Fetal lung liquid volume is usually determined by using radio-iodinated serum albumin (RISA) or blue dextran (BD) as volume tracers. We tested the reliability of both tracers at 124 (G124) and 142 days of gestation (G142; term = G147) when the labels were employed simultaneously. We measured the proportion of label bound reversibly to the lung, or apparently lost from the lung compartment, by washing out the lung with saline and 5% albumin. At G124, volume estimates with the two labels were similar. At G142, the volume estimate with BD (36.3 ± 8.7 ml/kg of body wt) was higher (P < 0.05) than with RISA (22.3 ± 3.5 ml/kg). This difference resulted from reversible binding of BD, because 5% albumin washout released 38.5 ± 4.0% of the BD added at the start of the experiment but a lesser amount of RISA (9.8 ± 0.7%; P < 0.05). At G142, when RISA was used alone, its reversible binding was 1.3 ± 0.2%. Background absorbance increased during experiments, giving rise to an apparent increase in BD concentration. We conclude that RISA is an effective tracer for lung liquid volume determination in the fetal lamb, whereas our findings of substantial epithelial binding of BD and large changes in background absorbance demonstrate that, under the conditions of our experiments, BD is a poor tracer close to term.
- fetal lamb
- radioiodinated serum albumin
- volume tracer
the fetal lung secretes a liquid that distends the future air space and plays a crucial role in promoting lung growth (1). Although lung liquid plays an important role in fetal development, it must be cleared from the lungs for effective gas exchange after birth. Recent evidence demonstrates that the bulk of this liquid leaves the lungs before birth (5), but the exact timing of this prenatal clearance is in dispute. Until recently, the only available published evidence suggested that lung liquid volume begins to decline several days before labor (11). However, a number of recent reports suggest that lung liquid volume continues to rise until the day before labor, reaching a level as high as 50 ml/kg of body weight (16,18, 22). A possible explanation for the contradictory findings about the volume of liquid present in the lungs of the late-gestation fetal lamb is that they result from differences in measurement techniques. In most studies of fetal lung liquid, its volume is determined by the indicator dilution technique, in which a known amount of tracer is completely mixed into the liquid occupying the airways and alveoli of the lung compartment. Volume is then determined by dividing the quantity of tracer added by its concentration in lung liquid. In the study reporting that lung liquid volume declines before labor (11), the indicator or tracer used was radioiodinated serum albumin (RISA), whereas blue dextran (BD) was used in studies reporting that lung liquid volume continues to rise until labor (16, 18, 22).
The dye moiety of BD (Cibacron Blue F3-GA) is known to have a high affinity for proteins (10), a property that has led to its frequent use in protein separation techniques (14, 19, 29). Accordingly, we suspected that BD may bind to cell-surface proteins on the apical surface of the maturing lung epithelium, just as it binds to the pulmonary vascular endothelium (25, 28). If this were so, BD would fail to satisfy a fundamental requirement of the indicator-dilution technique, namely, that the tracer should remain freely dispersed. As a result, use of BD would give rise to an overestimate of lung liquid volume. We, therefore, set up a study in which we evaluated BD and RISA in the measurement of lung liquid volume at a gestational age at which lung liquid volumes determined with the two tracers are similar [gestation day 124 (G124)] and at a gestational age at which reported volumes are widely different (G142). We were especially interested in how much of each tracer remained freely available during experimental conditions in vivo. Whereas most tracer validation studies have searched for tracer “lost” from the compartment of interest, in our novel approach we focused particularly on the amount of tracer that could be retrieved from the lung compartment itself, because it is the freely available or retrievable tracer that is the basis for the volume calculation.
MATERIALS AND METHODS
Ethical approval was obtained from the Monash University Standing Committee on Ethics in Animal Experimentation for all in vivo experiments.
Twenty fetuses of pregnant Border-Leicester ewes were instrumented at G120 or G133 (term = G147) for study at G124 or G142. An additional three fetuses were instrumented on the day of an acute study, either at G141 or G144. In all operations, anesthesia was induced with intravenous thiopental sodium (1–2 g; Pentothal, Abbott) and maintained with 66% N2O-1–2% halothane-32–33% O2.
