1985. chest. hipoxemia in acute pulmonary embolism

CHEST / 88 / 6 I DECEMBER, 1985 829

Hypoxemia in Acute Pulmonary EmbolismYann Huet, M.D.;* Francots Lemaire, M.D.;*

Christian Brun-Buisson, M.D.;* William A. Knaus, M.D.;*t

Bernard Teisseire, Ph.D.;1: Didier Payen, M.D.;* and

Didier Mathieu, M.D.�

Most patients with severe, acute pulmonary embolism (PE)

have arterial hypoxemia. To further define the respectiveroles of ventilation to perfusion (VA/Q) mismatch and in-

trapulmonary shunt in the mechanism of hypoxemia, weused both right heart catheterization and the six inert gas

elimination technique in seven patients with severe, acutePE (mean vascular obstruction, 55 percent) and hypoxemia(mean Pa08, 67±11 mm Hg). None had previous cardiopul-monary disease, and all were studied within the first ten

days of initial symptoms. Increased calculated venous ad-mixture (mean QV�t/QT 16.6 ± 5.1 percent) was present inall patients. The relative contributions of VA/() mismatching

and shunt to this venous admixture varied, however, accord-ing to pulmonary radiographic abnormalities and the timeelapsed from initial symptoms to the gas exchange study,

cute pulmonary embolism (PE) remains one of the

more common diagnoses of acutely ill, hospi-

talized patients and one of the more frequent complica-

tions to occur during hospitalization for other disor-

ders.’2 Hypoxemia is a consistent and important

clinical feature of PE � Its pathophysiology and pre-

cise clinical significance in man, however, have not

been clearly established. Ventilation/perfusion (VA/a)

abnormalities,45 impaired diffusion,3 and intrapulmon-

ary shunting, as well as right-to-left intracardiac

shunts,4’6’7 all have been implicated, but the exact role

of each remains unclear.

Although hypoxemia has been related to the degree

of pulmonary vascular obstruction (PVO),8 there are

reports of massive PE with normal arterial oxygen

tension (Pa02).#{176}No clear relationship between other

clinical characteristics and the PaO, has been found.

Variations in the timing and methods used to assess

hypoxemia in previous studies may have contributed to

this confusion. The major gas exchange studies in

humans were performed long, frequently months,

*SeMce de Reanimation M#{233}dicale, H#{244}pital Henri Mondor and

Universit#{233} Paris-Val de Mane, France.tICU Research Unit, The George Washington University Medical

Center, Washington, DC.�Laboratoire des echanges gazeux (INSERM U 138), H#{244}pitalHenri

Mondor and Universit#{233} Paris-Val de Mane.§Service de Radiologie, H#{244}pital Henri Mondor and Universit#{233}

Paris-Val de Marne.This work was supported in part by a grant from the D#{233}partementde Pharmacologie Clinique, H#{244}pital Henri Mondor, Creteil.Presented at the Annual Meeting of the American Thoracic Society,Anaheim, Calif, May, 1985.

Manuscript received April 2; revision accepted July 15.

Although all patients had some degree of VA/(I mismatch,the two patients studied early (ie, <48 hours following acute

FE) had normal chest x-ray film findings and no significantshunt; VA/� mismatching accounted for most of the hypox-emia. In the others a shunt (3 to 17 percent of cardiac output)was recorded along with radiographic evidence of atelec-

tasis or infiltrates and accounted for most of the venousadmixture in one. In all patients, a low mixed venous oxygentension (27 ± 5 mm Hg) additionally contributed to thehypoxemia. Our findings suggest that the initial hypoxemia

of acute FE is caused by an altered distribution of ventilationto perfusion. Intrapulmonary shunting contributes signifi-

cantly to hypoxemia only when atelectasis or another cause

of lung volume loss develops.

after the PE. In one of these, Wilson et al’#{176}suggested

that atelectasis and shunt accounted for most of the

hypoxemia, whereas Kafer2 attributed the major role to

VA/Q mismatching. In these studies the investigators

were unable to examine directly the relative contribu-

tions of “A/a mismatching vs shunt at different times

after the embolus. Both studies also used the oxygen

method to assess hypoxemia, and the degree of in-

trapulmonary shunting was not measured but inferred

using the alveolar-arterial oxygen difference.

Today, bedside right heart catheterization gives an

accurate evaluation of the contribution of cardiac

factors to the hypoxemia.” The recent introduction by

Wagner et al’2 of the multiple inert gas elimination

technique allows a more precise evaluation of VA/�

distributions than was possible using the oxygen

method alone.

In this study, we combined these two techniques to

study seven patients with recent, angiographically

documented acute PE. Our aim was to determine the

mechanism of their hypoxemia by relating the timing

of our investigation and each patient’s chest x-ray

abnormalities to the results of their gas exchange

study.

