LUNG WATER AND VASCULAR PERMEABILITY/Bri^/iam et al. brant. New York, Pergamon Press, 1963, pp 47-74 Rosen MR, Hordof AJ, Hodess AB, Verosky M, VuUiemoz Y: Ouabain-induced changes in electrophysiologic properties of neonatal, young and adult canine cardiac Purkinje fibers. J Pharmacol Exp Ther 194: 255-263, 1975 Schlettwein-Gsell D: Survival curves of an old age rat colony. Gerontologia 16: 111-115, 1970 Schwartz A: Is the cell membrane Na+, K+-ATPase enzyme system the pharmacological receptor for digitalis? Circ Res 39: 2-7, 1976 Scott WJ Jr, Beliles RP, Silverman HI: The comparative acute toxicity of two cardiac glycosides in adult and newborn rats.
523
Toxicol Appl Pharmacol 20: 599-601, 1971 Snedecor GW: Statistical Methods Applied to Experiments in Agriculture and Biology, ed 5. Ames, Iowa, Iowa State University Press, 1957 Speirs RL: Potentiation of contraction after rest in the isolated rat ventricule preparation. Nature 184: 66-67, 1959 Umbreit WW, Burris RH, Stauffer JF (eds): Manometric Techniques and Related Methods for the Study of Tissue Metabolism. Minneapolis, Burgess, 1945, p 194 Weisfeldt ML, Loeven WA, Shock NW: Resting and active mechanical properties of trabeculae carneae from aged male rats. Am J Physiol 220: 1921-1927, 1971
Indicator Dilution Lung Water and Vascular Permeability in Humans Effects of Pulmonary Vascular Pressure KENNETH L. BRIGHAM, JAMES D. SNELL, JR., THOMAS R. HARRIS, SUSAN MARSHALL, JAMES HAYNES, RONALD E. BOWERS, AND JAMES PERRY
SUMMARY To see whether a multiple-indicator dilution technique would measure lung vascular permeability and extravascular lung water (EVLW) in humans, we did indicator studies during cardiac catheterization in 18 patients with various degrees of stable heart failure. A mixture of "Cr-erythrocytes, '"I-albumin, 'H-water, and u C-urea was injected into the right atrium, and serial blood samples were taken from an arterial catheter. From the time-concentration curves we calculated cardiac output, 14 C-urea permeability-surface area product (PS) (by both integral extraction and a Krogh-convolution model), and EVLW (by both mean transit time and a Krogh-convolution model). We also calculated lung microvascular pressure (Pmv) from pulmonary artery and pulmonary artery wedge or left atrial pressures, and measured hematocrit, plasma protein concentration, lung vital capacity, total lung capacity (TLC), carbon monoxide-diffusing capacity, and alveolar volume (VA). 14C-urea PS correlated well with VA1/S (r - 0.62, P - 0.019). Urea PS/V A l / l did not correlate with Pmv (r - 0.36, P - NS), hematocrit (r - -0.07, P = NS), or cardiac output (r - 0.36, P - NS). EVLW/TLC correlated with Pra» (r = 0.51, P - 0.02) and even better with P mv - plasma oncotic pressure (r - 0.63, P - 0.007). We therefore conclude that "C-urea PS is a measure of lung vascular permeability in humans, and that, as in animals, permeability is unaffected by PmT. EVLW may be a more useful measure of lung water in humans than previously thought, when interpreted in light of the measurable forces affecting fluid exchange. Circ Res 44: 523-530, 1979
SOMETIMES excess fluid accumulates in the lung primarily because of changes in the lung circulation (Brigham et al., 1974). When humans develop pulmonary edema while lung vascular pressures are low, it is inferred that lung exchanging vessels are leaking too much fluid and protein (increased permeability; see Robin et al., 1968, 1973). DiagnosFrom the Pulmonary Circulation Center, Vanderbilt University School of Medicine, Nashville, Tennessee. Supported by National Heart, Lung, and Blood Institute Grant HL 19153 (SCOR in Pulmonary Vascular Diseases). This work was done during Dr. Brigham'B tenure as an Established Investigator of the American Heart Association and during Dr. Haynes' tenure as a National Heart, Lung, and Blood Institute Multidisciplinary Lung Research Trainee (Grant T32 HL 07132). Address for reprints: Dr. Kenneth L Bngham, A-5102, Vanderbilt University Hospital, Nashville, Tennessee 37232. Received April 20, 1978; accepted for publication October 26, 1978.
ing increased permeability in this way poses two problems. First, the diagnosis cannot be made until enough lung fluid has accumulated to make pulmonary edema obvious on chest x-ray or physical examination. Second, the diagnosis does not involve a measurement of lung vascular permeability per se, because there has been no way to make that measurement. In animals, we found a multiple-indicator method useful for measuring lung water and vascular permeability (Harris et al., 1976; Brigham et al., 1977; Harris et al., 1978; McKeen et al., 1978b). We have now used the method in humans with normal and increased pulmonary vascular pressures due to stable heart failure. We found that the lung permeability-surface area product (PS) for u C-urea correlated well with alveolar volume (VA) but not with
Downloaded from http://circres.ahajournals.org/ by guest on March 11, 2015
524
CIRCULATION RESEARCH
vascular pressures, and that extravascular lung water, when normalized to total lung capacity, correlated well with the difference between lung microvascular pressure (Pmv) and plasma oncotic pressure. This is consistent with the results of our studies on animals, and suggests that the multipleindicator method may help determine both the kind and amount of lung edema in humans.