The maternal abdomen was opened in the midline, the fetal head was delivered through a uterine incision, and the neck was opened in the ventral midline. The carotid artery was catheterized nonocclusively by using a Teflon cannula-Tygon tubing assembly. Two wide-bore Silastic catheters (internal diameter 3 mm, outer diameter 6 mm; Sil-Med, Taunton, MA) were introduced through a single tracheostomy, with one catheter directed rostrally and the other caudally and the ends connected to form a liquid-filled loop. A liquid-filled catheter open to the amniotic space was secured to the skin of the fetal neck. The fetus was returned to the uterus, which was carefully closed to avoid leakage of amniotic fluid. The ewe's abdominal wall was sutured, and all catheters were tunneled subcutaneously to the animal's flank where they exited through a small incision. All animals received postoperative analgesia (50 mg Finadyne, Schering-Plough) and daily intramuscular antibiotic treatment (500 mg procaine penicillin, 500 mg dihydrostreptomycin). Animals were allowed to recover for ≥4 days before observations commenced.
Preparation of Tracers
RISA was prepared by iodination (125I) of BSA by using Iodo-Gen (Pierce, Rockford, IL) using the technique of Fraker and Speck (12) no more than 6 wk before use. To remove free125I, the tracer was either mixed with Amberlite IRA-400 (Sigma Chemical) and allowed to stand for 15 min immediately before use, or it was passed over a Sephadex G-25 column (Column PD-10, Pharmacia Biotech). Specific activity of RISA approximated 20 μC/ml, and the albumin concentration was ∼0.5%. In each experiment, 0.1–0.2 ml of stock solution were diluted with 3–5 ml of saline (0.9%) to which fetal serum (0.5–1 ml) was added. As a result, the concentration of unlabeled or “cold” albumin exceeded that of RISA by at least an order of magnitude, with the aim of limiting any loss of label that would result from albumin binding to the lung. The tracer was then passed through a Millipore filter (0.22-μm Millex-GV filter unit; Millipore, Bedford, MA) to ensure sterility and to exclude amberlite from the sample.
A solution of BD was obtained by dissolving dry BD powder (Sigma Chemical) in lung liquid maintained at 40°C. Complete dissolution was achieved by continuous stirring for 15 min.
Measurement of Tracer Concentration
The concentration of RISA was measured on a gamma counter (1282 Compugamma, LKB, Wallac) in 0.5-ml samples counted for 10 min on three successive runs through a counting window set between 20 and 100 KeV, and the average of the three runs was used in calculations.
Calculation of Liquid Secretion Rate and Volume
Lung liquid volume was calculated according to established techniques (7). Slopes of lung liquid volume against time were calculated by using the least squares method, after values obtained during the first 20 min of the experiment were removed, because these may be affected by incomplete mixing. Lung liquid volume at the time of interruption of the tracheal loop (time 0) was obtained by extrapolation.
Physical and Chemical Properties of BD in Vitro
The time taken for dissolution of BD was established by adding BD (2 mg/ml) to a sample of water or lung liquid that was maintained at 40°C and stirred continuously. After the addition of BD, samples were taken from the mixture at 5, 10, 15, 30, and 60 min, as well as 24 h later. After 15 min, absorbance had reached >97% of the value determined at 24 h, and this time was independent of the solvent. As proof of the complete dissolution of BD at the higher concentrations used in vivo (see below), we sought to spin out undissolved BD at a concentration of up to 6 mg/ml after 15, 30, 45, and 60 min of stirring. Absorbance levels before and after spinning for 60 min at 2,000 g were not different for any duration of stirring.
BD absorbance was directly proportional to its concentration between 0 and 3 mg/ml and independent of solvent, with the slope of the relationship between absorbance and BD concentration being similar for H2O, 0.9% NaCl, and 5% BSA. By contrast, the absorbance of BD was reduced by 21.7% when amiloride hydrochloride (amiloride, Sigma Chemical) was present at a concentration of 10−4 M in water.
BD absorbance was not affected when incubation temperature was reduced from 40 to 4°C, nor was the absorbance of BD altered if samples were stored in darkness or exposed for different time periods (1–180 min) to strong artificial light to mimic laboratory conditions.