Patients

PATIENTS AND METHODS

Seven consecutive patients admitted to our Intensive Care Unit

(ICU) for hemodynamic monitoring and treatment of acute PE were

studied. Their individual clinical characteristics are in Table 1. None

had any significant preexisting cardiopulmonary disease. One pa-

Downloaded From: http://journal.publications.chestnet.org/ on 08/10/2013

Table 1-Clinical Features and Chest X-ray Findings in 7 Patients with Acute Pulmonary Embolism

830 Hypoxemla In Acute Pulmonary Embolism (Huet of a!)

Clinical Features

. Studyl-

Patient Age,

Hours Since

Occurrence Precipitating

.

Pulmonary

Vascular

X-ray Fi ndings at Time of

..

Elevated Platelike

No. * yr/Sex of PE Cause Obstructiont Diaphragm Atelectasis Infiltrates

1 4IJF 24 Immobilization

Pelvic surgery

58 + 0 0

2 41/M 48 Lower limb

fracture

60 0 0 0

3 40/F 84 Immobilization

Hepatic cancer

65 + + + + 0

4 69/F 102 Lower limb

fracture

56 0 + +

5 50/M 128 Immobilization

Neurosurgery

35 + + + + +

6 61/M 192 Immobilization

Abdominal surgery

65 + + + + + +

7 50/M 216 Lower limb

fracture

50 ++ ++ ++

*Numbering according to time elapsed since initial symptoms of embolism.

tIn % of pulmonary vascular bed, estimated according to index described by Tibutt et al.#{176}

�Notation as follows: 0 sign absent; + minimal changes; + + marked and/or multiple changes.

tient had suffered a pulmonary embolism five yeas before but had no

residual effects. The diagnosis of PE was made on clinical history andthe results of a pulmonary angiogram perfurmed within a few hoursof ICU admission. The degree of PVO was determined using the

index described by Tibutt et al.#{176}A Swan-Ganz catheter and radialartery line were already in place. The gas exchange study was

perfbrmed on the first day in the ICU, approximately two hours after

the pulmonary angiogram. The delay between the first symptom ofPE (acute chest pain and/or dyspnea) and the inert gas study was

carefully determined and ranged between one and nine days (Table1). The patients were receiving only heparin therapy at the time ofthe study. No vasoactive drug was used. After the study, we gave

urokinase to all except patient 5, who was recovering from recentneurosurgery.

Methods

The gas exchange study was done with patients in a semirecum-

bent position. All were breathing spontaneously via a face mask with

humidified room air or supplemental oxygen (patients 3 and 7) as

required to maintain an adequate oxygen arterial partial pressure.

The masks were tightly fitted to the patient’s face so that no air leaks

occurred. Patients breathed through a circuit consisting of an 0,

blender, a humidifier, and a specially designed heated mixing box via

a one-way valve. The expiratory tubing was also heated to prevent

condensation and extraction of the more soluble gas, ie, acetone.

In each patient, the following measurements and calculations

were made: (1) minute ventilation WE) and tidal volume (VT) using a

calibrated Wright spirometer; (2) fraction of inspired 0, (Flo,) by

sampling gas in the inspiratory line of the circuit; (3) right atrial,

pulmonary, and systemic arterial pressures (Statham P23 DB trans-

ducer and Honeywell recorder); (4) cardiac index (CI), derived from

cardiac output (CO) measured by the thermodilution technique

perfurmed in triplicate (CO computer 9520, A. Edwards Laborato-

ries); (5) arterial and mixed venous oxygen tension and carbon

dioxide tension (PaO,, P�O,, PaCO,) (ABL 30, Radiometer

Copenhagen) and 0, saturation (SaC), and S�O,) (OSMII; Radi-

ometer); (6) total pulmonary resistance (TPR) calculated as mean

pulmonary arterial pressure (PAP) divided by cardiac index (pulmo-

nary capillary wedge pressure measurements were possible only in

two patients, most likely because of extensive pulmonary arterial

obstruction); (7) venous admixture (QVAJQT) using the oxygen

method and computed with the standard equation of Berggren; (8)

physiologic dead space assessed by the Bohr dead space for carbon

dioxide calculated as (arterial-mixed expired) divided by PaCO,.

Anatomic dead space was assumed from body weight (1 mI/Ib) plus

instrumental dead space (80 ml); (9) distribution of”.&/(� ratios using

the multiple inert gas elimination technique, as described by Evans

and Wagner.’4

A brief summary of this last technique follows. Six inert gases

(sulfur hexafluoride [SF61, ethane, cyclopropane, halopropane,

ether, and acetone) were equilibrated in a 5 percent dextrose

solution and infused intravenously (IV) at a constant rate of 5 mI/mm

for 25 minutes. Then, simultaneous 15-mI blood samples were drawn

from the pulmonary artery and a systemic artery for analysis of the

inert gas concentrations. At the same time, 50 ml of expired gas was

sampled. The blood samples were equilibrated with nitrogen in a

heated bath for 45 minutes.

