B-type natriuretic peptide as a predictor of severity in admitted patients with acute exacerbation of COPD

Background patients with acute AECOPD show an abrupt worsening of baseline ventricular function and pulmonary hypertension, using clear predictors or markers for severity in such patients is critical to be early stratified and properly managed. Aim Assess the value of BNP as A predictor for severity in patients with AECOPD represented by need for intensive care admission, invasive and non-invasive mechanical ventilation, pulmonary hypertension, hospital and 3 months mortality. Study design Prospective study Methods A prospective Cohort study in tertiary level hospital conducted on 88 patients with AECOPD and 88 healthy control subjects, patients were divided into 2 subgroups: ICU admitted and ward admitted Patient were subjected to clinical, electrocardiographic, radiological and laboratory evaluation and observation of the clinical course during admission and 3 months following hospital discharge. Results The study revealed higher BNP in AECOPD patients compared with healthy control subjects and in ICU admitted compared with ward patients (P, 0.001). positive correlation with age, smoking index, Paco2, SPAP, RVD, need for ICU ,IMV, hospital stay and overall Mortality (r.coefficient: 0.398, 0.533, 0.605, 0.635, 0.732, -0.617, 0.577, 0.728 0.030, respectivly) ,we revealed negative correlation with ABGs parameters (Pao2, PH and o2 saturation), with r. coefficient of (-0.616, -0.609, -0.630, respectively), linear regression revelled that BNP is significant predictors for ICU admission ,ROC curve revealed that BNP more than 425pg/ml had sensitivity , specificity of (70.8% and, 100%) to predict need for ICU admission. Conclusion BNP may be considered as an accessible, useful, non-invasive and low-cost marker of severity COPD exacerbations.


Introduction
Pulmonary hypertension (PH) and heart failure are common comorbidities in patients (of about 20-30%) with stable chronic obstructive pulmonary disease (COPD) [1]. Cardiac dysfunction may trigger exacerbation in up to 25% [2]. Although recognition of cardiac comorbidities is very important for the assessment of severity and prognosis. Similarities in symptoms and signs are a challenge [3]. Echocardiography is useful in confirming heart failure; however, it requires specialized training unavailable in primary care and thus may not be ideal in some settings [4].
B-type natriuretic peptide (BNP) is secreted from the myocardium [5], with a higher level in pulmonary arterial hypertension and hypoxemia presenting a strong correlation with the functional impairment and the hemodynamic sequels [6,7]. BNP is elevated in COPD without heart failure. It has been suggested that it may arise from both left and right heart, chronic hypoxemia, and secondary PH [8,9]; in COPD exacerbation the levels of BNP are greater than in stable patients and the explanation for their rising may be more multifactorial. Accordingly, BNP may accurately reflect and predict the severity of acute exacerbation of chronic obstructive pulmonary disease (AECOPD) [10].

Aim
The aims of the study were to estimate the significance of BNP during AECOPDs as an important marker of severity and as an indicator to suspect patients who will require more aggressive treatment, such as intensive care management, indications for mechanical ventilation (invasive or noninvasive), hospital admission period, and mortality.

Ethics
The study was accepted by the local ethics committee and signed consent was obtained from all patients or their guardians.

Patients
This study was conducted at the respiratory department and respiratory critical care unit in three referral medical centers, starting from July 2015 till June 2016. This study was conducted on 88 COPD patients admitted with acute exacerbation (16 women and 72 men), their ages range from 50 to 74 years, and on 88 healthy control participants (22 women and 66 men).The patients were divided into two groups: 40 (45.5%) patients with severe exacerbation admitted in the ward and 48 (54.5%) patients with life-threatening exacerbation admitted in the ICU, six (6.8%) patients died during hospitalization where four (4.5%) patients died within 3 months from the date of discharge.

Inclusion criteria
According to the guidelines of the Global Initiative for Obstructive Lung Diseases [11] (1) Severe exacerbations: include two categories as follows.
The distribution of BNP in the different patient subgroup and control patients is described using the boxplot; a statistically significant level was considered if P is less than 0.05. (8) The value of BNP as a predictive of severity of exacerbation was studied by assessing the receiver operating characteristic (ROC), sensitivity, specificity, positive, and negative predicted values.