The possibility that BD might sediment or spin out was assessed in samples of BD (0–2.5 mg/ml) diluted in either normal saline (n = 13) or 5% BSA (n = 13). Centrifugation at 2,000g for different time periods up to a maximum of 60 min did not change absorbance of BD in saline or in lung liquid (see background absorbancebelow). However, when BD in 5% BSA was spun, we found a small but significant 7% reduction in absorbance after 60 min at 2,000g. This small reduction in absorbance in 5% BSA would have no effect on the magnitude of what we call the “free” fraction of BD (see Tracer Availability), but it would cause the “bound” fraction of BD to be slightly underestimated and the “irrecoverable” fraction of BD to be correspondingly overestimated.
Characteristics of BD and RISA in Vivo
We employed four experimental protocols. The first (BD+RISA) aimed to compare BD and RISA as lung liquid volume tracers in studies utilizing both tracers either at G124 (n = 5) or G142 (n = 7). The second (RISA only) investigated the accuracy of RISA when used alone in fetuses at G142 (n = 7). The third protocol (BD only), in which BD was used alone at G142, was performed in a single animal to check whether the high binding of BD to the lung, as demonstrated in the BD+RISA protocol, was dependent on the presence of RISA. These three protocols consisted of the same four experimental periods, each lasting ∼90 min. Lung liquid volume and secretion rate were determined with BD over the first period. After RISA was added, volume and secretion rate were determined simultaneously with both tracers during the second period, which was followed by two washout periods: the first with saline and the second with 5% BSA. A fourth protocol (BD+RISA: acute preparation) was performed to address the objection that the high binding of BD that we observed in the lung in late gestation results from infection, giving rise to an abnormal increase in epithelial cell surface proteins or to an increase in the particulate content of lung liquid. Accordingly, we performed acute studies in three late-gestation fetuses immediately after completion of sterile surgery.
Protocol 1: BD+RISA: Comparison of BD and RISA in simultaneous use.
experimental period 1: addition of bd.
The limb of the tracheal loop directed toward the lung was connected to a temperature-controlled (40°C) glass burette that was closed at its top by a rubber stopper that was penetrated by an 18-gauge needle to which a Millipore filter was attached. This allowed liquid to be drained from, and reinstilled into, the lung under sterile conditions while minimizing evaporative water loss. A volume of 50–100 ml of liquid was obtained from the lungs, and a small sample was taken to measure the background absorbance and radioactivity of lung liquid. BD (250–300 mg) was then mixed by stirring for 15 min into a sample of lung liquid (50–100 ml), and a specimen (3 ml) was withheld before the remaining liquid was accurately weighed (Mettler AE 166 delta range) and reinstilled into the lung. The tracer was mixed with lung liquid by repeated cycles of drainage and reinstillation over at least 30 min by using a maximum hydrostatic pressure of ±15 cmH2O. Subsequently, between 12 and 15 samples (1 ml) were taken at 5- to 10-min intervals, and absorbance at 620 nm was measured.
EXPERIMENTAL PERIOD 2: ADDITION OF RISA.
After draining lung liquid into the burette, we again withdrew 50–100 ml of this liquid, now containing BD, to which we added RISA. After thorough mixing, a 1.5-ml sample was taken to determine radioactivity level, and the volume of liquid remaining was accurately weighed before it was reinstilled and mixed into the lung over a 30-min period as described above. An additional 12–15 samples (each 1.5 ml) were taken at 5- to 10-min intervals for determination of absorbance and gamma radiation.
EXPERIMENTAL PERIOD 3: SALINE WASHOUT.
We then sought to retrieve both tracers from the lung by repeated washout. The procedure entailed draining as much liquid as possible from the lungs and replacing it with ∼20 ml/kg of isotonic saline at 40°C. This was repeated seven times and resulted at the final wash in BD and RISA levels close to background.
EXPERIMENTAL PERIOD 4: ALBUMIN WASHOUT.
After the saline washes, we performed another series of seven washes with 5% BSA. Albumin has a very high affinity for BD (2, 10, 14, 30), and we expected it would release any tracer bound to the epithelium, just as it releases BD from the pulmonary endothelium (25, 28).
We further detailed whether either volume tracer had crossed the pulmonary epithelium by sampling fetal carotid artery blood (n = 5) before and 5, 10, 20, 30, 60, 120, and 240 min after introducing the tracer to the lung. At the end of the experiment, all animals were killed, and maternal blood and the following fetal and maternal organ samples were analyzed for the presence of tracer: lung, thyroid, liver, kidney, heart, muscle, and skin.
