INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Bronchoalveolar lavage (BAL) is a diagnostic technique routinely used to sample cells, proteins, and other constituents of the lung epithelial lining fluid (ELF). Standard bronchoalveolar lavage usually involves instillation of 20 to 50 ml of saline through a fiberoptic bronchoscope into a pulmonary subsegment followed by immediate aspiration and fluid retrieval. As many as five aliquots of saline are sequentially instilled and reaspirated. The most important shortcoming of standard BAL is that it does not allow for the measurement of cell and solute concentrations in the ELF alone. Consequently, the interpretation of concentrations of retrieved substances is difficult. This is due to several factors: (1) the retrieved fluid represents dilution of a variable magnitude of ELF by saline, (2) a significant exchange of fluid may occur between the lung tissues and the air space during lavage (1), and (3) the instilled fluid is only partially retrieved. This may be due either to a loss of fluid outside the lavaged lung subsegment or to sequestration of lavage fluid within the lavaged territory, suggesting a partial sampling of the alveolar and airway surface. Also, a significant exchange of solute with the lung tissues occurs during lavage, which is usually inversely proportional to the molecular size of the solute. One approach for estimating epithelial lining fluid volume (V_ELF) is the measurement in the recovered lavage fluid of a solute whose concentration is at equilibrium in ELF and plasma. Rennard and colleagues (2) measured endogenous urea concentration in the recovered lavage fluid to estimate V_ELF. Because urea diffuses rapidly across cell membranes, it was assumed that the concentration of this solute in ELF was similar to that in plasma. However, because urea rapidly diffuses into the BAL fluid during the lavage procedure, the use of this indicator was seen to lead to an overestimation of V_ELF by this method (3, 4).

Our objective was to develop a new technique that would allow estimation of V_ELF for a given lung territory by BAL, using equilibration between plasma and ELF of a solute that, unlike urea, would be restricted to the extracellular space (5) and not subjected to active transport. We used technetium labeled diethylenetriamine pentaacetic acid (^99mTc-DTPA, molecular weight = 492 Da), infused at constant activity (cpm/ ml) in the blood for 215 min prior to and during BAL. The use of ^99mTc-DTPA for measurement of BAL dilution of ELF was first suggested by Feng and colleagues (6). These investigators found that equilibration half time of ^99mTc-DTPA between plasma and ELF of rat lungs is less than 1 h. We tested the hypothesis that because of the small size of ^99mTc-DTPA, its radioactivity would reach equilibrium between plasma and the ELF of the lavaged lung segment. ^99mTc-DTPA activity in the ELF prior to lavage could then be estimated by ^99mTc-DTPA plasma activity. The ELF volume could be calculated knowing the ^99mTc-DTPA activity in the retrieved lavage fluid and in plasma. Exchange of ^99mTc-DTPA occurring during BAL could be corrected for by constructing a best-fit line for ^99mTc-DTPA activity in lavage fluid as a function of time from the beginning of lavage with back-extrapolation to time zero, to find the ^99mTc-DTPA activity in ELF prior to lavage, when no net exchange of ^99mTc-DTPA had yet occurred. We used ¹²⁵I-labeled albumin as an alveolar indicator diluted in the lavage fluid prior to instillation. This allowed us to measure the total volume of fluid present in the lavaged lung segment and to calculate the volume of fluid exchanged with the lung tissues during each lavage. Finally, we used an original "saturation BAL" (SBAL), a sequential lavage technique that allowed us to progressively saturate the lavaged lung segment with lavage fluid and to approach 100% retrieval of the instilled lavage fluid.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal Preparation

The experiments were performed in nine anesthetized dogs. A 20-Gauge catheter (Venflon) was placed in a forelimb vein. The dogs were anesthetized intravenously with 100 to 200 mg sodium pentothal (Nesdonal; Rhône-Poulenc-Rorer, Paris, France), intubated, and mechanically ventilated with 1% halothane (Halothane Belamont; France). FI_O2 was adjusted to maintain arterial blood gases within normal range in injury experiments. A 7.5F Swan Ganz catheter (Baxter, Paris, France) was placed in a pulmonary artery via the jugular vein by the Seldinger method using sterile technique in Group 2 control experiments. In addition, a carotid artery polyethylene catheter was placed in Group 2 injury experiments.

