INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The pathogenesis of acute respiratory distress syndrome (ARDS) remains poorly understood. Consequently, current mortality rates of ~ 50% to 60% in patients with ARDS differ little from those of three decades ago, when the condition was first described. A key factor in preventing better patient management in ARDS is the inability to detect and quantify early damage to the alveolocapillary membrane. Thus, ventilatory regimens that aggravate the existing high-permeability edema characteristic of the injured lung continue to be inadvertently used (1), and antiinflammatory therapies continue to be poorly prescribed (2).

Although numerous methods for monitoring lung permeability have been assessed, none has reached clinical utility. Whereas measurement of the flux of radiotracers into or out of the lungs has shown promise for measuring lung permeability, logistical problems have precluded their use (3). Consequently, attention has turned to the utility of biologic markers (4), either in blood or in the epithelial lining fluid, for this purpose. However, although the plasma protein content in epithelial lining fluid may specifically reflect changes in lung permeability, bronchoalveolar lavage (BAL), which is necessary for the measurement of this protein content, has associated risks and may also present logistical difficulties. In addition, proteins in the epithelial lining fluid are only slowly cleared from the alveolus, and therefore lack the required sensitivity for monitoring lung permeability. On the other hand, because ARDS is invariably associated with systemic inflammation and a related increase in vascular permeability, cellular and soluble inflammatory markers also lack the required specificity.

Whereas increased permeability and the flux of plasma proteins into the alveolar compartment impairs surfactant function and sets in train a vicious cycle of biophysical events that lead to worsening respiratory failure (5), our work suggests that the flux of surfactant proteins out of the air spaces and into the circulation may fortuitously provide a sensitive means of noninvasively monitoring the lung (6, 7). Arguably, surfactant proteins are uniquely suited as biomarkers of lung injury. They are found in appreciable amounts only in the alveolus, and are synthesized and secreted in high concentrations directly into the hypophase, where the surface area of the epithelium is ~ 50 times that of the body and only ~ 0.1 to 0.2 µm thick. Based on estimates of surfactant protein concentration in the hypophase, the alveolocapillary concentration gradient is normally on the order of ~ 1:1,500 for immunoreactive surfactant protein-B (SP-B) and ~ 1:7,000 for immunoreactive surfactant protein-A (SP-A) (8). Consequently, it is little wonder that these proteins diffuse down the gradient, and that they occur at high circulating concentrations in patients with impaired alveolocapillary permeability (6, 7).

However, circulating levels of lung proteins will be influenced not only by the degree of alveolocapillary permeability and rate of entry of these proteins into the circulation, but also by their rate of clearance. Virtually nothing is known of plasma protein clearance in patients with acute lung injury (ALI), but because ARDS is invariably associated with systemic inflammation and increased vascular permeability, plasma protein clearance is presumably also altered in this condition.

Since the diagnostic utility of SP-A and SP-B as markers of lung permeability demands a sound understanding of the kinetics of their movement into and clearance from the circulation, we compared plasma SP-A and SP-B concentration with various physiologic parameters, including indices of lung dysfunction (blood oxygenation and lung injury score [LIS]) and glomerular filtration rate (GFR) (of plasma creatinine and cystatin C) in matching arterial and mixed venous blood samples from 37 patients with acute respiratory failure. In addition, we compared our findings for SP-A and SP-B with those for Clara cell secretory protein (CC16), an independent, low-molecular-weight airway protein that is known to be freely eliminated by glomerular filtration (9, 10).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients and Sample Collection

The study was approved by the Ethics Review Committee for Clinical Investigation of the Flinders Medical Centre. Informed consent for blood sampling was obtained directly from the patients or from their closest relative.

Blood sampling and plasma preparation. Blood was collected from 37 consecutive consenting patients (gender: 25 males and 12 females, with a median age of 67 yr [range 18 to 79 yr]) requiring assisted ventilation for acute respiratory failure. The clinical course of one patient was consistent with a diagnosis of acute cardiogenic pulmonary edema, whereas the remaining patients had severe hypoxemia and diffuse lung infiltrate not due to left ventricular failure (Table 1). Lung injury was scored according to the method of Murray and associates (11). Despite acceptable blood pressure (BP), acute tubular necrosis resulting from localized hypoperfusion is a frequent complication of ALI (12). At the time of blood sampling, eight of the 37 patients in our study had undergone at least 6 h of continuous veno-veno-hemodiafiltration (CVVHD) (Filtral AN 69 HF membrane; Hospal, France).