Methods General We studied subjects of both sexes, aged 19-71 years, who were hemodynamically stable and were undergoing cardiac catheterization for diagnosis and evaluation of heart disease. Each subject lay supine on a table in the catheterization laboratory while catheters were placed under fluoroscopic observation through an antecubital vein into the pulmonary artery and through a femoral artery into the abdominal aorta. In three subjects, the left atrium was catheterized transseptally to record pressure there. Strain gauges were zeroed at the mid thoracic level, and pulmonary artery (Ppa) and pulmonary arterial wedge (Pp,w) or left atrial (Pu) pressures were recorded. P mv was calculated [Pmv = Ppa + 0.4 (Ppa - Pu or Ppaw); Brigham et al., 1974]. The venous catheter was then withdrawn into the right atrium. Three milliliters of a mixture of 51Cr-erythrocytes, l25I-human serum albumin, u C-urea, and 3H-water in saline were injected as a bolus through the venous catheter, and 30 arterial blood samples were collected at 1.5-second intervals by allowing blood to flow from the arterial catheter into heparinized tubes mounted on a rotating disc collector. Total protein concentration (biuret) and hematocrit were measured on samples of heparinized arterial blood drawn prior to the study. Throughout this study we complied with DHEW and institutional rules for protection of human subjects. Isotopic Methods Ten milliliters of venous blood were drawn from each subject into acid-citrate-dextrose anticoagulant prior to the study. To label erythrocytes, the blood was incubated with BICr-sodium chromate for 30 minutes, and the cells were washed once with 0.89% sodium chloride solution (normal saline). The labeled cells were resuspended in normal saline to original hematocrit, and 125I-human serum albumin, u C-urea, and 3H-water were added to make the injectate. Three milliliters of this mixture contained (in /zCi): 5lCr, 15; 125I, 10; 3H, 40; and UC, 45. We measured radioactivity in a 0.5-ml portion of each arterial blood sample and in the injectate diluted 1/51 in the subject's blood drawn before the study. Activities of 51Cr and 125I were measured in a gamma spectrometer (Packard Instruments); activities of MC and 3H were measured in a liquid scintillation counter (Beckman Instruments) after
VOL. 44, No. 4, APRIL 1979
ethanol precipitation of the proteins. In each case, overlap was subtracted. Indicator Dilution Calculations The time-concentration curves, normalized to the injected activity for each isotope, were plotted. The details of analysis are discussed in the Appendix. Cardiac output was calculated as the inverse of the area under the 6lCr curve, corrected by extrapolation for recirculation (Chinard et al., 1962). We used a Krogh-convolution circulatory model to calculate extravascular lung water (EVLW) volume and u C-urea PS. We have described these calculations in detail in the literature (Harris et al., 1976; Rowlett and Harris, 1975) and compared them with several other methods for interpreting indicator dilution data (Harris et al., 1976). For comparison, we also calculated EVLW volume by the more standard mean transit time method (Chinard et al., 1962), and MC-urea PS from the integral of extraction to the peak of the reference curve (Crone and Garlick, 1969). In all calculations, the intravascular reference curve was a composite of the red cell and albumin curves, weighted for hematocrit and water content of red cells and plasma (Goresky et al., 1969; see Appendix). Other Methods We measured lung volumes using a Collins spirometer, functional residual lung capacity by a closed-circuit helium dilution method (Hathirat et al., 1970), and single-breath carbon monoxide diffusing capacity in the Pulmonary Function Laboratory within 24 hours of the indicator study. From these data, we calculated total lung capacity and VA using the equation of Cotes (1975). Plasma oncotic pressure was calculated by the formula of Landis and Pappenheimer (1963) from the measured total plasma protein concentration. Statistics We calculated means, standard errors, and correlation coefficients by standard formulas (Snedecor and Cochran, 1967). The coefficient of variation (CV) for the curve fitting was computed as follows:
^-v?^F/[?
(1)
where n is the number of points, and P is 2 for 3Hwater data and 1 for MC-urea. CDO and DDP are the observed and predicted diffusing tracer relative concentrations, respectively.
Results Figure 1 shows a typical set of indicator curves. The relationships among the four indicators are like those we have previously reported for animals and humans (Harris et al., 1976; Brigham et al., 1977b; Harris et al., 1978; Brigham et al., 1976a). There is
Downloaded from http://circres.ahajournals.org/ by guest on March 11, 2015
LUNG WATER AND VASCULAR PERMEABILITY/Brjg/iam et al. 2
Q
VCH-4 C-ur«o PS-IIOmi/nc
•—• o--o I—I
K
EVLW'37lm
5l
Cr-«r»throcrtt» I-olbu«iin l4 C-ur«0 3 H-wotir
l25
1.5RELATIVE ARTERIAL BLOOO | RAOIOACTIVITY '
0 5-
0 5
10 15 20 TIME AFTER INJECTION d e c )
1 Representative set of indicator curves in a human subject.
FIGURE
25
dilution
some intravascular separation of red cells and albumin reflected in a higher peak and steeper downslope for the red cell curve. The urea curve falls between the intravascular curves and that of the flow-limited tracer, tritiated water. Figure 2 shows the indicator curves (composite intravascular, UCurea, and 3H-water) for subject VCH-6. The ability of the mathematical model to fit concentration data and extraction is shown by the dotted lines. The CV for the two fits was 0.104 for 3H-water data and 0.0232 for MC-urea data. The model fits all but the earliest extraction point where the relative concenPt VCH-6 —•— COMPOSITE INTRAVASCULAR INDICATOR + M C-UREA DATA £, * H - WATER DATA MODEL PREDICTIONS
0
2
4
6
8
10
12
14
16
trations were very low (0.05 for reference indicator and 0.039 for 14C-urea data). Table 1 lists the clinical data for all of the subjects studied. Table 2 lists the pressure, plasma protein, and pulmonary function values. Table 3 lists the indicator dilution data. Values for Pp.w ranged from 4-35 torr. There was also a broad range of cardiac outputs, hematocrits, EVLW volumes, MC-urea PS, and total lung capacities. Although the primary data show widely varying urea PS values, these values correlated quite well with measured total lung capacity (r = 0.52, P = 0.03 for model PS values). The best normalizing variable for PS would be exchanging vessel surface area. If this approximates alveolar surface area, as has been suggested (Weibel, 1973), then PS should correlate best with VA 2/3 as a result of the relationship between volume and surface area of a sphere. In fact, urea PS did correlate better with VA2/3, as is illustrated in Figure 3. When normalized to VA2/3, urea PS did not correlate significantly with hematocrit (r = -0.07, P = NS), cardiac output (r = 0.39, P = NS), or P mv (r = 0.36, P = NS). The relationship between urea PS and Pmv is illustrated in Figure 4. Although mean transit time extravascular water space is significantly correlated with model-based values, (r = 0.79, P < 0.001), there is an average difference of 25% between the two values. This variation is quite different from the close agreement between the two techniques in a series of multipleindicator studies on sheep (Harris et al., 1978). The probable cause is illustrated in Figure 5, in which normalized composite intravascular and 3H-water concentrations are plotted as a function of dimensionless time for sheep and human data. The intravascular curves match except on the downslope. The 3H-water curves are dissimilar. There are five points past the peak before recirculation in the curves for sheep, but only three in the curve for humans. This is because human and sheep cardiac outputs are similar (60 ml/sec in VCH-10 vs. 63.8 for the sheep), but their water volumes are quite different (for the sheep, model EVLW volume was 236 ml and the mean transit time value was 241 ml). This results in fewer downslope points from which to extrapolate the curves for humans and compromises the accuracy of the mean transit time calculation. When indicator dilution EVLW volume (EVLW model values) was normalized to total lung capacity, it correlated significantly with Pmv (r = 0.51, P = 0.016). The correlation of EVLW with pressure was improved by including plasma oncotic pressure. Figure 6 shows EVLW as a function of the difference between P mv and plasma oncotic pressure, illustrating the good correlation between those variables.