The equilibrated blood and the expired gas samples were analyzed

for SF6 by an electron capture detector (Packard 429; Packard

Instrument Co) and for the five other gases by a flame ionization

detector (Packard). The blood solubility coefficient of each gas wasdetermined for each patient. The relative concentrations of the inert

gases in arterial and mixed venous samples were derived from the

equilibrated samples and the measured solubilities. The error of

measurement of SF6 determined in our laboratory was 1.3 percent.

and less than 1.8 percent for the other five gases.�S

Retention-solubility and excretion-solubility curves were then

constructed. From these, the best approximation of the ventilation-

perfusion distributions was determined, using the ridge regression

technique.’4 The mean residual sum of square was 3.2 ± 2.0, indicat-ing compatibility between the distribution curves and the inert gas

ram data.’4 A computer-assisted analysis of the distribution of the

VA/Q ratios allowed the determination of shunt (Qs/Qt); ie, Y..�JQ

<0.005; low VA/Q; ie, 0.005<VA/Q<0.1; high VA/Q; je, 10<VA/Q

<100; and dead space (Vu/VT); ie, VA/Q<100. The mean peak valuesof blood flow and ventilation distribution were determined, as well as

the SD of the distribution calculated with a natural log scale for the

abcissa (log SD), which is taken as an index of dispersion of blood flow

and ventilation. The ventilation to perfusion distributions were

combined with the mixed venous blood gases, cardiac output, VE,

and PaO, in the 50-compartment model’4 to predict the arterialpartial pressure of oxygen.


Patient

No.

HR,

beats/mm

RAP

mm Hg

PAP

mm Hg

SAP

mm Hg

CI,

Limin#{149}m’

TPR,

mm Hgfmin/m’/L

1 105 2 20 100 4.76 4.2

2 110 12 24 70 1.91 12.5

3 145 15 29 60 2.58 11.2

4 90 10 32 87 2.54 12.6

5 103 2 24 76 2.25 10.7

6 84 3 20 103 2.65 7.5

7 110 3 19 106 3.2 5.9

Mean 107 7 24 86 2.84 9.2

±SD 20 5 5 18 0.94 3.3

*HR = heat rate; �XP = mean right atrial pressure; = mean pulmonary arterial pressure; = mean systemic arterial pressure; CI =

cardiac index; TPR = total pulmonary resistance.

Table 2-Hemodynamic Data*

CHEST I 88 I 6 I DECEMBER, 1985 831

A standard chest x-ray film was also obtained on the day of the

study. Analysis of the films was made by two independent observers

who did not know the patient’s clinical characteristics. According to

Fleischner,” they described each film with emphasis on the follow-

ing: (1) elevated hemidiaphragm were marked present or absent; (2)

platelike atelectasis and areas of consolidation (“infiltrates”) were

scored 0 for no abnormality, + for slight opacities, and + + for

marked and/or multiple opacities. There was close agreement

between the two observers on all films.

All the data are expressed as mean ± SD.

RESULTS

In all tables and figures the patients are listed from 1

to 7 corresponding to the time elapsed from initial

symptoms of PE to the gas exchange study. The chest

x-ray findings are recorded in Table 1. The two patients

studied within two days of initial symptoms had no

parenchymal radiographic abnormalities. Localized

atelectasis and/or small infiltrates were common in

patients studied later. Pulmonary arterial obstruction

was substantial in all patients (mean value, 55 percent

of the pulmonary arterial vascular bed). There was no

obvious geographic relationship between radiologic

findings and site of embolus.

The standard hemodynamic data are shown in Table

2. Right atrial pressure ranged from 2 to 15 mm Hg.

Mean pulmonary arterial pressure was only moder-

ately increased (24 ± 5 mm Hg). Cardiac index was

moderately low (2.84±0.96 L/minm’) and notably

depressed in patient 2 (1.91 L/min#{149}m’). Total pulmo-

nary resistance was markedly elevated (9.2 ± 3.2

mmHg/L.min.m’). The hemodynamic impact of PE

was further evidenced by the increased arteriovenous

oxygen difference (mean 5.7 ± 1.5) with a calculated

mean oxygen consumption of 165±39 mllminm’. No

patient, however, had a systolic blood pressure less

than 90 mm Hg.

Standard gas exchanges are presented in Table 3. A

mild hypoxemia (mean 67 ± 11mm Hg) was recorded in

all patients. The alveolar-arterial 02 difference wasenlarged, but did not correlate to the degree of pul-

monary vascular obstruction or to the PaO,. Calculated

venous admixture (0, method) was elevated in each

patient, and ranged from 10.5 to 25.6 percent of cardiac

output (mean, 16.6 ± 5.1). Mixed venous oxygen partial

pressure and saturation were low (mean, 27 ± 5 mm Hg

and 52 ± 10 percent, respectively), as could be ex-

pected from the decreased cardiac index. The Bohr

dead space was markedly high, ranging from 44 to 75

percent of minute ventilation. All of the patients had an

arterial hypocarbia, and respiratory alkalosis was noted

in five of seven patients (mean pH, 7.47±0.04).

