Type III Procollagen Peptide in the Adult Respiratory Distress Syndrome: Association of Increased Peptide Levels in Bronchoalveolar Lavage Fluid with Increased Risk for Death

Methods

Patients

All patients between the ages of 18 and 72 years who were admitted to intensive care units at Harborview Medical Center [Seattle, Washington] between September 1986 and April 1991 were screened prospectively for the onset of the adult respiratory distress syndrome. Patients were screened using the following criteria: 1) for critical hypoxemia with cutoff PaO₂/FIO₂ ratios of 150 mm Hg or less or of 200 mm Hg or less using 5 cm H₂O or more of positive end-expiratory pressure; 2) for diffuse parenchymal infiltrates involving at least three quadrants on chest radiographs; 3) for pulmonary artery wedge pressure [when available] of 18 mm Hg or less or no clinical evidence of congestive heart failure; and 4) no other obvious explanation for these findings [17, 18]. Because of the possible risk for complications related to bronchoalveolar lavage, patients with the adult respiratory distress syndrome were excluded if they met any of the following criteria: 1) PaO₂ less than 80 mm Hg with FIO₂ of 1.0; 2) evidence of acute ischemic heart disease; 3) severe hypotension (systolic blood pressure < 90 mm Hg); 4) cardiac dysrhythmias (heart rate > 140 beats/min or complex ventricular ectopy); 5) sustained increased intracranial pressure greater than 20 mm Hg; and 6) endotracheal tube internal diameter less than 7.0 mm. Patients were not excluded because of high minute ventilation, high levels of positive end-expiratory pressure, or presence of barotrauma. Informed consent was obtained from either the patient or the legal surrogate. The study was approved by the University of Washington Human Review Committee.

Before bronchoalveolar lavage was done, the following clinical data were obtained: levels of FIO₂ and PaO₂, static compliance, and level of positive end-expiratory pressure. These data were used to calculate a modified lung injury score for the adult respiratory distress syndrome, as described by Murray and colleagues [19], except that a chest radiograph score was not included. However, all patients had alveolar infiltrates in three or four quadrants. Thus, the Murray acute lung injury score would be 0.75 to 1.0 points greater than our modified score. Patients in our study with lung injury scores of 1.75 or more met Murray criteria for severe lung injury.

Risk factors associated with development of the adult respiratory distress syndrome were defined as previously described [17]. These included the sepsis syndrome, trauma, aspiration of gastric contents, drug overdose, and multiple transfusions. Trauma risk was defined as the presence of multiple long bone or pelvic fractures, pulmonary contusion, or trauma-associated multiple transfusions (≥ 15 units in 24 hours of emergency resuscitation) [17]. For this analysis, we combined clinical risks for aspiration of gastric contents, drug overdose, and multiple transfusions without trauma into the category “other risks.” Risk factors were identified prospectively when the patient was entered into the study.

The first 18 patients in the study had a single bronchoalveolar lavage. Subsequently, we attempted to do lavage serially at days 3, 7, and 14 after the onset of the adult respiratory distress syndrome unless the patient died, was extubated, or became too unstable to tolerate a lavage, as indicated by the criteria above. Patients were followed until death or hospital discharge. Survival was defined as discharge from hospital. Organ failure and cause of death, as defined by Montgomery and colleagues [18], were analyzed in a subset of patients enrolled during 1990.

Bronchoalveolar Lavage and Analyses

All patients were intubated at the time of lavage. They were ventilated with FIO₂ levels of 1.0 for 10 to 15 minutes before and during the procedure. An adaptor was placed on the patient's endotracheal tube, and a fiberoptic bronchoscope was passed through the endotracheal tube into the lower airway and was wedged into a subsegment of either the right middle lobe or lingula. Five 30-mL aliquots of sterile pyrogen-free 0.9% NaCl at room temperature were instilled (150 mL total) and recovered by gentle suction. Six normal volunteers had lavage using a similar technique. Serum samples also were obtained from patients at the time of lavage and were stored at −70°C.

Lavage samples were transported immediately to the laboratory for analysis. The fluid was filtered through gauze moistened with 0.9% NaCl, and the total recovered volume was recorded. Total cell counts were done in a hemacytometer, and differential cell counts were done on cytospin preparations stained with Diff-Quick (American Scientific Products, McGaw Park, Illinois). Cell viability was measured by trypan blue exclusion. After an aliquot was taken for cellular analysis, the lavage fluid was centrifuged at 200 g for 15 minutes to pellet the cells. Aliquots of the supernatant lavage fluid were put into polypropylene tubes and stored at −70°C. Total protein was measured on an aliquot of the supernatant using either the modified Lowry method [20] or the bicinchoninic acid method [21].