Protocol 2: RISA only: Evaluation of RISA used alone as a volume tracer.
This protocol, performed only at G142 (n = 7), incorporated all four steps described in the previous protocol, except that no BD was used. During experimental period 1 no tracer was present, and during experimental period 2 RISA was added. This experiment provided an estimate of the free, bound, and irrecoverable tracer that could be compared with the values obtained in the BD+RISA protocol. Any adverse effect of BD on RISA as a volume tracer, and especially whether the presence of BD altered the amount of bound and irrecoverable RISA, would be revealed by the difference in results obtained between the two protocols.
In preliminary experiments employing RISA alone, the initially clear lung liquid draining into the burette appeared to become more opaque during the course of the experiment. To establish whether this change in opacity increases background absorbance, which would falsely elevate the BD concentration of the liquid, we measured absorbance at 620 nm in every third sample of periods 1 and2, and in every sample ofperiods 3 and4 during the RISA-only protocol. To examine whether particulate components in lung liquid may cause this effect, we then centrifuged the same samples for 60 min at 2,000g before measuring absorbance again.
To examine whether BD binds to the material responsible for increasing the opacity of lung liquid, BD was added to selected samples, and absorbance was measured before and after centrifugation at 2,000g. For this purpose we chose the third sample of lung liquid (n = 7) taken inperiod 1 of the RISA-only protocol, because this sample had a high-background absorbance. Absorbance of the sample was measured before and after centrifugation. Then an amount of BD was added to each sample to give a concentration ranging between 0.5 and 2.5 mg/ml and was mixed thoroughly, and absorbance was measured again. The samples were then centrifuged as stated above, and absorbance was remeasured. If BD were to bind to this material, then we would predict that the change in absorbance on centrifugation would be greater in the presence of BD than in its absence.
Protocol 3: BD only: Evaluation of BD used alone as a volume tracer.
In this protocol, all four steps described inprotocols 1 and2 were performed in a single animal at G142, except that no RISA was used. This experiment was performed to test whether the presence of RISA affects the amount of BD that is bound to the lung or is irrecoverable by washout.
Protocol 4: BD+RISA: Acute preparation.
This experiment was designed to evaluate the possibility that infection introduced at the time of surgery caused a response in the lungs over ensuing days that led to a high level of BD binding in protocols 1 and 3. Surgery was performed as in the earlier protocols under sterile conditions in three fetuses (2 at G141 and 1 at G144), and the volume determination was performed immediately afterward. The protocol used here began at experimentalperiod 2, as described inexperimental period 2 : addition of risa . BD and RISA were added simultaneously to ∼50 ml of lung liquid, and samples were taken to provide an estimate of lung liquid volume with both tracers. At the end of the volume determination, all lung liquid was returned to the lung, the trachea was clamped, the ewe and fetus were killed, and the lung was removed and weighed. Seven saline washes were then performed, followed by seven washes with 5% albumin. Samples of each wash were taken for determination of absorbance and radioactivity. The lungs were dried to constant weight over a period of ∼2 wk, and the weight was expressed in grams per kilogram of body weight. Using the factor scaling dry lung weight to normal wet parenchymal weight in the newborn lung (5.29; see Ref. 5), we could estimate lung tissue weight and obtain the volume of liquid in the air space by subtracting lung tissue weight from total wet weight at postmortem. This value represents the volume of liquid in the lung at the time the fetus was killed and could be compared with the volume estimated with BD and with RISA at the same time.
Data Analysis and Presentation
Raw tracer concentration and volume of samples were used to calculate the amount of tracer removed with each sample, and, knowing the total amount added, we calculated the pool of tracer remaining in the fetus at each step of the protocol. This value was expressed as a fraction of the pool introduced at the start of the experiment. For Fig. 4, we recalculated lung liquid volumes, taking into account our finding that background absorbance changes during an experiment; for this calculation we used the absorbance of centrifuged samples. In addition, we recalculated lung liquid volume using only the fraction of the BD that was freely available in the lung compartment. Comparisons among groups were performed by using Student's paired and unpairedt-tests for normally distributed data; otherwise, the Wilcoxon signed ranks test was used. ANOVA was used to compare the lung liquid volume estimates derived in the three acute studies. Results are expressed as means ± SE unless otherwise stated. Differences were considered significant whenP < 0.05.