Experimental Protocol

Shortly before each experiment, 5 mCi of ^99mTc-DTPA was prepared as a 0.5 mCi/ml solution in saline using a commercial kit (Pentacis; Cis Bio International, Paris, France). After a 15 min baseline a 1.3 ml bolus of the ^99mTc-DTPA solution was injected followed immediately by a 2 ml/h infusion for the entire duration of the experiment. An initial series of experiments (Group 1, n = 6) was performed to assess the equilibration of ^99mTc-DTPA between plasma and ELF. In Group 1 experiments, six sequential lavages were performed at 30 and at 90 min after the beginning of ^99mTc-DTPA infusion in order to compare the estimated V_ELF at the respective equilibration times. The serial lavages at 90 min of ^99mTc-DTPA infusion were performed in a different lung territory than those performed at 30 min. A second study (Group 2) was performed in two separate experiments: control and after oleic-acid-induced lung injury. In Group 2 control experiments (n = 9), 215 min after the beginning of the ^99mTc-DTPA infusion, eight sequential lavages were performed. In Group 2 injury experiments (n = 8), 20 min after the beginning of the ^99mTc-DTPA infusion, lung injury was induced by injection of 0.08 ml/kg oleic acid into the superior vena cava through the proximal lumen of the Swan Ganz catheter. Every 2 min 0.2-ml boluses of oleic acid were injected. Each bolus was followed by a 5-ml saline rinse of the catheter lumen. Eight sequential lavages were also performed 215 min after the beginning of the ^99mTc-DTPA infusion. At the end of the BAL procedure, a thoracotomy was performed, and the pedicle of the lung contralateral to the lavaged lung was ligated. The lung was removed and was immediately frozen in liquid nitrogen for later measurement of extravascular lung water. The animal was then killed by intracardiac injection of KCl.

Heart rate, pulmonary arterial pressure (Ppa), pulmonary capillary pressure (Ppc), and pulmonary wedge pressure (Ppw) were monitored throughout both Group 2 control and injury experiments. Systemic arterial pressure, blood gases, and hematocrit were monitored throughout Group 2 injury experiments. Pulmonary capillary pressure was measured by analysis of pulmonary arterial postocclusion pressure profile as described by Hakim and colleagues (7). Briefly, the pulmonary artery was occluded by rapid inflation of the Swan-Ganz catheter balloon. During acquisition of each occlusion pressure-time profile, the ventilator was turned off for approximately 10 s. The post-occlusion pressure profile was digitized and stored in a computer for later analysis. Ppc was calculated by extrapolation of a monoexponential fit of the arterial pressure profile, starting 0.3 s postocclusion, back to the instant of occlusion. Three occlusion pressure profiles were obtained and averaged for each data point.

Extravascular lung water was measured by a gravimetric method (8). Lung lobes were weighed and homogenized with equal weights of distilled water in a Waring blender. Two samples of the homogenate, two samples of blood, and one sample of the supernatant of the centrifuged homogenate were weighed, dried in an oven, and weighed again. The blood content of the lung was determined by measuring the hemoglobin concentration in blood and in the supernatant using a spectrophotometer (OSM3; Radiometer, Copenhagen, Denmark). Extravascular lung water was expressed as the extravascular lung water/blood-free dry weight ratio (EVLW/BFDW).

Saturation BAL Procedure

Prior to each experiment a batch BAL solution of 500 ml was prepared by adding ¹²⁵I-albumin (Cis Bio International) to sterile physiologic saline at a final activity of 4 µCi/L. An inflatable latex balloon was adapted to the extremity of a fiberoptic bronchoscope (Fujinon; Tokyo, Japan). Prior to each BAL, the bronchoscope was inserted into the endotracheal tube through an air-tight adapter made in our laboratory, and its distal end was randomly placed in a segmental bronchus of the right or left lung. A pressure transducer was connected to the bronchoscope lumen. The balloon was inflated and the seal provided by the balloon was tested by serial injections of 10-ml boluses of air into the bronchoscope lumen. The seal was considered satisfactory when successive pressure increments induced by air injections were found stable. When the pressure in the lung segment exceeded 10 cm H₂O for a total injected air volume of 60 ml, the bronchoscope was repositioned in another segment, or the instilled BAL fluid volume was reduced. The lavage fluid was contained in a closed circuit at all times (Figure 1), and the lavage fluid pressure was monitored constantly to prevent overdistension or excessive suction of the lavaged lung segment; 20 to 60 ml of lavage fluid were initially instilled in approximately 15 s, and 2 to 3 ml of air were injected to clear the operating channel dead space, followed by rapid reaspiration. Fluid retrieval was visually monitored through the bronchoscope, and suction was stopped as soon as the bronchus tended to narrow. A 3-ml sample was taken from the retrieved lavage fluid. Fresh lavage fluid was added to the retrieved fluid so that the total would equal the volume of fluid that was initially instilled. This fluid was then reinstilled, and the lavage cycle was repeated six (Group 1 experiments) to eight (Group 2 experiments) times. The volume of instilled fluid was therefore identical from one lavage cycle to the next. ^99mTc-DTPA activity was measured in 1 ml samples of each retrieved lavage fluid immediately after the end of the experiment using a gamma counter (Novelec; Meylan, France). Corrections were made for background noise, the decay of ^99mTc-DTPA, and crossover between the ¹²⁵I and the ^99mTc channels. ¹²⁵I-albumin activity was measured approximately 1 to 4 wk later, when ^99mTc-DTPA activity had decayed (^99mTc physical half-life = 6.05 h).