18 20 22
TIME AFTER INJECTION (SEC)
2 Multiple indicator data and theoretical predictions for subject VCH-6. FIGURE
525
Discussion At least in animals, when lung vascular permeability to fluid and protein is increased, there may
Downloaded from http://circres.ahajournals.org/ by guest on March 11, 2015
526
CIRCULATION RESEARCH
TABLE
VOL. 44, No. 4, APRIL 1979
1 Clinical Data for All Subjects
Subject no.
Duration of symptoms (years)
Age (years)
Sex
Ht.(inches)
Wt(kg)
VCH-1
60
M
72
80.0
2
VCH-2
57
M
68
75.0
1
VCH-3
19
F
62
43.6
5
VCH-4
54
F
64
64.1
6
VCH-5 VCH-6 VCH-7 VCH-8 VCH-9
34 49 36 42 49
M F M M M
70 63 66 68 71
65.5 63.6 48.6 72.3 86.8
1 0.5 14 6 1
VCH-10
35
M
65
80.9
VCH-11
60
F
64
55.4
12
VCH-12 VCH-13 VCH-14
38 71 25
F F M
62 62 66
50.9 69.1 64.5
1 1 13
VCH-15 VCH-16 VCH-17 VCH-18
56 53 27
M M M M
66 69 68 68
60.4 66.4 77.7 73.0
1 13 9 4
51
be very large increases in transvascular fluid filtration without substantial lung fluid accumulation (Brigham et al., 1974). Pulmonary edema [increased lung water content (Visscher et al., 1956)] occurs late in the pathogenetic sequence of events (Staub, 1974). If a reliable measure of the integrity of exchanging vessels could be made in humans, it should be possible to detect increased vascular permeabilTABLE
Diagnosis
Cardiomyopathy Ischemia Ventricular aneurysm Mitral stenosis Mitral insufficiency Mitral stenosis Mitral insufficiency Mitral insufficiency Pericardial effusion Mitral stenosis Mitral stenosis Mitral stenosis Mitral insufficiency Cardiomyopathy Mitral insufficiency Mitral stenosis Mitral insufficiency Normal Aortic stenosis Aortic insufficiency Mitral insufficiency Aortic insufficiency Mitral stenosis Aortic insufficiency Ischemia Aortic stenosis
0.25
ity early, before alveoli flood and lung function deteriorates. We have shown that the lung vascular uC-urea PS calculated from single pass indicator dilution curves is unaffected by increased pulmonary vascular pressures in dogs (Harris et al., 1976) and sheep (Harris et al., 1978), but increases after injecting alloxan in dogs (Harris et al., 1976) or his-
2 Pressure, Hematocrit, Plasma Protein, and Pulmonary Function Data for All Subjects Total plasm* protein
Mean pressure (torr )
Subject no.
VCH-1 VCH-2 VCH-3 VCH-4 VCH-5 VCH-6 VCH-7 VCH-8 VCH-9 VCH-10 VCH-11 VCH-12 VCH-13 VCH-14 VCH-15 VCH-16 VCH-17 VCH-18
Pulmonary artery
Pulmonary artery wedge
25 17 44 44 60 16 38 20 31 60 21 8 17 15 20 24 12 15
13 11 — 27 25 10 — 13 23 35 19 4 — 10 7 20 7 10
Left atrium
20 — — 23 — — — — 21 — — — —
HcL
0.34 0.50 0.39 0.38 0.34 0.40 0.46 0.40 0.46 0.50 0.45 0.36 0.38 0.42 0.55 0.40 0.50 0.39
Concentration (g/dl)
Oncotic pressure (torr)
VC(ml)
TLC (ml)
DLco (ml/min X torr)
VA(ml)
6.70
24.0
2392
5817
4.9
4970
6.90 7.35 5.44 7.08 6.05 6.75 7.22 7.80 6.45 5.84 5.46 6.40 7.16 7.10 7.07 7.80
25.1 27.6 17.6 26.1 20.6 24.2 26.9 30.4 22.6 19.5 17.7 22.4 26.5 26.2 26.0 30.4
1415 3010 4152 2441 2792 3255 2726 3348 2889 2903 2640 3715 3429 3581 5712
2540 4574 5834 4167 3726 6368 5587 4330 4924 4266 4066 4663 6018 6602 6552 6371
— 16.6 13.8 16.8 15.5 28.3 28.6 27.8 19.1 17.9 20.1 29.6 33.2 34.0 — 21.9
4402 5453 4022 2924 5737 4756 3617 4537 3302 3561 4578 5753 6209 — 4005
4470
VC = Lung vital capacity; TLC — total lung capacity, DLco — lung carbon monoxide diffusing capacity, VA — alveolar volume; Hct. — hematocrit; • measurement* not made.