We conducted experiments in 20 chronically instrumented healthy fetuses, as indicated by their weights, arterial blood gases, and acid-base status (Table 1), and postmortem examination. An additional three fetuses were studied immediately after acute instrumentation.
Figure 1,top, shows tracer concentration (absorbance or counts ⋅ min−1 ⋅ ml−1) in each sample of lung liquid plotted against time. Secretion of lung liquid caused tracer concentration to fall gradually inperiods 1 and2 of the experiment. Inperiod 3, successive washes of the lung with saline resulted in a rapid and exponential fall of tracer concentration to a value close to zero. In period 4, the addition of 5% BSA resulted initially in a substantial increase in BD concentration and a minimal increase in RISA. Subsequent BSA washes led to the concentration of both tracers falling exponentially toward zero.
In Fig. 1, bottom, the decline in each tracer pool remaining in the fetus is plotted against time. Removal of samples during periods 1 and2 slowly reduced the tracer pool size, until a substantial fall in the pool occurred after the last sample inperiod 2, when we removed as much liquid as possible from the lungs. Saline washes inperiod 3 exponentially decreased the pool to a first-plateau level. A second plateau was reached with the 5% BSA washout in period 4. The fraction of the pool that lies above the first plateau represents a freely available fraction that is readily retrieved by sampling and saline washout (free fraction), whereas the fraction between the first and second plateau represents a bound fraction that is unavailable until it is released by 5% BSA washout. The tracer pool that lies below the second plateau could not be recovered from the lung compartment (irrecoverable fraction).
Average results for the seven experiments are illustrated in Fig.2. The fraction of both BD and RISA that was bound was four- to fivefold greater at G142 than at G124. At both gestations, significantly more BD than RISA was bound to the pulmonary epithelium: thus at G124, 11.3 ± 1.9% of BD was bound compared with 2.2 ± 0.04% of RISA (P < 0.05), whereas at G142, 38.5 ± 4.0% of BD was bound compared with 9.8 ± 0.7% of RISA (P < 0.05). When RISA was used alone, its bound fraction was significantly smaller (seeEffect of BD on RISA binding).
In the BD+RISA protocol, at G124 the irrecoverable BD (2.9 ± 3.3%) was not different from the value at G142 (5.1 ± 3.8%), nor was the irrecoverable fraction of RISA different between G124 (4.8 ± 0.3%) and G142 (8.2 ± 1.6%). In the RISA-only protocol, 4.9 ± 0.8% of the added RISA was irrecoverable, a value that was not significantly different from the irrecoverable RISA at G142 in the BD+RISA protocol.
For BD, we were unable to determine the location of the irrecoverable fraction. This may be explained by technical limitations, because we could not detect BD in blood even when we added it to a sample of whole blood in vitro at a concentration of 2 mg/ml, nor was BD detectable when the blood sample was centrifuged and the serum analyzed for BD. In the case of RISA, very low 125I gamma activity was found, in decreasing order of activity per gram of wet tissue weight, in fetal thyroid, fetal lung, fetal liver, fetal blood, and maternal thyroid.
Interaction Between Tracers
Effect of RISA on BD binding.
To test whether the binding of BD (38.5 ± 4.0%) might be dependent on the presence of RISA, we performed a single experiment at G142 using only BD as the tracer (BD only). In this experiment, 48.1% of the BD introduced at the start of the experiment was released by 5% BSA washout, a level of binding within the range observed in the BD+RISA protocol.
Effect of BD on RISA binding.
To test whether binding of RISA required the presence of BD, we compared the bound fraction of RISA at G142 obtained in the presence of BD (BD+RISA protocol) with the same fraction determined in the absence of BD (RISA-only protocol). As shown in Fig. 2, the proportion of RISA bound was approximately eightfold less in the absence of BD compared with the proportion bound in its presence (1.3 ± 0.2 vs. 9.8 ± 0.7%; P < 0.05).
