Journal Club Podcast #45: August 2018
A look at the use of BNP to differentiate cardiac from non-cardiac causes of syncope...
Click Tabs Below to Expand
Article 1: Pfister R, Hagemeister J, Esser S, Hellmich M, Erdmann E, Schneider CA. NT-pro-BNP for diagnostic and prognostic evaluation in patients hospitalized for syncope. Int J Cardiol. 2012 Mar 8;155(2):268-72. Answer Key.
Article 2: Reed MJ, Newby DE, Coull AJ, Prescott RJ, Jacques KG, Gray AJ. The ROSE (risk stratification of syncope in the emergency department) study. J Am Coll Cardiol. 2010 Feb 23;55(8):713-21. Answer Key.
Article 3: Reed MJ, Newby DE, Coull AJ, Jacques KG, Prescott RJ, Gray AJ. Role of brain natriuretic peptide (BNP) in risk stratification of adult syncope. Emerg Med J.2007 Nov;24(11):769-73. Answer Key.
Article 4: Isbitan A, Hawatmeh A, Elnahar Y, et al. Utility of brain natriuretic peptide assay as a predictor of short term outcomes in patients presenting with syncope to the emergency department. Cardiovasc Diagn Ther. 2016 Jun;6(3):234-40. Answer Key.
You are working an evening shift in your ED as the senior resident when you encounter Mr. Drop, a 58-year-old male presenting with syncope. He was sitting at the dinner table, getting ready to eat, when he suddenly lost consciousness and fell face-first into his mashed potatoes. He had no seizure activity and awoke within a few seconds. His wife helped him clean the mashed potatoes off his face and brought him to the ED. His past medical history includes hypertension and diet-controlled diabetes, and he endorses a family history of coronary artery disease in both parents. He denies chest pain, shortness of breath, or palpitations, and his vital signs and physical exam are currently normal.
You go to put in orders, including an ECG, BMP, CBC, and troponin, when your large Russian attending, whose hairstyle reminds you of Eddie Munster, suggests you order a BNP, “to help determine if this is cardiac syncope.” In your years of training, this is the first time anyone has suggested such a thing, but not wanting to anger this dangerous-looking attending, you comply without question. Later, when you’re certain he’s not looking, you open up PubMed and begin searching the evidence to see if there is actually any correlation between BNP levels and cariogenic syncope…
Population: Adult patients presenting to the ED with syncope of unclear etiology
Intervention: BNP testing
Comparison: Standard ED workup
Outcome: Diagnostic accuracy of BNP for potentially life-threatening causes of syncope (e.g. arrhythmia or structural cardiac disease), impact of BNP testing on admission rates, impact of BNP on diagnostic testing.
PubMed was searched using the terms “(BNP OR “brain natriuretic peptide”) AND syncope” (https://tinyurl.com/yc4ta5e9). This resulted in 35 citations, from which the 4 most relevant articles were chosen.
Syncope is a common presenting complaint in the ED with a large number of potential underlying causes, ranging from the very benign (e.g. vasovagal syncope, orthostasis) to potentially life-threatening causes (e.g. cardiac arrhythmia). While studies have demonstrated a broad range for rates of serious outcomes, as wide as 1.2% to 36.2%, there is very little objective data to help the emergency physician determine who is safe for discharge and who requires admission to prevent morbidity and mortality. Several clinical prediction rules have been developed to assist with disposition decisions; unfortunately, none of these has performed well when external validation has been attempted (Birnbaum 2008, Serrano 2010, Safari 2016). As a result, investigators continue to search for potential prognostic factors to help differentiate those syncope patients at high risk of adverse outcome from those safe for discharge.