Blood was sampled at ~ 7.00 A.M. via both an indwelling arterial catheter and a pulmonary artery catheter. The blood was immediately centrifuged in lithium heparin tubes at 5,000 rpm for 5 min at room temperature (Megafuge; Heraeus-Christ, Osterode, Germany), and the plasma was stored at 20° C for batch analysis.

Clinical measurements. In addition to recording standard clinical parameters (6), we measured pulmonary capillary pressure (PCP) and cardiac output (CO) (determined by the thermodilution method) whenever possible on each occasion that blood was sampled (6).

Analytical Methods

Samples were assayed in a blind, randomized manner.

Glomerular filtration. Cystatin C is a cysteine protease inhibitor produced by most nucleated cells at a constant rate that is purportedly not altered by inflammation (14). Because it is freely filtered and catabolized in the proximal tubules, plasma cystatin C has recently been proposed as a more sensitive marker of the GFR than plasma creatinine, which increases only when the GFR has fallen by ~ 50% (13, 14).

Plasma creatinine was measured with Jaffé's technique (10). Plasma cystatin C was measured in serum with a latex immunoassay (LIA) (15).

CC16. CC16 was measured with an LIA (9). The assay uses the rabbit antiprotein 1 antibody (Dakopatts, Glostrup, Denmark) and protein purified in our laboratory as a standard. When pooled normal sera is fractionated by fast protein liquid chromatography on Sephacryl S-200 (Pharmacia Biotechnology, Uppsala, Sweden), CC16 elutes as a single component with an apparent M_r of ~ 16 kD, and is indistinguishable from the native protein.

To avoid possible interference by complement, rheumatoid factor, or chylomicrons, the plasma samples were pretreated by heating at 56° C for 30 min and by the addition of polyethylene glycol (16%, 1:1 [vol/vol] and trichloroacetic acid (10%, 1:40 [vol/vol]). After overnight precipitation at 4° C, the samples were centrifuged (2,000 × g for 10 min) and CC16 was determined in the supernatants. All samples were analyzed in duplicate at two different dilutions. The assay has a detection limit of 0.5 ng/ml and an average analytical recovery of 95%, with the intra- and interassay coefficients of variation ranging from 5 to 10% (9, 16).

Plasma SP-A and SP-B. SP-A and SP-B in plasma were first freed from any associated components with ethylenediamine tetraacetic acid (EDTA), sodium dodecylsulfate (SDS), and Triton X-100 (6, 7). The proteins were then measured with inhibition-type enzyme-linked immunosorbent assays (ELISAs) using Po-A and B, which are polyclonal antibodies raised against alveolar proteinosis-derived SP-A and mature SP-B, respectively (6, 7). Absorbance was measured at 405 nm using a Dynatech MR5000 reader (Dynatech Laboratories, Chantilly, VA). An AssayZap program (Biosoft, Ferguson, MO) was used to generate a standard curve and to compute the concentration of SP-A and SP-B in each sample. All samples were assayed in duplicate at four serial dilutions. Standards, assayed in quadruplicate, were included in each ELISA plate at eight serial dilutions (SP-A: 1.95 to 250.00 ng/ml; SP-B: 7.8 to 1,000.0 ng/ml, r > 0.99).

Statistics

We used nonparametric statistics, since plasma SP-A and SP-B are not normally distributed (7). Results are expressed as median (range). The Mann-Whitney U test or Wilcoxon's matched-pairs signed-ranks test was used for all comparisons. For comparisons with n > 20, significance was determined through the Mann-Whitney U test from the normal deviate Z, with reference to normal frequency distribution tables. The association between measured variables was tested with Spearman's rank order correlation test.