Changing Background Absorbance
To assess whether the background absorbance of lung liquid is constant throughout the experiment, we measured absorbance in every third sample taken in periods 1 and2 and in every sample duringperiods 3 and4 in the RISA-only protocol at G142 (n = 7). Absorbance of lung liquid increased from a mean of 0.31 ± 0.05 absorbance units (AU) in the sample taken at the experiment onset to a maximum of 0.63 ± 0.11 AU by the third sample, by which time ∼45 min of mixing had occurred. Thereafter, a slow downward trend was noted (Fig. 3). Interestingly, the observed increase in absorbance was present not only at 620 nm, but over the whole spectrum from 325 to 900 nm. Centrifugation of the samples for 60 min at 2,000 g reduced absorbance of all samples to values ranging from 0.18 to 0.23 AU at 620 nm (Fig. 3); this range is close to the absorbance of water [0.18 ± 0.01 (SD) AU], saline (0.18 ± 0.01 AU), and 5% BSA (0.19 ± 0.01 AU). Centrifugation for ≤30 min at 2,000g, however, did not completely remove the increase in background that occurred during the course of an experiment, and centrifugation for 10 min at 250g, as used in some studies employing BD for lung liquid volume determination (13, 21), reduced the rise in background absorbance by only ∼50%.
To determine whether BD binds to the material causing a rise in background absorbance, we added BD to lung liquid samples with high-background absorbance and measured absorbance before and after centrifugation. This procedure was performed forsample L3 in each experiment in the RISA-only protocol. The reduction in absorbance obtained by centrifugation in the presence of BD (0.44 ± 0.12 AU,n = 7) was not different from the reduction obtained by centrifugation before BD was added (0.44 ± 0.11 AU, n = 7), indicating that BD did not spin down with whatever material is responsible for the rise in background absorbance.
Estimates of Lung Liquid Volume and Secretion Rate
In the BD+RISA protocol, at G124 (n = 5), lung liquid volumes calculated without correcting for the fraction of tracer that is bound, or for changes in background absorbance, were 29.3 ± 3.9 ml/kg for BD and 29.3 ± 3.2 ml/kg for RISA. By contrast, at G142 (n = 7), calculated volumes obtained with BD were >60% greater (P < 0.05) than those obtained with RISA (36.3 ± 8.7 vs. 22.3 ± 3.5 ml/kg; Fig.4).
The volumes given above with each tracer were obtained with comparable methodology to those reported in all other studies that used these tracers in the fetal lamb. However, it is also useful to examine how correction for bound and irrecoverable tracer affects estimated lung liquid volume. When BD concentrations were corrected for the background effect and for the fact that only a fraction of the added tracer is free, volumes obtained with BD at G124 (31.5 ± 2.8 ml/kg) did not differ significantly from volumes obtained with RISA. By contrast, at G142, calculated volumes obtained with BD (29.7 ± 3.1 ml/kg) remained greater than those obtained with RISA by >30% (P < 0.01; see Fig. 4).
BD+RISA in Acute Preparation
As in the BD+RISA protocol in chronically prepared fetuses, a substantial proportion of each tracer was bound in three acute studies performed immediately after sterile operation (Table2). This finding demonstrates that lung infection is not responsible for BD binding in our chronically prepared fetuses. Lung liquid volume determined from the wet and dry weights of the lung, and from indicator dilution by using BD and RISA, is also shown in Table 2. When lung liquid volume was calculated without correcting for the bound and irrecoverable tracer, both tracers grossly overestimated lung liquid volume as determined from wet and dry lung weights. When only the free amount of tracer was used in calculating the volume of liquid in the lung just before the fetus was killed, in each of the three fetuses studied, the volume derived with BD was greater than that derived with RISA, but there was no statistical difference between the two estimates (Table 2). Although further data would be needed before reaching a conclusion, in this limited series the volumes estimated with both tracers were not significantly different from those calculated from lung weight (Table2).
Two lines of evidence support the idea that prenatal clearance of liquid from the fetal lungs is important for postnatal gas exchange. First, newborn infants born by elective cesarean section frequently have an excess of liquid in the lungs and typically develop a form of respiratory distress that is usually referred to as “wet lung” (15, 24, 27). Second, physiological experiments show that removal of approximately one-half of the liquid present in the fetal lungs before cesarean delivery speeds the normal rise in arterial O2 levels that occurs immediately after birth (6).