Type III Procollagen Peptide Analysis

The concentration of procollagen III in bronchoalveolar lavage fluid specimens or serum was determined by radioimmunoassay (RIAgnost PAP, Behringwerke, Marburg, Germany) according to the manufacturer's instructions using 20 µL of lavage fluid or serum. The radioimmunoassay for procollagen III was linear over a range of 0.4 to 9.5 U/mL. Serum control samples provided by the manufacturer contained 1.6 to 1.7 U/mL. Samples in which the procollagen III concentration was greater than the standard detection range were diluted 1:4 in 0.9% NaCl. Samples in which procollagen III concentrations were less than the detection range were assigned a value of 0.4 U/mL for subsequent data analysis.

To determine the variability of repeat determinations, five lavage samples were analyzed in triplicate. The variability of triplicate determinations of procollagen III concentration measured in five different lavage samples was 5% or less. In one analysis, an aliquot of the lavage sample was subjected to a total of three freeze-thaw cycles to determine stability of the peptide under these conditions. No increase in the variability of measured values was detected.

Cross-sectional and Serial Analyses

Lavage fluid procollagen III concentrations in survivors and nonsurvivors initially were compared using the nonparametric Wilcoxon rank-sum test [22]. We then calculated the relative risk (RR) for fatality in patients with a lavage fluid procollagen III level of 1.75 U/mL or more compared with those with values less than 1.75 U/mL; 95% CIs were determined using the method of Rothman [23]. We also did a stratified analysis to determine if the relation between procollagen III concentration and fatality differed by risk factor for the adult respiratory distress syndrome (sepsis, trauma, other) or degree of lung injury, using the lung injury score. The difference in RRs for death among risk groups for the adult respiratory distress syndrome and lung injury severity groups was assessed by the Breslow-Day test of homogeneity [24].

We did logistic regression analyses of data obtained on days 3, 7, and 14 after onset of the adult respiratory distress syndrome to measure the effect of procollagen III levels on the risk for death while controlling for the effect of demographic and physiologic variables that, independent of procollagen III, might be associated with fatality and increased levels of procollagen III. Variables included in the multivariate model were age; sex; risk group for the adult respiratory distress syndrome; lung injury score; and lavage fluid neutrophil, macrophage, and total protein concentrations. Sex (using men as the reference group) and risk group for the adult respiratory distress syndrome (using sepsis as the reference group) were analyzed as categorical variables. For analysis of lavage fluid cell and protein concentrations, logarithmic transformations (log₁₀) were done. Variables were retained in the final model if an improvement in fit was noted, as gauged by the deviance and likelihood ratio statistic [25].

Because the outcome of interest is frequent (41% mortality rate in the study population), the estimate of the RR for dying that is generated by the logistic regression (odds ratio) is inflated. Therefore, we used coefficients derived from the regression analysis to calculate the probability and RR for dying. The probability was calculated using mean values for the study population [25].

We analyzed serial lavage data by creating two summary measures for each patient: the maximum procollagen III levels attained and the rate of change of procollagen III (the slope). Survivors and nonsurvivors were compared using the Wilcoxon test.

Results

Patients

Lavage fluid samples were available from 117 of 337 (35%) patients prospectively identified as having the adult respiratory distress syndrome between September 1986 and April 1991. Patients enrolled in this study Table 1 were similar to the 221 persons who were not enrolled according to risk group and sex distribution, but they were slightly younger (mean age, 41 years compared with 43 years; P < 0.3) and had a lower fatality rate (41% compared with 57%; P < 0.001); in trauma patients, injury severity scores [26] were nearly identical (mean score, 28). Reasons for nonenrollment in the study were examined in detail for January 1988 through April 1991 [27]. During this period, 107 of our 117 patients had lavage, and 164 additional patients were identified as having the adult respiratory distress syndrome but were not enrolled. Reasons for exclusion were medical instability, as described above (n = 33); death before day 3 (n = 26); extubation before day 3 (n = 10); consent denied (n = 36); missed inclusion, usually because the investigator was not available (n = 27); and other reasons, such as enrollment of the patient in another study (n = 32).

