In this study, the levels of plasma BNP in patients with early ALI were modestly elevated and the range of distribution was wide. However, there was no difference in BNP levels in patients with or without RV dilatation or dysfunction and no relationship between BNP and mortality.
Increased levels of plasma BNP in patients with ALI/ARDS have been previously reported by other authors in several observational studies [8–11]. However, it is not clear what pathophysiological mechanisms are primarily responsible for the increased BNP levels in this patient population. Pulmonary hypertension causing right heart strain, leading to release of BNP from the right ventricular myocardium has been the most commonly implicated mechanism [21, 22]. Several other mechanisms have been proposed. Hypoxia has been shown to increase cardiac gene expression of BNP [23, 24] and decrease lung expression of the NPR-C clearance receptor leading to increased plasma levels of BNP in animal models . Transcription of the BNP gene has been described not only in cardiac myocytes but also in the lung . Thus, it has been suggested that BNP is released in lung tissue in response to pulmonary capillary leakage .
Pulmonary hypertension with RV dysfunction is a well-recognized complication of ALI in mechanically ventilated patients [28–30]. The incidence of cor pulmonale, historically documented to be up to 60% , has decreased with the introduction of low tidal volume lung-protective ventilation, but it is still reported to be approximately 25% in an article published in 2001 . There is evidence from other patient populations to support the hypothesis that elevated BNP levels in patients with ALI are caused by RV strain. In patients with isolated RV dysfunction due to variety of conditions, BNP levels have been shown to be elevated. For example, patients with chronic respiratory failure who develop cor pulmonale have significantly higher BNP levels compared with patients with chronic respiratory failure without cor pulmonale or controls [32, 33]. In patients with idiopathic pulmonary hypertension, BNP was elevated and was correlated with the severity of RV dysfunction and outcome [5, 6]. Similar relationships have been demonstrated in patients with pulmonary embolism complicated by RV dysfunction, where BNP levels were significantly higher and predictive of mortality [4, 34, 35].
However, in contrast to those findings, our study showed no difference in the plasma levels of BNP in patients with or without RV dilatation. Furthermore, there was no correlation between systolic pulmonary artery pressure and BNP levels. The different findings may be explained by the timing of measurements obtained. Pulmonary hypertension with subsequent RV dilatation and dysfunction in mechanically ventilated patients with ALI is a result of a combination of factors. These include abnormalities of pulmonary blood flow due to formation of microthrombi in the pulmonary vasculature, hypoxemic vasoconstriction, and positive end-expiratory pressure. In our study, BNP levels and echocardiographic measurements were performed early in the course of the disease (as soon as possible after the diagnosis of ALI was made), thus potentially minimizing the effect of these factors on BNP levels, pulmonary artery pressures, and RV geometry and function. However, although the systolic pulmonary artery pressures were significantly elevated and BNP levels were markedly elevated, there was no relationship between these two variables. Additionally, BNP did not correlate with RV dilatation. Thus, our study suggests that BNP elevation in the early stages of ALI may not be caused by RV strain alone.
BNP has been established to be a predictor of mortality in a variety of chronic and acute conditions, including congestive heart failure, coronary artery disease, acute coronary syndromes [36, 37], and acute pulmonary embolism . In critically ill patients, the prognostic value of elevated BNP is less clear. In several studies, BNP has been predictive of outcome in patients with cardiogenic and septic shock [7, 38, 39]. However, in a mixed population of patients who present with severe sepsis and septic shock, the results are inconsistent; some studies have shown BNP to be predictive of mortality , others have not . Similarly, in patients presenting with hypoxic respiratory failure due to CHF or ALI, the studies have shown conflicting results. Jefic et al.  showed no relationship of BNP with mortality in 41 critically ill patients with respiratory failure (909 ± 264 in survivors vs. 841 ± 171 in nonsurvivors). Rana et al.  in a study of 204 patients who presented with pulmonary edema found that BNP levels did not differ between survivors and nonsurvivors (median 528 vs. 774, p = 0.24; O. Gajic, personal communication). Our data are consistent with those findings. In contrast, in a study by Karmpaliotis et al. , BNP showed a strong graded relationship with mortality risk in 79 subjects admitted to the ICU with hypoxic respiratory failure. In the subgroup of patients with ALI (n = 51), this relationship did not reach statistical significance but the trend was present (p = 0.07). We are unable to fully explain the discrepancies between these studies, but these may be partially attributed to different patient populations, study designs, and statistical analyses. We found interesting that Karmpaliotis et al. elected to analyze the mortality data using tertiles of BNP; however, using this method to analyze our data did not change our results. Also, in their study, 52% of the patients with ALI were in shock, and BNP has been shown to predict mortality in patients with shock. Because the authors did not stratify for the presence of shock, it is possible that shock could have accounted for the significant relationship with mortality.
RV dysfunction has been associated with an increased risk of death in patients with ALI [43–46]. In our study, we did not find a relationship between RV dilatation as a measure of RV dysfunction and mortality. However, compared with other studies that have shown this association, our study was modest in size. Furthermore, in addition to receiving lung protective ventilation, our patients also received relatively "RV protective" ventilation, as has been suggested by Bouferrache and Vieillard-Baron . Specifically, our patients received protocolized low tidal volume ventilation with a target plateau pressure of 30 cmH2O or less and relatively low PEEP (following the protocol described in ) and had minimal hypercapnia (only one subject had a paCO2 > 50 mmHg). Thus, perhaps the impact of ALI on RV dysfunction and associated mortality was reduced by our overall ventilatory approach, despite the fact that the ventilatory approach was not specifically modified after the detection of RV dysfunction by echocardiography in a protocolized fashion in this study.
The strength of our study includes its prospective design, rigorous collection of clinical and hemodynamic variables, and blinded interpretation of echocardiograms. However, some limitations should be mentioned. First, because our study was single-center and prospective, the sample size was modest and may limit our conclusions. Second, we used only a single measurement of BNP. Both echocardiography and BNP levels were obtained as soon as feasible after the diagnosis of ALI and every effort was made to coordinate these measurements. Because BNP has half-life of approximately 20 minutes and is known to fluctuate with changes in loading conditions, serial measurements of BNP may have been more useful. However, previous studies have shown that daily BNP levels in ICU patients do not change significantly [11, 41].
In summary, in patients with acute lung injury the plasma levels of BNP are increased, yet the reasons for this increase remain unclear. In this study, BNP levels were elevated regardless of right ventricular dilatation or dysfunction and an elevated BNP level was not predictive of mortality in this population of patients with ALI.