Prognostic Value of Increased Plasma Levels of Brain Natriuretic Peptide in Patients With Septic Shock

ABSTRACT: Our objective was to investigate the plasma levels of brain and atrial natriuretic peptides (BNP and ANP, respectively) in patients with septic shock/severe sepsis and to study the association of BNP and ANP levels with hemodynamic parameters, severity of the disease, and prognosis of those patients. This is a prospective case series study of 22 patients with septic shock, 11 patients with severe sepsis, and 20 healthy volunteers at the Department of Emergency and Critical Care Medicine, Nara Medical University Hospital, Japan. Blood collection was performed on admission and on days 1, 2, and 4. Plasma BNP and ANP levels were measured by radioimmunoassay. Right atrial pressure, mean pulmonary arterial pressure, pulmonary arterial wedge pressure, and left ventricular stroke work index were determined using a thermodilution catheter. Acute Physiological and Chronic Health Evaluation II scores were calculated. Plasma levels of BNP and ANP were markedly elevated in patients with septic shock/severe sepsis compared with controls (BNP, 7 ± 0.3 pg mL^-1; ANP, 13 ± 1 pg mL^-1). In patients with septic shock, both BNP and ANP peaked on day 2 (BNP, 987 ± 160 pg mL^-1; ANP, 103 ± 17 pg mL^-1). Plasma levels of BNP on day 2 in patients with septic shock significantly correlated with right atrial pressure (r = 0.744, P < 0.01), mean pulmonary arterial pressure (r = 0.670, P < 0.01), pulmonary arterial wedge pressure (r = 0.709, P < 0.01), left ventricular stroke work index (r = -0.552, P < 0.05), Acute Physiological and Chronic Health Evaluation II score (r = 0.581, P < 0.01), and poor prognosis (P < 0.05). The optimal cutoff point for predicting mortality in patients with septic shock was a BNP level of 650 pg mL^-1 on day 2, in which sensitivity and specificity were 92% and 80%, respectively. Increased plasma levels of BNP may reflect not only the severity of myocardial depression but also the disease severity and could be of prognostic value in patients with septic shock.

INTRODUCTION

Severe sepsis and septic shock are common problems in intensive care unit, and the mortality rate in patients with septic shock is still high despite advanced therapeutic interventions. The main cause of death in patients with sepsis is multiple organ failure and refractory hypotension, which are frequently complicated by impaired tissue perfusion and myocardial depression. Myocardial depression is usually characterized by reversible biventricular dilatation and decreased ejection fraction, which may result in poor response of the patients to fluid resuscitation and catecholamine infusion (^1-3). Several reports have indicated progressive myocardial depression is associated with mortality even in the early phase of septic shock (^{4, 5}). Brain natriuretic peptide (BNP), also known as B-type natriuretic peptide, and atrial natriuretic peptide (ANP) belong to a natriuretic peptide family, which possess potent diuretic, natriuretic, and vasorelaxing properties. BNP and ANP are mainly secreted from the cardiac ventricle and the cardiac atrium, respectively, in response to fluid overload or mechanical stretch (^6-10). They are involved in the pathogenesis of diverse cardiovascular diseases, including congestive heart failure and myocardial infarction. Increased plasma levels of these peptides have been accepted as a diagnostic marker or a prognostic predictor in several cardiac diseases (^11-17). Recently, plasma BNP and ANP levels are reported to be increased in critically ill patients without cardiovascular diseases (^18-20).

Implications of these natriuretic peptides in the pathogenesis of sepsis are still unclear. Increased plasma levels of ANP have been reported in sepsis in some studies (^21-26). Hartemink et al. (²⁵) showed that circulating ANP could be a marker of myocardial depression in patients with septic shock. Compared with ANP, the significance of BNP in septic shock has been rarely reported (^{22, 26, 27}). The value of these natriuretic peptide as a predictor of mortality in patients with septic shock especially remains a matter of controversy.

To investigate the possible predictive value of BNP and ANP in sepsis in a prospective study, we measured sequential plasma levels of BNP and ANP in patients with severe sepsis or septic shock and compared those levels with their hemodynamics, severity of illness, and mortality.