The inert gas study revealed different patterns of

VA/Q distribution in relation to the timing of the study

and the presence of x-ray abnormalities. Shunt, ie,

Table 3-Standard Gas Exchange in 7 Patients with Acute Pulmonary Embolism

Patient

No.

Weight,

kg

Height,

cm

PaO,,*

mm Hg

SaO,,

%

PaCO,,

mm Hg pH

P(A-a)0,,

mm Hg

P�O,,

mm Hg

S�O,,

mm Hg

(>VA/(>r,

%

VD/VT

(Bohr), %

1 45 162 66 94 34 7.45 42 34 65 19.3 57.6

2 73 180 60 93.5 27 7.52 55 25 45 15.8 66

3 71 170 90 95 29 7.40 88 25 43 10.5 75

4 62 157 62 92.5 26 7.49 53 21 44 15.3 62

5 68 180 65 92.5 31 7.49 43 25 45 15.8 47

6 72 170 57 89.5 37 7.42 44 29 57 25.6 62

7 80 171 70 93.5 26 7.51 112 32 65 18.5 44

Mean 67.3 170 67 93 30 7.47 - 27 52 16.6 59

±SD 10 8 11 2 4 0.04 - 5 10 5.1 10

*�4.ll patients breathing room air, except 3 and 7 (FIo,= 0.30). Abbreviations: PaO, = arterial partial pressure of oxygen; SaO, - arterial

oxyhemoglobin saturation; PaCO, = arterial carbon dioxide tension; P(A-a)O, = alveolar-arterial oxygen tension difference; P�O, = mixedvenous partial pressure of oxygen; Sc�o, = Mixed venous oxyhemoglobin saturation; QVA/QT = venous admixture (oxygen method); VD/VT

dead space (Bohr).


Perfusion Distribution, Ventilation Distribution,

Perfusion to Ventilation Anatomic

%CO %VE. . �,

Pa02/PaO,, Unventilated to Unperfused Dead Log SD of Log SD of

VE, VT, CO, Meas/Pred, Lung,t Lung,1 Space,* Mean VAJQ Blood Flow Mean VA/Q Ventilation

Limn ml L/mn mm Hg % % % Ratio Distribution Ratio Distribution

1 11 730 5.3 66/72 0 56.8 23.3 0.58 0.838 1.58 1.129

2 20 910 3 60/56 0.3 53 24.8 1.65 1.235 5.45 0.877

3 31 910 4.4 90/106 4.6 63.4 24.4 1.22 1.163 5.01 1.031

4 21 820 5.7 62168 3.2 55.1 24.5 1.39 1.142 2.19 0.877

5 14 700 6.1 65/63 4.2 40.5 30.8 1.03 0.842 1.96 0.723

6 13 650 5.4 57/66 8.8 58.4 32.5 0.89 0.815 1.26 0.494

7 17 770 6.3 70/66 16.9 35.4 31.2 1.66 0.691 2.68 0.676

Mean 18 784 5.2 67/71 5.4 51.8 1.20 0.950 2.88 0.83±SD 7 101 1.1 11/15 5.9 10.1 0.40 0.210 1.67 0.22

#{176}Abbreviations: iTE = expired ventilation; VT = tidal volume; CO = cardiac output.

t’7A/( 0.005.�‘A/() 100.§Anatomic dead space assumed from weight (1 ml/lb) plus instrumental dead space.

a

Sc

[

0

Atelectasis

o 0 ++ + + ++ ++

Infiltrat#{149}s ++ �1+

o 0 0 + ++

.15

.10 -

.fl.ii.flI I I I I I F I

1 2 3 4 5 6 7 8 9 Days

FIGURE 1. Association between shunt measured with the six inert gas technique and x-ray findings in relation

to time elapsed since the occurrence of PE. X-ray changes scored as: 0 = no abnormality; + = slightopacities; + + = marked and/or multiple opacities. It is clear that a significant shunt is present only when

chest x-ray is abnormal. Conversely, normal chest films were obtained in patients with no detectable shunt

and studied within 48 hours after embolism.

Table 4-Inert Gas Data#{176}

832 Hypoxemla In Acute Pulmonary Embolism (Huot of a!)

perfusion to the unventilated lung, was negligible in

patients 1 and 2 but accounted for a low but significant

part of cardiac output (3.2 to 8.8 percent) in patients 3

to 6, and to as much as 17 percent of cardiac output in

patient 7 (Table 4). There was no relationship between

venous admixture measured with the 02 method and

Qs/�t (six inert gases method).