Downloaded from http://circres.ahajournals.org/ by guest on March 11, 2015
LUNG WATER AND VASCULAR PERMEABILITY/Br^/iam et al. TABLE
527
3 Indicator Dilution Data for all Subjects M
C-Urea PS (ml/sec)
Subject no.
Cardiac output (ml/ sec)
VCH-l VCH-2 VCH-3 VCH-4 VCH-5 VCH-6 VCH-7 VCH-8 VCH-9 VCH-10 VCH-11 VCH-12 VCH-13 VCH-14 VCH-15 VCH-16 VCH-17 VCH-18
68 76 31 73 60 81 51 66 63 60 51 58 56 55 37 57 129 91
CVfor
EVLW volume (ml)
l4
Integral extraction
Kroghconvolution
C-urea model fitting
8.8 7.4 5.4 12.9 13.2 7.9 4.6 19.0 7.9 10.9
8.1 4.8 4.2 9.3 11.6 7.1 3.4 13.6 10.3 7.5 5.9 5.0 4.3 4.6 1.7 22.8 19.6 1.1
0.0294 0.0427 0.0580 0.0770 0.0530 0.0232 0.0296 0.1040 0.0855 0.0820 0.0499 0.0931 0.0899 0.0780 0.0758 0.0932 0.0679 0.1300
7.4
6.6 5.9 3.7 2.0 23.3 24.5 5.7
Mean transit time
351 645 407 322 675 132 * 454 380 498 394 322 334 362 99 574 248 229
Kroghconvolution
270 444
251 315 473 199 614 415 290 385 354 213
t
394 167 649 170 180
CVfor 'H-water curve fitting
0.107 0.132 0.114 0.132 0.118 0.104 0.0725 0.120 0.266 0.107 0.139 0.254 0.284 0.117 0.135 0.0681 0.120
* Insufficient data points for exponential extrapolation of 1H-water curve. t Inaccurate relative normalization of 'H-water data.
tamine in sheep (Harris et al., 1978). Alloxan causes pulmonary edema in dogs due to increased vascular permeability (Staub et al., 1967), and lung lymph measurements in sheep indicate that histamine increases pulmonary vascular permeability (Brigham and Owen, 1975; Brigham et al., 1976a). Thus in animals the uC-urea PS detects increased permeability in the lung circulation and distinguishes edema due to high pressure from that due to increased permeability (Harris et al., 1978). Pulmonary vascular uC-urea PS also is quantitatively consistent with protein permeability calculated from lung lymph measurements (Harris and Brigham, 1977). The first step in evaluating the utility of the indicator dilution lung vascular urea PS in humans was to demonstrate its feasibility and to see whether it was affected by pulmonary vascular pressures. Thus we chose subjects with apparently normal
lung vascular permeability, but with a broad range of pulmonary vascular pressures due to various degrees of heart failure. Although the actual urea PS values varied widely in this group, there was good correlation with VA2/3. When urea PS was normalized to VA 2/3 , there was no significant correlation with pulmonary vascular pressures. This result suggests that, as in animals, urea permeability is unaffected by vascular pressures in humans. Since indicator dilution methods measure only diffusive transport [i.e., "true" permeability (Kedem and Katchalsky, 1958)], hydrostatic pressure should have negligible effects unless permeability actually increases in response to high vascular pressures (Pietra et al., 1969). The bulk of available evidence COMPOSITE INTRAVA3CULAR DATA - - . _ - P t VCH-10 —O--SHEEP S
— A — P I VCH-10 —A—SHEEP
7.50n 100-
r - .36 p« NS
Q 6.25-
S
H-WATER DATA
9080-
5.00-
70-
3 75-
60-
2.50-
40-
50-
30-
1.25-
2010-
00
7.5
15.0
22.5
30.0
37 5
45 0
0 2 4 6 8 I 0 1.2 14 1.6 I 8 2.0 TIME/INTRAVASCULAR MEAN TRANSIT TIME
LUNG MICROVASCULAR PRESSURE (torr)
3 uC-urea PS (Krogh-convolution model values) as a function of VA2/3 in human subjects. FIGURE
FIGURE 4 VA2/3 as a
l4
C-urea PS (model values) normalized to function of lung Pmo (see text for calculation).
Downloaded from http://circres.ahajournals.org/ by guest on March 11, 2015
528
CIRCULATION RESEARCH O 1751
r=.63 p=.OO7 0 125EXTRAVASCULAR LUNG WATER TOTAL LUNG CAPACITY I m l / L ) X IO" J
0.100-
0.075-
0.050-
O.O25J -22.5
-15
7.5
-7.5
22.5
LUNG MICHOVASCULAR PRESSURE PLASMA ONCOTIC PRESSURE (torr)
5 Normalized indicator composite intravascular and 'H-water curves from experiments on human subjects and sheep. The extravascular lung water in human subjects was 498 ml by mean transit time, and 385 ml by the model calculation. FIGURE
suggests that this does not occur with pressures in the range seen in intact organisms (Brigham and Owen, 1975a; Brigham et al., 1976a; Erdmann et al., 1975; Brigham and Owen, 1975b). Recruitment of more microvessels at higher pressures also might increase PS due to an increase in surface area. Our data do not show this, possibly because we studied supine subjects for whom, even at normal pressures, none of the lung is in zone I. The best normalizing variable for PS would be exchanging vessel surface area in each subject, but that cannot be measured. Although the relationship between lung volume and exchanging vessel surface area may not be linear, they should relate; that is, bigger lungs should have more surface area. If alveolar surface area approximates exchanging vessel 24r=.615 p=.015
201612840175
200
225
250
275
300 325
(ALVEOLAR VOLUME, ml l^J
FIGURE 6 Extravascular lung water (Krogh-convolution model values) normalized to total lung capacity as a function of the difference between lung PmD and plasma oncotic pressure in human subjects. The unfilled circle and bars represent the average normal data (Goresky et al., 1975), given a normal value on the abscissa.
