INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The pathophysiological role of actin leaked from damaged cells is being increasingly scrutinized as a major contributor to the propagation of organ injury after a self-limited insult (1). Actin is present in high concentrations in most eukaryotic cells, constituting more than 10% of cytosolic protein. As a result of tissue injury, large quantities of actin are released locally (2) and may impede the downstream microcirculation, further damaging already compromised organs (6). Actin filaments may directly obstruct small vessels and initiate the clotting cascade (9, 10). Alternatively or additionally, actin monomers can bind DNase and inhibit digestion of any DNA simultaneously released from ruptured cells (11); as demonstrated in sputum from patients with cystic fibrosis, this process can result in a local increase in viscosity caused by uncut, entangled strands of DNA (11).

Plasma gelsolin and vitamin D-binding protein function as actin-scavenging proteins (1). After binding actin, these complexes are cleared from the circulation, thus protecting the host from further injury (15, 16). Gelsolin is a calcium-dependent actin-binding protein that severs actin filaments and caps their fast-growing "barbed" ends; it also nucleates and thereby sequesters actin monomers into short filaments (17, 18). Because gelsolin has a higher affinity for actin than does DNase, gelsolin promotes the enzymatic hydrolysis of DNA by DNase in the presence of actin (11).

Decreased levels of vitamin D-binding protein in shock-like states, presumably resulting from an inability of synthesis to keep pace with the accelerated clearance of actin complexes, have been associated with an increased mortality rate (3, 19). A similar relationship is assumed to exist between gelsolin levels and outcomes in critically ill patients (20), but data supporting such a correlation are less extensive. We measured serum levels of gelsolin throughout the hospitalization of a patient with complicated Plasmodium falciparum malaria and found that increasing gelsolin concentration heralded the onset of clinical recovery (8).

On the basis of these observations, we decided to investigate the effects of trauma on the levels of circulating gelsolin shortly after injury. More specifically, our study examined the prognostic value of plasma gelsolin levels at the time of hospital admission in predicting subsequent clinical events, such as the later development of the acute respiratory distress syndrome (ARDS).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection and Plasma Procurement

As part of a separate ongoing protocol to study the effects of major trauma on interleukin levels approved by our Institutional Review Board, blood samples were collected into ethylenediaminetetraacetic acid (EDTA)-containing tubes from 13 nonconsecutive patients admitted to the Trauma Intensive Care Unit (TICU) with an Acute Injury Score (AIS) of  3 for at least one anatomical region (22). All samples were drawn within 4 h of the estimated time of injury. Specimens were centrifuged at 1.2 × 10³ × g for 5 min at 4° C. The plasma was removed and diluted between 1:40 and 1:200 with gel sample buffer and frozen at 60° C. Similarly processed specimens were also obtained from 11 healthy volunteers. Three separate samples were obtained from patient 3 during the first hospital day.

Subsequently we collected admission blood samples from 52 additional patients presenting to the trauma admitting area. These patients were identified in the same way as the first group, or through the hematology laboratory in the hospital, and were classified into two groups on the basis of whether they were admitted to the Trauma Intensive Care Unit (TICU group) or not (non-TICU group). This approach enabled us to accrue patients with various types and severity of trauma and thereby study the differential utility of gelsolin measurements over a broad spectrum of traumatic injuries.

Measurement of Gelsolin Levels by Quantitative Western Blotting

Specimens were applied to 10% sodium dodecyl sulfate (SDS)-polyacrylamide minigels, transferred onto Immobilon-P membrane (Millipore, Bedford, MA), and stained with a monoclonal antibody to human plasma gelsolin (GS-2C4; Sigma, St. Louis, MO), followed by a peroxidase-conjugated monoclonal antibody against mouse IgG (Sigma). Enhanced chemiluminescence (Amersham, Arlington Heights, IL) was used to detect the gelsolin bands. Purified human plasma gelsolin (obtained from Cytoskeleton, Denver, CO) was diluted with gel sample buffer and loaded as standards in at least four lanes of each gel in a concentration range comparable to the gelsolin concentrations in the samples (determined by trial and error). The starting concentration of gelsolin, before it was mixed with gel sample buffer, was 1 mg/ml as determined by the Bradford assay (29), using bovine serum albumin to generate the standard curve (reagents from Bio-Rad, Hercules, CA). Rainbow-colored molecular weight markers (Amersham-Pharmacia Biotech, Piscataway, NJ), ranging from 14,000 to 220,000, were used to confirm the adequacy of the transfer from gel to blotting membrane.

