes in velocity of blood flow throughpulmonary capillaries during and after the ma-neuver would not significantly alter the end-capil-lary oxygen content, as equilibrium between al-veolar and capillary oxygen tension is achieved sorapidly (1, 10). The decreased pulmonary bloodflow of systemic shock, for example, or the in-creased pulmonary flow in congenital left-to-rightshunts, do not per se produce abnormal pulmonaryvenous oxygen saturation.

The explanation of our findings which we be-lieve to be correct is that a redistribution of bloodflow occurs within the pulmonary vascular bedduring the maneuver. If in the resting state asmall proportion of the pulmonary blood flowwere distributed to unventilated shunt vessels, andif during the maneuver the flow through such ves-sels were diminished, the oxygen saturation of themixed arterial blood would rise and the magnitudeof the rise would be determined by the amountby which such shunt flow was reduced relative tothe total flow. Similarly, an increase in the rela-tive flow through the shunt pathways followingrelease of the maneuver would cause a fall in thearterial saturation.

There is no direct evidence in the present re-sults to indicate which particular feature of thehemodynamic responses to the Valsalva maneuveris responsible for the redistribution of flow. Asdescribed by Lee, Matthews and Sharpey-Schafer(11), both transmural pressure and flow in thepulmonary artery fall progressively during themaneuver and overshoot after release before re-turning to their resting levels. In Figure 2, thefall in transmural pressure correlates closely inpattern with the rise in arterial saturation, and theovershoot in pressure with the saturation fall. Thesame correlation between changes in transmuralpulmonary artery pressure and in arterial satura-tion is found in patients with mitral stenosis onexercise (12) and on change of posture (13), inpatients with atrial septal defect and pulmonaryhypertension on exercise (14), and in dogs fol-

1920

P'ULMOUNARYART1ERIOVEV

lowing injection of glass beads into the pulmonaryartery (15). It is tentatively suggested therefore,that transmural pressure changes may influencepatency of the intrapulmonary shunts, at leastunder the conditions studied here. Good evidencealready exists that shunt vessels may be directlyinfluenced by acetylcholine (16) and by the oxy-

gen tension of inhaled gas (15).All the six patients who showed saturation rises

greater than the normal range had the clinicalsigns of pulmonary hypertension and this was

confirmed by direct measurement in five (pul-monary artery systolic pressure above 50 mm.

Hg). It is possible that the pulmonary hyperten-sion itself may have been related etiologically tothe presence in these patients of a large shunt flow.The Patient J. P. in Figure 3 (Group 2), whoshowed a fall of 7.2 per cent after release of themaneuver, may have had a patent foramen ovale.It is interesting, however, that he had sufferedone year previously from an episode of cor pul-monale caused by obesity and hypoventilation;since then he had lost over 100 pounds in weightand had made a complete clinical recovery. Analternative explanation for the saturation fall isthat it indicated the persistence of pulmonaryshunt vessels developed during the earlier episodeof cor pulmonale.

Uniformity of the hemodynamic response to theValsalva maneuver must be assumed before thepresent findings can be compared quantitatively.Obvious exceptions occur in congestive heartfailure, where little or no response is seen (7),and in emphysema, where greatly increased re-

sponses may occur (17). In the absence ofemphysema, however, hemodynamic changes are

rarely greater than normal and it is safe to as-

sume that large saturation changes indicate largeshunt flows.

The flow through shunt vessels at rest can beestimated from the rise in saturation and thecardiac output. In the group of normal subjects,the average rise of 1.2 per cent in saturation in-dicates a shunt flow at rest equal to about 1 percent of the total pulmonary flow. In two of thesix patients in whom a larger saturation rise was

found, the cardiac output was measured and simi-lar calculations indicated shunt flows at rest equalto 15 per cent and 20 per cent of the pulmonaryflows.