View larger version (19K): [in this window] [in a new window]   Figure 1.   Lavage fluid circuit. Fresh lavage fluid was contained in a reservoir (L). A modified 60 ml syringe was used to instill fluid. Fluid was instilled (i) and recovered (r) through separate tubing. The lavage syringe was connected to a vacuum pump (V) via a trap (t). Three- and four-way stop-cocks (x) were used to control fluid flow through the circuit. Lavage fluid samples were taken through a sample port (s) at the end of each fluid recovery. Lavage fluid pressure was monitored continuously by a pressure transducer (P).

Calculations

Prior to BAL, V_ELF (ml) is the volume of ELF contained in the alveoli of the lavaged lung segment. During each lavage cycle, V_i,n is the volume of lavage fluid instilled in the lung segment, V_r,n is the volume of lavage fluid recovered at the end of fluid retrieval, and V_n is the volume of residual lavage fluid that is not recovered remaining in the lavaged lung segment. V_e,n is the net volume of fluid exchanged between the alveoli and the lung tissues. V_a,n is the volume of fresh lavage fluid that is added during each lavage cycle. Eight sequential lavage cycles were performed; the cycle number being designated by n. The residual fluid volume in the lavaged lung segment at the end of each lavage cycle is described by the following equation:

(1)

where:

(2)

when n = 1 (first lavage cycle):

(3)

The quantity (cpm) of ¹²⁵I-albumin remaining in the lavaged lung segment at the end of each lavage cycle is described by Equation 4:

(4)

where: C_r,n (cpm/ml) is the ¹²⁵I-albumin activity in the retrieved lavage fluid, C_r,n1 is the ¹²⁵I-albumin activity in the retrieved lavage fluid of the preceding lavage, and C_i,n is the ¹²⁵I-albumin activity in the instilled lavage fluid.

V_i,n, V_a,n and V_r,n were known, and V_n was calculated, using C_r,n. Although the algebraic sum: V_ELF + V_e,1 can be calculated for the first lavage (when n = 1), neither V_ELF nor V_e,1 can be calculated separately on the basis of known parameters.

In order to measure V_ELF, ^99mTc-DTPA was infused at constant plasma activity for 215 min prior to and during BAL. It was assumed that the ^99mTc-DTPA activity reached equilibrium between plasma and ELF prior to BAL. Q₀ is the quantity (cpm) of ^99mTc-DTPA contained in V_ELF prior to BAL:

(5)

where C₀ (cpm/ml) is the ^99mTc-DTPA radioactivity in the ELF prior to lavage. C₀ was estimated from the ^99mTc-DTPA plasma activity C_p (cpm/ml) assuming equilibrium between the two compartments at the time of lavage. However, when the earliest data were obtained at the end of the first lavage cycle, both net exchange of fluid and of ^99mTc-DTPA from plasma towards the BAL fluid had already taken place. Therefore Q₀ and consequently V_ELF could not be calculated on the basis of known parameters.

Q_n (cpm/ml), or the total quantity of ^99mTc-DTPA in the lavage fluid at each lavage cycle was calculated as follows:

(6)

where C_n is the activity (cpm/ml) of ^99mTc-DTPA in the retrieved lavage fluid and is a measured parameter. V_t,n (ml) is the total volume of fluid present in the lavaged lung segment at end-instillation and was defined as: V_t,n = V_n + V_r,n. Based on the observation that Q_n increases at a constant rate in the lavage fluid throughout the sequential lavages, we adjusted a best-fit line by least squares regression analysis to Q_n plotted as a function of time (min) from the beginning of the first instillation. Back-extrapolation of this best-fit line to time zero, when no net exchange of ^99mTc-DTPA had yet occurred, allowed us to find Q₀. Q₀ was then used to calculate V_ELF by Equation 5.