Figure 1 correlates the shunt to the x-ray abnormali-

ties, and shows that a significant shunt occurred only

when atelectasis and/or infiltrates were seen radio-

graphically. Figure 2 shows the VAJQ distribution for

each patient and a normal control. Analysis of the \TA/�

distribution shows that the mean peak of blood flow,

although not markedly different from 1 (1.2 ± 0.4), had

a large dispersion (0.95 ± 0.21). Thus, a significant part

of perfusion was distributed to lung units with “A/(�

ratios lower than 1 (between 1 and 0.1), despite the

absence of a definite zone of very low VAJQ ratios. The

measured PaO, values were closely related to the PaO,

values predicted from the PiIO, and the VA/�

distributions (Table 4).

In all patients, ventilation to the unperfused areas


CO 4.4

SHUNT

1.1%1.4

YE 74

©DEAD SPACE

26.5%

C

E

IS0

000

0z4

I0,01 01 10

1.4

1.0

1.4

1.0

0.6

0.2

IA

1.0

o.e

0.2

/.‘

/ / \\�1/ .�‘.__J�

ii” 001 0.1 10 100

VE 31DEAD SPACE

63.4%

1�

/�‘��OI O�I�’ � 10 500

E

SHUNT

4.2%

0.6

0.2

1.4

1.0

0.6

0.2

1.4

1.0

0.8

0.2

YE 14

DEAD SPACE

40.5%

0

CO 5.7 (�) VE 21

SHUNT DEAD SPACE3i% 55.5%

W. I I ‘I I0 0.01 0.1 1 10 100

CO 5.4 © YE 13

SHUNT DEAD SPACE8.8% 58.4%

1.4

to

08

0.2

1.4

1.0

0.6

0.2

CO 6.3 YE I?

SHUNT DEAD SPACE

16.9% 35.4%

0:01 � � i#{243}o10 100

I.8


IS0

000

0z4

C03 VE 20

SHUNT DEAD SPACE0.3% 53.0%

I’

001 0.1 1 1�O tOO

CO 5.3

SHuNT0.0%

CO 4.4

SHUNT4.6%

VENTILATION - PERFUSION PATIO

A

1/III

-�V� #{149}#{149}‘t Io 0.01 0.1

YE 11DEAD SPACE

56.8%

VENTILATION-PERFUSION RATIO

FIGURE 2 (A, upper, and B, lower). Distribution of ventilation and perfusion, according to the ventilation!

perfusion ratios in a normal subject (0) and 7 patients with acute PE (ito 7). Co = cardiac output (Llmin); VE

= minute ventilation (Limmn). Normally, perfusion and ventilation are distributed to VA1Q around 1, with

tall and narrow unique peak. No shunt exists. By contrast, patients 1 and 2, studied within the 48 first hours,

have no shunt but a marked enlargement of the perfusion peak. Thus, a significant part of perfusion isdistributed to VA/Q between land 0.1. The ventilation peak is shifted to the right. Dead space is markedly

increased. Patients 6 and 7, studied later, have a normal mode of VAJQ ratio distribution, with a narrowed

perfusion peak around 1 and a shunt of 9 and 17 percent, respectively.

(dead space) was markedly high (52±10 percent) and

accounted for as much as 63 percent of YE in patient 3

(Table 4). Anatomic dead space assumed from body

weight was markedly smaller than inert gas dead space

(27 percent vs 52 percent, respectively). As a conse-

quence of the increased dead space, minute ventilation


834 Hypoxemla in Acute Pulmonary Embolism (Huet of a!)

(‘�‘E) was consistently increased (mean VE 17±7 L/min)

and was above 20 IJmin in patient 2. The mean peak of

ventilation was also consistently shifted to the high

VA/� ratios (mean, 2.88 ± 1.67), with a large dispersion

of the distribution (.83 ± .22).

The VA/Q diagrams in Figure 2 make it possible to

contrast these different patterns of VA/� distribution

and to compare them to normal. For example, in

patient 1, who was studied within 24 hours of the

embolic event and had normal chest x-ray film find-

ings, there is substantial dispersion of the perfusion

peak but no shunt. In patient 6, studied one week afterthe onset of symptoms and presenting with atelectasis

and infiltrates chest x-ray film, we found a shunt of 9

percent. His VA/Q distribution, however, is close to

normal, with a mean value centered on a VA/Q ratio of

approximately 1.