Results
The study included 88 patients and 88 control patients. Comparisons between the ICU-admitted and the ward-admitted subgroups are studied in Table 2 Table 2). The length of hospital stay, and indication for IMV and NIMV were seen higher significantly in the ICU group compared with the patient group (Table 2); there were no significant differences regarding hospital mortality and 3 months mortality between the ICU and the ward patients with a P value of 0.24 and 0.49, respectively. Figure 1 illustrates a boxplot showing a highly significant difference (P<0.0001) distribution of BNP (pg/ml) for all patients, controls, ICU patients, and ward patients Table 3 shows the correlations between BNP levels and relevant variables in all patients; there was a strong association represented by highly significant positive correlation with age (r=0.398 with P=0.007) and smoking index (r=0.533 with P<0.0001). The table also showed a significant negative correlation with  PaO2, PH, and O2 saturation (r=−0.616, −0.609 and 0.630, respectively) and a positive correlation with PaCO 2 with an r coefficient of 0.605. The study also showed a highly significant positive correlation with SPAP, RVD with an r. coefficient of 0.635 and 0.732, respectively with P value less than 0.0001. There was a strong association between BNP and management place, indication for IMV, the period of hospital admission and the overall mortality with an r coefficient of −0.617, 0.577, 0.728 0.030, respectively, with a P value of 0.0001. In contrast there was no significant correlation with sex, BMI, NIMV, hospital mortality, and 3 months mortality.

Predictors of ICU admission and mortality
Multivariate linear regression analysis (Table 4) revealed that PaO 2 , PaCO 2 , pH, management place, SPAP, IMV, NIMV, and length of hospital stay were significant predictors for hospital mortality but only low O 2 saturation and RVD were significant predictors for 3 months mortality. It also showed that sex, smoking index, BNP, pH, O 2 saturation, and length of hospital stay remained significant predictors for the need for ICU admission. Univariate analysis (Table 5) showed that smoking index, BNP, PaO 2 , PaCO 2 , pH, O 2 saturation, SPAP, RVD, and length of hospital stay were significant predictors for the need for ICU admission.  Sensitivity and specificity of B-type natriuretic peptide Table 6 and Fig. 2 illustrate the ROC of the serum BNP to predict the indication for ICU admission showing the area lying under the ROC curve of 0.869 (95% confidence interval: 0.733-0.951), used to obtain the optimal cutoff point of BNP levels as a predictor for ICU admission (with a maximal specificity and sensitivity). In an optimal cutoff point of 425 pg/ml, the sensitivity was 70.8%, specificity was 100%, positive predictive value was 100%, and the negative predictive value was 74.1% (P<0.0001). The ROC of BNP levels for the prediction of the need for IMV (Table 6 and Fig. 2) showed also that the area lying under the ROC curve of 0.814 (95% confidence interval: 0.668-0.915), we noticed that the cutoff point of more than 780 pg/ml had a sensitivity of 69.2% and a specificity of 90.3%, with a positive predictive value of 75% and the negative predictive value was (87.5%) (P=0.0001). The role of BNP as a predictor for hospital and 3 months after discharge mortality is studied in Table 6, revealing that BNP with a cutoff point of 980 and 1545 pg/ml, respectively, had a sensitivity of 66.7 and 50%, respectively, and specificity of 87.8 and 87.6, respectively, P value: 0.148 and 0.552, respectively.