Multiple linear regression analysis of the plasma variables. Levels of plasma CC16, SP-A, and SP-B must be determined from both their rate of entry into and clearance from the circulation. CC16 is known to be freely eliminated by glomerular filtration, which, as we previously suggested, may also be a major route of systemic clearance of SP-A and SP-B. The GFR may in part be influenced by CO and mean systemic arterial BP, and reflected in the plasma creatinine concentration. Plasma CC16, SP-A, and SP-B are also derived from the lung, and depend in part on the alveolocapillary permeability. Arguably, this is best reflected by the LIS and is influenced by PCP (7). Additionally, there is an age and gender bias in the levels of many circulating antigens (10, 17). Therefore, we hypothesized that independent variables influencing plasma CC16, SP-A, and SP-B might include CO, BP, plasma creatinine, LIS, PCP, age, and gender.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Plasma Variables and Indices of Renal and Lung Function

CVVHD. Plasma creatinine was appreciably increased in the eight dialyzed patients (p = 0.038). In all other regards the dialyzed and nondialyzed patients were similar (Table 1).

CO and BP. Plasma creatinine and cystatin C concentrations were directly related to BP, and the concentration of CC16 was inversely related to CO (Table 2). The plasma concentraton of cystatin C was also inversely related to CO, but only when the dialyzed patients were excluded.

Plasma variables and glomerular filtration. Whereas plasma cystatin C (Figure 1) and CC16 concentrations were related to that of plasma creatinine, the concentrations of SP-A and SP-B were not. Neither concentrations of SP-A nor those of SP-B were related to concentrations of CC16 or cystatin C, although the concentration of SP-A was weakly related to that of SP-B (r_s = 0.288, p < 0.05, n = 37).

View larger version (10K): [in this window] [in a new window]   Figure 1.   Relationship between the plasma concentrations of creatinine and cystatin C in patients with acute respiratory failure, some of whom (open circles) had undergone at least 6 h of CVVHD for acute renal failure.

Plasma variables and lung function.

Blood oxygenation. Plasma concentrations of SP-B and CC16 were inversely related to Pa_O2/FI_O2 (Table 2). Similarly, concentrations of both proteins were directly related to blood oxygenation when corrected for the influence of Pa_CO2 and expressed as the ratio of the alveolar oxygen tension over the arterial oxygen tension (A/a). The plasma SP-A concentration was also inversely related to Pa_O2/FI_O2, and directly to A/a, but only when dialyzed patients were included.

LIS. Plasma concentrations of SP-B (Figure 2), and to a lesser extent those of SP-A, were related to LIS (Table 2). The relationships appeared to be exponentional, in accord with our previous findings in relation to Pa_O2/FI_O2 and static respiratory system compliance (6, 7), and with the weighting that these parameters have on LIS.

View larger version (9K): [in this window] [in a new window]   Figure 2.   Relationship between LIS and immunoreactive SP-B in patients with acute respiratory failure. Open circles indicate patients who had undergone at least 6 h of CVVHD for acute renal failure.

Arterial-Venous Concentration Difference

Concentrations of cystatin C, CC16, SP-A, and SP-B were all lower in mixed venous plasma samples than in matching arterial samples both with and without inclusion of the dialyzed patients (p < 0.001 in all cases). The A-V gradients were related to the arterial levels of all four proteins (Table 2). Neither plasma creatinine nor cystatin C concentrations were related to the absolute or relative A-V gradients of CC16, SP-A, or SP-B.

If it is assumed that the average blood volume is 5 L, with 85% of the blood residing in the systemic circulation and a hematocrit of 48%, then the mean transit time (MTT = 4.25/ CO × 60 s) between the aorta and the pulmonary artery is 35.7 s (17.2  72.9 s).

The plasma extraction ratio (ER) (i.e., the body less the lungs) is (Table 3):

(1)

the plasma clearance (Cl), defined as ER × plasma flow (or 0.48 × CO in the case being considered), is:

(2)

and the half-life of a protein circulating in the plasma (t_1/2), described by the distribution volume (V) and indirectly by Cl, is:

(3)

We have assumed that plasma constitutes 20% of extracellular fluid and that interstitial fluid constitutes the remaining 80%. The distribution volume is 2.4 L if the proteins partition themselves solely intravascularly and 12 L if the proteins partition themselves both intravascularly and across the extravascular space (Table 3). Matching pairs with A/V ratios  1 were not included in these estimates. Plasma clearance of CC16, SP-A, and SP-B was not related to clearance of either creatinine or cystatin C.