VOL. 44, No. 4, APRIL 1979
surface area, then, given a spherical lung model, PS ought to correlate with VA 2/3 . The good correlation of C-urea PS with VA2/3 in these studies gives us some confidence in the accuracy of the PS measurement and provides a good normalizing variable in this group. Unfortunately, VA measured by
523
Toxicol Appl Pharmacol 20: 599-601, 1971 Snedecor GW: Statistical Methods Applied to Experiments in Agriculture and Biology, ed 5. Ames, Iowa, Iowa State University Press, 1957 Speirs RL: Potentiation of contraction after rest in the isolated rat ventricule preparation. Nature 184: 66-67, 1959 Umbreit WW, Burris RH, Stauffer JF (eds): Manometric Techniques and Related Methods for the Study of Tissue Metabolism. Minneapolis, Burgess, 1945, p 194 Weisfeldt ML, Loeven WA, Shock NW: Resting and active mechanical properties of trabeculae carneae from aged male rats. Am J Physiol 220: 1921-1927, 1971
Indicator Dilution Lung Water and Vascular Permeability in Humans Effects of Pulmonary Vascular Pressure KENNETH L. BRIGHAM, JAMES D. SNELL, JR., THOMAS R. HARRIS, SUSAN MARSHALL, JAMES HAYNES, RONALD E. BOWERS, AND JAMES PERRY
SUMMARY To see whether a multiple-indicator dilution technique would measure lung vascular permeability and extravascular lung water (EVLW) in humans, we did indicator studies during cardiac catheterization in 18 patients with various degrees of stable heart failure. A mixture of "Cr-erythrocytes, '"I-albumin, 'H-water, and u C-urea was injected into the right atrium, and serial blood samples were taken from an arterial catheter. From the time-concentration curves we calculated cardiac output, 14 C-urea permeability-surface area product (PS) (by both integral extraction and a Krogh-convolution model), and EVLW (by both mean transit time and a Krogh-convolution model). We also calculated lung microvascular pressure (Pmv) from pulmonary artery and pulmonary artery wedge or left atrial pressures, and measured hematocrit, plasma protein concentration, lung vital capacity, total lung capacity (TLC), carbon monoxide-diffusing capacity, and alveolar volume (VA). 14C-urea PS correlated well with VA1/S (r - 0.62, P - 0.019). Urea PS/V A l / l did not correlate with Pmv (r - 0.36, P - NS), hematocrit (r - -0.07, P = NS), or cardiac output (r - 0.36, P - NS). EVLW/TLC correlated with Pra» (r = 0.51, P - 0.02) and even better with P mv - plasma oncotic pressure (r - 0.63, P - 0.007). We therefore conclude that "C-urea PS is a measure of lung vascular permeability in humans, and that, as in animals, permeability is unaffected by PmT. EVLW may be a more useful measure of lung water in humans than previously thought, when interpreted in light of the measurable forces affecting fluid exchange. Circ Res 44: 523-530, 1979
SOMETIMES excess fluid accumulates in the lung primarily because of changes in the lung circulation (Brigham et al., 1974). When humans develop pulmonary edema while lung vascular pressures are low, it is inferred that lung exchanging vessels are leaking too much fluid and protein (increased permeability; see Robin et al., 1968, 1973). DiagnosFrom the Pulmonary Circulation Center, Vanderbilt University School of Medicine, Nashville, Tennessee. Supported by National Heart, Lung, and Blood Institute Grant HL 19153 (SCOR in Pulmonary Vascular Diseases). This work was done during Dr. Brigham'B tenure as an Established Investigator of the American Heart Association and during Dr. Haynes' tenure as a National Heart, Lung, and Blood Institute Multidisciplinary Lung Research Trainee (Grant T32 HL 07132). Address for reprints: Dr. Kenneth L Bngham, A-5102, Vanderbilt University Hospital, Nashville, Tennessee 37232. Received April 20, 1978; accepted for publication October 26, 1978.
ing increased permeability in this way poses two problems. First, the diagnosis cannot be made until enough lung fluid has accumulated to make pulmonary edema obvious on chest x-ray or physical examination. Second, the diagnosis does not involve a measurement of lung vascular permeability per se, because there has been no way to make that measurement. In animals, we found a multiple-indicator method useful for measuring lung water and vascular permeability (Harris et al., 1976; Brigham et al., 1977; Harris et al., 1978; McKeen et al., 1978b). We have now used the method in humans with normal and increased pulmonary vascular pressures due to stable heart failure. We found that the lung permeability-surface area product (PS) for u C-urea correlated well with alveolar volume (VA) but not with
Downloaded from http://circres.ahajournals.org/ by guest on March 11, 2015
524
CIRCULATION RESEARCH
vascular pressures, and that extravascular lung water, when normalized to total lung capacity, correlated well with the difference between lung microvascular pressure (Pmv) and plasma oncotic pressure. This is consistent with the results of our studies on animals, and suggests that the multipleindicator method may help determine both the kind and amount of lung edema in humans.