Sepsis and major trauma were risk factors for the adult respiratory distress syndrome in most of our study patients (n = 72). Other risk factors (gastric aspiration, drug overdose, and massive transfusion) were present in 45 patients. Impairment of gas exchange and lung mechanics was severe as indicated by PaO₂/FIO₂ ratios and modified lung injury scores. Modified lung injury scores measured 3 days after onset of the adult respiratory distress syndrome were similar in survivors and nonsurvivors but were higher in nonsurvivors at days 7 and 14 after onset of the syndrome (day 7, P < 0.01; day 14, P = 0.08). Lavage fluid neutrophil and protein concentrations were also higher in nonsurvivors at day 3.

A total of 196 bronchoalveolar lavage procedures were done in 117 patients. Eighty-three lavages were done 3 days after onset of the adult respiratory distress syndrome, 73 lavages were done 7 days after onset, and 40 lavages were done 14 days after onset. Fifty-six patients had one lavage procedure, 43 had two lavage procedures, and 18 had three serial lavage procedures on days 3, 7, and 14 after onset of the adult respiratory distress syndrome. Of 83 patients who had lavage on day 3, 43 patients also had lavage on day 7. Of 40 patients who dropped out, 2 died, 15 were extubated, 11 were “missed” because serial lavages were not initially planned, and the rest were missed for other reasons. On day 7, the 73 patients who had lavage included 30 new patients who met eligibility and safety criteria.

Univariate Analysis

At all times, the median procollagen III level was higher in patients who died (Figure 1). The median procollagen III level at day 3 was 3.8-fold higher in patients who died than in those who survived (P < 0.001). At 7 days, median values were 3.2-fold higher (P = 0.002); at 14 days, values were 2.5-fold higher (P = 0.09). The median procollagen III value for the entire population of patients (survivors and nonsurvivors) was 1.75 U/mL. Lavage samples from six normal volunteers contained undetectable levels of procollagen III.

The unadjusted RR for death was approximately 2.5 times greater in patients with lavage fluid procollagen III levels of 1.75 U/mL or more compared with those with levels less than 1.75 U/mL (Figure 2). To estimate the diagnostic accuracy of procollagen III levels, we analyzed the sensitivity and false-positive rate using different threshold values. On day 7, cutoff values of 1.75 to 2.75 U/mL had a sensitivity of 70% to 80% and a false-positive rate of approximately 40%. On days 3 and 14, the sensitivity of these procollagen III levels was decreased (38% to 62%), but the false-positive rate was decreased (17% to 29%).

When patients were analyzed according to risk group for the adult respiratory distress syndrome, increases in the risk for death associated with higher procollagen III levels were seen in all groups (Table 2). Consistent with higher fatality rates in patients with sepsis, procollagen III levels tended to be higher in this group. We also stratified patient populations on each day by lung injury score (<2 compared with ≥ 2). On all days after the onset of the adult respiratory distress syndrome, the RR for death was higher in patients with increased procollagen III levels regardless of the modified lung injury score (Figure 3). The combined effect of increased lung injury scores and levels of lavage procollagen III was especially apparent on day 7. A fatality rate of 73% was observed when both risk factors were present compared with a 33% fatality rate associated with increased levels of procollagen III alone or a 31% fatality rate with increased lung injury scores alone. The fatality rate when neither risk factor was present was 15%. The difference in RR between the high and low lung injury score groups was, however, not statistically significant on any of the days.

Multivariate Analysis

We did a multivariate analysis that included variables that might be associated independently with increased risk for death or might account independently for the increased lavage procollagen III concentration. In a preliminary analysis, we did not detect a clear linear relation between increasing lung injury score or procollagen III concentration and fatality. Therefore, these variables were treated categorically (lung injury score ≥ 2 compared with <2 and procollagen III levels ≥ 1.75 U/mL compared with <1.75 U/mL). Further categorization of these variables (for example, procollagen III levels <1.75, 1.76 to 2.50, and >2.5) did not improve the model.

Three days after onset of the adult respiratory distress syndrome, addition of other variables to the model only minimally decreased the effect of increased procollagen III concentration, as indicated by the small decreases in the RR associated with increased procollagen III (Table 3). Increased lavage fluid protein also was independently associated with increased risk for death at day 3. Similar patterns were seen in analyses for days 7 and 14. As expected, an increased risk for death was noted in patients with sepsis compared with patients with trauma or other causes for the adult respiratory distress syndrome on day 3 or day 7 [17]. Seven days after onset of the adult respiratory distress syndrome, lung injury scores of 2 or more were associated with increased risk for death (P = 0.07). The risk for death associated with increased procollagen III levels was still increased on day 14 but was decreased in patients with higher lavage fluid macrophage concentrations (P = 0.03).