Although it is generally accepted that liquid is cleared from the lungs before delivery, just when this clearance occurs is under debate. In one study, it was reported that lung liquid volume begins to decline in the last days of gestation in the lamb (11), suggesting that the process of adapting the lungs for postnatal function is triggered before labor. By contrast, recent reports present evidence that the volume of liquid in the fetal lungs continues to rise until the day before labor (16, 18, 22). On the basis of these reports, the mechanism that results in a decline in lung liquid before delivery would be initiated only after the start of labor. Such starkly differing results led us to test whether the discrepancy might be explained by the only methodological difference among the studies, with one using RISA as the volume tracer (11) and the others using BD (16, 18, 22). We hypothesized either that RISA underestimates volume in late gestation, or that BD overestimates it. Our results clearly show that when BD and RISA are used together at G124, the two tracers produce similar estimates of lung liquid volume. By contrast, in chronically prepared fetuses at G142, lung liquid volume estimated with BD exceeds that derived from RISA by an average of 63%. We show that this overestimate springs from a number of weaknesses in the BD technique, the chief one being that a large proportion (average of 38.5%) of the BD added at the start of the experiment binds to the interior of the lung. When used in the presence of BD, a much smaller proportion of RISA (10.8%) also binds to the pulmonary epithelium, an effect that would cause overestimation of lung liquid volume derived from RISA. However, as we argue later, this seems likely to result from the radioiodinated albumin binding to the epithelium in association with BD, and when RISA is used alone only a tiny amount (1.3%) of radiolabel binds to the epithelium. Finally, the results obtained in three acutely prepared fetuses demonstrate that the high binding of BD that we report in near-term fetuses is not associated with infection.
Our findings show that, in our fetuses at G142, BD fails to satisfy a key requirement of the indicator dilution technique; that is, the tracer must remain freely available in the compartment being studied so that its dilution accurately reflects the volume of liquid in that compartment. We have also shown that the use of BD near term does not satisfy a second major requirement of the indicator-dilution technique, namely, that the background level of the chosen tracer must remain constant over the course of an experiment. This background effect is most easily seen from the results obtained when samples taken in the RISA-only protocol were analyzed for absorbance at 620 nm. Such changes in background absorbance after the start of an experiment may have a substantial effect on the estimated lung liquid volume. The magnitude and direction of the effect depend on the extent of mixing before the first sample is taken for measurement of background, because it is this value that is subtracted from all later samples. In addition, whether the sample is centrifuged will also determine the size of the error in the volume estimate. Because reports generally provide no information on either issue, the size of the problem in earlier publications is impossible to assess.
The two main sources of error inherent in the use of BD as a volume tracer in the lung of the late-gestation lamb fetus are very large. These errors, however, have potentially opposing effects on estimated lung liquid volume, with binding giving rise to an overestimate of lung liquid volume and increased background giving rise to a potential underestimate of volume. Accordingly, by chance some estimates of lung liquid volume reported in studies that used BD in late gestation might be close to the correct value. We may obtain insight into the magnitude of the error arising from binding and changing background by examining the difference between volume estimates derived from RISA and those derived from BD, either with or without correction for the two errors. As shown in Fig. 4, when no correction is applied, estimated lung liquid volume derived with BD at G142 exceeds the volume derived with RISA by an average of 63%. It is, therefore, clear that under our experimental conditions the binding effect exceeds the tendency for volume to be underestimated as a consequence of the background effect. Allowing for both sources of error, that is, by using the absorbance of centrifuged samples, and by using only the free fraction of BD in calculations, estimated lung liquid volume at the time the trachea was first connected to the burette at the start of the experiment still exceeds that derived from RISA by >30%.
Whereas the foregoing analysis provides insight into the magnitude of the errors that result from use of the BD technique near term in the fetal lamb, a number of issues remain. It is not obvious why the free fraction of BD and RISA can differ considerably at G124 and yet the volumes estimated with the two tracers do not differ significantly. Equally, it is not obvious why volumes calculated with BD at G142, after correcting for the two errors, do not equal those derived with RISA. One factor that may be important is the kinetics of BD binding to the epithelium. For example, slow binding throughout an experiment would mimic a high secretion rate, steepening the slope of the line that is subtended back to the start of the experiment to estimate lung liquid volume, thereby lowering the estimated lung liquid volume. This possibility is supported by our finding in acute preparations that BD, after allowing for bound and irrecoverable fractions, gives rise to lung liquid volume estimates reasonably similar to those derived from RISA at the end of the experiment, i.e., after binding is likely to be complete (Table 2).