The integrated intensity of each band on the Western blot, minus the adjacent background signal, was quantified by an imaging acquisition and analysis system (Ambis, San Diego, CA). Plots of band intensity versus known amounts of gelsolin were fitted to linear and logarithmic curves (Cricket Graph 1.3.2; Criket Software, Media, PA); the type of curve that fit the data points better (with a correlation coefficient  0.92) was then used to derive the gelsolin concentrations of the samples. A standard curve from a typical experiment is presented in Figure 1A.

View larger version (54K): [in this window] [in a new window]   View larger version (72K): [in this window] [in a new window]   View larger version (16K): [in this window] [in a new window]   Figure 1.   (A) A standard curve for determination of plasma gelsolin concentrations taken from a typical experiment using quantitative Western blotting. In the example shown here, human plasma gelsolin at a starting concentration of 1 mg/ml was diluted 1:800 with gel sample buffer. Volumes of 1, 2, 4, 6, 8, and 10 µl were loaded on a 10% SDS-polyacrylamide minigel as standards and transferred to blotting membrane. The "integrated intensity" of each band on the Western blot, minus the adjacent background signal, was plotted versus the known amounts of gelsolin and fitted to a straight line (r = 0.988). The graph was used as a standard curve to determine the unknown gelsolin levels in simultaneously assayed patient samples. (B) Plasma gelsolin levels were significantly lower in patients after major blunt trauma than in normal volunteers. The concentration of gelsolin in plasma samples taken from 13 patients shortly after major trauma and 11 healthy controls was measured by quantitative Western blotting. The graph plots the average gelsolin concentration from duplicate experiments for each sample. The horizontal red lines identify the mean gelsolin levels for the patient and control groups. (C ) There was a significant correlation between admission gelsolin and albumin levels. In an analysis of 49 patients (including the original 13 cases) for whom simultaneous albumin and gelsolin measurements were available, the albumin concentration directly correlated with the gelsolin level (p < 0.001). However, when compared with values in uninjured persons, the proportional and absolute reduction in gelsolin concentrations substantially exceeded the decrease in albumin levels. The blue circles and red crosses denote those patients who developed ARDS or expired, respectively.

Detection of Actin in Plasma

Western blotting, as described above, was performed on plasma samples from patients 1-8, using monoclonal antibodies against a synthetic actin C-terminal peptide (clone AC-40; Sigma), recognizing an epitope conserved in all actin isoforms. Known concentrations of pyrene-labeled rabbit skeletal muscle actin, determined from the absorbance in storage buffer (5 mM triethanolamine, 0.3 mM CaCl₂, 0.68 mM ATP, 0.1 mM EDTA, 0.02% NaN₃, pH 7.5) at a wavelength of 280 nm, were used to determine the sensitivity of the assay.

Actin was purified from acetone powder extracted from the skeletal muscle of New Zealand White rabbits, by the method of Spudich and Watt (30). Actin was labeled with N-(1-pyrene)iodoacetamide (Molecular Probes, Eugene, OR) as described by Murray and coworkers (31). The pyrenyl-G-actin was applied to a Sephacryl S-300 (Pharmacia, Piscataway, NJ) gel-filtration column and then maintained in storage buffer at 4° C.