Methods General We studied subjects of both sexes, aged 19-71 years, who were hemodynamically stable and were undergoing cardiac catheterization for diagnosis and evaluation of heart disease. Each subject lay supine on a table in the catheterization laboratory while catheters were placed under fluoroscopic observation through an antecubital vein into the pulmonary artery and through a femoral artery into the abdominal aorta. In three subjects, the left atrium was catheterized transseptally to record pressure there. Strain gauges were zeroed at the mid thoracic level, and pulmonary artery (Ppa) and pulmonary arterial wedge (Pp,w) or left atrial (Pu) pressures were recorded. P mv was calculated [Pmv = Ppa + 0.4 (Ppa - Pu or Ppaw); Brigham et al., 1974]. The venous catheter was then withdrawn into the right atrium. Three milliliters of a mixture of 51Cr-erythrocytes, l25I-human serum albumin, u C-urea, and 3H-water in saline were injected as a bolus through the venous catheter, and 30 arterial blood samples were collected at 1.5-second intervals by allowing blood to flow from the arterial catheter into heparinized tubes mounted on a rotating disc collector. Total protein concentration (biuret) and hematocrit were measured on samples of heparinized arterial blood drawn prior to the study. Throughout this study we complied with DHEW and institutional rules for protection of human subjects. Isotopic Methods Ten milliliters of venous blood were drawn from each subject into acid-citrate-dextrose anticoagulant prior to the study. To label erythrocytes, the blood was incubated with BICr-sodium chromate for 30 minutes, and the cells were washed once with 0.89% sodium chloride solution (normal saline). The labeled cells were resuspended in normal saline to original hematocrit, and 125I-human serum albumin, u C-urea, and 3H-water were added to make the injectate. Three milliliters of this mixture contained (in /zCi): 5lCr, 15; 125I, 10; 3H, 40; and UC, 45. We measured radioactivity in a 0.5-ml portion of each arterial blood sample and in the injectate diluted 1/51 in the subject's blood drawn before the study. Activities of 51Cr and 125I were measured in a gamma spectrometer (Packard Instruments); activities of MC and 3H were measured in a liquid scintillation counter (Beckman Instruments) after
VOL. 44, No. 4, APRIL 1979
ethanol precipitation of the proteins. In each case, overlap was subtracted. Indicator Dilution Calculations The time-concentration curves, normalized to the injected activity for each isotope, were plotted. The details of analysis are discussed in the Appendix. Cardiac output was calculated as the inverse of the area under the 6lCr curve, corrected by extrapolation for recirculation (Chinard et al., 1962). We used a Krogh-convolution circulatory model to calculate extravascular lung water (EVLW) volume and u C-urea PS. We have described these calculations in detail in the literature (Harris et al., 1976; Rowlett and Harris, 1975) and compared them with several other methods for interpreting indicator dilution data (Harris et al., 1976). For comparison, we also calculated EVLW volume by the more standard mean transit time method (Chinard et al., 1962), and MC-urea PS from the integral of extraction to the peak of the reference curve (Crone and Garlick, 1969). In all calculations, the intravascular reference curve was a composite of the red cell and albumin curves, weighted for hematocrit and water content of red cells and plasma (Goresky et al., 1969; see Appendix). Other Methods We measured lung volumes using a Collins spirometer, functional residual lung capacity by a closed-circuit helium dilution method (Hathirat et al., 1970), and single-breath carbon monoxide diffusing capacity in the Pulmonary Function Laboratory within 24 hours of the indicator study. From these data, we calculated total lung capacity and VA using the equation of Cotes (1975). Plasma oncotic pressure was calculated by the formula of Landis and Pappenheimer (1963) from the measured total plasma protein concentration. Statistics We calculated means, standard errors, and correlation coefficients by standard formulas (Snedecor and Cochran, 1967). The coefficient of variation (CV) for the curve fitting was computed as follows:
^-v?^F/[?
(1)
where n is the number of points, and P is 2 for 3Hwater data and 1 for MC-urea. CDO and DDP are the observed and predicted diffusing tracer relative concentrations, respectively.
Results Figure 1 shows a typical set of indicator curves. The relationships among the four indicators are like those we have previously reported for animals and humans (Harris et al., 1976; Brigham et al., 1977b; Harris et al., 1978; Brigham et al., 1976a). There is
Downloaded from http://circres.ahajournals.org/ by guest on March 11, 2015
LUNG WATER AND VASCULAR PERMEABILITY/Brjg/iam et al. 2
Q
VCH-4 C-ur«o PS-IIOmi/nc
•—• o--o I—I
K
EVLW'37lm
5l
Cr-«r»throcrtt» I-olbu«iin l4 C-ur«0 3 H-wotir
l25
1.5RELATIVE ARTERIAL BLOOO | RAOIOACTIVITY '
0 5-
0 5
10 15 20 TIME AFTER INJECTION d e c )
1 Representative set of indicator curves in a human subject.
FIGURE
25
dilution
some intravascular separation of red cells and albumin reflected in a higher peak and steeper downslope for the red cell curve. The urea curve falls between the intravascular curves and that of the flow-limited tracer, tritiated water. Figure 2 shows the indicator curves (composite intravascular, UCurea, and 3H-water) for subject VCH-6. The ability of the mathematical model to fit concentration data and extraction is shown by the dotted lines. The CV for the two fits was 0.104 for 3H-water data and 0.0232 for MC-urea data. The model fits all but the earliest extraction point where the relative concenPt VCH-6 —•— COMPOSITE INTRAVASCULAR INDICATOR + M C-UREA DATA £, * H - WATER DATA MODEL PREDICTIONS
0
2
4
6
8
10
12
14
16
trations were very low (0.05 for reference indicator and 0.039 for 14C-urea data). Table 1 lists the clinical data for all of the subjects studied. Table 2 lists the pressure, plasma protein, and pulmonary function values. Table 3 lists the indicator dilution data. Values for Pp.w ranged from 4-35 torr. There was also a broad range of cardiac outputs, hematocrits, EVLW volumes, MC-urea PS, and total lung capacities. Although the primary data show widely varying urea PS values, these values correlated quite well with measured total lung capacity (r = 0.52, P = 0.03 for model PS values). The best normalizing variable for PS would be exchanging vessel surface area. If this approximates alveolar surface area, as has been suggested (Weibel, 1973), then PS should correlate best with VA 2/3 as a result of the relationship between volume and surface area of a sphere. In fact, urea PS did correlate better with VA2/3, as is illustrated in Figure 3. When normalized to VA2/3, urea PS did not correlate significantly with hematocrit (r = -0.07, P = NS), cardiac output (r = 0.39, P = NS), or P mv (r = 0.36, P = NS). The relationship between urea PS and Pmv is illustrated in Figure 4. Although mean transit time extravascular water space is significantly correlated with model-based values, (r = 0.79, P < 0.001), there is an average difference of 25% between the two values. This variation is quite different from the close agreement between the two techniques in a series of multipleindicator studies on sheep (Harris et al., 1978). The probable cause is illustrated in Figure 5, in which normalized composite intravascular and 3H-water concentrations are plotted as a function of dimensionless time for sheep and human data. The intravascular curves match except on the downslope. The 3H-water curves are dissimilar. There are five points past the peak before recirculation in the curves for sheep, but only three in the curve for humans. This is because human and sheep cardiac outputs are similar (60 ml/sec in VCH-10 vs. 63.8 for the sheep), but their water volumes are quite different (for the sheep, model EVLW volume was 236 ml and the mean transit time value was 241 ml). This results in fewer downslope points from which to extrapolate the curves for humans and compromises the accuracy of the mean transit time calculation. When indicator dilution EVLW volume (EVLW model values) was normalized to total lung capacity, it correlated significantly with Pmv (r = 0.51, P = 0.016). The correlation of EVLW with pressure was improved by including plasma oncotic pressure. Figure 6 shows EVLW as a function of the difference between P mv and plasma oncotic pressure, illustrating the good correlation between those variables.
