Natriuretic peptides to differentiate constrictive pericarditis and restrictive cardiomyopathy: A systematic review and meta‐analysis

Abstract Previous studies have shown that natriuretic peptide levels are increased in patients with restrictive cardiomyopathy (RCM) but not in constrictive pericarditis (CP). We performed a systematic review and meta‐analysis to evaluate the diagnostic utility of B‐type natriuretic peptide (BNP) and N‐terminal pro‐brain natriuretic peptide (NT‐proBNP) to differentiate CP and RCM. We searched electronic databases from inception to January 07, 2021. Studies involving adult patients that assessed the utility of natriuretic peptides to differentiate CP and RCM were included. All meta‐analyses were performed using a random‐effects model. Seven studies (four case‐control and three cohorts) involving 204 patients were included. The mean age ranged between 25.7 and 64.1 years and 77% of patients were men. BNP levels were significantly lower (standardized median difference [SMD], −1.48; 95% confidence interval [CI], −2.33 to −0.63) in patients with CP compared to RCM. The pooled area under the curve (AUC) of the BNP level was 0.81 (95% CI, 0.70–0.92). NT‐proBNP (SMD, −0.86; 95% CI, −1.38 to −0.33) and log NT‐proBNP (SMD, −1.89; 95% CI, −2.59 to −1.20) levels were significantly lower in patients with CP compared to RCM. Our review shows that BNP and NT‐proBNP levels were significantly lower in patients with CP compared to RCM. The pooled AUC of BNP level showed a good diagnostic accuracy to differentiate both conditions.

pro-brain natriuretic peptide (NT-proBNP) to differentiate patients with CP and RCM.

| METHODS
This review was reported according to the 2009 PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement. 6

| Search strategy
We searched in the following electronic databases: PubMed, Embase, Scopus, and Web of Science. The search was conducted from inception to January 07, 2021. The complete search strategy is available in Table S1. There were no restrictions on language or publication date.
Additionally, we conducted a hand-searching of reference lists of all included studies and relevant reviews to identify further studies.

| Eligibility criteria
Studies involving adult patients (≥18 years old) that evaluated the utility of natriuretic peptides to differentiate CP and RCM were included. We excluded conference abstracts, animal studies, editorials, commentaries, systematic reviews, and narrative reviews.

| Study selection
We downloaded all articles from electronic search to EndNote X8 software and duplicate records were removed. Titles and abstracts were independently screened by two review authors (CDA and JSC) to identify relevant studies. The same review authors (CDA and JSC) independently evaluated the full text of the articles. Any disagreement on title/abstract and full-text selection was resolved through consensus.

| Data extraction
The data from each study were independently extracted by two review authors (CDA and JSC) using standardized data extraction and any disagreement was resolved through consensus. If additional data was needed, the corresponding author was contacted through email.
We extracted the following information: first author name, year of publication, country, study design, population, sample size, age, sex, etiology, and diagnosis of CP and RCM, and natriuretic peptide levels.

| Risk of bias assessment
The Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool was used to evaluate the quality of diagnostic accuracy studies. 7 This tool includes the evaluation of the risk of bias (four domains) and concerns about applicability (three domains). Each domain will be judged as "low," "high," or "unclear." The risk of bias of case-control and cohort studies was assessed using the Newcastle-Ottawa Scale (NOS) tool. 8 Each study was classified as low risk of bias (8-9 points), moderate risk of bias (5-7 points), and high risk of bias (0-4 points). Two review authors (CDA and JSC) independently perform assessments and any disagreement was resolved by consensus.

| Statistical analysis
For diagnostic accuracy studies, when individual patient data were available, we displayed a receiver operating characteristic (ROC) curve for each study. In addition, we used a random-effects model to calculate the pooled area under the curve (AUC) with their 95% confidence interval (CI) of BNP level. For case-control studies, the Dersimonian-Laird random-effects model was performed to pool standardized mean difference (SMD) with their 95% CIs of BNP, NT-proBNP, and log NT-proBNP levels. The SMD was chosen because there was variation in the type of assay used for the measurement of natriuretic peptides. Log NT-proBNP represents the logarithmic transformation of the NT-proBNP levels reported in the studies. We pooled NT-proBNP and log NT-proBNP separately because some studies only reported log NT-proBNP and it was not possible to convert to NT-proBNP. Heterogeneity among studies was evaluated using the χ 2 test (threshold p < .10) and I² statistic. Heterogeneity was defined as low if I 2 < 30%, moderate if I 2 = 30%-60%, and high if I² > 60%. Meta-analyses were conducted using the software R 3.6.3 and the web IPD Meta-Analysis of Diagnostic Accuracy. 9 A two-tailed p < .05 was considered statistically significant.

