| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Experimental Studies |
1 From the Department of Radiology (P.S., J.S., P.A.G.) and the Laboratory of Physiology (M.D., P.L., C.M., R.N.), Erasme University Hospital, Route de Lennik, 808-1070 Brussels, Belgium. R.N. supported in part by Fonds de Recherche Scientifique et Médicale grants 9.4513.94 and 3.4517.95. Received March 16, 1998; revision requested May 13; final revision received September 9; accepted October 22. Address reprint requests to P.A.G.
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Left atrial pressure, pulmonary arterial pressure, effective pulmonary capillary pressure, and extravascular lung water were measured in six dogs before and during progressive increase of effective pulmonary capillary pressure. A thin-section CT scan was obtained at each step. Lung attenuation and subjective assessments of ground-glass opacification were compared with hemodynamic variables and extravascular lung water.
RESULTS: Ground-glass opacification was identified when effective pulmonary capillary pressure equaled critical pulmonary capillary pressure. Extravascular lung water increased, and the distribution curve of lung attenuation coefficients shifted to higher attenuation from the second measurement at an effective pulmonary capillary pressure greater than the critical pulmonary capillary pressure. Attenuation was highly correlated (r = 0.98, P < .001) with extravascular lung water; ground-glass opacification was detected before a significant (P = .615, analysis of variance) increase in extravascular lung water.
CONCLUSION: Thin-section CT depicts ground-glass opacification when effective pulmonary capillary pressure equals critical pulmonary capillary pressure and before a detectable increase in extravascular lung water. Attenuation reflects extravascular lung water.
Index terms: Computed tomography (CT), thin-section, 60.12111 • Heart, failure, 51.71 • Lung, CT, 60.12111 • Lung, fluid, 60.71 • Lung, function, 60.91
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Based on a canine model of hydrostatic pulmonary edema, the present study aimed (a) to determine the respective roles of effective pulmonary capillary pressure, critical pulmonary capillary pressure, and extravascular lung water in the occurrence of ground-glass opacification, (b) to validate attenuation measurements in the quantification of pulmonary edema, and (c) to compare these measurements with subjective assessments of ground-glass opacification.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Hemodynamic measurements.—A balloon-tipped pulmonary artery catheter (model 131H-7F; Baxter Healthcare, Irvine, Calif) was inserted via the right external jugular vein and positioned with pressure monitoring in a branch of the pulmonary artery for measurements of pulmonary arterial pressure, occluded pulmonary arterial pressure, right atrial pressure, and central temperature, for estimation of effective pulmonary capillary pressure from the pressure decay curve after balloon occlusion, and for mixed venous blood sampling. A polyethylene fiber-optic thermistor catheter (Cold Pulsion; Pulsion Medical System, Munich, Germany) was inserted into the thoracic aorta via the right femoral artery for the measurements of cardiac output (Q), systemic blood pressure, and arterial blood sampling.
A left lateral thoracotomy was performed to insert a 16-F Foley balloon catheter and a polyethylene fluid-filled catheter into the left atrium for the manipulation and measurement of left atrial pressure. The dog's chest was tightly closed, a chest tube was placed, and pleural air was evacuated. Thereafter, positive end expiratory pressure was relieved. A balloon catheter (Percor Sata-DL 10.5F; Datascope, Paramus, NJ) was inserted into the inferior vena cava through a right femoral venotomy. Inflation of this balloon produced an adjustable decrease in Q by reducing venous return. To maintain a constant Q throughout the experiment, and since it was expected that hydrostatic lung edema might be associated with a decrease in Q, it was decided to maintain Q at a moderately subnormal value from the start. Thrombus formation along the balloon catheter was prevented with the intravenous injection of 100 units of heparin sodium (Heparine B; Braun, Melsungen, Germany) per kilogram of body weight just before its insertion.