;NOUS SHUNTSIN MAN 1921

The relation of the shunt vessels discussed toalveolar capillaries and to the bronchial circulationcannot be evaluated in the present study. If, how-ever, the distribution of pulmonary blood flow be-tween these vessels and ventilated alveolar capil-laries is altered by simple changes in pressure orflow in the pulmonary artery, this would suggestthat the two pathways differ anatomically and thatthe shunt vessel concerned is not, in fact, an un-ventilated capillary.

Patients with congenital cardiac shunts

In the patients of Group 3, the changes recordedin arterial saturation must represent redistribu-tion of blood flow both within the lungs and acrossthe intracardiac shunt, no distinction between thetwo being possible by the present method. Changesin the intracardiac shunt obviously predominatedin some instances, producing much larger changesin saturation than were seen in patients of Groups1 and 2. When the intracardiac shunt was pre-dominantly from left to right, however, the satura-tion changes were often no larger than those seenin normal subjects. For this reason we cannotagree with the suggestion of Lee and Gimlette (8)that arterial saturation changes detected by theear oximeter following release of the Valsalvamaneuver are helpful in the diagnosis of atrialseptal defect.

If pulmonary hypertension behaves in this groupof patients as is suggested in those of Group 2, alarge arteriovenous shunt may develop in thelungs. This intrapulmonary shunt may explainthe pulmonary venous desaturation described incases of atrial septal defect with pulmonary hyper-tension (14, 18), the arterial desaturation per-sisting after ligation of a patent ductus arteriosuscomplicated by pulmonary hypertension (19), andthe presence of arterial desaturation in cases ofventricular septal defect with pulmonary hyper-tension insufficient to cause a reversal of the intra-cardiac shunt (20). This possibility should beconsidered before assuming that arterial desatura-tion indicates reversal of an intracardiac shunt.

SUMMARY

Arterial saturation changes have been recordedduring the performance of the Valsalva maneuver

'nTTT 'RAt-%KT A T-.'Xr A

ANTHONYD. JOSE AND WILLIAM R. MILNOR

in 38 subjects, using a cuvette densitometer to givea direct and continuous record of oxygen satura-tion from arterial blood.

A consistent pattern of change in saturation wasfound in patients without intracardiac shunts, con-sisting of a rise in saturation during the maneuverand a fall in saturation following release of themaneuver. In nine normal subjects, the satura-tion rise averaged 1.2 per cent, while in six of thepatients with cardiac or pulmonary disease, satura-tion rises of from 3.9 to 7.3 per cent occurredduring the maneuver.

The most probable explanation for these find-ings is that the saturation rise results from a de-crease in blood flow through pulmonary arterio-venous shunts, secondary to the hemodynamiceffects of the Valsalva maneuver, and that thosepatients who showed a large rise in saturation hadan abnormally large flow through such shunts atrest. The observations are consistent with thehypothesis that transmural pulmonary artery pres-sure is an important factor controlling flowthrough these shunts, and that an abnormally largeshunt flow may develop in patients with chronicpulmonary hypertension.

Evidence of shunt flow at rest was found ineight of nine normal subjects and averaged ap-proximately 1 per cent of total pulmonary bloodflow.

In 14 patients with cardiac or pulmonary dis-ease but without intracardiac shunts or signs ofcongestive heart failure, each case showed evi-dence of shunt flow at rest. The shunt was ab-normally large in six of the 14, amounting to 20per cent of total pulmonary flow in one subject.Two patients with congestive heart failure showedno change in saturation, presumably because theValsalva maneuver produced no hemodynamicchange in the pulmonary circulation.

The arterial saturation changes seen in 12 pa-tients with congenital intracardiac shunts repre-sented a summation of changes in shunt flowwithin the lungs and across the congenital defect,frequently dominated by the latter.

ACKNOWLEDGMENTS

We are indebted to Doctors Richard L. Riley andRichard H. Shepard for much helpful discussion and ad-

vice during the course of this work, and to Miss AMaryLou Sparger for technical assistance.

REFERENCES

1. Lilienthal, J. L., Jr., Riley, R. L., Proemmel, D. D.,and Franke, R. E. An experimental analysis inman of the oxygen pressure gradient from alveolarair to arterial blood during rest and exercise atsea level and at altitude. Amer. J. Physiol. 1946,147, 199.