MATERIALS AND METHODS

Study population

All the patients were admitted to the Department of Emergency and Critical Care Medicine, Nara Medical University Hospital, from May 1998 to April 2000. Enrolled patients fulfilled the clinical criteria of severe sepsis or septic shock, which were defined by the Members of the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference Committee (²⁸). Patients were managed according to the standard protocol for sepsis, including antibiotics, pressors, respiratory support, and surgical intervention if indicated. Patients with a history of congestive heart failure or chronic renal failure were excluded from the study. All patients were followed until death or for 28 days after being enrolled in the present study. Patients were considered as "survivors" when they survived 28 days. This study was approved by the Ethics Committee of Nara Medical University Hospital, and all blood samples were drawn after written informed consent was obtained from the patients or relatives. Twenty blood samples were obtained from healthy volunteers (10 men and 10 women) as controls.

Sample collection

All blood samples were obtained with EDTA per 2 Na (1 mg mL^-1) and aprotinin (500 kIU mL^-1 Trasylol; Bayer, Leverkusen, Germany) on admission and on days 1, 2, and 4. Plasma was separated by centrifugation at 3000 rpm for 10 min at 4°C and was stored at -80°C until assayed.

Radioimmunoassay for BNP and ANP

Plasma BNP and ANP were measured by a specific radioimmunoassay using Shiono RIA BNP and ANP (Shionogi Co, Ltd, Osaka, Japan) with an intra-assay coefficient of variation of 5.3% or 5.8%, an interassay coefficient of variation of 5.9% or 11.9%, and lower detection limit of 2 or 5 pg mL^-1, respectively. These measurement systems are highly specific to BNP or ANP, with no cross-reactivity. The physicians were blinded from the BNP and ANP levels for 28 days after enrollment.

Clinical data

The clinical status of the patients was evaluated using the Acute Physiology and Chronic Health Evaluation (APACHE) II system on admission (²⁹). Sepsis-related Organ Failure Assessment (SOFA) scores were also determined everyday by summing the scores for the grades (from 0 to 4) of failure of six organ systems (pulmonary, cardiovascular, renal, hepatic, hematological, and central nervous systems), with maximum score of 24 (³⁰). The peak SOFA score during the first 28 days after admission was used to grade organ failure in each patient. Criteria of acute lung injury (ALI) as defined by the 1994 American-European Consensus Conference (³¹) was used.

Measurement of hemodynamic parameters

A pulmonary artery flotation catheter (7-Fr Swan-Ganz TM catheter; Baxter Healthcare, Rouland Lake, Ill) was placed at the discretion of an intensive care team. When the catheter was placed, heart rate (HR), mean arterial blood pressure (mABP), right atrial pressure (RAP), mean pulmonary arterial pressure (mPAP), and pulmonary arterial wedge pressure (PAWP) were measured on admission, days 1, 2, and 4, at the same time blood samples were obtained. Systemic vascular resistance index (SVRI), left ventricular stroke work index (LVSWI), and cardiac index (CI) were calculated accordingly.

Statistical analysis

Continuous variables were analyzed for normal distribution by using Kolmogorov-Smirnov test and expressed as the mean ± SEM. The plasma levels of BNP, ANP, and hemodynamic parameters were compared over the time course using repeated measure analysis of variance followed by Dunnett test. Student t test for continuous variables and the χ² test for categorical variables were used to examine differences on each study day between the two subgroups (septic shock versus severe sepsis, nonsurvivor versus survivor). Pearson correlation coefficients were used for the correlation analysis. We considered Ps of <0.05 (two-tailed) to be statistically significant. The data were analyzed by using Statview 5.0 (SAS Institute, Cary, NC).

RESULTS

Clinical characteristics of patients

We enrolled 33 patients with severe sepsis (11 patients) or septic shock (22 patients). All patients with septic shock and 6 of 11 patients with severe sepsis were mechanically ventilated. Continuous hemodiafiltration was started within the first 2 days in 9 of 22 patients with septic shock and 3 of 11 patients with severe sepsis. A pulmonary artery flotation catheter was inserted in 19 of 22 patients with septic shock. Demographic data of these patients are shown in Table 1. APACHE-II score (septic shock versus severe sepsis, 27.4 ± 1.5 versus 16.0 ± 1.6, respectively; P < 0.01), SOFA score (11.9 ± 1.0 vs. 9.5 ± 0.8; P < 0.01), and mortality (nonsurvivors/survivors + nonsurvivors; 12/22 vs. 1/11; P < 0.01) were significantly higher in septic shock than those in severe sepsis (Table 1). All patients with septic shock required blood pressure support by intravenous infusion of dopamine, dobutamine, or norepinephrine to maintain systolic blood pressure above 90 mmHg. There was no difference in the type of pressors or the duration on the pressors between survivors and nonsurvivors with septic shock.