DIscUssIoN

Hypoxemia has been consistently demonstrated in

acute PE,345’#{176}but its etiology remains unclear. We

believed that part of the difficulty in discovering its

cause was related to variations in the selection of

patients or in the severity of the embolism and to the

long delay between initial symptoms and clinical

investigation.5 In addition, past studies have in-

terpreted hypoxemia in regard to associated changes in

P(A-a)0, and venous admixture.’#{176} The clinical signifi-

cance of these measurements are difficult to deter-

mine, since they are influenced by shunt, VAJ�

mismatching, and impaired diffusion.’7

In this study, we examined a group of patients with

large (mean vascular obstruction 55 percent) but

nonlethal acute pulmonary embolism shortly after

their first symptoms. All were hypoxemic (mean PaO,,

67 ± II), and each had moderate hemodynamic depres-

sion from the acute embolism, but none required

vasoactive drugs. We combined right heart catheteri-

zation with the multiple inert gas elimination tech-

nique to determine the relative impact of pulmonary

shunt, ‘s’A/Q mismatch, impaired diffusion, and cardiac

factors on the hypoxemia. We found that their hypox-

emia was caused by different pulmonary mechanisms,

depending on chest x-ray findings and on the time

elapsed since the PE. In patients studied early follow-

ing embolism, hypoxemia was mainly due to perfusion

of lung units with low VA/C ratios (ie, VA/�

mismatching), without a significant contribution from

perfusion of blood through unventilated lung units, ie,

shunting (Fig 1 and 2). Shunt became a significant

contributor to hypoxemia only if atelectasis or other

causes of lung volume loss developed. In our series

these characteristics were found in patients studied 48

hours or more after the initial symptoms.

Three major mechanisms have been proposed to

explain the pulmonary contribution to the hypoxemia

following acute PE: (1) a decreased 0, diffusing capac-

ity across the pulmonary membrane;3’8 (2) ventilation

to perfusion mismatching;46 and (3) shunting of blood,

at either the intracardiac or the intrapulmonary

level.�7’#{176} An embolus could theoretically cause a

decrease in diffusing capacity by reducing the surface

area of the alveolar capillary membrane available for

gas exchange or by decreasing the transit time of blood

flow through a reduced pulmonary vascular bed. But

the measured PaO, values were nearly identical to the

PaO, values predicted from the actual ‘,�A/�

distributions. This suggests that impaired diffusion was

not a cause for hypoxemia in our patients, a finding in

agreement with several recent experimental and clini-

cal studies.’�”

In contrast, in most of our patients an increased

venous admixture (0, method) in the absence of a

parallel increase in Qs/�t (inert gases method) indi-

cated that VAIQ mismatching accounted for a substan-

tial portion of their hypoxemia. This was most apparent

in patients studied early, eg, 1 and 2, in whom virtually

no shunt was found despite a venous admixture of 19

and 16 percent of cardiac output. The inert gas study

confirmed the VA/� mismatching in these patients by

documenting increased dispersion of perfusion, with-

out a definite zone of very low � (Fig 2). The low

VA/� ratios may be explained on the basis of an

increased perfusion to the nonembolic lung.” This was

the conclusion from a recent automated computer

analysis of ventilation-perfusion lung scans in II pa-

tients with acute pulmonary emboli and normal chest

x-ray findings.” In each of these patients the analysis

showed lung areas remote from those affected by the

emboli that had an abnormal matching of ventilation to

perfusion with relative overperfusion.

It has also been suggested that this \TA/� mismatch

could be caused by selective pneumoconstriction in

areas contiguous to the embolic lung that remain per-

fused.” This has been demonstrated in experimental

canine thromboembolism and is attributed to reflex

bronchoconstriction or pneumoconstriction.’4 Bron-

choconstriction has also been consistently shown in

human pulmonary embolism,4 and a shift of ventilation

away from the unperfused lung has been demonstrated

in man after the temporary occlusion of one pulmonary

artery.” This bronchoconstriction or pneumoconstric-

tion may then be responsible for subsequent loss of

lung volume in the embolic lung.9

Our data suggest that when such volume loss occurs,

shunt enters the clinical picture. The high incidence of

discoid atelectasis in patients with measurable shunts

has already been emphasized.’#{176} In experimental PE

studies, shunt as a major contributor to hypoxemia is

usually reported when alveolar collapse is seen at

autopsy. Wilson et al’#{176}found that they could tem-

porarily lower their patients’ P(A-a)O, if they were



encouraged to breathe 80 to ifi percent of their

predicted inspiratory capacity, thus decreasing the

amount of focal alveolar collapse. Similarly, no shunt

was demonstrated in the study by Caldini6 when

mechanical ventilation with periodic sighs was used. In

the patients we studied, clinically significant shunting

was found only in those with atelectasis and infiltrates

(Fig 1). It may be that significant loss of lung volume

and alveolar collapse developed in lung areas corre-

sponding to those of pulmonary vascular obstruction

and that early reperfusion of these zones produced

shunting.

Pulmonary edema from alteration of pulmonary

alveolar-capillary permeability or from increased blood

flow to the still-perfused lung is an alternative explana-

tion for shunting in acute PE, and alveolar edema has

repeatedly been shown in animal experiments.’6’7

However, convincing clinical reports of PE with acute

pulmonary edema in patients free of left heart disease

are distinctly rare.” None of our patients had x-ray

evidence of acute edema in nonembolic areas of the

lung. Subclinical focal pulmonary edema, however,

cannot be excluded in the areas of consolidation.