Actin Elongation Assay

The elongation rates of 1 µM pyrenyl-G-actin in polymerization buffer (containing 25 mM Tris-HCl, 138 mM KCl, 2 mM MgCl₂, 0.2 mM CaCl₂, and 1 mM ATP, pH 7.4) in the presence and absence of plasma were determined from the increase in pyrenylactin fluorescence (excitation wavelength at 370 nm; emission wavelength at 410 nm) in a Perkin-Elmer (Norwalk, CT) LS5 fluorimeter. Fluorescence readings were zeroed by subtracting the fluorescence of pyrenyl-G-actin immediately on addition to the polymerization buffer.

In this assay, pyrenylactin fluorescence increases ~ 20-fold when pyrenyl-G-actin is incorporated into filaments (F-actin). Hence, the rate of actin polymerization is directly related to the increase in fluorescence over time. The lag phase preceding filament elongation reflects the rate-limiting nucleation step in actin polymerization. Spontaneous nucleation can be bypassed by the addition of a nucleating protein such as gelsolin; the initial elongation rate is then directly proportional to the concentration of nuclei. Because gelsolin is the principal actin-nucleating protein in plasma (17), the relative concentration of gelsolin can be ascertained by comparing the elongation rates (i.e., the slopes of fluorescence versus time) with plasma from different patients.

Definitions of Acute Injury Score, Injury Severity Score, Revised Injury Score, and Acute Respiratory Distress Syndrome

Patients were retrospectively scored as to the extent of their injuries at the time of admission using published criteria for the Acute Injury Score (AIS) (n = 65 patients), Injury Severity Score (ISS) (n = 11), and Revised Injury Score (RIS) (n = 11) (22). Patients were grouped for tabulation according to whether or not their AISs were  2 (moderate or less) versus  3 (serious or worse) in each of three body sites (head and neck; chest; abdomen and pelvis), as assessed by one of the investigators (K.C.M.). The formal diagnosis of ARDS was based on the caregiver's clinical assessment and subsequently confirmed by one of the investigators (K.C.M.) as fulfilling the criteria of the American-European consensus conference (32, 33). All of our patients with ARDS had otherwise unexplained bilateral alveolar infiltrates and Pa_O2/FI_O2 < 150.

Measurement of Tumor Necrosis Factor

Plasma concentrations of tumor necrosis factor  (TNF-) were measured in samples from the 11 controls and the first 13 patients by an enzyme-linked immunosorbent assay (ELISA) (Quantikine HS; R&D Systems, Minneapolis, MN). All specimens were run in duplicate. This quantitative sandwich ELISA using alkaline phosphatase in its amplification system was calibrated against Escherichia coli-expressed recombinant human TNF-.

Other Measurements

For the first 49 plasma specimens used for gelsolin determinations, albumin concentrations were measured by the clinical chemistry laboratory at Cooper Hospital (Camden, NJ). By the time albumin assays were performed, the samples had been thawed and refrozen multiple times. Measurements of hemoglobin concentration and prothrombin time obtained on the day of admission to our hospital were retrievable from the clinical hematology laboratory for most of the patients.

Statistical Methods

Statistical comparisons of mean gelsolin levels in different groups, using independent (unpaired) two-tailed Student t tests, were executed by Excel 5.0 software (Microsoft, Redmond, WA). The Fisher exact test (performed by Instat software [GraphPad, San Diego, CA]) was used to compare the frequency of complications in patients with normal and depressed gelsolin concentrations. Pearson correlation coefficients testing the degree of association between gelsolin and other clinical (ISS, RIS) and laboratory (hemoglobin, albumin, and TNF- levels) indices were also calculated by Instat.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Plasma Gelsolin Levels in Patients Shortly after Major Trauma

All 13 patients included in our original series were admitted to the intensive care unit of our level I Trauma Center shortly after experiencing serious trauma. By definition, these patients had an AIS  3 in at least one organ system. For the 11 patients for whom composite scores were available, the mean (median; range) ISS and RIS were 22 (20; 9-43) and 4.1 (4.1; 2.2-6.9), respectively (22). All patients received crystalloid volume expansion, and some were transfused. The plasma samples used for determination of gelsolin levels were obtained before or during the initial transfusions. The mean (± standard deviation) plasma gelsolin concentration in our 11 healthy volunteers was 517 (± 134) mg/L, compared with the trauma patients, whose levels near the time of their admissions averaged 261 (± 104) mg/L (p = 0.0001) (Figure 1B).