Statistics

Group data are summarized as mean ± SEM. The Mann-Whitney test was used to compare data between groups and to compare EVLW/ BFDW between the Group 2 injury data and the laboratory reference for normal dog lung. Wilcoxon's test was used to compare V_ELF at 30 and 90 min in Group 1 experiments. Differences with a p < 0.05 were considered as significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Arterial Blood Gases

In Group 2 injury experiments, Pa_O2 dropped from 218.6 ± 9.5 mm Hg prior to lung injury to 118 ± 16.5 mm Hg in the 100 to 135 min period after oleic acid injection. Pa_O2 further decreased to 81.2 ± 11.8 during lavage, then it rose to 97.1 ± 23.6 after removal of the fiberoptic bronchoscope at the end of the experiment.

Evidence for Saturation

The average instilled volume was 58.9 ± 1.2 and 55 ± 2.7 ml in Group 2 control and injury experiments, respectively. The ratio of retrieved to instilled fluid volume V_r,n/V_i,n (Figure 2) approached 100% after the fourth lavage, i.e., no further trapping of lavage fluid occurred in either Group 2 control or injury experiments. Mean fluid recovery for the last four lavages was 98 ± 1 and 95 ± 2% in Group 2 control and injury experiments, respectively. At the first lavage, more fluid was recovered in Group 2 injury experiments (71 ± 7%) than in Group 2 control experiments (45 ± 5%). Fluid retrieval at saturation did not depend on the anatomic localization of the lavaged segment. The average ratio of retrieved to instilled volume at saturation did not change whether SBAL was performed in the left or the right lung, or whether a dependent or a nondependent segment were chosen. In Group 1 experiments, average V_r,n/V_i,n at saturation was 104% in dependent lung segments (10 SBALs) and 101% in nondependent lung segments (2 SBALs).

View larger version (10K): [in this window] [in a new window]   Figure 2.   The ratio of retrieved to instilled fluid volume expressed as a function of lavage number.

In parallel to V_r,n/V_i,n, the total volume of lavage fluid-ELF mixture present in the lavaged segment at the end of instillation (V_t,n, ml) approached a steady state after the fourth lavage in Group 2 control experiments (Figure 3). In Group 2 injury experiments, V_t,n tended to increase beyond the fourth lavage; however, this increase was not statistically significant. Mean _t,n at saturation, in other words for the last four lavages, or V_{t,n(n = 5,8)} was 104.8 ± 5.5 ml in Group 2 control and 100.2 ± 6.3 ml in Group 2 injury experiments.

View larger version (13K): [in this window] [in a new window]   Figure 3.   The total volume of lavage fluid-ELF mixture present in the lavaged segment at end-instillation (Vt,n, ml) expressed as a function of lavage number.

^99mTc-DTPA Activity in Plasma and Retrieved SBAL Fluid

^99mTc-DTPA average plasma activity (Figure 4) was (cpm/ml · 10⁵) 2.8 ± 2.3 in Group 2 control and 2.7 ± 4.8 in Group 2 injury experiments, respectively. ^99mTc-DTPA plasma activity was stable during lavage in Group 2 control experiments; however, it increased during lavage in Group 2 injury experiments (2.7 ± 0.5 to 3.1 ± 0.7 cpm/ml · 10⁵). This elevation was probably due to a fall in glomerular filtration in these acutely ill animals.

View larger version (12K): [in this window] [in a new window]   Figure 4.   99mTc-DTPA average plasma activity (cpm/ml × 105) versus time (min).

^99mTc-DTPA average activity (cpm/ml · 10 ³) in the retrieved lavage fluid rose progressively from the first to the eighth lavage, from 9.2 ± 1.5 to 22.8 ± 3.5 in Group 2 control experiments and from 99.9 ± 11.5 to 135.2 ± 18.1 in Group 2 injury experiments.

^99mTc-DTPA Quantity (Q_n) in Lavage Fluid

Q_n (cpm · 10⁶) progressively increased in the lavage fluid both in Group 2 control (Figure 5) and Group 2 injury (Figure 6) experiments. A best-fit line was adjusted to Q_n expressed as a function of time from the beginning of the serial lavages. The average regression coefficients (r) for the adjusted lines were 0.99 ± 0.01 (range; 0.94 to 1.00) and 0.94 ± 0.02 (range; 0.81 to 0.99) for Group 2 control and injury experiments, respectively. Back extrapolation of the best-fit line to time 0 yielded Q₀ or the quantity of ^99mTc-DTPA in ELF prior to lavage.

View larger version (25K): [in this window] [in a new window]   Figure 5.   Total amount of 99mTc-DTPA present in the lavaged lung segment (Qn, cpm × 106) versus time in Group 2 control (C) experiments (see key on right for symbol identification). Best-fit lines were adjusted to Qn = f (t).