Remaining explanations for right-to-left shunting in

acute PE are intracardiac shunts from the opening of a

patent foramen ovale6’7 or intrapulmonary shunts from

the opening of precapillary shunts.3 The opening of a

patent foramen ovale cannot be entirely ruled out in

our patients, but their right atrial pressures were in the

normal range, especially in those with the largest

shunts. The opening of precapillary shunts never has

been demonstrated in patients or in recent experimen-

tal studies,” and no relationship was found between

shunt and PAP in our patients.

As already mentioned, however, the magnitude of V

A/� mismatching and shunt were moderate and dispro-

portionate to the degree of hypoxemia recorded in our

patients. This suggests that an additional factor may

worsen the hypoxemia. Except patient 1, all of our

patients suffered from hemodynamic changes associ-

ated with their acute PE. Because of their low cardiac

output, they had low P�O, (27 ± 5 mm Hg), and this no

doubt contributed to their hypoxemia. The potential

role of a low P�O, for any level of shunt or low VA/� has

recently been emphasized by Dantzker and Bower.’6

Assuming that VA/� distributions would remain un-

changed, these authors calculated that a marked in-

crease in predicted PaO, would result from an increase

of Pi�O, to 40mm Hg. This explains why PaO, values in

this study were significantly reduced despite relatively

mild degrees of’�/� mismatch and shunt. Therefore,

depression of cardiovascular function is a major con-

tributor to the hypoxemia in acute PE.

Dead space was consistently increased in all pa-

tients. Consequently, minute ventilation was mark-

edly increased, up to 31 L in patient 3, and the peak of

ventilation shifted to the right, with a mean value of

2.88. Measured dead space, using both the expired

CO, (Bohr, 59 ± 10 percent) and the excretion of

soluble gases (inert gas dead space, 52±10 percent)

was higher than the predicted anatomic dead space

(27 ± 4 percent), corresponding to the ventilation of

unperfused lung. In addition to the dead space,

patients studied within the first 48 hours following PE

had a supplemental mode of ventilation, with high

VA/Q ratios (Fig 2). In dog experiments, Dantzker et al

have already demonstrated the early occurrence of

such a high mode, with VA/Q between 10 and 100, that

disappeared after a few hours.

The major limitation to our findings is the small

number of patients studied and that, for practical

reasons, the inert gas study could not be repeated

within individuals to document changes in VA/Q and

shunt over time. A larger number of patients would

have permitted the use of stronger statistical testing to

demonstrate the relationship between atelectasis and

shunting. Repeated determinations of gas exchange in

the same patient would have further supported the

observations made between individuals. Neverthe-

less, we believe that the results in these individuals are

firm enough to suggest that when an acutely ill patient

has acute PE, initial hypoxemia is caused by VA/Qmismatch. Only when atelectasis or other indications

of lung volume loss develop should the hypoxemia be

attributed to shunting. Initial therapeutic measures

therefore should be aimed at preserving or, if possible,

increasing the functional residual capacity.” Depend-

ing on the clinical circumstances, rapid initial support

of cardiovascular function may also be a critical part of

management.

CONCLUSION

The primary cause of the initial hypoxemia in acute

pulmonary embolism appears to be a disturbed pat-

tern of ventilation to perfusion. Simultaneous depres-

sion of cardiac function further contributes to hypox-

emia, but intrapulmonary shunting does not become

important until atelectasis or another cause of lung

volume loss develop.

ACKNOWLEDGMENT: We are indebted to Robert Herigault andFrancoise Veyssi#{232}re for her excellent technical assistance, and toMrs. Fiorella Delcampe and Maryse Gu#{233}rinfor preparation of themanuscript.

REFERENCES

1 Dalen JE, Alpert JS. Natural history of pulmonary embolism.

Prog Cardiovasc Dis 1975; 17:259-70

2 Moser KM. Pulmonary embolism. Am Rev Respir Dis 1977;

115:829-52

3 Robin ED, Forkner CE, Bromberg PA, Croteau Jr, Travis DM.

Alveolar gas exchange in clinical pulmonary embolism. N EngI JMed 1960; 262:283-87