Pulmonary and systemic vascular pressures were measured by using pressure transducers (Statham P50; Gould, Oxnard, Calif). The pressure transducers were zero referenced at the midchest, and vascular pressures were measured at end expiration. The heart rate was determined from a continuously monitored electrocardiographic lead. Q and extravascular lung water were determined with injection of 0°C 10 mL/kg 0.9% sodium chloride (ICG-Pulsion; Pulsion Medical Systems) containing 1 mg/mL indocyanine green dye into the central circulation as a bolus by using the proximal port of the pulmonary artery catheter and with the computation of the thermal and green dye dilution curves from data measured with the aortic polytheylene fiber-optic thermistor catheter (5). Q and extravascular lung water each were calculated as the mean of three measurements. Respiration was suspended during these measurements to avoid alveolar dispersion of the thermal indicator. Arterial and mixed venous blood gases were measured with an automated analyzer (model ABL 2; Radiometer, Copenhagen, Denmark) immediately after acquisition of the samples and were corrected for temperature.
Venous admixture (QVA/QT) was calculated by using the standard formula ([capillary O2 content - arterial O2 content]/[capillary O2 content - mixed venous O2 content]), with the capillary oxygen content estimated by using the calculated alveolar partial pressure of oxygen and with the oxygen saturation determined from the nomogram by Rossing and Cain (6).
Critical pulmonary capillary pressure, that is, the pulmonary capillary pressure above which the lungs become edematous, was calculated (7) at each time point from the colloidal osmotic pressure as follows: critical pulmonary capillary pressure = 1.55 x colloidal osmotic pressure - 6.8. The colloidal osmotic pressure was measured with a reflectometer in a centrifuged blood sample. The pulmonary vascular pressure signals were sampled at 200 Hz by using an analog-digital converter (model RTI 800; Analog Device, Norwood, Mass), were stored, and were analyzed with a personal computer. Effective pulmonary capillary pressure was computed in triplicate from the pulmonary arterial pressure decay curves after inflation of the balloon in the pulmonary artery. For these measurements, the dog was disconnected from the ventilator for 8 seconds. Time zero was defined as the time when pulmonary arterial pressure began to deviate from the normal wave. A monoexponential curve was fitted to a set of data 0.2–2.0 seconds after the occlusion and was extrapolated back toward time 0 + 150 msec (8). The time lag after the occlusion was chosen because it has been shown to correspond to the delay from the beginning of the occlusion to zero flow reached in the capillaries (9).
Sequence of measurements.—As soon as the dogs were in steady-state conditions (stable heart rate, systemic arterial pressure, and pulmonary arterial pressure for 30 min), left atrial pressure was manipulated by means of stepwise inflation of the left atrial balloon to obtain a sequence of effective pulmonary capillary pressures progressively increased to be equal to or above the critical pulmonary capillary pressure, as summarized in Table 1. At every step, the dog's condition was allowed to stabilize for 5–10 minutes, and a thin-section CT scan was obtained immediately after the measurements of pulmonary arterial pressure, effective pulmonary capillary pressure, left atrial pressure, colloidal osmotic pressure, Q, extravascular lung water, and blood gases. A complete set of hemodynamic and blood gas measurements was obtained in about 5 minutes. Thin-section CT was performed during a breath hold at 30 cm H2O plateau inspiratory pressure. It was estimated that at this level of inspiratory pressure, the animals would be at total lung capacity throughout the experiment. Only one thin-section CT scan was obtained at each time point.
|
Subjective assessment.—First, two independent radiologists (P.S., P.A.G.) skilled in thin-section CT imaging independently assessed the CT images for the presence, the severity, and the extent of ground-glass opacification. Images were scored according to a system adapted from that of Remy-Jardin et al (10). With this system, the (a) severity and (b) extent of ground-glass opacification are scored section-by-section, with scores ranging from (a) 0, or not present, to 3, or high visual severity, and (b) 4, or extent greater than 75%, respectively. Both these scores are combined with a homogeneity score—where a score of 0 indicates that ground-glass opacification is not present, a score of 1 indicates a homogeneous distribution of ground-glass opacification, and a score of 2 indicates it is heterogeneous—and a localization score—where a score of 0 indicates that ground-glass opacification is not present, a score of 1 indicates it is in a dependent lung region, and a score of 2 indicates it is not in a dependent lung region—for a global ground-glass opacification score ranging from a minimum of 0 to a maximum of 11 (10). This first scoring evaluation was performed dog by dog, with all the thin-section CT images classified in the sequence of measurements in the experimental protocol. For this scoring session, a normal reference image was thus always available and the entire sequence of images was available to the observer, so that each image could be compared with the preceding and the following images.