Serial Analysis

We analyzed the rate of change in procollagen III levels (the slope) and the maximum level of procollagen attained in patients who had two or three lavages (n = 61). In survivors, procollagen III levels tended to remain constant (median slope, 0 U/mL per day), whereas in nonsurvivors, the slope increased (median slope, 0.1 U/mL per day; P = 0.29). The maximum procollagen III level attained was higher in patients who died (median level, 6.2 U/mL compared with 2.0 U/mL; P < 0.001).

Forty-three patients had lavage on days 3 and 7 after onset of the adult respiratory distress syndrome. In patients with procollagen III levels less than 1.75 U/mL on both days, the fatality rate was 20% (3 of 15), whereas in patients with values of 1.75 U/mL or more on both days, the fatality rate was 72% (13 of 18) (RR, 3.6; 95% CI, 2.1 to 6.2). In patients whose procollagen III level went from high to low, the fatality rate was 25% (1 of 4). In patients whose values went from low to high, the fatality rate was 50% (3 of 6).

Relation between Serum and Lavage Fluid Procollagen III Levels

To determine whether increased serum procollagen III levels might be reflected in bronchoalveolar lavage specimens, we compared procollagen III concentrations in serum and lavage samples concurrently collected at 7 days after onset of the adult respiratory distress syndrome. We detected only a weak correlation (r = 0.22; P = 0.15; Spearman rank-order test) between lavage fluid and serum levels. Serum values (median) tended to be lower in survivors than in nonsurvivors (2.55 U/mL compared with 3.75 U/mL; P =0.08).

Relation of Lavage Fluid Procollagen III Concentration to Organ Failure and Physiologic Status

To determine if the correlation of increased procollagen III levels with fatality could be the consequence of on-going respiratory failure that contributed to death, we reviewed in detail clinical data from all study patients enrolled during 1990 who had lavage on day 7. This subset included 25 patients; 10 died. Respiratory failure resolved in all survivors. Death was caused by sepsis in 3 patients, multiple organ dysfunction in 2, brain death in 2, hepatic failure in 1, respiratory failure in 1, and withdrawal of support in 1. However, 9 of 10 patients who died had respiratory failure at the time of death. One patient died because of central nervous system problems weeks after resolution of a severe case of the adult respiratory distress syndrome.

To determine if procollagen III levels correlated with physiologic status, we examined the relation between procollagen III levels and the degree of physiologic lung impairment as measured by the modified lung injury score. Positive correlations were observed on day 7 (r = 0.35, P = 0.002; n = 72) and day 14 (r = 0.50, P = 0.001; n = 32).

Discussion

Previous studies have suggested that lavage fluid or serum procollagen III levels may be useful in assessing disease activity or prognosis in fibrotic lung disease and the adult respiratory distress syndrome. A correlation between lavage fluid procollagen III concentrations and clinical indices of severity was reported recently in patients with idiopathic pulmonary fibrosis [10], who as a group had substantially greater procollagen III concentrations than those in a control group. Serum procollagen III concentration also correlated with clinical severity indices even though serum levels for the patient group were not statistically different from those for the control group.

Serum procollagen III levels in patients with sarcoidosis have been shown to be increased and correlated with disease progression as measured by radiographic stage [7]. Increased serum procollagen III concentrations were also reported [11] in eight patients with the adult respiratory distress syndrome and other patients requiring mechanical ventilation for other nonrespiratory disorders. Serial analysis of serum procollagen III concentrations in a large cohort of patients with severe trauma showed statistically significant higher values in nonsurvivors than in survivors [12]. The highest values were observed in patients with the adult respiratory distress syndrome, but increased bilirubin and creatinine levels also explained much of the variability in serum procollagen III concentration. Injury severity score [26] was also correlated with serum procollagen III concentration. These investigators concluded that the serum procollagen III concentration cannot be used as a specific marker for the adult respiratory distress syndrome or lung fibrosis in patients with trauma. Our data support this conclusion because no correlation could be detected between lavage fluid and serum procollagen III levels in our patients, and we suspect that impaired hepatic and renal clearance as well as wide-spread tissue damage and inflammation would tend to obscure the biological significance of serum procollagen III levels in patients with the adult respiratory distress syndrome in whom multiple organ dysfunction is frequently present.