View larger version (21K): [in this window] [in a new window]   Figure 6.   Total amount of 99mTc-DTPA present in the lavaged lung segment (Qn, cpm × 106) versus time in Group 2 injury (I) experiments (see key on right for symbol identification). Best-fit lines were adjusted to Qn = f (t).

We used a piecewise cubic polynomial model (also called quadratic regression spline) to represent our Q_n data (9), and to calculate the zero intercept. This method does not require any assumptions concerning the linearity of the data. The V_ELF calculated by this method was not different from the values obtained by simple linear regression, by Wilcoxon's test. For the sake of simplicity, we have reported only the linear regression method, which is less sophisticated and easier to apply.

Epithelial Lining Fluid Volume

The estimated V_ELF, expressed as a percentage of the mean lavage fluid volume present in the lung segment at saturation (_{t,n(n = 5,8)}), was 1.4 ± 0.4% at 30 min and 2.1 ± 0.5% at 90 min from the beginning of ^99mTc-DTPA infusion in Group 1 experiments. V_ELF /_{t,n(n = 5,8)} was 1.7 ± 0.4% in Group 2 control experiments. V_ELF at 30 and at 90 min in Group 1 experiments were not significantly different, and neither values were significantly different from V_ELF in Group 2 control experiments, which was measured 215 min after the ^99mTc-DTPA infusion was begun. It should be noted that in the case of lung injury, V_ELF is actually the ELF-edema fluid mixture volume. V_ELF /_{t,n(n = 5,8)} was 25.0 ± 4.4% in Group 2 injury experiments, which was significantly higher than Group 2 control experiments (p < 0.05) by the Mann-Whitney test (Table 1). Individual values of V_ELF and V_ELF /_{t,n(n = 5,8)} are summarized in Table 2.

Net Exchanged Fluid Volume

The net exchanged fluid volume (V_e,n) could be calculated for all lavages except the first lavage. The average rate of this net fluid exchange expressed as a percentage of _{t,n(n = 5,8)}, was 0.37 ± 0.13% per lavage in Group 2 control experiments. If it is assumed that the average net fluid volume that is exchanged during the first lavage is similar to that of following lavages, then the net volume of fluid exchanged during the first lavage represented 318.7 ± 116.5% of V_ELF in Group 2 control experiments.

The average rate of net fluid exchange expressed as a percentage of _{t,n(n = 5,8)}, was 1.4 ± 0.7% per lavage in Group 2 injury experiments. Similarly, if the average net exchanged fluid volume during the first lavage is similar to the following lavages, the net volume of fluid exchanged during the first lavage represented 35.9 ± 14.6% of the V_ELF-edema fluid mixture volume in Group 2 injury experiments.

Pulmonary Vascular Pressures

Ppa and Ppc were stable throughout Group 2 control experiments and were not affected by lavage. In Group 2 injury experiments, Ppa and Ppc rose after oleic acid injection, then decreased back to baseline 45 min later. Mean values of Ppa, Ppc, and Ppw during lavage are summarized in Table 1. Wedge pressure (Ppw) was lower in Group 2 injury experiments. Ppc in Group 2 injury experiments was not different from that in control experiments, confirming that pulmonary edema in the injury experiments was due to increased permeability.

Extravascular Lung Water

The EVLW/BFDW was 8.6 ± 0.9 in Group 2 injury experiments, which was significantly higher than our laboratory reference value for normal dog lung of 3.5 ± 0.4 (Table 1). Both EVLW/BFDW and the V_ELF-edema fluid mixture volume significantly increased in all of the Group 2 injury experiments (Figure 7), however the correlation between the two parameters was not statistically significant.

View larger version (9K): [in this window] [in a new window]   Figure 7.   Changes in VELF/Vt,n(n = 5,8) and in EVLW/BFDW in Group 2 control and injury experiments. Both parameters increased in the presence of an elevation in capillary-alveolar permeability. However, the relative increase in VELF/Vt,n(n = 5,8) (expressed as %; open circles) was higher than that of EVLW/BFDW (closed crosses).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The goal of the present study was to design a method to measure V_ELF based on the equilibration between plasma and ELF of ^99mTc-DTPA. This technique requires progressive saturation of the lavaged lung segment or lobe using an original lavage protocol or SBAL. SBAL will be discussed first, followed by the measurement of V_t,n using ¹²⁵I-albumin as an alveolar indicator. The equilibration of ^99mTc-DTPA between plasma and ELF will be discussed thereafter.

