NT-proBNP levels in the ICU do not predict the etiology of respiratory distress

Introduction

BNP and NT-proBNP are hormones released by the ventricular myocardium in response to pressure and volume overload. Their levels help differentiate cardiac from noncardiac etiologies of dyspnea. For example, NT-proBNP above 1000 pg/ml is as a cutoff value indicating CHF is a more likely cause of dyspnea in non-ICU patients. The applicability of this test to a mixed ICU population is unclear and there is no known correlation between PCWP and BNP levels in these patients. In addition, there is no known correlation between elevated NT-proBNP and the incidence of respiratory failure.

Methods

A retrospective review of our combined medical/surgical ICU between 1 July 2004 and 30 June 2005 using Project IMPACT (Cerner Corporation) to generate a list of patients admitted for a variety of potential causes of dyspnea along with NT-proBNP levels. The Kruskal–Wallis test was used to determine significant differences between five specific diagnoses; CHF/ pulmonary edema (both cardiogenic and noncardiogenic), pneumonia, COPD exacerbation, ARF and sepsis. A receiver–operator curve (ROC) was used to examine the sensitivity and specificity of CHF at different NT-proBNP cutoff values.

Results

A total of 47 patients out of 199 (23.6%) surveyed had NT-proBNP measured. Forty-two of 199 (21.1%) fit into the diagnostic categories used for comparison. The Kruskal–Wallis test showed among the five diagnoses that only CHF and pneumonia had significantly different NT-proBNP levels (P = 0.0025). The difference in NT-proBNP levels between patients with and without respiratory failure was not significant. Results are presented in Table 1. The ROC analysis showed a ROC AUC of 0.774, but to achieve a specificity >0.9 requires a cutoff value of approximately 15,000 pg/ml.

Table 1 (abstract P189)

Conclusion

The generally accepted cutoff for NT-proBNP supporting the diagnosis of CHF is 1000 pg/ml. The average values in this ICU population (14,114.6 pg/ml) are well above this cutoff value; 85% were above 1000 pg/ml but only 12/47 (25.5%) of the patients were diagnosed with CHF. This suggests either a higher cutoff value may be required to properly utilize the test in an ICU population or it should be abandoned completely for patients admitted to the ICU, especially in light of such large standard deviations about the mean. Moreover, the data suggest that NT-proBNP does not play a role in differentiating the etiology of respiratory compromise.