4 Sasahara AA, Cannilla JE, Morse RL, Sidd JJ, Tremblay GM.

Clinical and physiologic studies in pulmonary thromboem-


836 Hypoxemia in Acute Pulmonary Embolism (Huef of a!)

bolism. Am J Cardiol 1967; 20:10-20

5 Kafer ER. Respiratory function in pulmonary thromboembolic

disease. Am J Med 1969; 47:904-15

6 Caldini P. Pulmonary hemodynamics and arterial oxygen satura-

tion in pulmonary embolism. J AppI Physiol 1965; 20:184-90

7 Gazzanica AR, Dalen JE. Paradoxical embolism: its pathophysi-

ology and clinical recognition. Ann Surg 1970; 171:137-41

8 McIntyre KM, Sasahara AA. The hemodynamic response to pul-

monary embolism in patients without prior cardiopulmonary

disease. Am J Cardiol 1971; 28:288-94

9 Jardin F, Bardet J, Sanchez A, Blanchet F, Bourdarias JP,

Margairaz A. Massive pulmonary embolism without arterial

hypoxemia. Intensive Care Med 1977; 3:77-80

10 Wilson JE, Pierce AK, Johnson RL, Wings ER, Hazull WR,

Curry CC, et al. Hypoxemia in pulmonary embolism: a clinical

study. J Clin Invest 1971; 50:481-91

ii Jardin F, Gurdjian F, Desfonds P, Fouilladieu JL, Margairaz A.

Hemodynamic factors influencing arterial hypoxemia in massive

pulmonary embolism with circulatory failure. Circulation 1979;

59:909-12

12 Wagner PD, Laravuso RB, Uhl R, West JB. Continuous distribu-

tions of ventilation-perfusion ratios in normal subjects breathing

air and 100 per cent 0,. J Clin Invest 1974; 54:54-68

13 ‘flbutt DA, Fletcher EWL, Thomas ML, Sutton GC, Miller

GAH. Evaluation of a method fur quantifying the angiographic

severity of major pulmonary embolism. Am J Radiol 1975; 125:

895-99

14 Evans JW, Wagner PD. Limits on VA/Q distribution from

analysis experimental inert gas elimination. J Appi Physiol 1977;

42:889-98

15 Matamis D, Lemaire F, Harf A, Teisseire B, Brun Buisson C.

Redistribution of pulmonary blood flow induced by positive end

expiratory pressure and dopamine infusion in acute respiratory

failure. Am Rev Respir Dis 1984; 10:39-44

16 Fleischner FG. Observations on the radiologic changes in pul-

monary embolism. In: Sasahara AA, Stein M, eds. Pulmonary

embolic disease. New York: Crone & Stratton, 1965:206-13

17 Wagner PD, West JB. Ventilation-perfusion relationships. In:

West JB, ed. Pulmonary gas exchange (vol 1). New York:

Academic Press, 1980:219-62

18 Williams MH. Mechanical vs reflex effects of diffuse pulmonary

embolism in anesthetized dogs. Circ Res 1956; 4:325-31

19 Levy SE, Simmons DH. Mechanism of arterial hypoxemia

following pulmonary thrombo-embolism in dogs. J Appl Physiol

1975; 39:41-46

20 Dantzker DR, Wagner PD, Tornabene VW, Alazraki NP, West

JB. Gas exchange after pulmonary thromboembolization in dogs.

Circ Res 1978; 42:92-103

21 D’Alonzo GE, Bower JS, Dettart P, Dantzker PD: The mecha-

nisms of abnormal gas exchange in acute massive pulmonary

embolism. Am Rev Respir Dis 1983; 128:170-72

22 Meignan M, Harf A, Oliveira L, Simonneau C, Cinotti L,

Duroux P, et al. Overperfusion of non embolic lung as a cause of

hypoxemia in pulmonary embolism [Abstract]. Am Rev Respir

Dis 1983; 116:A127

23 Burton CH, Seed WA, Vernon P. Observations on the mecha-

nism of hypoxemia in acute minor pulmonary embolism. Br Med

J 1984; 289:276-79

24 Levy SE, Simmons DH. Redistribution of alveolar ventilation

following pulmonary thromboembolism in the dog. J Appl

Physiol 1974; 36:60-68

25 Even F, Sors H, Reynaud F, Safran D, Duroux F, Ruff F

L’hypoventilation alveolaire r#{233}gionale par constriction hypocap-

nique des voies a#{233}riennes. Bull Eur Physiopathol Respir 1979;

15:145-66

26 Young I, Mazzone RW, Wagner PD. Identification of functional

lung unit in the dog by graded vascular embolization. J AppI

Physiol 1980; 49:132-41

27 Johnson A, Malik AB. Effects of different size microemboli in

lung fluid and protein exchange. J AppI Physiol 1981; 51:461-64

28 Manier C, Mora B, Castaing Y, Guenard H. Oed#{232}me pulmonaire

au cours d’une embolie pulmonaire. Bull Eur Physiopathol

Respir 1984; 20:55-60

29 Cheney PW, Paulin J, Fevens BS, Allen D. Effect of pulmonary

microembolism on arterio venous shunt flow. J Thorac Cardiovasc

Surg 1978; 75:473-77

30 Dantzker DR, Bower JS. Mechanisms of gas exchange abnormal-

ity in patients with chronic obliterative pulmonary vascular

disease. J Clin Invest 1977; 59:203- 16


1985. chest. hipoxemia in acute pulmonary embolism

Documents