SBAL

We designed our lavage protocol in order (1) to isolate the lavaged lung segment by adapting an inflatable cuff at the distal extremity of the fiberoptic bronchoscope; (2) to instill the least final volume of lavage fluid after n sequential lavages to obtain saturation of the lavaged segment, therefore to minimally dilute the soluble constituents of ELF; (3) to obtain saturation of the lung segment progressively, without inducing hyperinflation of the lung segment, which could increase permeability to solute (10). Precautions were taken in our lavage protocol to avoid hyperinflation. Our results show that almost all of the instilled fluid volume was recovered from the fourth lavage on in both Group 2 control and injury experiments. The mean percentage of recovered fluid at the first lavage was significantly higher in Group 2 injury experiments probably because of the accumulation of edema fluid in the lavaged lung segment. In parallel, the mean volume of fluid present in the lavaged segment at end instillation (V_t,n) plateaued after the fourth lavage. These observations suggest that the lung segment is progressively saturated; this may not mean that the lung segment is literally full but rather that all airway and alveolar surfaces that are accessible to lavage fluid come in contact with the instilled fluid during the first four sequential lavages. Once this is achieved, fluid recovery is near total. During standard BAL, part of the instilled fluid is sequestered in the alveoli and small airways by capillarity. As the lavage process is repeated, the additional instilled fluid either replaces the air or occupies new territories within the lung segment, which again trap part of that fluid. The latter point represents a significant problem in quantifying components of the recovered lavage fluid because the size of the lung territory that is effectively lavaged is not known.

The validity of the SBAL method relies on thorough mixing of the instilled lavage fluid with ELF. There is little objective data on fluid mixing within the airways and alveoli during BAL. Recently, Butler and Tsuda (11) sequentially instilled excised rat lungs with an ultra-low-viscosity polymerizeable silicone fluid of two different colors in equal volumes. The fluid density was adjusted to that of normal saline. After three to five ventilatory cycles, a complete mix of the two different colors was achieved at the acinar level on light microscopy examination of the lung cross section. Although ventilatory and instillation-retrieval cycles are different, these findings suggest that convection of lavage fluid through the instillation-recovery sequence contributes to mixing of the fluid within the airways and alveoli.

Measurement of V_t,n

In the standard lavage procedure the total volume of fluid present in the lung segment at end-instillation (V_t,n) is not known. V_t,n is not equal to the instilled volume (V_i,n) but rather is the algebraic sum of the instilled fluid volume, the net exchanged fluid volume, and the ELF volume. As mentioned above (see RESULTS) the net volume of exchanged fluid in normal lung is significant compared with V_ELF. We used a macromolecular indicator (¹²⁵I-albumin) diluted in the lavage fluid to measure the volume of fluid present in the lavaged lung segment. This method is valid as long as no significant exchange of ¹²⁵I-albumin occurs across the capillary-alveolar barrier. Gee and Staub (12) observed bulk flow of ¹²⁵I-albumin from the alveoli into the interstitium along with formation of peribronchial cuffs after instillation of Krebs-Henseleit buffer in the left lower lobe of anesthetized dogs. However, the instilled volumes were much larger (249 to 292 ml) and mean lavage fluid pressure (11 ± 4 cm H₂O) was higher than in the present study, suggesting an injury to the alveolar epithelium (13). Matthay and colleagues (14) later found that after instillation of Ringer lactate (75 ml) containing tracer amounts of ¹²⁵I-albumin in a lower lobe in sheep, less than 2% of the ¹²⁵I-albumin added to the instilled fluid left the air spaces after the 4-h experimental period, although 50% of the fluid was evacuated after 3 h. Similarly, in normal dog lung, Berthiaume and colleagues (15) observed that ¹²⁵I-albumin evacuation is slow (approximately 1% per hour). Therefore, considering that the duration of the eight sequential lavages in our study averaged 25 ± 2.7 min in Group 2 control experiments, probably less than 1% of ¹²⁵I-albumin left the alveoli during lavage. Passage of ¹²⁵I-albumin into the interstitium is probably more significant in lung injury where epithelial permeability is increased. Quantification of this passage would require simultaneous measurement of ¹²⁵I activity in lymph and plasma after SBAL in injured lungs. A significant passage of ¹²⁵I-albumin into the interstitium would result in an overestimation of V_t,n.