Hemoglobin concentrations in the samples from our first 13 patients did not correlate with the gelsolin concentrations; in fact, the two patients with the lowest gelsolin levels had normal hemoglobin values. This finding makes it unlikely that the reductions observed in serum gelsolin levels were due to fluid shifts/volume repletion with dilution of intravascular solutes. No patient had a recognized coagulopathy on presentation. However, the sickest patients in our first series had reduced plasma albumin levels. In a later analysis of 49 patients (including the original 13 cases) for whom simultaneous measurements were available, the albumin concentration directly correlated with the gelsolin level (p < 0.001) (Figure 1C). Although the significant correlation between albumin and gelsolin levels is consistent with nonspecific leakage of protein out of the intravascular space in severely traumatized persons, the proportional and absolute decline in gelsolin concentrations greatly exceeded the decrement in albumin levels. Finally, the two sickest patients in the initial group had relatively high plasma concentrations of TNF- (> 3 pg/ml) at the time of admission; nevertheless, we could not discern any consistent relationship between the levels of gelsolin and TNF- in our analysis of the original 13 patients.

Because we had collected three different plasma specimens from patient 3 during the first hospital day, we decided to investigate how quickly gelsolin levels changed as the patient was stabilized in the hospital. At least for this patient, plasma gelsolin concentrations (189  253  221 mg/L at 0, 9, and 15 h, respectively, after admission) did not change dramatically during the first hospital day. This single observation is consistent with our unpublished data on patients with serious acute medical illnesses, in whom depressed gelsolin levels rose slowly over days as the patients improved. Thus the exact timing of sample procurement may not be critical.

Plasma Actin Levels in Patients Shortly after Major Trauma

Because actin in patient plasma has been detected in complexes with gelsolin and vitamin D-binding protein (2, 7, 21), we examined plasma from our first eight patients to see if the actin concentration and/or the molar ratio of actin to gelsolin would provide additional prognostic information. By comparison to serial twofold dilutions of purified rabbit skeletal muscle actin of known concentration, we estimated the sensitivity of our Western blot assay to be less than 2 ng of actin. Actin was not detected in any of our patient samples. After adjustment for the 8-µl aliquots of patient plasma at a 1:40 dilution applied to this gel, we calculated that the concentration of actin in all patient samples was less than 8 mg/L (< 0.2 µM). Accordingly, because every patient had a gelsolin level between 1 and 4 µM, only a small fraction, if any, of the circulating gelsolin in our patients could have been complexed to actin.

Relationship between Admission Gelsolin Levels and Hospital Course

Figure 2A demonstrates that each of the four patients with admission gelsolin levels > 2 standard deviations (SD) below the mean value for healthy volunteers required prolonged mechanical ventilation in the TICU for  13 d. In contrast, five of the surviving six patients whose gelsolin concentrations were low but still < 2 SD below the normal range remained in the TICU less than 1 wk. The relative risk of prolonged mechanical ventilation in the TICU for patients with markedly depressed admission gelsolin levels was 3.5 (95% confidence interval, 1.1-11.3; one-tailed p = 0.0455 by Fisher exact test). Neither these aspects of the patients' hospital courses (Figure 2B) nor their admission gelsolin levels (Figure 2C) correlated with the ISS (or RIS).