18 20 22
TIME AFTER INJECTION (SEC)
2 Multiple indicator data and theoretical predictions for subject VCH-6. FIGURE
525
Discussion At least in animals, when lung vascular permeability to fluid and protein is increased, there may
Downloaded from http://circres.ahajournals.org/ by guest on March 11, 2015
526
CIRCULATION RESEARCH
TABLE
VOL. 44, No. 4, APRIL 1979
1 Clinical Data for All Subjects
Subject no.
Duration of symptoms (years)
Age (years)
Sex
Ht.(inches)
Wt(kg)
VCH-1
60
M
72
80.0
2
VCH-2
57
M
68
75.0
1
VCH-3
19
F
62
43.6
5
VCH-4
54
F
64
64.1
6
VCH-5 VCH-6 VCH-7 VCH-8 VCH-9
34 49 36 42 49
M F M M M
70 63 66 68 71
65.5 63.6 48.6 72.3 86.8
1 0.5 14 6 1
VCH-10
35
M
65
80.9
VCH-11
60
F
64
55.4
12
VCH-12 VCH-13 VCH-14
38 71 25
F F M
62 62 66
50.9 69.1 64.5
1 1 13
VCH-15 VCH-16 VCH-17 VCH-18
56 53 27
M M M M
66 69 68 68
60.4 66.4 77.7 73.0
1 13 9 4
51
be very large increases in transvascular fluid filtration without substantial lung fluid accumulation (Brigham et al., 1974). Pulmonary edema [increased lung water content (Visscher et al., 1956)] occurs late in the pathogenetic sequence of events (Staub, 1974). If a reliable measure of the integrity of exchanging vessels could be made in humans, it should be possible to detect increased vascular permeabilTABLE
Diagnosis
Cardiomyopathy Ischemia Ventricular aneurysm Mitral stenosis Mitral insufficiency Mitral stenosis Mitral insufficiency Mitral insufficiency Pericardial effusion Mitral stenosis Mitral stenosis Mitral stenosis Mitral insufficiency Cardiomyopathy Mitral insufficiency Mitral stenosis Mitral insufficiency Normal Aortic stenosis Aortic insufficiency Mitral insufficiency Aortic insufficiency Mitral stenosis Aortic insufficiency Ischemia Aortic stenosis
0.25
ity early, before alveoli flood and lung function deteriorates. We have shown that the lung vascular uC-urea PS calculated from single pass indicator dilution curves is unaffected by increased pulmonary vascular pressures in dogs (Harris et al., 1976) and sheep (Harris et al., 1978), but increases after injecting alloxan in dogs (Harris et al., 1976) or his-
2 Pressure, Hematocrit, Plasma Protein, and Pulmonary Function Data for All Subjects Total plasm* protein
Mean pressure (torr )
Subject no.
VCH-1 VCH-2 VCH-3 VCH-4 VCH-5 VCH-6 VCH-7 VCH-8 VCH-9 VCH-10 VCH-11 VCH-12 VCH-13 VCH-14 VCH-15 VCH-16 VCH-17 VCH-18
Pulmonary artery
Pulmonary artery wedge
25 17 44 44 60 16 38 20 31 60 21 8 17 15 20 24 12 15
13 11 — 27 25 10 — 13 23 35 19 4 — 10 7 20 7 10
Left atrium
20 — — 23 — — — — 21 — — — —
HcL
0.34 0.50 0.39 0.38 0.34 0.40 0.46 0.40 0.46 0.50 0.45 0.36 0.38 0.42 0.55 0.40 0.50 0.39
Concentration (g/dl)
Oncotic pressure (torr)
VC(ml)
TLC (ml)
DLco (ml/min X torr)
VA(ml)
6.70
24.0
2392
5817
4.9
4970
6.90 7.35 5.44 7.08 6.05 6.75 7.22 7.80 6.45 5.84 5.46 6.40 7.16 7.10 7.07 7.80
25.1 27.6 17.6 26.1 20.6 24.2 26.9 30.4 22.6 19.5 17.7 22.4 26.5 26.2 26.0 30.4
1415 3010 4152 2441 2792 3255 2726 3348 2889 2903 2640 3715 3429 3581 5712
2540 4574 5834 4167 3726 6368 5587 4330 4924 4266 4066 4663 6018 6602 6552 6371
— 16.6 13.8 16.8 15.5 28.3 28.6 27.8 19.1 17.9 20.1 29.6 33.2 34.0 — 21.9
4402 5453 4022 2924 5737 4756 3617 4537 3302 3561 4578 5753 6209 — 4005
4470
VC = Lung vital capacity; TLC — total lung capacity, DLco — lung carbon monoxide diffusing capacity, VA — alveolar volume; Hct. — hematocrit; • measurement* not made.
Downloaded from http://circres.ahajournals.org/ by guest on March 11, 2015
LUNG WATER AND VASCULAR PERMEABILITY/Br^/iam et al. TABLE
527
3 Indicator Dilution Data for all Subjects M
C-Urea PS (ml/sec)
Subject no.