2. Prinzmetal, M., Ornitz, E. M., Jr., Simkin, B., andBergman, H. C. Arterio-venous anastomoses inliver, spleen, and lungs. Amer. J. Physiol. 1948,152, 48.

3. Tobin, C. E., and Zariquiey, M. 0. Arterio-venousshunts in the human lung. Proc. Soc. exp. Biol.(N. Y.) 1950, 75, 827.

4. Sabiston, D. C., Jr., Khouri, E. M., and Gregg, D. E.Use and application of the cuvette densitometeras an oximeter. Circulat. Res. 1957, 5, 125.

5. Wood, E. H., and Geraci, J. E. Photoelectric deter-mination of arterial oxygen saturation in man.J. Lab. clin. Med. 1949, 34, 387.

6. Elisberg, E. I., Goldberg, H., and Snider, G. L.Value of intraoral pressure as a measure of intra-pleural pressure in man. J. appl. Physiol. 1951,4, 171.

7. Sharpey-Schafer, E. P. Effect of Valsalva's maneu-ver on the normal and failing circulation. Brit.med. J. 1955, 1, 693.

8. Lee, G. de J., and Gimlette, T. M. D. A simple testfor interatrial communications. Brit. med. J.1957, 1, 1278.

9. Thompson, T., and Evans, W. Paradoxical embo-lism. Quart. J. Med. 1930, 23, 135.

10. Farhi, L. E., and Riley, R. L. Graphic analysis ofmoment-to-moment changes in blood passingthrough the pulmonary capillary, including a dem-onstration of three graphic methods for estimatingthe mean alveolar capillary diffusion gradient(Bohr integration). J. appl. Physiol. 1957, 10,179.

11. Lee, G. de J., Matthews, M. B., and Sharpey-Schafer,E. P. The effect of the Valsalva maneuver onthe systemic and pulmonary arterial pressure inman. Brit. Heart J. 1954, 16, 311.

12. Donald, K. W., Bishop, J. M., and Wade, 0. L. Astudy of minute to minute changes of arterio-venous oxygen content difference, oxygen uptakeand cardiac output and rate of achievement of asteady state during exercise in rheumatic heartdisease. J. clin. Invest. 1954, 33, 1146.

13. Cardus, D., Mackinnon, J., and Wade, G. Circula-tory effects of changing position in mitral disease.Brit. Heart J. 1958, 20, 233.

14. Luchsinger, P. C., Moser, K. M., and Buhlmann, A.Physiologic definition of pulmonary vascular status

1922

PULMONARYARTERIOVENOUSSHUNTSIN MAN

in atrial septal defect (abstract). Circulation1957, 16, 912.

15. Niden, A. H., and Aviado, D. M., Jr. Effects ofpulmonary embolism on the pulmonary circulationwith special reference to arteriovenous shunts inthe lung. Circulat. Res. 1956, 4, 67.

16. Wood, P. Pulmonary hypertension with specialreference to the vasoconstrictive factor. Brit.Heart J. 1958, 20, 557.

17. Stucki, P., and Luthy, E. Les variations de latension arterielle chez l'emphysemateux lors del'empreuve de valsalva. Arch. Mal. Coeur. 1955,48, 441.

18. Jonsson, B. in Diagnosis of Congenital Heart Disease,A Clinical and Technical Study by the CardiologicTeam of the Pediatric Clinic, Karolinska Sjuk-huset, Erica Odelberg, Trans. Chicago, Year BookPublishing Company, 1955, pp. 285 and 338.

19. Limon Lason, R. Pulmonary hypertension in patentductus arteriosus in Pulmonary Circulation, W.Adams and I. Veith, Eds. New York, Grune andStratton Inc., 1959, p. 216.

20. Mannheimer, E., Ikkos, D., and Jonsson, B. Prog-nosis of isolated ventricular septal defects. Brit.Heart J. 1957, 19, 333.

1923