View larger version (16K): [in this window] [in a new window]   View larger version (13K): [in this window] [in a new window]   View larger version (11K): [in this window] [in a new window]   View larger version (16K): [in this window] [in a new window]   Figure 2.   (A) Both the duration of mechanical ventilation and time spent in the Trauma Intensive Care Unit (TICU) were inversely related to the plasma gelsolin level obtained shortly after presentation. All patients were mechanically ventilated; days intubated denotes the time from admission until mechanical ventilation was discontinued. The vertical red line identifies the concentration of gelsolin exactly 2 standard deviations (SD) below the mean value of the control group. The solid red diamonds show the duration of mechanical ventilation, while the open blue circles represent the length of stay in the TICU. The arrowheads point to the three patients who expired within 30 d of admission; a fourth patient (not denoted in the figure by an arrowhead ) died months after his accident from unrelated causes. (B) The Injury Severity Score on admission was not closely related to the duration of mechanical ventilation or the time spent in the Trauma Intensive Care Unit (TICU). The Injury Severity Score at the time of admission did not predict the subsequent hospital course in the 11 patients in whom it was ascertained. There was likewise no association between gelsolin levels and the Revised Trauma Score in this small group of patients. (C ) The Injury Severity Score did not correlate with gelsolin levels at the time of admission. No consistent relationship between the Injury Severity Score and gelsolin levels at the time of admission was discernible in the 11 patients for whom both indices were available. (D) The length of hospital stay was inversely related to the plasma gelsolin concentration at the time of admission. The length of hospitalization for all four patients with gelsolin levels < 250 mg/L (> 2 SD below the control mean) approximated two or more weeks. The vertical red line identifies the concentration of gelsolin exactly 2 SD below the mean value of the control group. The blue circles and red crosses specify patients who developed ARDS or died, respectively. The sole patient whose hospital stay exceeded 7 wk suffered from multiple medical problems that predated his accident.

Only 2 (patients 1 and 2) of our original 13 patients were judged to have developed ARDS based on widely accepted published criteria (32, 33). Their gelsolin levels were substantially and significantly lower than those of the other 11 patients (69 versus 296 mg/L, p = 0.0011). Although three patients died early during their hospitalizations, only the death of patient 1 was directly attributable to multisystem organ dysfunction resulting from his injuries; an additional patient with multiple comorbid medical problems (patient 9) died months after his accident.

Figure 2D reinforces our earlier observations by demonstrating that the length of stay in the hospital was also inversely correlated with plasma gelsolin concentrations on admission. Patient 9, whose hospital course extended beyond 7 wk, had multiple medical problems, including chronic renal failure necessitating dialysis, predating his accident.

Relative Plasma Gelsolin Levels Can Be Quickly Measured by a Functional Assay

Gelsolin is the primary actin-nucleating protein in plasma (17). This property can be exploited to assess the relative gelsolin concentration in different plasma samples (10, 20), as illustrated in Figure 3A. Under control conditions, without plasma, polymerization occurred slowly because nucleation is the rate-limiting step. The addition of equal volumes of plasma from the sickest patient (patient 1), a patient with a relatively uncomplicated course (patient 8), and a healthy volunteer (E) resulted in linear elongation rates without a lag phase. The slopes of these lines (indicating the number of nuclei under each condition) were roughly proportional to the gelsolin levels observed with quantitative Western blotting. For illustrative purposes, a blot of serial twofold dilutions of plasma (previously diluted 1:50) is presented in Figure 3B for the three specimens used in the functional assay; actual sample volumes applied to the gel were 8, 4, 2, and 1 µl.