Cardiac output (ml/ sec)
VCH-l VCH-2 VCH-3 VCH-4 VCH-5 VCH-6 VCH-7 VCH-8 VCH-9 VCH-10 VCH-11 VCH-12 VCH-13 VCH-14 VCH-15 VCH-16 VCH-17 VCH-18
68 76 31 73 60 81 51 66 63 60 51 58 56 55 37 57 129 91
CVfor
EVLW volume (ml)
l4
Integral extraction
Kroghconvolution
C-urea model fitting
8.8 7.4 5.4 12.9 13.2 7.9 4.6 19.0 7.9 10.9
8.1 4.8 4.2 9.3 11.6 7.1 3.4 13.6 10.3 7.5 5.9 5.0 4.3 4.6 1.7 22.8 19.6 1.1
0.0294 0.0427 0.0580 0.0770 0.0530 0.0232 0.0296 0.1040 0.0855 0.0820 0.0499 0.0931 0.0899 0.0780 0.0758 0.0932 0.0679 0.1300
7.4
6.6 5.9 3.7 2.0 23.3 24.5 5.7
Mean transit time
351 645 407 322 675 132 * 454 380 498 394 322 334 362 99 574 248 229
Kroghconvolution
270 444
251 315 473 199 614 415 290 385 354 213
t
394 167 649 170 180
CVfor 'H-water curve fitting
0.107 0.132 0.114 0.132 0.118 0.104 0.0725 0.120 0.266 0.107 0.139 0.254 0.284 0.117 0.135 0.0681 0.120
* Insufficient data points for exponential extrapolation of 1H-water curve. t Inaccurate relative normalization of 'H-water data.
tamine in sheep (Harris et al., 1978). Alloxan causes pulmonary edema in dogs due to increased vascular permeability (Staub et al., 1967), and lung lymph measurements in sheep indicate that histamine increases pulmonary vascular permeability (Brigham and Owen, 1975; Brigham et al., 1976a). Thus in animals the uC-urea PS detects increased permeability in the lung circulation and distinguishes edema due to high pressure from that due to increased permeability (Harris et al., 1978). Pulmonary vascular uC-urea PS also is quantitatively consistent with protein permeability calculated from lung lymph measurements (Harris and Brigham, 1977). The first step in evaluating the utility of the indicator dilution lung vascular urea PS in humans was to demonstrate its feasibility and to see whether it was affected by pulmonary vascular pressures. Thus we chose subjects with apparently normal
lung vascular permeability, but with a broad range of pulmonary vascular pressures due to various degrees of heart failure. Although the actual urea PS values varied widely in this group, there was good correlation with VA2/3. When urea PS was normalized to VA 2/3 , there was no significant correlation with pulmonary vascular pressures. This result suggests that, as in animals, urea permeability is unaffected by vascular pressures in humans. Since indicator dilution methods measure only diffusive transport [i.e., "true" permeability (Kedem and Katchalsky, 1958)], hydrostatic pressure should have negligible effects unless permeability actually increases in response to high vascular pressures (Pietra et al., 1969). The bulk of available evidence COMPOSITE INTRAVA3CULAR DATA - - . _ - P t VCH-10 —O--SHEEP S
— A — P I VCH-10 —A—SHEEP
7.50n 100-
r - .36 p« NS
Q 6.25-
S
H-WATER DATA
9080-
5.00-
70-
3 75-
60-
2.50-
40-
50-
30-
1.25-
2010-
00
7.5
15.0
22.5
30.0
37 5
45 0
0 2 4 6 8 I 0 1.2 14 1.6 I 8 2.0 TIME/INTRAVASCULAR MEAN TRANSIT TIME
LUNG MICROVASCULAR PRESSURE (torr)
3 uC-urea PS (Krogh-convolution model values) as a function of VA2/3 in human subjects. FIGURE
FIGURE 4 VA2/3 as a
l4
C-urea PS (model values) normalized to function of lung Pmo (see text for calculation).
Downloaded from http://circres.ahajournals.org/ by guest on March 11, 2015
528
CIRCULATION RESEARCH O 1751
r=.63 p=.OO7 0 125EXTRAVASCULAR LUNG WATER TOTAL LUNG CAPACITY I m l / L ) X IO" J
0.100-
0.075-
0.050-
O.O25J -22.5
-15
7.5
-7.5
22.5
LUNG MICHOVASCULAR PRESSURE PLASMA ONCOTIC PRESSURE (torr)
5 Normalized indicator composite intravascular and 'H-water curves from experiments on human subjects and sheep. The extravascular lung water in human subjects was 498 ml by mean transit time, and 385 ml by the model calculation. FIGURE
suggests that this does not occur with pressures in the range seen in intact organisms (Brigham and Owen, 1975a; Brigham et al., 1976a; Erdmann et al., 1975; Brigham and Owen, 1975b). Recruitment of more microvessels at higher pressures also might increase PS due to an increase in surface area. Our data do not show this, possibly because we studied supine subjects for whom, even at normal pressures, none of the lung is in zone I. The best normalizing variable for PS would be exchanging vessel surface area in each subject, but that cannot be measured. Although the relationship between lung volume and exchanging vessel surface area may not be linear, they should relate; that is, bigger lungs should have more surface area. If alveolar surface area approximates exchanging vessel 24r=.615 p=.015
201612840175
200
225
250
275
300 325
(ALVEOLAR VOLUME, ml l^J
FIGURE 6 Extravascular lung water (Krogh-convolution model values) normalized to total lung capacity as a function of the difference between lung PmD and plasma oncotic pressure in human subjects. The unfilled circle and bars represent the average normal data (Goresky et al., 1975), given a normal value on the abscissa.
VOL. 44, No. 4, APRIL 1979
surface area, then, given a spherical lung model, PS ought to correlate with VA 2/3 . The good correlation of C-urea PS with VA2/3 in these studies gives us some confidence in the accuracy of the PS measurement and provides a good normalizing variable in this group. Unfortunately, VA measured by