Table 1

Hemodynamic assessments

Serial values of hemodynamic parameters measured on admission, days 1, 2, and 4 in survivors and nonsurvivors with septic shock who were inserted a pulmonary artery flotation catheter were shown in Table 2. To determine the factors that correlate with mortality, we compared hemodynamic variables between survivors and nonsurvivors. No significant difference was observed in HR, mABP, RAP, mPAP, PAWP, and SVRI between survivors and nonsurvivors. On the other hand, LVSWI, which is one of the indices for cardiac performance, on days 2 (P < 0.05) and 4 (P < 0.05) were significantly higher in survivors than those in nonsurvivors.

Table 2

Plasma levels of BNP or ANP

Figure 1 illustrates the changes in plasma levels of BNP (A) and ANP (B), respectively, in patients with septic shock and severe sepsis. On admission, the plasma level of BNP in patients with septic shock (346.6 ± 69.5 pg mL^-1) was remarkably elevated compared with control (7.4 ± 0.3 pg mL^-1, P < 0.01). The plasma level of BNP peaked on day 2 (987.1 ± 159.9 pg mL^-1) and decreased gradually afterward. The changing pattern of the ANP plasma level was essentially the same as that of BNP. On admission, the plasma ANP level of patients with septic shock (62.9 ± 25.7 pg mL^-1) was significantly elevated compared with control (13.2 ± 0.6 pg mL^-1, P < 0.01), peaked on day 2 (120.5 ± 17.9 pg mL^-1), and decreased gradually afterward. The plasma levels of BNP in patients with severe sepsis (on admission, 91.5 ± 25.9 pg mL^-1; day 1, 100.4 ± 23.4 pg mL^-1; day 2, 101.3 ± 25.7 pg mL^-1; day 4, 99.6 ± 32.5 pg mL^-1; P < 0.01) and ANP (on admission, 17.0 ± 4.2 pg mL^-1; day 1, 27.4 ± 6.1 pg mL^-1; day 2, 22.4 ± 4.7 pg mL^-1; day 4, 21.5 ± 3.6 pg mL^-1; P < 0.01) significantly increased compared with controls but did not change significantly over the study period.

Fig. 1

Correlation of plasma BNP or ANP levels with hemodynamic parameters

The correlations of plasma BNP or ANP levels with hemodynamic parameters obtained by using a pulmonary artery flotation catheter in patients with septic shock were examined. Plasma ANP levels did not correlate with hemodynamic parameters throughout the study. HR, mABP, and CI did not correlate with plasma BNP levels throughout the study in patients with septic shock, whereas the following parameters significantly correlated with plasma levels of BNP on its respective days: RAP (day 1, r = 0.520, P < 0.05; day 2, r = 0.744, P < 0.01), mPAP (day 1, r = 0.490, P < 0.05; day 2, r = 0.670, P < 0.01), PAWP (on admission, r = 0.509, P < 0.05; day 1, r = 0.587, P < 0.01; day 2, r = 0.709, P < 0.01; day 4, r = 0.642, P < 0.01), and LVSWI (day 2, r = -0.552, P < 0.05).

Correlation of plasma BNP or ANP levels with disease severity

On days 1, 2, and 4, the plasma level of BNP in patients with septic shock, not severe sepsis, significantly correlated with the APACHE-II score (day 1, r = 0.509, P < 0.05; day 2; r = 0.581 (Fig. 2), P < 0.01; day 4, r = 0.553, P < 0.05). Plasma level of BNP did not correlate with the peak SOFA score in septic shock. Plasma level of ANP did not correlate with the APACHE-II or the peak SOFA scores.