The results of our study show that lavage fluid procollagen III levels were frequently increased in a large cohort of patients with the adult respiratory distress syndrome and extend previous observations by showing a strong association between increased lavage fluid procollagen III concentration and fatal outcome. The increased risk for fatal outcome associated with high procollagen III levels was independent of disease severity as assessed by physiologic dysfunction and abnormal gas exchange. Moreover, the risk associated with procollagen III levels was independent of other putative lavage fluid markers of disease activity, such as neutrophil or alveolar macrophage concentrations and total protein concentration, for predicting death in the adult respiratory distress syndrome. As early as 3 days after onset of the adult respiratory distress syndrome, increased lavage fluid procollagen III levels were associated independently with approximately a 2.5-fold increased risk for death; in a similar manner, increased risk was associated with increased procollagen III values at later times.

Our analysis of serial data suggests that repeating the measurements may add additional prognostic information. The utility of the prognostic information provided by procollagen III concentration for clinical decision making remains to be shown, but the ability to identify patients at exceedingly high risk (for example, those with 70% mortality rates as shown in Figure 3) clearly would be useful in planning future intervention trials. Lavage fluid procollagen III levels also might be used in assessing the equality of patients in intervention trials, and observational studies might be improved by using procollagen III concentrations to control for severity. Although the measurement of procollagen III levels in bronchoalveolar lavage samples was easily and reproducibly done using a commercial assay kit, most of the experience with this and other assays has been in measurements of blood levels. Therefore, standardization of lavage procedures is necessary, and further assessment of measurements using other similar assays is needed for potential clinical applications. Further, the cutoff value for procollagen III (1.75 U/mL) that we used in this analysis was the median value in our study population, and the performance of this value, therefore, should be validated in other populations.

One limitation of our study is that patient selection and entry criteria may have biased or influenced the results. To have the lavage procedure, patients had to be intubated, meet “safety” requirements, and give consent. Also, lavage was done at predetermined intervals regardless of the patient's clinical course. Thus, our study probably excluded the most severely ill patients who did not meet “safety” criteria or who died before the scheduled lavage. This is evident from the higher mortality rate in patients not enrolled during the study period (57% compared with 41%). Similarly, patients who were extubated before the second lavage could not be evaluated in the serial analysis. These patient drop-out patterns and exclusion criteria limited our analysis of serial procollagen III concentrations because the least severely affected patients and the most severely affected patients were not available for observation. It is possible that even more striking differences in procollagen III values over time would be noted if the extremes of the population were analyzed.

The biological basis for increased procollagen III levels in lavage fluid is probably caused, in part, by increased synthesis of type III procollagen in lung [28]. Evidence that serum or lavage fluid levels of procollagen III reflect type III procollagen synthesis rates is indirect but compelling, because many diseases and disorders that involve fibrosis of various tissues and organs consistently are associated with increased procollagen III levels in serum [7-10, 12-15]. A recent report [16] described an association of increased lavage fluid procollagen III levels with interstitial and intra-alveolar fibrosis in a few patients with the adult respiratory distress syndrome.

The strong association between procollagen III in lavage fluid and fatal outcome suggests that increased lung collagen synthesis is an important determinant of outcome from the adult respiratory distress syndrome. The mechanisms whereby increased lung collagen synthesis might lead to death were not specifically identified by our study, but one might reasonably speculate that increased collagen synthesis results in lung fibrosis and prolonged or irreversible respiratory failure, requiring prolonged ventilatory support that places patients at increased risk for lethal complications, including infection, barotrauma, insupportable failure of gas exchange, and other organ failure. We found a statistically significant correlation between procollagen III levels and disturbance of lung function. Analysis of organ dysfunction in our study as well as in a previous study from our institution indicates that ongoing respiratory failure is present in most fatal cases of the adult respiratory distress syndrome, whereas death is infrequent among patients who no longer require ventilatory support [18].

In its early stages, the adult respiratory distress syndrome is characterized as an inflammatory disease resulting in abnormalities of alveolar permeability and gas exchange. On the basis of our results, we suggest that increased lung collagen synthesis is also a relatively early event in the adult respiratory distress syndrome and may represent an important mechanism limiting restoration of normal lung function. Other investigators [29, 30] have suggested that therapies aimed at terminating the fibrotic response to acute lung injury might promote effective lung repair. Although a causal link between increased collagen synthesis and mortality in the adult respiratory distress syndrome is speculative, our data tend to support this concept. Further, analysis of lavage fluid procollagen III levels might be useful in identifying patients who would benefit from therapies aimed at modulating a maladaptive fibroproliferative response.