Multiple Linear Regression Analysis of Plasma Variables

Multiple linear regression analysis was done only on samples for which there was a complete data set. We hypothesized that arterial plasma CC16, SP-A, and SP-B concentrations would be influenced by a patient's CO, plasma creatinine, LIS, PCP, age, and gender (Table 4). With the exception of plasma creatinine, which was directly related to age (r_s = 0.366, p < 0.05), and CO, which was directly related to gender (r_s = 0.303, p < 0.05) when the dialysed patients were excluded (n = 25), none of the hypothesized determinants was related to each other. When the independent variables with the least significance (p > 0.05) were sequentially omitted, stepwise regression analysis indicated that with the dialyzed patients included:

CC16 = 42,301 + 12,547 × creatinine (p = 0.069) (R² = 0.092, p = 0.069)

SP-A = 22.66-34.87 × CO (p = 0.006) + 141.63 × LIS (p = 0.009) + 208.03 × gender (p = 0.005) (R² = 0.376, p = 0.002)

SP-B = 2,171.7 + 3,136.6 × LIS (p = 0.015) (R² = 0.157, p = 0.015)

and without the dialyzed patients:

CC16 = 44,306 + 11,847 × creatinine (p = 0.147) (R² = 0.076, p = 0.147)

SP-A = 61.63-31.41 × CO (p = 0.022) + 146.37 × LIS (p = 0.016) + 202.80 × gender (p = 0.015) (R² = 0.389, p < 0.007)

SP-B = 3,357.1 + 3,658.9 × LIS (p = 0.018) (R² = 0.189, p = 0.018)

Since the A-V gradient was related to the arterial level of protein, not surprisingly, the same independent variables were important in its determination (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Despite its vast surface area and thinness, the alveolocapillary membrane is normally extraordinarily effective in partitioning plasma and lung proteins. However, because the alveolar epithelium may be more permeable than previously thought (8), it is inevitable that proteins move down their concentration gradients across this barrier, particularly when permeability is compromised, as in ALI. We have confirmed that plasma concentrations of SP-A and SP-B are increased in ARDS patients and have further reported that concentrations of these proteins are also increased in patients with lower LIS. We now also report that plasma concentrations of SP-A and SP-B are directly related to LIS, in addition to reflecting blood oxygenation and static respiratory system compliance (6, 7). Although the plasma concentration of CC16 was related to blood oxygenation, it was not related to LIS. On the other hand, both plasma CC16 and cystatin C concentrations were related to that of creatinine, but plasma SP-A and SP-B concentrations were not. In most cases, concentrations of cystatin C, CC16, SP-A, and SP-B were reduced in mixed venous plasma. The A-V gradients cystatin C, CC16, SP-A, and SP-B were directly related to their arterial levels. Our study confirms that all three lung proteins are acutely cleared from the circulation of patients with acute respiratory failure, and we conclude that whereas the plasma concentrations of CC16 depends on GFR, the plasma concentrations of SP-A and SP-B reflect lung function independent of this variable.

Lung Dysfunction: Breaching the Pulmonary-Vascular Barrier

Since plasma proteins normally reach the alveolus as part of a bidirectional protein flux that restricts the passage of large molecules in a manner related logarithmically to their M_r (19- 21), we have previously argued that the relative amounts of CC16, SP-A, and SP-B that enter the circulation are a function of their molecular size and concentration in the alveolus. For a given LIS, the levels of circulating SP-A and SP-B were similar to those we have previously reported (6, 7). Regardless of patients' hemodialysis status, SP-B strongly correlated with the A/a ratio, Pa_O2/FI_O2, and LIS. On the other hand, SP-A levels were ~ 10-fold lower, and the correlations observed for SP-B, although evident, were weaker, in accord with SP-B being a more sensitive marker of lung damage (7). By way of contrast, although CC16 was weakly related to blood oxygenation in our patients, its concentration was slightly lower than normal (16, 22, 23) and was not related to LIS. CC16 was also weakly inversely related to CO. Whereas this relationship may be fortuitous, since CC16 is secreted into the airways rather than into the alveolus, it may only slowly enter the circulation, with the result that its circulating concentration also reflects the rate of perfusion. Because Clara cells are particularly sensitive to injury, the inflammation associated with ALI may not only increase lung permeability, but may also destroy these cells, hence decreasing the synthesis and secretion of CC16 (22). Since CC16 is a potent inhibitor of cytosolic phospholipase A₂ (23), a key enzyme in the production of lysophosphatidylcholine, prostaglandins, and leukotrienes, its decreased synthesis and secretion may exacerbate the ensuing lung injury.