Fig. 2

Correlation of plasma BNP or ANP levels with ALI and amount of administered fluid

We studied whether the presence of ALI affected the plasma BNP and ANP levels because there was a previous report showing that the plasma levels of BNP and ANP levels in patients with ALI were elevated. We found that 5 of 22 patients with septic shock were complicated with ALI. There are no statistically significant differences observed between the patients with and without ALI in the plasma BNP and ANP levels on day 2 (BNP, 1048.0 ± 124.5 pg mL^-1 vs. 969.2 ± 205.5 pg mL^-1, P = 0.842; ANP, 107.6 ± 39.4 pg mL^-1 vs. 101.6 ± 19.0 pg mL^-1; P = 0.885). Septic patients usually receive large amounts of fluid to maintain blood pressure. The effect of excessive fluid on the elevation of BNP or ANP also could not be excluded. In the present study, we did not find any association between the administered amount of fluid during the first 2 days (7869 ± 1275 mL for 2 days) and BNP or ANP levels (r = 0.075, P = 0.739, and r = 0.335, P = 0.113, respectively).

Plasma levels of BNP or ANP in survivors and nonsurvivors in septic shock

Twelve of 22 patients with septic shock and 1 of 11 patients with severe sepsis died (Table 1). The differences in plasma levels of BNP and ANP in septic shock were investigated between survivors and nonsurvivors. The plasma BNP levels on days 1 (P < 0.05), 2 (P < 0.05, Fig. 3), and 4 (P < 0.05), and APACHE-II score were significantly higher in nonsurvivors compared with those in survivors. Receivers operating characteristic curves, showing the sensitivity and specificity of plasma BNP and ANP in discriminating nonsurvivors from survivors, were constructed. The cutoff values of BNP and ANP on day 2 that gave the best sensitivity and specificity were 650 pg mL^-1 (sensitivity, 92%; specificity, 80%) with area under the curve of 0.85 and 95 pg mL^-1 (sensitivity, 60%; specificity, 50%) with area under the curve of 0.54, respectively.

Fig. 3

DISCUSSION

In recent years, there have been accumulating data supporting the pivotal roles of BNP and ANP in the diagnosis and management of cardiovascular diseases such as congestive heart failure and myocardial infarction. The preferred use of BNP than ANP could be justified because of the localization of these peptides in cardiac tissue. ANP is released from the cardiac atria, whereas BNP is released from both cardiac atria and ventricle in response to increased ventricular diastolic pressure and volume expansion. Thus, it is thought that BNP levels are more useful than ANP in making diagnosis and predicting patient outcome in cardiovascular diseases (^7-17). Recently, it has been reported that BNP is synthesized and secreted by macrophages and cardiac tissue (³²) and that production of BNP was stimulated by interleukin 1β in cardiac myocytes (³³). These findings suggested that BNP may be closely associated with the pathophysiology of inflammation and cardiac dysfunction. Taken together with the report that myocardial depression is typically involved in sepsis (^{1-5, 22, 27}), it is interesting to know whether BNP could be a predictor in sepsis as a marker reflecting both severity of inflammation and myocardial depression.

In the present study, the mean plasma level of BNP in septic shock was dramatically elevated. It was 47 times higher on admission, and it was 133 times higher on day 2 than those in controls. Of note is that plasma level of BNP in nonsurvivors with septic shock on day 2 was even higher and reached 177 times higher than controls, whereas that in survivors with septic shock was only 81 times higher. This extreme elevation of plasma BNP levels was comparable with or exceeding the levels found in patients with most severe forms of cardiac disease (^{8, 10, 13, 15-17}). This may suggest that BNP plays a role in the pathophysiology of septic shock. Although plasma ANP levels were also elevated in patients with septic shock, this elevation was much less marked than that of BNP. Considering these findings, BNP could be more closely involved in the pathophysiology of septic shock rather than ANP.