View larger version (20K): [in this window] [in a new window]   View larger version (46K): [in this window] [in a new window]   Figure 3.   (A) The initial elongation rate of pyrenylactin, reflecting the number of gelsolin nuclei, was proportional to the plasma gelsolin concentration measured by quantitative Western blotting. Pyrenylactin fluorescence increases dramatically as pyrene-labeled monomeric actin is incorporated into filaments, and thus the increase in fluorescence with time can be used to determine the rate of actin polymerization. Gelsolin (supplied from the added plasma in this experiment) nucleates actin polymerization; in the presence of plasma, the initial elongation rate is a direct function of the plasma gelsolin concentration. The black squares depict the polymerization rate in the absence of added plasma; the red, magenta, and blue diamonds show the rate of polymerization in the presence of plasma from patient #1, patient #8, and control E. The slopes of the lines for each sample, shown at right, were roughly proportional to the corresponding gelsolin levels measured by quantitative Western blotting, given in brackets subjacent to each line, on the right. (B) A Western blot shows the relative gelsolin levels of the samples used in the elongation assay. Serial twofold dilutions of plasma previously diluted 1:50 with gel sample buffer were probed with monoclonal antibodies against human plasma gelsolin. The actual sample volumes loaded on the gel were 8, 4, 2, and 1 µl for each specimen, as marked above the lanes. The same three specimens (#1, #8, E) used for the functional assay presented in (A) are shown on the blot. Patient #1 was the sickest patient; he developed ARDS and subsequently died. Patient #8 was much less ill. Visual inspection reveals that the gelsolin concentration in control E was approximately twice that of patient #8, which was in turn much higher than the level of patient #1. These relative estimations of gelsolin concentrations were consistent with the ratio of the slopes (i.e., elongation rates) for the same three samples depicted in (A).

Subsequent Analysis of 65 Patients after Trauma of Varying Severity

After analysis of the initial 13 patients, described in detail above, we studied another 52 patients experiencing a wide range of traumatic injuries of various degrees of severity (Table 1 groups all 65 patients by whether they were admitted to the Trauma Intensive Care Unit). The less seriously ill patients in the non-TICU group provided another control group similar in demographics to the sicker TICU patients. ARDS supervened in 8 of 10 patients who experienced this complication within 3 d of admission and was not related to nosocomial complications in any patient. The incidence of ARDS was 32% (6 of 19) for patients experiencing major chest trauma (AIS  3), compared with 9% (4 of 46) for those without major chest trauma (AIS  2) (p = 0.0522 by the Fisher exact test).

Figure 4 summarizes the relationship of gelsolin levels at the time of hospital admission to "bad" clinical outcomes for all our patients. As noted, these complications often developed days later. The mean gelsolin levels on presentation were 193 ± 150 versus 400 ± 140 mg/L for patients with and without ARDS, respectively (p < 0.0001). Six of the 10 patients with ARDS had admission gelsolin levels < 250 mg/L (> 2 SD below the mean of our control group). Gelsolin levels at presentation in every patient who developed ARDS were < 380 mg/L (> 1 SD below the control mean). Conversely, 6 of 13 patients (46%) with admission gelsolin levels less than 250 mg/L developed ARDS, compared with none of the 20 patients whose levels fell within the normal range (> 380 mg/L) (p = 0.0015 by the Fisher exact test). Including all 65 patients, 6 of the 10 patients who developed ARDS had admission gelsolin levels less than 250 mg/L, compared with only 7 of the 55 patients without ARDS (p = 0.0028 by the Fisher exact test).

View larger version (17K): [in this window] [in a new window]   Figure 4.   Plasma gelsolin levels in all 65 patients obtained shortly after traumatic injuries of various types and severity. Concentrations of gelsolin were determined by quantitative Western blotting in plasma samples taken from 65 patients shortly after traumatic injuries of various types and severity. Blue circles and red crosses identify patients who subsequently developed ARDS or expired, respectively. The horizontal solid black line identifies the mean gelsolin level for our control group; the horizontal dotted blue and red lines represent the gelsolin level 1 and 2 SD, respectively, below the control mean.

Plasma gelsolin levels at presentation in survivors versus nonsurvivors were 398 ± 99 and 259 ± 95 mg/L, respectively (p < 0.0001). The corresponding mortality rates for patients with gelsolin levels < 250 and > 380 mg of gelsolin per liter were 46 and 10%, respectively. The 32 patients with intermediate gelsolin levels experienced a mortality rate of 16%; 4 of the 32 patients (13%) in this stratum developed ARDS.

