Higher incidence of acute respiratory distress syndrome in cardiac surgical patients with elevated serum procalcitonin concentration: a prospective cohort study

Epidemiological studies have shown that cardiac surgery with cardiopulmonary bypass (CPB) is a known risk factor for acute respiratory distress syndrome (ARDS) [1‐5]. Although being a highly sterile type of surgery, cardiac surgery with CPB can lead to a systemic inflammatory response with the activation of multiple inflammatory pathways involving both cellular elements and soluble proteins [6, 7]. The possible causes of the inflammatory response are the exposure of blood to non-physiological surfaces, ischemia–reperfusion injury due to aortic clamping and extracorporeal circulation [8], as well as translocation of endotoxins from gut to the bloodstream [9], which can activate inflammatory cascades similar to those observed in sepsis. Cytokines such as interleukin (IL)-6, IL-8, tumor necrosis factor, C-reactive protein, lipoprotein-binding protein and procalcitonin (PCT) may play an important role in the immune reaction.

PCT is initially described as an early, sensitive and specific marker for sepsis associated with bacterial infection [10]. However, it also increases in clinical situations without infections such as major surgery, burns, or trauma [11]. Previous studies suggested that the concentration of serum PCT increased at the end of CPB, reaching its peak on the first day and then declined rapidly [6, 12]. Previous studies also suggested that significant elevation of PCT level can be observed when complications (including pulmonary complications) presented [6, 13, 14]. However, ARDS was not differentiated from other kinds of pulmonary edema in these studies [6, 13, 14]. It is important to do so because patients undergoing cardiac surgery are at high risk of developing cardiogenic pulmonary edema, which is probably due to cardiac dysfunction or fluid overload, whereas the development of ARDS is predominantly attributed to the systemic inflammatory response. It is still unknown whether high PCT concentration indicates a high risk of developing ARDS. In the present study, our aim was to validate whether patients with higher PCT concentrations have a higher incidence of ARDS.

We enrolled 296 patients between October 2014 and October 2017, 64 patients were assigned to the PCT-elevated cohort and 232 patients were assigned to the control cohort; 76 patients were excluded (Fig. 1). PCT concentration was 16.23 ± 5.9 ng/mL in the PCT-elevated cohort, and 2.70 ± 1.43 ng/mL in the control cohort (p < 0.001). There was no significant difference in demographic characteristics, disease history or type of surgery between the two cohorts. There was longer duration of operation (305.7 ± 63.5 min versus 264.4 ± 51.1 min, p < 0.001), CPB (147.2 ± 32 min versus 135.6 ± 22.3 min, p = 0.008) and aortic clamping (84.5 ± 17.6 min versus 78.4 ± 18.2 min, p = 0.017) in the PCT-elevated cohort than in the control cohort. There were significant correlations between PCT levels and both CBP and operation time (r = 0.430 and 0.528, respectively; all p < 0.001). There was no significant difference in intraoperative fluid balance or amount of transfusion between the two cohorts (Table 1).

The PCT-elevated cohort had significantly lower P/F ratio than the control cohort (334 ± 88 versus 375 ± 56, p = 0.001). The incidence of ARDS was significantly higher in the PCT-elevated cohort than in the control cohort (21.9% versus 5.6%, p < 0.001; Fig. 2a). The incidence of moderate-to-severe ARDS (i.e., P/F ratio < 200) was also significantly higher in the PCT-elevated cohort than in the control cohort (10.9% versus 0.4%, p < 0.001; Fig. 2b). The hazard ratio of ARDS at 7 days in the PCT-elevated cohort, as compared with the control cohort, was 6.8 (95% confidence interval: 2.7 to 17.4). The hazard ratio of moderate-to-severe ARDS in the PCT-elevated cohort was 57.3 (95% confidence interval: 10.4 to 316.3). Among the ARDS patients, six patients were diagnosed with pneumonia (4 in the PCT-elevated cohort and 2 in the control cohort).

There were longer ICU stay and mechanical ventilation duration in the PCT-elevated cohort than in the control cohort (Table 2). There was no significant difference in mortality and renal failure rate between the two cohorts. There was also no significant difference in postoperative BNP, 1st-postoperative day fluid balance, white blood cell count and VIS between the two cohorts (Table 2).

After ROC analysis, PCT had an AUC of 0.734 (95% CI 0.680 to 0.783, p < 0.001) for ARDS prediction (Fig. 3). The optimal cut-off value of PCT was 6.5 ng/ml, yielding a sensitivity of 59.3% and a specificity of 81.4% for ARDS; whereas for moderate-to-severe ARDS, the sensitivity was 100% and specificity 79.9%. The performance of the ROC curve is summarized in Table 3.

PCT level, age, gender, operation duration, CPB time, aortic clamping time and the fluid balance on the first postoperative day were included in multivariate logistic regression analysis. As a result of multivariate logistic regression analysis, only PCT concentration was extracted as an independent risk factor of ARDS (OR = 1.165, 95% confidence interval 1.106 to 1.226).

Although initially described as an early marker for sepsis associated with bacterial infection [10], PCT can also elevate after cardiac surgery that is considered highly sterile [6, 13, 14]. It has been shown that PCT liberation predominantly depended on the use of CPB in cardiac surgery [18]. It is also known that the development of ARDS is associated with the use of CPB [4]. However, it is still unknown whether high PCT indicates a higher incidence of ARDS. In this prospective cohort study, we provided data to support the hypothesis that patients with elevated PCT have a higher incidence of ARDS.