Equilibration of ^99mTc-DTPA between Plasma and ELF

An ideal indicator for estimating the dilution of ELF by BAL should diffuse across the alveolar epithelium sufficiently rapidly so that equilibrium between plasma and ELF may be reached prior to lavage (6). On the other hand, the smaller the indicator, the more easily it will diffuse from the interstitium into the alveoli during the lavage procedure. Because these two properties cannot be dissociated, it seems that such an "ideal" indicator does not exist. Our method was based on the consideration that if the exchange of an indicator between the interstitium and the alveoli during lavage is slow enough, the kinetics of this exchange can be measured through sequential lavages. We selected ^99mTc-DTPA as a diffusible indicator because of its small size, which allows us to assume that it rapidly reaches equilibrium between plasma and ELF. Moreover, ^99mTc-DTPA is restricted to the extracellular space and is not subject to active transport (5). In order to determine whether ^99mTc-DTPA in the ELF was in complete equilibrium with plasma, we performed preliminary (Group 1) experiments in which sequential lavages were performed at 30 and at 90 min from the beginning of ^99mTc-DTPA infusion. If ^99mTc-DTPA is not at equilibrium at a given time interval after the beginning of infusion, it should continue to diffuse along its concentration gradient from the plasma to the ELF beyond that time. Because Q₀ represents the quantity of ^99mTc-DTPA that has diffused from the plasma to the alveoli prior to lavage, in the absence of equilibrium, Q₀ (and consequently the calculated V_ELF) should be smaller at an earlier time interval from the beginning of ^99mTc-DTPA infusion. However, according to our data V_ELF is not significantly different at 215 min (Group 2 control) than at 90 min (Group 1), suggesting equilibrium of ^99mTc-DTPA between plasma and ELF after 90 min of intravenous infusion. V_ELF seemed smaller at 30 min of ^99mTc-DTPA infusion than at 90 min. The difference was not statistically significant; however, the lack of significance may be due to the small number of experiments.

Our linear model of Q_n as a function of time is based on the empirical observation that Q_n constantly increases in the lavage fluid, during the sequential lavages, i.e., for approximately 30 min after the initial instillation. This is due to an increase of the activity of ^99mTc-DTPA (C_r,n) in the lavage fluid because Q_n continues to increase even when V_t,n has reached a plateau. This suggests that there is net exchange of ^99mTc-DTPA from plasma into the alveoli. The fact that ^99mTc-DTPA passes into the alveoli during lavage may be explained by the fact that the dilution of ELF by BAL fluid by a factor of approximately 50 to 70 disrupts the equilibrium in ^99mTc-DTPA activity between the interstitium and ELF, resulting in ^99mTc-DTPA exchange down a new concentration gradient, possibly through a diffusive mechanism. Conversely, it can be expected that the ^99mTc-DTPA activity in the new dilution space will progressively reach equilibrium with plasma, and the ^99mTc-DTPA exchange will slow down and eventually stop. Increases in the activity of ^99mTc-DTPA in the collected air-space fluid could represent access of BAL fluid to more of the exchange surface area within the lung segment. Although this is a possible phenomenon during the first three lavages, it becomes much less likely at saturation when the lavage fluid volume in the lung segment reaches a plateau, if a relationship is assumed between the volume of fluid in the lung and exchange surface area. However, ^99mTc-DTPA activity in the lavage fluid continues to increase steadily beyond that point, suggesting mainly an exchange phenomenon.

Passage of ^99mTc-DTPA into the lung segment can be accelerated if epithelial permeability is increased by the lavage process. It should be pointed out that changes in permeability induced by the lavage itself would theoretically affect the slope of Q_n (expressed as a function of time), but not its zero-time intercept (Q₀) which is used to calculate V_ELF.