In total, 10 of the 13 patients (77%) with gelsolin levels at the time of admission more than 2 SD below the control mean had "bad" outcomes," defined as mechanical ventilation for  13 days in the Trauma Intensive Care Unit, ARDS, and/or death.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Circulating actin leaked from injured cells may be directly toxic to pulmonary endothelial cells and obstruct the microcirculation of the lungs. Oleic acid-induced lung injury in rats has been associated with detectable actin in the blood, mostly complexed to vitamin D-binding (Gc) protein and gelsolin (4). Both the total and free levels of these actin-scavenging proteins were markedly reduced after acute lung injury, presumably owing to the more rapid clearance of the complexes (15). Rapid infusions of exogenous G-actin into Wistar rats likewise reduced levels of Gc protein and caused pulmonary endothelial and vascular injury (6). Platelet thrombi enmeshed in a dense network of F-actin bundles were observed in the arterial and capillary circulation of the lungs. These angiopathic changes could be averted if G-actin was first incubated with Gc protein. In patients with ARDS plasma gelsolin concentrations were lower (and gelsolin-actin complexes sometimes detectable) than in healthy adults and patients with uncomplicated bacterial pneumonia (7). We demonstrated a temporal correlation between the evolution of multiorgan failure and gelsolin levels in an individual patient with complicated malaria (8).

The present study focused on the clinical implications of plasma gelsolin levels obtained at the time of admission to a trauma intensive care unit. We demonstrated that, although almost all patients have decreased gelsolin levels after major blunt trauma (34), the lowest levels were associated with more complicated and prolonged hospital courses. In our original small series, the predictive value of a low admission gelsolin concentration (> 2 SD below the normal mean) for prolonged mechanical ventilation ( 2 wk) was 100% (four of four patients). Relatively preserved gelsolin levels did not guarantee an uneventful course, although five of the six survivors with nearly normal gelsolin levels were intubated for less than 1 wk. Four of our 13 patients expired, but only one death (patient 1, with the lowest gelsolin level) could be attributed to multiorgan failure directly resulting from traumatic injuries.

Subsequent observations of 65 patients with traumatic injuries of different types and intensities confirmed the strong correlation between plasma gelsolin levels at presentation and such important clinical complications as the development of ARDS and survival. Still unsettled issues concern the particular or incremental utility of gelsolin levels as timely prognostic indicators, the generalization of these findings to diverse clinical situations (34), and the biological consequences resulting from hypogelsolinemia.

Concurrently with our investigation of trauma patients, we measured gelsolin levels in patients who developed ARDS subsequent to pneumococcal pneumonia (n = 2), pancreatitis (n = 1), and candidemia (n = 1); in all four cases, gelsolin levels nadired below 100 mg/L early in the clinical course. These data suggest that low gelsolin levels may be a sensitive correlate of ARDS in many different settings. In a blinded survey of 16 patients undergoing allogeneic bone marrow transplantation (of whom 8 died with acute respiratory failure and 8 had uneventful posttransplantation courses), we correctly identified 7 of 8 in each group solely on the basis of the presence or absence of any gelsolin level < 100 mg/L during the first month posttransplantation (35). Gelsolin levels may be a discriminative index in this context. Admittedly, one cannot ascertain from the small numbers of patients with ARDS in our studies whether gelsolin levels provide an earlier or more reliable marker of the development of ARDS than do conventional parameters; on the other hand, there are biologically plausible reasons to implicate gelsolin as a key physiological mediator in the evolution of ARDS.

Do extracellular actin and plasma gelsolin play adversarial roles in the pathogenesis of multiorgan dysfunction that often follows severe traumatic injuries? More generally, is plasma gelsolin a critical factor in preventing further secondary organ damage nonspecifically elicited by a variety of insults? Unraveling the role of plasma actin-binding proteins in limiting host injury has both therapeutic and prognostic implications (36- 39). For example, local applications or systemic infusions of recombinant human gelsolin (40) might avert ARDS after trauma or sepsis if administered during the "window of opportunity." Of course, fluctuations in gelsolin levels may not be causally related to the status of a patient. However, even if simply an epiphenomenon with regard to pathogenesis, plasma gelsolin levels shortly after severe trauma may have prognostic significance beyond standard parameters that could impact on the degree, duration, and cost of intensive care.