| Study selection
Our electronic search retrieved 162 articles. After the removal of duplicates, 83 articles were screened by title/abstract, and of those, 70 were excluded. After a full-text assessment of 13 remaining articles, six were excluded due to other populations (4), conference abstract (1), and commentary (1). Finally, seven articles 3,4,10-14 were selected ( Figure 1).

| Study characteristics
The characteristics of the seven included studies (n = 204) are shown in Table 1. Four studies had a case-control design and three studies were cohorts. Seventy-seven percent of patients were men and the mean age ranged between 25.7 and 64.1 years. Four studies were conducted in the United States of America.
In five studies, the diagnosis of CP was defined on the basis of surgical findings, while in the rest, it was made by cardiac catheterization. The most common etiologies of CP were idiopathic (49%), postcardiac surgery (27%), and tuberculosis (20%) across six studies.
In two studies, the diagnosis of RCM was defined according to cardiac catheterization findings. In rest of the studies, it was based on echocardiography and/or endomyocardial biopsy. The etiology of RCM was reported in three studies. Endomyocardial fibrosis (40%) and cardiac amyloidosis (37%) were the most frequent etiological diagnosis. Only one study reported data on the impact of kidney function on natriuretic peptide levels. 13 The type of assay used for BNP measurement was the Biosite BNP assay in two studies and the ADVIA Centaur BNP assay in two studies. In three studies, the assay used for NT-proBNP was the Roche Diagnostics assay. Overall, four studies reported data on BNP, two studies on NT-proBNP, and two studies on log NT-proBNP. No studies reported information on log BNP.

| Risk of bias assessment
The QUADAS-2 scores for the risk of bias and applicability concerns are shown in Figure S1. Overall, the risk of bias was high or unclear for almost all domains. In contrast, concerns regarding applicability were generally low for most studies. The NOS tool for case-control and cohort studies showed a moderate risk of bias for all studies (Tables S2 and S3).

| BNP levels
In four studies (two case-control and two cohorts, n = 97), the BNP levels were significantly lower (SMD, −1.48; 95% CI, −2.33 to −0.63; I 2 = 59%) in patients with CP compared to RCM ( Figure 2). In three studies with individual patient data, the ROC curves are shown in

| NT-proBNP levels
In two studies (two case-control, n = 67), the NT-proBNP levels were   Natriuretic peptides are widely used biomarkers in routine clinical practice and have an established diagnostic and prognostic role in patients with heart failure. 15 In addition, natriuretic peptides have also been shown to be useful in other cardiac diseases such as valvular heart disease, coronary artery disease, and cardiomyopathies, particularly RCM, which is characterized by substantial diastolic dysfunction due to intrinsic myocardial disease. 15 In contrast, although CP is also characterized by abnormal ventricular diastolic filling, it is primarily caused by a disease of the pericardium not involving the myocardium unless in mixed forms or advanced cases. 2 Given that the myocardium is usually normal in CP and the myocardial stretch is impeded by the pericardial constraint, 4 it has been suggested that measurement of natriuretic peptides could be valuable in the diagnostic work-up to differentiate constrictive from restrictive physiology. 16 On the basis of our results, CP should be suspected when normal or slightly increased levels of natriuretic peptides are detected in patients with unexplained heart failure, especially when other physical and echocardiographic findings coexist.
However, more studies with larger sample sizes are still needed to confirm the diagnostic utility of natriuretic peptides in CP.
BNP and NT-proBNP are secreted by cardiomyocytes predominantly in response to wall stress. 15 Although plasma levels of BNP and NT-proBNP are similar in normal individuals, NT-proBNP rises more than BNP in patients with heart failure. 17 We found that the levels of the two natriuretic peptides are significantly reduced in patients with CP in comparison to RCM; therefore, both biomarkers seem to be equally useful, although BNP was the most studied. On the contrary, there is some variation in their values according to the following patient's characteristics: age, sex, body mass index, and kidney function. 17 Accordingly, plasma concentrations of natriuretic peptides tend to increase with age, to be higher in women, lower in obese people, and higher in patients with kidney dysfunction. Unfortunately, only one study provided data on kidney function. Reddy et al. 13