The pathophysiology of ARDS following CPB has not been completely defined yet. It is believed that the use of CPB is associated with the activation of multiple inflammatory pathways [7]. A few proinflammatory cytokines elevate in response to tissue injury, transfusion and endotoxic challenge, and lead to systemic proinflammatory status [19‐22]. This systemic proinflammatory status results in neutrophil influx [22], pulmonary coagulopathy [22], capillary endothelial damage [23] and increased pulmonary vascular permeability [24], which are the characteristic findings in ARDS. In this case, an indicator of systemic inflammatory response might be helpful for alarming the development of ARDS. Although the mechanism has not yet been completely clarified, it is believed that bacterial translocation from the gastrointestinal tract and the associated release of endotoxin during CPB may play an important role in stimulating the liberation of PCT [18]. Therefore, PCT might serve as an indicator of the inflammatory response caused by CPB, and thereby warning the development of ARDS. Moreover, patients with higher PCT levels have significantly longer durations of surgery and CPB in our study, and (weak) positive correlations were also found. This is somehow explained by that a longer CPB time would lead to more proinflammatory cytokine release and a more profound inflammatory response. For instance, it has been reported that the magnitude of complement activation is proportional to the duration of CPB [25].

We used a cut-off value of 7.0 ng/mL for patient assignment in this study. This cut-off value was obtained from our previous retrospective case–control study in a similar study population, and more than 100 cases were enrolled in that study between 2010 and 2014 (unpublished data). Therefore, the present study was actually a prospective validation of elevated PCT as a predictor of ARDS. Using the data of this study we performed ROC curve analysis again and obtained an optimal cut-off value of 6.5 ng/mL. Two patients who were in the control cohort should be assigned to the PCT-elevated cohort if 6.5 ng/mL was chosen as the cut-off value, and the sensitivity would be slightly improved (51.9% to 59.3%) with no difference in specificity. Considering the tiny difference between the two cut-off values and the improved performance, we reported the performance of the ROC curve base on our new cut-off value (using the cut-off value of 6.5 ng/mL).

Our data suggest a very high negative predictive value of PCT. This means that we can use PCT as a tool to rule out patients who are unlikely to develop ARDS. However, our data suggest that the positive predictive value was not high for either ARDS or moderate-to-severe ARDS. This is probably because of the low morbidity of ARDS. In our study, the overall incidence of ARDS was 9.1%; this is comparable to the previous report (10.2%), which used the term “acute lung injury” [5]. The incidence of moderate-to-severe ARDS was 2.7%, which was also close to previous studies (that defined ARDS as a PaO_2/FiO₂ ratio < 200) [1, 2, 26]. Nevertheless, provided high mortality of ARDS, it is still good to have a tool to warn the possibility of developing ARDS, especially in those with risk factors such as massive transfusion, previous cardiac surgery, long-time CPB [2, 4], etc.

However, our patients had obviously lower mortality (1/27) compared with the overall ARDS population, who have a mortality of 20% to 50% [16]. The possible explanation is that ARDS is a heterogeneous syndrome that encompasses lung injury from both direct and indirect sources. A sepsis-like inflammatory response and endothelial damage caused by CPB leads to an extra-pulmonary ARDS, which is more homogeneous and recruitable. Therefore, in spite of severe reductions in oxygenation, these patients are less likely to develop ventilator-induced lung injury which worsens the lung injury and is associated with death. Once the inflammatory response ceases and the endothelial damage recovers, patients can rapidly recover. One more reason is that our patients were having elective surgery, which means that they had a better underlying health condition.

There were 6 patients who developed pneumonia, which happened after the sampling of PCT. This means that pneumonia was less likely be the reason for the elevation of PCT. However, the PCT-elevated cohort had a higher incidence of pneumonia as well. This is probably because of the longer duration of mechanical ventilation in the PCT elevation cohort, which is the risk factor of ventilator-associated pneumonia. Also, the pre-existing inflammatory lung edema may also attribute to the development of ventilator-associated pneumonia.

Although PCT has been used to predict postoperative pulmonary complications in patients undergoing cardiac surgery, ARDS was not differentiated from other kinds of pulmonary edema in previous studies [6, 13, 14]. It is important to distinguish ARDS patients because cardiac surgical patients are at high risk of developing cardiogenic pulmonary edema, whereas the development of ARDS is predominantly attributed to a systemic inflammatory response. In the present study, we used echocardiography or pulmonary artery catheter to rule out cardiogenic pulmonary edema (the numbers of patients diagnosed by echocardiography were 43 [67.2%] in the PCT-elevated cohort and 165 [71.1%] in the control cohort). This is an important strength of this study. We also measured Nt-pro-BNP as an indicator of cardiac function. The possible mechanism of production of BNP is the increase in regional ventricular wall stretch due to local depression of myocardial contraction [27]. There was no difference in BNP between the two cohorts in this study, suggesting a comparable cardiac function between cohorts. To the best of our knowledge, this is also a study with the largest sample size to investigate the value of PCT in predicting post-cardiosurgical ARDS.

Cardiac surgical patients with elevated PCT concentration have a higher incidence of ARDS. Elevated PCT may serve as a warning signal of postoperative ARDS in patients undergoing cardiac surgery with CPB.