Glomerular Filtration

The filtration of proteins across the renal glomerular capillary membrane occurs with an approximate log-log relationship to M_r (17, 24, 25). Although steric considerations also play a major role in such filtration, most proteins smaller than 50 kD normally pass relatively freely through the glomerulus, and virtually all (95% to 99%) are taken up and catabolized by fusion with lysosomes in the heterolysosomes of cells in the proximal tubules (24, 25). In blood, native SP-A (~ 650 kD) appears to complex with IgG and IgM (M_r < ~ 200 kD) (6, 26). Although such complexes would usually be too large to be filtered, increased glomerular permeability and glomerular proteinuria is a frequent complication of ARDS (27, 28). Since we have detected immunoreactive SP-A and SP-B in the urine of patients with ARDS (7), and since the concentration of CC16 is increased in predialysis serum from patients on maintenance hemodialysis (10), we anticipated that glomerular filtration may be a major route of systemic clearance for these proteins. However, only plasma CC16 was related to plasma creatinine and none were related to cystatin C. This does not appear to be a type II error, since multiple linear regression analysis suggests that GFR, as reflected by plasma creatinine, accounts for only ~ 10% of plasma CC16 clearance. In accord with this, clearance rates for cystatin C, CC16, SP-A, and SP-B were ~ 0.5 L/min, which is considerably greater than the normal GFR (~ 120 ml/min). Thus, whereas the kidney normally accounts for between 30% and 80% of the plasma clearance of low-molecular-weight proteins (24), this may be considerably altered in ALI. However, this is not to say that glomerular filtration does not contribute to plasma clearance in ALI.

CVVHD achieves a dialysis rate of ~ 20 ml/min, which is similar to the GFR of many of our nondialysed patients, and has an M_r cutoff of ~ 30 kD (29). Whereas hemofiltration may clear large amounts of mid-M_r molecules, partly because of the large volumes of ultrafiltrate produced, this is not the case with CVVHD. In contrast to both CVVHD and hemofiltration, maintenance hemodialysis uses bioincompatible membranes with low M_r cutoff values (~ 5 to 7 kD), which clear little protein, despite having high flow rates. In addition, given that renal clearance does not appear to have been a major pathway of excretion in our patients, it is not surprising that plasma creatinine, cystatin C, CC16, SP-A, and SP-B levels in patients requiring acute dialysis were compatible with those of nondialyzed patients. In accord with this, Mokrzycki and Kaplan (30) have shown that protein losses during CVVHD are negligible (i.e., < 20 mg/L of ultrafiltrate). The ultrafiltration rates in our patients varied between 850 and 1,100 ml/h. Whereas ultrafiltration membranes can bind protein, the Filtral AN 69 HF-type membrane does so only to a very small extent, and the electronegativity of the acrylonitrile/sodium methallyl sulfonate copolymer of the membrane repels anionic proteins at the membrane interface, in a manner analogous to the action of the sialoproteins of the glomerulus (personal communication, C. Hodgson, Gambro, Melbourne for Hospal Industrie, Meyzieu, France). Consequently, it is highly unlikely that nonspecific protein binding to the ultrafiltration membrane contributes significantly to the plasma clearance of CC16, SP-A, or SP-B. Clearly, all three lung proteins must also be cleared by other tissues. Consistent with this is the finding in numerous studies that in the absence of a normal GFR, creatinine and proteins otherwise cleared by the kidney are removed by unknown pathways that may partly or even totally compensate for renal protein clearance (10, 24). Consequently, although plasma cystatin C has been purported to be a more sensitive marker of renal function than creatinine (13, 14), it may be inappropriate in ALI patients.