Discussion
COPD is considered as a leading cause of mortality and morbidity with rising prevalence during the past years [12]. The development of PH in COPD develops when the airflow limitation is severe and is accompanied by chronic hypoxemia as a result of small pulmonary arterioles vasoconstriction and synthesis of endothelin from endothelial cells [13]. Pathogenesis for PH in COPD includes chronic hypoxia, emphysema, systemic and pulmonary inflammation, and capillaries destruction associated with hyperinflation [14,15]. COPD comorbidities include respiratory acidosis, ischemic heart disease, cardiomyopathy, and heart failure with cor pulmonale [16].
The result of this study showed that the serum BNP level is significantly elevated in admitted COPD patients in comparison to control patients (median: 204.0 vs. 61.0 pg/ml, P<0.0001). Moreover, patients admitted in the ICU had a higher BNP level compared with patients admitted in wards (median: 835.0 vs. 67.5 pg/ml, P<0.0001). Stolz et al. [9] studied COPD patients in exacerbation and after recovery and noticed that the level of BNP was greater during the exacerbation in comparison to the period following treatment, explaining the drop in BNP by the improvement of hypoxemia during recovery. It was observed that some COPD patients developed higher PH during acute respiratory failure of up to 20 mmHg above the baseline that improves after the treatment of the exacerbation [17]. The secretion of BNP is triggered by the following mechanisms: hypoxemia, PH, cor pulmonale, hyperinflation, left ventricular dysfunction, valvular dysfunction, and associated inflammation [18].
In this study, we demonstrated a significant correlation with blood gas parameters (PaO 2 , PaCO 2 , O 2 saturation) and echocardiographic data (SPAP and RVD), The results of our study are in agreement with the results of the Chi et al. [19] study  which assessed the COPD patients in different stages of and reported high significant correlation between BNP serum level and systolic PAP and PaCO 2 . In contrast Stolz et al. [9] showed that there was not any significant correlation between the level of serum BNP and PaO 2 , PaCO 2 , or oxygen saturation, explaining that missing correlation by the absence of other triggering factors that increase BNP release. Mallwany et al. [20] found no significant association between NT-pro BNP level and both PaO 2 and PaCO 2 explaining their finding by oxygen supplementation during the time of admission. Same results were seen by another study by Bando et al. [21] attributing their results to oxygen supplementation as well. Lang et al. [22] found significant inverse correlation between BNP and the PaO 2 only.
Diagnosis of cor pulmonale and pulmonary artery hypertension in COPD patients is very important for prognosis assessment and the indication for long-term oxygen, as cardiac catheterization is an invasive technique and echocardiography is difficult because of tachypnea and hyperinflation; so our results support the use of BNP as an indicator for PH and right ventricle dilatation and accordingly for the prediction of bad prognosis.
We also revealed a close correlation between BNP and ICU admission. Linear regression analysis demonstrated that BNP is an independent strong predictor for ICU admission (P<0.0001). The ROC curve revealed that BNP more than 425 pg/ml had a sensitivity of 70.8% while the specificity is 100% for ICU admission with a positive predictive value of 100%, but the negative predictive value was70%. In our study, we found a strong correlation with the indication for IMV, hospitalization period, and overall mortality with no significant correlation with hospital mortality and 3 months mortality. The ROC curve showed that BNP more than 780 pg/ml had a sensitivity and specificity of 69.2 and 90.3%, respectively, to predict the indication for IMV while BNP more than 980 pg/ml had a sensitivity of 66.7%, while the specificity is 87.8%, as a predictor for hospital  mortality and BNP more than 1545 pg/ml had a sensitivity of 50% and the specificity is 87.6%, for 3 months after discharge mortality. Our study showed results similar to Stolz et al. [9] who failed to predict short-term and long-term mortality rates in AECOPD patients, while in contrast Hoiseth et al. [23], who found that BNP is considered significant and independent predictor of mortality in AECOPD, but we should consider that they did not exclude patients with a history of previous heart failure from their study.
The poor correlation and weak sensitivity between BNP and hospital mortality could be attributed cautiously to many factors including: low patient numbers, reduced statistical power, high-risk factors, IMV complications, hospital-acquired infections, and comorbidities in COPD as causes for death rather than cor pulmonale and severe PH while the poor correlation to the 3 months mortality may be attributed to the short period of follow-up.
Conclusion BNP may be used as an available, valuable, low-cost, and noninvasive predictor for the severity grading of COPD exacerbations. BNP is higher in hospitaladmitted patients with AECOPD and in patients admitted to the ICU, in patients that indicated IMV, length of hospitalization, BNP may be considered a significant predictor for ICU admission, indication for invasive mechanical ventilation, and long periods of hospitalization.

Limitations and weakness
The low number of studied patients and the short period for follow-up limited the accuracy of estimated results.

Recommendations
Large sample studies are needed to evaluate the value of BNP-guided management for AECOPD to improve prognosis, outcome, length of hospitalization, and mortality.