Brain natriuretic peptide (BNP) and its analogs are typically elevated in the setting of structural heart disease, but has also been show to increase following certain cardiac arrhythmias. As a result, BNP has been proposed as a possible test to help differentiate cardiogenic from non-cardiogenic causes of syncope. A small, prospective, observational study conducted in Scotland in 2007 (Reed 2007) demonstrated a significant association between BNP values and serious adverse outcome. Unfortunately, the test characteristics for this association were rather poor. To predict a serious outcome at 3 months (including death, MI, life-threatening arrhythmia, implantation of a pacemaker or defibrillator, PE, CVA, intracranial hemorrhage, need for blood transfusion or need for acute surgical procedure or endoscopic intervention), a BNP > 100 had a positive likelihood ratio (LR+) of 2.21 and a negative likelihood ratio (LR-) of 0.48, suggesting that neither a positive nor negative test at this threshold would have any significant impact on post-test probability. For a threshold of 1000, BNP had a LR- of 0.51, which again would be of little help. While the LR+ was infinity, there were only 3 patients with a BNP this high. A larger study would need to confirm the potential utility of a BNP this elevated, though it would likely impact only a small percent of patients presenting with syncope (most of whom would likely be admitted anyway).
Based on this observed association, these authors conducted an additional study (Reed 2010) in which they attempted to derive and validate a clinical decision rule employing BNP results as one of the criteria. They developed the ROSE rule, which employed the following criteria:
BNP level ≥ 300 pg/mL
Bradycardia ≤ 50 in ED or pre-hospital
Rectal examination with fecal occult blood
Anemia (hemoglobin ≤ 90 g/L)
Chest pain associated with syncope
ECG showing Q-wave (not in lead III)
Saturation ≤ 94% on room air
This rule was derived in 529 ED patients presenting with syncope. In the derivation cohort, the rule had a LR+ of 3.5 and LR- of 0.1 , suggesting that it may be useful in reducing the probability of a serious outcome in patients with none of the criteria, with a small increase in the probability in patients for whom the rule is positive. The rule performed slightly less well in the validation cohort (N = 550), with a LR+ of 2.5 and LR- of 0.2. Unfortunately, these results are skewed by several sources of bias: only 13% of patients underwent rectal examination (which was a component of the rule); a non-consecutive sample of patients was enrolled, with 35% and 40% of eligible patients not approached for enrollment in the derivation and validation cohorts, respectively; and the rule was derived and validated at a single study site, requiring additional validation at multiple sites before it can be used safely. Previously, the San Francisco syncope rule was validated at a single site, but then failed to validate when tested at additional sites.
Two additional prospective studies evaluating the diagnostic accuracy of BNP or its analogs were reviewed. The first of these (Pfister 2012) enrolled patients with syncope admitted to a cardiac unit in Cologne, Germany. They evaluated the accuracy of NT-pro-BNP for identifying patients with an arrhythmia or structural cardiac/cardiopulmonary abnormality. With 161 patients enrolled, the study found a LR+ of 1.86 and LR- of 0.20. The second study (Isbitan 2016) enrolled patients at two emergency departments in New Jersey and evaluated the accuracy of BNP at determining “serious outcomes” (which were similar to those listed for the two studies by Reed). Using a cutoff of 250, the authors found a LR+ of 5.02 and LR- of 0.57. Again, these two studies demonstrate rather poor positive and negative likelihood ratios which would not be expected to have a significant effect on probability. Additionally, the first study likely suffers from spectrum bias, as it was conducted on a cohort of patients already admitted to a cardiac unit rather than undifferentiated ED patients.
Overall, the current evidence for the use of BNP in the work-up of syncope is rather limited and not very promising. All of the studies were limited by the lack of a clear gold standard in the evaluation of syncope and by a non-uniform application of additional testing leading to differential verification bias and partial verification bias. The diagnostic accuracy of BNP in these studies was poor, with a LR- ranging from 0.2 to 0.57 and LR+ ranging from 1.86 to 5.02; although LR+ was infinity when a high enough BNP threshold was used, such a cutoff would likely be of little value, as it would be unlikely to change management and would apply to a rather small number of patients. The ROSE rule, while promising on derivation, had a LR- of 0.2 when validated; additional validation in other clinical settings would be needed to determine whether it would be reasonable to use. For now, BNP and the ROSE rule should not be used to determine disposition or direct further evaluation of patients presenting to the ED with syncope.