Arterial-Mixed Venous Gradient: Rate of Systemic Clearance

Although the vascular endothelium generally restricts the movement of molecules larger than albumin (M_r = 67 kD, hydrodynamic radius ~ 3.5 nm), low-M_r proteins diffuse down their concentration gradients between vascular and extravascular compartments. Indeed, CC16 has been detected in pleural, ascites, synovial, cerebrospinal, and gastric fluids (15, 31, 32). In ascites and synovial fluid the concentration of CC16 is comparable with that in serum (15, 31). Although SP-A-IgG and SP-A-IgM complexes would normally be too large to cross the endothelium, there is compelling evidence that ALI is associated with a generalized inflammatory-mediated increase in vascular permeability, in addition to that in the lung (33, 34). Immunoreactive SP-B (~ 26 and ~ 42 kD), and possibly also immunoreactive SP-A, may also partition across systemic tissues. Therefore, the extravascular compartment for these proteins is potentially markedly greater than the vascular compartment. Tissue stores of SP-A and SP-B can presumably reenter the circulation when plasma levels of these proteins fall. In doing so, they complicate the proteins' plasma kinetics.

Cleared proteins may generally be excreted, metabolized, or catabolized to amino acids and recycled. Kreuzfelder and coworkers (27) have shown that ALI is commonly associated with mixed glomerular-tubular proteinuria resulting from glomerular endothelial and proximal tubular epithelial dysfunction. Similarly, Pallister and associates (28) have shown a ~ sevenfold increase in urine albumin excretion in ALI patients. Although we do not know the relative extent to which protein catabolism and metabolism are altered in ALI, Finn and coworkers (35) recently showed a daily protein loss of ~ 76 g in critically ill trauma and sepsis patients. Since the body is normally protein neutral with a daily intake of ~ 100 g, this daily protein loss represents nearly a twofold increase in total protein metabolism and/or excretion, despite the sedentary state of such patients.

We estimated the half-lives of the lung proteins in the circulation on the basis of their A-V gradients and a number of assumptions, including that of a steady state. Since we have no idea of the rate of flux of these proteins, or the size of other tissue pools, we estimated their half-lives on the basis of total extracellular fluid volume and intravascular fluid volume. Since the A-V gradient for all three lung proteins was directly related to their arterial concentrations, their kinetics are likely to be nonlinear and suggest that their plasma clearance is not saturated. Normally, the plasma half-lives for various small-M_r (< 4 kD) peptide hormones are between 2 and 10 min (36), whereas those for ₂-microglobulin (~ 12 kD) and leptin (~ 16 kD) have been estimated as ~ 125 min (37) and ~ 25 min (38, 39), respectively. Similarly, the half-life of chicken cystatin in the rat circulation has been estimated as ~ 75 min (40). Our estimates of the plasma half-lives of cystatin C, CC16, SP-A, and SP-B are considerably shorter than those based on similar sized proteins, in accord with the generally increased vascular permeability (27, 28, 33, 34) and metabolic state (35) of ALI patients. Indeed, the relative A-V difference, at least for plasma cystatin C and SP-B, tended to reflect LIS.

Multiple Linear Regression Analysis

We analyzed the arterial levels of the lung proteins by stepwise regression analysis, using age, gender, CO, BP, plasma creatinine, LIS, and PCP as independent variables. We found that GFR, as reflected by the plasma creatinine concentration, was the only major determinant of the plasma concentration of CC16. On the other hand, alveolocapillary permeability, as reflected by LIS, was the only major determinant of the plasma concentration of SP-B. Although alveolocapillary permeability was also a major determinant of the plasma concentration of SP-A, that of SP-A was also directly correlated with gender and inversely correlated with CO. Whereas immunoreactive SP-B (~ 26 and ~ 42 kD) is small and presumably readily crosses the alveolocapillary membrane, the size of SP-A (~ 650 kD) may restrict its movement down its concentration gradient, with the result that its content in blood is also determined by the rate of its perfusion.

In summary, we have shown that CC16, SP-A, and SP-B are rapidly cleared from the circulation of patients with acute respiratory failure. Our findings suggest that whereas the plasma concentration of CC16 is influenced by GFR, the plasma concentrations of SP-A and SP-B reflect LIS largely independently of this variable. Plasma levels of the lung proteins will depend on the rates of both their entry into and clearance from the circulation. Although we do not know the precise rates for either protein, the A-V gradients for SP-A and SP-B were directly related to their arterial levels, suggesting that temporal increases in their plasma levels are due to increases in alveolocapillary permeability that surpass any increases in plasma clearance.