Second, to avoid the visual reference given by the first images obtained with normal hemodynamic conditions, a second reading was performed 3 months after the first reading. Unlike in the first reading, all the images were assessed in random order by three readers, two of whom (P.S., P.A.G.) had already been involved in the first reading. Images were evaluated according to the mentioned scoring system (10) but without the homogeneity component. This second score thus ranged from 0 to 9.
Objective assessment.—Objective assessment of lung attenuation was performed by using semiautomatic software (PULMO CT; Siemens) (11). This software automatically delineates the lung parenchyma on CT images and calculates the distribution curves of lung attenuation coefficients (11). The threshold value to isolate the soft-tissue–lung interface was 200 HU. If certain regions were to be included or excluded, contours were drawn manually. In our study, data obtained with this device were transferred to a personal computer for calculations of means, medians, modes, and SDs of the distribution curves.
Statistical Analysis
All results are expressed as mean ± SD. The hemodynamic data, blood gas results, parameters of distribution curves of lung attenuation coefficients, and visual scores were compared by using the analysis of variance for repeated measurements. When the F ratio with the analysis of variance resulted in a P value less than .05, specific comparisons were made by using modified Student t tests (12). Correlations were calculated by using least squares regression analysis (two-stage analysis for longitudinal data). Inter- and intraobserver agreements were assessed by calculating the Cohen 
coefficient. Agreements were classified as mild ( > 0.40), good ( > 0.60), or excellent ( > 0.80) (13). Statistical significance was set at the P less than .05 level.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Thin-section CT scan appearances of hydrostatic pulmonary edema have been previously described for patients with congestive heart failure (3) and for experimental animal models of hydrostatic pulmonary edema (14,15). Distributions of attenuation measurements on thin-section CT scans in patients with congestive heart failure have been reported as compared with those in healthy subjects (16). However, to our knowledge, there has been no previous study in which the hemodynamic determinants of hydrostatic lung edema varied in a controlled fashion have been correlated with estimates of lung fluid content and subjective and objective assessments of thin-section CT scans.
In the present experiments, the mean lung attenuation increased by a maximum of 64% at the highest values of extravascular lung water. This is in agreement with an increase of 69% reported by Hedlund et al (17), who induced lung edema by using a controlled increase in left atrial pressure up to 30 mm Hg in dogs. These authors made CT measurements during suspended ventilation at functional residual capacity, and CT sections were 1 cm thick. We used CT sections that were 1 mm thick and that were obtained at deep inspiration, two necessary conditions for the detection of ground-glass opacification (1,2).
Thin-section CT has become a major imaging method in the clinical assessment of infiltrative lung diseases, but, to our knowledge, until now only a few studies have attempted the attenuation measurements. Using semiautomatic computer software similar to what we used, Hartley et al (18) showed that, compared with data in healthy control subjects, the distribution curve of the attenuation coefficients was shifted to higher attenuation in patients with pulmonary fibrosis and in asbestos-exposed workers with a chest radiograph displaying an International Labor Office major category 1 abnormality. A similar shift was reported by Beinert et al (19) in a group of patients with idiopathic pulmonary fibrosis as compared with data in a group of healthy subjects. However, in both studies, the technique of measuring attenuation was not validated by comparing it to a reference method.