Previous Estimations of V_ELF by BAL

Several investigators in the past have attempted to estimate V_ELF by BAL using indicator dilution methods. One approach to estimating V_ELF is to introduce an indicator with a known concentration in the fluid used for lavage. Baughman and colleagues (16) used methylene blue dye in the lavage fluid to estimate the dilution of lavage fluid by ELF in humans. They reported that V_ELF represented 35 ± 15% of the total fluid present in the lavaged lung segment. However, their calculations of V_ELF were overestimated because of the loss of methylene blue indicator, possibly through diffusion into the interstitium and the bloodstream, or chemical reduction to its colorless form (17). Peterson and colleagues (18) used technetium pertechnetate (^99mTcO₄) as a lavage fluid dilution indicator to estimate V_ELF in sheep. They estimated a V_ELF of 1.6 ± 1.0 ml for a single lung segment. They used a rewash lavage method where the same fluid was reinstilled four consecutive times. Because ^99mTcO₄ rapidly diffuses out of the alveolar compartment, these investigators used an extrapolation method to calculate ^99mTcO₄ in the initial lavage fluid, prior to exchange. In this technique, changes in the ^99mTcO₄ concentration may be due to either exchange of indicator or exchange of fluid between the alveoli and the bloodstream, neither of which can be estimated separately. Also, because only 18% of the total instilled lavage fluid volume was recovered, it is possible that the ELF of the lung segment was only partially mixed and therefore partially sampled. Finally, Stephens and colleagues (19) used a non-diffusible indicator; ¹²⁵I-albumin, to estimate V_ELF in postnatal sheep lungs. In their studies the lungs were saturated with fluid. The instilled fluid was actively removed from the alveoli; however, these investigators were able to calculate the total volume of fluid remaining in the lungs from the activity of ¹²⁵I-albumin. Using back-extrapolation of best-fit line for remaining lung fluid volume against time, the total volume of fluid, i.e., instilled fluid volume plus V_ELF, was calculated at the time of instillation, prior to fluid exchange between the alveoli and the lung tissues and the bloodstream. However, because V_ELF was very small compared with the instilled fluid volume, small errors in the back-extrapolation method were enough to make individual measurements of V_ELF inaccurate, the values of which ranged from 2.87 to 3.79 ml/kg body weight. Using a similar analysis of the Group 2 control data in the present study, we found an estimated V_ELF of 1 ± 0.6% of V_t,n, which ranged from 0.6 to 4.1% (20), confirming that the above method is less accurate than our method for individual estimations of V_ELF.

It is difficult to compare our calculated values of V_ELF with those previously reported in the literature because of differences in the lavage procedure. Ideally, V_ELF should be expressed as a function of the considered airway surface area. Unfortunately, surface area cannot be measured independently, particularly in vivo. If an assumption is made that the volume of fluid in the lung segment at saturation is directly related to the alveolar surface area coming in contact with it, then V_ELF may be expressed as a function of the fluid volume present in the lavaged lung segment. On the basis of work by Taylor and Guyton (21) showing that 85% of the instilled fluid entered the alveoli, Goetzman and Visscher (22) estimated the alveolar surface area of fluid-filled alveoli from the airway fluid volume, according to a model assuming spherical alveoli that are either filled or empty. Using a similar model, and assuming an alveolar radius of 45 µm (23), we calculated a mean alveolar surface area of 5.3 · 10⁴ cm², which came in contact with the lavage fluid at saturation. The estimated thickness of the ELF layer in our study would therefore be 0.30 µm in normal lung. ELF thickness values of 0.20 µm measured by morphometric techniques have been reported in rats (24). Because the above model requires several assumptions, we chose to express V_ELF as a function of _{t,n(n = 5,8)} or the average volume of fluid present in the lung segment at saturation. On the other hand, if the lung volume of dogs with an average weight of 14.7 kg is estimated as 984 ml (25), total lung ELF can be estimated as 16.7 ml or 1.13 ml/kg.

V_ELF was increased in all animals with oleic-acid-induced lung injury. The variation coefficient of V_ELF was similar in Group 2 control and injury experiments. The large increase in EVLW/BFDW reflected the presence of edema in injured lung. However, no correlation was found between the EVLW/ BFDW ratio and V_ELF in Group 2 injury experiments. This may be due to several factors. First, anatomic distribution of injury and edema is inhomogeneous in the oleic acid model (26). Second, in order to avoid complicated calculations to correct for the instilled fluid present in the lavaged lung, we chose to measure the extravascular lung water of the contralateral lung, which may have been another cause of variation. Also, EVLW was not measured in Group 2 control experiments because the same animals underwent injury experiments later on. The reference value for EVLW/BFDW was obtained in other control experiments in similar conditions. Finally, an increase in EVLW reflects both an increase in ELF (alveolar edema) and in interstitial fluid (interstitial edema). In normal lung, fluid exchange may occur in a three-compartment system (plasma  interstitium  ELF), where the permeability of the epithelial barrier is much lower than that of the endothelial barrier (27). The comparatively low permeability of the epithelial membrane is presumably responsible for the fact that fluid will first accumulate in the interstitium before entering the alveoli. In the presence of lung injury where the permeability of both epithelial and endothelial membranes are elevated, this system is simplified to two compartments (plasma  ELF), and fluid moves more freely from plasma into the alveoli. Consequently, EVLW is a poor reflection of the fluid contained in the alveoli, particularly in normal lung where V_ELF is very small compared with the total EVLW.

In summary, our SBAL technique allowed us to reliably measure ELF volume, which was expressed as a percentage of total fluid volume present in the lung segment at saturation. V_ELF values were reproducible from one subject to another, and the presence of edema fluid in the injured lung was detected. In the future, simplifications of this technique such as reduction of the number of sequential saturation lavages to only four and validation of nonradioactive indicators may allow trials of this technique in human subjects.