An explanation must also be given for the observation at G142 that 9.8% of the RISA added to the lung is released by washout with 5% BSA when RISA and BD are used simultaneously, and yet, when RISA is used alone, only 1.3% of the added RISA is released by albumin washout. We propose that this finding is a consequence of the well-known high affinity of BD for proteins (2, 10, 14, 30), resulting in a molecule of BD binding reversibly both to the pulmonary epithelium and to the albumin of the RISA tracer. In this scenario, BD would remove RISA from lung liquid during the period of volume determination, followed by the release of both molecules by competitive displacement when unlabeled albumin in higher concentration (5%) is used to wash out the lungs. In the RISA-only protocol, the sites of attachment to the lung provided via BD are not available, and any binding of RISA to the epithelium must be direct; under these conditions, the bulk of any RISA that binds to the epithelium will derive from the cold albumin added along with the RISA tracer at the start of the experiment.
A final issue for explanation is that the only previous study that fully documents a comparison of BD and RISA as volume tracers in the fetal lung reported that they produced similar volume estimates (4). There may be a number of explanations for this result, one being that many of the fetuses used were studied earlier than G142. Although our results allow us to say no more than that the increased binding occurs at some time between G124 and G142, it is conceivable that binding and background errors remain small until fetuses approach G142. In addition, in the earlier validation study (4) many of the fetuses, and perhaps all of them, were studied at more than one gestational age. When multiple studies are carried out, most of the epithelial binding sites for BD may already be occupied when BD is introduced into the lung liquid at the start of an experiment, with the result that almost all of the added BD might remain free within the liquid compartment. Under such circumstances, BD could provide an acceptable estimate of volume if the background effect were also minimized by the design of the experiment. For example, if the protocol called for lung liquid to be thoroughly mixed before the first sample was removed for determination of background absorbance, background absorbance might already have reached its peak value (see Fig. 3) before the addition of BD. Interestingly, even if multiple studies were performed in each fetus, the high affinity of BD for albumin would result in some of the RISA added during each experiment binding to the epithelium in association with BD, giving rise to an overestimate of volume with RISA.
The large increase in BD binding to the pulmonary epithelium between G124 and G142 demonstrates that binding sites increase in number or affinity between these ages. Given the high affinity of BD for proteins, higher binding could result from an ontogenetic increase in cell surface proteins (20, 23), and these may include epithelial Na+ channels, which are induced by cortisol and thyrotropin-releasing hormone near term in the fetal lamb (3).
Whereas the biggest error in the BD technique results from the bound fraction, both BD and RISA introduce error to volume estimates through a fraction that is irrecoverable by washout. In the BD+RISA protocol, this irrecoverable fraction for both tracers was <10%, giving rise to a theoretical volume overestimate of ∼11%. Where the irrecoverable tracer is located proved impossible for us to determine for BD, because we could not demonstrate its presence in blood. Thus the statement that there is no BD measurable in blood in other studies (8, 9, 17, 26) cannot be used as evidence that BD is confined entirely within the lung compartment. For RISA we established that radiolabel was present in fetal blood and a number of fetal and maternal tissues, including the thyroid, after RISA was added to the fetal lung liquid. Whether this label was in the form of free iodine or was still bound to albumin we did not attempt to determine, but it represented only 4.9 ± 0.8% of the added tracer in the RISA-only protocol, and it would, therefore, introduce only a small error to volume estimates.
In summary, we have shown that, when used alone, RISA satisfies the assumptions of the indicator dilution technique, both at G124 and near term. By contrast, BD binds strongly to the pulmonary epithelium of the fetal lamb near term. This binding, together with the rise in background absorbance that occurs during experiments, makes BD unreliable as a tracer for lung liquid volume in the near-term fetal lamb, at least under the experimental conditions we used. Thus the important question of whether lung liquid volume declines before labor, or continues to rise until labor begins, now needs to be carefully reexamined.
We are grateful to A. O'Connor for preparation of RISA.
Address for reprint requests and other correspondence: P. J. Berger, Institute of Reproduction and Development, Monash Medical Centre, 246 Clayton Rd., Clayton, Victoria 3168, Australia (E-mail:).
This research was supported by the National Health Medical Research Council of Australia, Swiss National Foundation, Litta Foundation, Ciba-Geigy Jubilee, Glaxo Wellcome, Roche Research Foundation, and Monash University.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
- Copyright © 1999 the American Physiological Society