In the present study, the attenuation measurements were highly correlated with an independent measurement of lung fluid content. It has long been shown that an increase in hydrostatic pressure is associated with the development of lung edema following essentially two stages (20). The first is interstitial edema, with perivascular and peribronchial engorgement and eventually with some widening of the interstitium of the alveolar wall. The second is alveolar flooding. This quantal sequence of alveolar fluid accumulation has been simulated in a lung fluid phantom, that is, a polyether sponge containing various amounts of fluid in three-dimensional air spaces separated by a thin lattice, but without the equivalent of an interstitial space (21). In this model, a shift in CT attenuation distributions parallels the appearance of fluid trapping (21). It is thus possible that ground-glass opacification reflects early interstitial edema preceding alveolar flooding at a stage not yet important enough to induce a significant increase in extravascular lung water as evaluated with the pulmonary extravascular thermal volume. However, another possibility is that ground-glass opacification reflects the increase in capillary blood volume (2). An acute increase in left atrial pressure to around 34 mm Hg has been shown to increase pulmonary capillary blood volume by about 40% (20).
We have subjectively assessed the severity and the extent of ground-glass opacification on thin-section CT scans. In the first scoring system, it was possible to compare the considered image against all images successively obtained in the same dog, including the normal image obtained when the left atrial pressure was normal. To overcome possible overestimation from these comparisons, we also scored the CT scans randomly and without considering the normal reference image. The evaluation of ground-glass opacification was not possible after the eighth step of the protocol because of the appearance of consolidation of lung parenchyma. Independent of the presence of the reference image, ground-glass opacification was detected before a significant increase in lung attenuation but also before a significant increase in extravascular lung water. Subjective recognition of ground-glass opacification before a significant increase in lung attenuation could be related to abnormalities that are not detectable by using attenuation measurements obtained from an entire lung CT section. In the present experiments, quantitative thin-section CT indexes became altered immediately after extravascular lung water increased, and the increasing attenuation was closely correlated with extravascular lung water. These indexes thus may be less accurate for the detection of the earliest stages of hydrostatic pulmonary edema but appear to be adequate for its quantitative assessment. However, it must be remembered that overestimation is a general feature when subjective assessment is performed (21–25). In addition, inter- and intraobserver agreements ranged, in the present study, from only mild to good.
To obtain thin-section CT scans at the earliest stages of hydrostatic lung edema, we have calculated the critical pulmonary capillary pressure by using the equation proposed by Drake et al (7). The critical pulmonary capillary pressure is the critical level at which extravascular lung water begins to accumulate. As expected, extravascular lung water increased as soon as the effective pulmonary capillary pressure was higher than the critical pulmonary capillary pressure. On the other hand, the arterial partial pressure of oxygen decreased as soon as the effective pulmonary capillary pressure became equal to the critical pulmonary capillary pressure. It is generally believed that the earliest stages of interstitial hydrostatic lung edema do not affect gas exchange (26). An increase in capillary blood volume, caused by increased left atrial pressure, tends to homogenize the distribution of pulmonary perfusion (27) and consequently tends to improve gas exchange (26). However, it is conceivable that peribronchial cuffing with interstitial edema could alter the distribution of ventilation and thereby generate low–ventilation-perfusion areas, which would explain the decrease in arterial partial pressure of oxygen in the early stages of pulmonary edema in our study.
The arterial occlusion technique has been extensively used in intact animals and in patients for the estimation of the effective pulmonary capillary pressure as a major determinant of fluid filtration and edema in the lung (28,29). To estimate the effective pulmonary capillary pressure, we used an improved method based on microvascular flow study findings reported by Gilbert and Hakim (9) that consists of a single exponential fitting of the pulmonary arterial pressure decay curve between 0.2 and 2.0 seconds after occlusion, and we extrapolated to the moment of occlusion plus 150 msec (8).
In the present experiment, the extravascular lung water was evaluated by using the thermal and green dye dilution method that had been shown to be accurate when compared with the classic gravimetric measurement of extravascular lung water in patients with pulmonary hydrostatic edema (5,30). A problem with the procedure is that reduced perfusion in edematous areas may decrease the extravascular diffusion of the thermal indicator, leading to an underestimation of extravascular lung water. This may occur not only in acute noncardiogenic lung injuries and in thromboembolism, but also when pulmonary vascular resistance increases in severe hydrostatic lung edema (5,30). In our dogs, it is very likely that perfusion became more homogeneous during the progressive increase in left atrial pressure (27). However, we observed an increase in pulmonary vascular resistance when left atrial pressure was returned to low baseline values; this was probably not associated with a significant decrease in thermal volume measured (in other words, extravascular lung water measured), since extravascular lung water remained closely correlated to attenuation when left atrial pressure was returned to baseline values.