Because BNP increases with left ventricular dysfunction (¹³), we hypothesized that the elevation of BNP is the result of myocardial depression due to sepsis. Because myocardial depression is frequently complicated by the increase of ventricular diastolic volume and pressure, which can be the stimuli for BNP release, it is physiologically plausible that myocardial depression leads to the release of BNP from the myocardium. It is also reasonable that BNP may be much more elevated than ANP in myocardial depression because BNP is mainly released from the ventricle, whereas ANP is mainly released from the atrium. The time course of BNP levels may also support the hypothesis that myocardial depression is the main cause of the BNP elevation in septic shock. We observed that plasma BNP levels in septic shock peaked on day 2 as myocardial depression occurred magnitudely (^{1, 2}). We also showed that plasma BNP levels were best correlated with hemodynamic parameters such as PAWP, RAP, and LVSWI. Interestingly, LVSWI has been reported to be one of indices of myocardial depression (^{4, 5}).

Plasma ANP levels in septic shock also peaked on day 2 but did not correlate with any hemodynamic parameters. Considering these findings, plasma BNP, not ANP, could be a preferred marker of myocardial depression in septic shock. These results are contrary to previous reports showing that increased plasma ANP levels in sepsis correlate with mPAP (^{18, 22}) and that plasma BNP levels in septic shock are increased only twice and correlated with CI (²⁶). Perhaps a difference in patient population may explain this discrepancy.

Although we speculate that myocardial depression is the major factor for the increase of BNP, the contribution of other factors could not be excluded. We speculate that severe inflammation may have a positive effect on BNP elevation. When patients with septic shock in this study were classified according to the Forrester classification (³⁴), the 12 patients that were in Forrester I subset (CI >2.2 L min^-1 m^-2, PAWP <18 mmHg) exhibited BNP levels that were 95 times higher than controls. On the other hand, BNP levels in patients with acute myocardial infarction or dilative cardiomyopathy with Forrester subset I showed only 4.5 to 38 times higher BNP levels than those in controls (^15-17). This suggests that factors other than myocardial depression could contribute to the elevation of BNP in sepsis. This is consistent with the hypothesis that intense inflammation in sepsis may induce the additional release or production of BNP. Also, whether decreased elimination of BNP and ANP can increase those levels in sepsis need to be considered. Although decreased cleavage of BNP and ANP because of reduced endopeptidase activity may cause the elevation of BNP and ANP, enhanced endopeptidase activity, which generally occurs in sepsis, makes this possibility highly unlikely. Furthermore, we found no correlation between BNP or ANP concentrations and serum creatinine concentrations in this series. A previous study demonstrated that the elimination of BNP and ANP by continuous hemodiafiltration was negligible (³⁵). Considering these findings decreased elimination of these peptides may not contribute to increase those levels in patients with septic shock so much.

If the elevation of BNP is caused by myocardial depression and severe inflammation, BNP levels may represent the degree of myocardial depression together with the general severity of sepsis, making it a desirable index for predicting prognosis in sepsis. Plasma levels of BNP in patients with septic shock significantly correlated with the APACHE-II score, showing that BNP level could be a marker of disease severity of sepsis. We observed that plasma BNP levels on day 2 in nonsurvivors with septic shock were significantly higher than in survivors with septic shock. The optimal cutoff point for predicting mortality was BNP level of 650 pg mL^-1 on day 2, in which sensitivity and specificity for predicting mortality in septic shock were 92% and 80%, respectively, with area under the curve of 0.85. At this cutoff value, BNP level on day 2 is a strong predictor of mortality in patients with septic shock. Recently, Witthaut et al. (²⁶) showed no difference in plasma BNP between survivors and nonsurvivors with septic shock, but Charpentier et al. (²⁷) demonstrated that increased BNP was associated with mortality in patients with severe sepsis and septic shock. Our conclusion is consistent with the finding of Charpentier et al., but because the size of both prospective studies are small, several lines of prospective studies are needed to use BNP as a simple, rapid, and readily available marker to predict mortality in septic shock. Also, the exact mechanism by which these natriuretic peptides are elevated remains to be elucidated. Studies using molecular biological techniques might clarify this point.

In conclusion, BNP plasma levels were significantly increased in patients with septic shock, perhaps because of the combination of myocardial depression and intense inflammation. Plasma levels of BNP might not only be a good marker of the severity of septic shock, including myocardial depression, but also a good predictor of mortality in patients with septic shock.

ACKNOWLEDGMENT

The authors thank Norio Kurumatani, MD, PhD, of Nara Medical University for his critical advice in the statistical analysis of this study.