In our dogs, a persistently high effective pulmonary capillary pressure after the return of left atrial pressure to baseline values may have resulted from a combination of direct vascular compression, lung volume changes due to decreased compliance of lung, and hypoxia. Hakim and Kelly (31), applying different techniques—including micropuncture, use of a small retrograde catheter, and arterial and venous occlusion—to isolated perfused dog lungs, have shown arterial occlusion pressure measurements in vessels with a diameter close to 100 µm. Thus, persistently increased effective pulmonary capillary pressure and pulmonary arterial pressure after the return of left atrial pressure to baseline values was probably, at least in part, caused by an increased resistance of small pulmonary arterioles.
It may be noted that, in spite of careful evacuation of pleural air at closure of the thorax, there was a persistence of small pneumothoraces (Fig 6). This would be associated with a less negative pleural pressure than normal but a pressure of a magnitude unlikely to affect the hemodynamic measurements. On the other hand, the section surface area of lung parenchyma remained unchanged throughout the experiments; thus, persistent small pneumothoraces most probably did not affect the present results.
In conclusion, our study findings suggest three things. First, thin-section CT can depict early hydrostatic pulmonary edema as ground-glass opacification as soon as the hemodynamic determinants of edema are reached. That is, thin-section CT can depict early hydrostatic pulmonary edema as ground-glass opacification as soon as the effective pulmonary capillary pressure is greater than the critical pulmonary capillary pressure. Second, attenuation measurements reflect the amount of extravascular lung water. Third, although well correlated with hemodynamic parameters, subjective assessment of lung attenuation has only moderate inter- and intraobserver agreements and results in possible overestimation at early stages of pulmonary edema. Further studies based on other canine models of pulmonary edema are needed to test whether these conclusions could be extended to edema from other origins.
Practical application: Our results offer the prospect of using thin-section CT for the noninvasive detection and quantification of early hemodynamic pulmonary edema in clinical practice.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Author contributions: Guarantors of integrity of entire study, R.N., P.A.G.; study concepts, P.S., P.A.G.; study design, P.S., M.D.; definition of intellectual content, R.N., P.A.G.; literature research, P.S.; experimental studies, P.S., M.D., P.L., C.M., P.A.G.; data acquisition, P.S., M.D., P.L., C.M., P.A.G.; data analysis, P.S., R.N., P.A.G.; statistical analysis, P.S., C.M.; manuscript preparation, P.S., P.A.G.; manuscript editing, P.S.; manuscript review, J.S., R.N., P.A.G.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. A. Guenette, B. C. Sporer, M. J. MacNutt, H. O. Coxson, A. W. Sheel, J. R. Mayo, and D. C. McKenzie Lung density is not altered following intense normobaric hypoxic interval training in competitive female cyclists J Appl Physiol, September 1, 2007; 103(3): 875 - 882. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Schueller-Weidekamm, E. Wassermann, H. Redl, M. Prokop, M. Zimpfer, C. Herold, P. Germann, and R. Ullrich Dynamic CT Measurement of Pulmonary Enhancement in Piglets with Experimental Acute Respiratory Distress Syndrome. Radiology, May 1, 2006; 239(2): 398 - 405. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. C. Parker and M. I. Townsley Evaluation of lung injury in rats and mice Am J Physiol Lung Cell Mol Physiol, February 1, 2004; 286(2): L231 - L246. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Scillia, S. A. Kafi, C. Melot, C. Keyzer, R. Naeije, and P. A. Gevenois Oleic Acid-induced Lung Injury: Thin-Section CT Evaluation in Dogs Radiology, June 1, 2001; 219(3): 724 - 731. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
