Methods

Serial cardiac biomarker measurement had been introduced at the department of pulmonology of VU Medical Center of Amsterdam (The Netherlands) since 2002 as part of routine clinical assessment. We retrospectively identified all patients who were diagnosed with World Health Organization group I PAH from November 2002 through September 2009. The study was approved by our institutional review board and ethics committee. All World Health Organization group I subgroups with ≥1 NT–pro-BNP measurement were included in this analysis (n = 198 patients). Thirty-seven of these had only a single measurement at baseline and these were not included in longitudinal analysis. Survival data were available for all patients until date of death or January 1, 2010. Follow-up was performed by regular visits to the outpatient clinic at 3- to 6-month intervals and by telephone contact.

Blood samples for NT–pro-BNP measurements were drawn from a peripheral vein in all patients while a patient was in a stable hemodynamic state before the 6-minute walk distance test was performed. Figure 1 shows available longitudinal data of NT–pro-BNP measurements for the 161 patients (determination at time of diagnosis). Because of the observational nature of this study the visiting patterns of the patients were rather variable. NT–pro-BNP plasma levels were analyzed with the Elecsys 1010 electrochemiluminescence immunoassay (Roche Diagnostics, The Netherlands) as described previously.

Patients with pulmonary arterial hypertension with blood samples available for N-terminal pro–B-type natriuretic peptide testing at baseline and follow-up visits.

Differences in baseline covariates between survivors and nonsurvivors were tested with the Mann–Whitney U test. Multivariable Cox regression was used to assess the association between baseline log ₁₀ (NT–pro-BNP) level and risk of death, adjusting for gender, age, New York Heart Association (NYHA) class, mRAP, and baseline eGFR.

To investigate the added value of serial NT–pro-BNP measurements, a 2-stage procedure was employed. In the first stage longitudinal information for each patient was summarized while accounting for the correlation in NT–pro-BNP serial measurements by a linear mixed-effects model with a random-intercepts and random-slopes structure. The aim of this model is to estimate the longitudinal evolution of NT–pro-BNP for each patient and account for biological variation. Typically, biomarker measurements such as NT–pro-BNP contain intrapatient “error,” which is also known as “biological variation.” More specifically, by biological variation we imply the fluctuations that are observed in the biomarker levels in human physiology. As shown in Figure 2 , patient B had a relatively stable profile of NT–pro-BNP levels over time; however, not all NT–pro-BNP measurements decrease exactly on the dashed line denoting this profile. Fluctuations of observed NT–pro-BNP levels around the profile denote biological variation. From a statistical point of view this variability should be taken into account to obtain optimal results because failure to account for this variation will lead to an underestimation of the effect size of NT–pro-BNP.

Examples of observed N-terminal pro–B-type natriuretic peptide measurements and estimated underlying true evolutions from the linear mixed-effects model for patients A and B. The slope quantifies the expected “average” change of the marker, not the observed change, where data scatter.

At the second stage the patient-specific intercepts (baseline value) and slopes were included as covariates in a similar multivariable Cox regression model as in the previous analysis, which also corrected for the effects of gender, age, NYHA class, mean right atrial pressure (mRAP), and baseline estimated glomerular filtration rate (eGFR).

Based on receiver operating characteristic curve analysis at 80 months, optimal cutpoints for baseline NT–pro-BNP and patient-specific intercept were determined as points for which the product of sensitivity and specificity attains its maximum. All analyses were performed in R 2.11.1 (2010) using JM 0.7, nlme 3.1, Survival 2.35, and Survival ROC 1.0.0 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Baseline characteristics of all 198 patients are presented in Table 1 . The population was predominantly women (71.2%) with a mean age of 54 ± 17 years. Fifty-six percent had idiopathic PAH and >58% were in modified NYHA/World Health Organization functional classes III and VI. PAH baseline therapies included prostacyclin analogs in 26 (13.1%), endothelin receptor antagonists in 118 (59.5%), and phosphodiesterase-5 inhibitors in 40 patients (20.2%). Eight patients received combination PAH therapies. Calcium channel blockers were used for PAH treatment in 5 patients (2.5%, data not shown).

During a mean follow-up period of 38 ± 23 months 54 patients (27.3%) died. Figure 3 shows the Kaplan–Meier estimate of survival function for the entire group. Overall survival estimates for our PAH population at 1 year, 3 years, and 4 years were 93% (95% confidence interval [CI] 90 to 97), 80% (95% CI 74 to 86), and 71% (95% CI 65 to 79), respectively. Pulmonary artery pressure and pulmonary vascular resistance (PVR) at baseline were similar for survivors and nonsurvivors. In contrast, those who died had significantly worse functional class and walked a significantly shorter distance in 6 minutes. Furthermore, their hemodynamic parameters (higher mRAP and lower cardiac output and mixed venous saturation) indicated more compromised right ventricular function.

Kaplan–Meier estimates and 95 confidence intervals (dashed lines) of survival from time of enrollment.

Baseline NT–pro-BNP levels of survivors and nonsurvivors were significantly different ( Table 1 ). Plasma NT–pro-BNP increased significantly (p <0.0001, Kruskal–Wallis test) with severity of NYHA functional class ( Figure 4 ) . Of our patients 147 had diagnostic assessment and simultaneous NT–pro-BNP measurements at baseline. NT–pro-BNP was positively correlated with mRAP (Spearman correlation coefficient 0.26, p <0.001), serum creatinine (Spearman correlation coefficient 0.48, p <0.001), PVR (Spearman correlation coefficient 0.54, p <0.001), and heart rate (Spearman correlation coefficient 0.28, p <0.001), whereas negative associations were found with cardiac index (Spearman correlation coefficient −0.43, p <0.001) and mixed venous oxygen saturation (Spearman correlation coefficient −0.42, p <0.001).

Relation between New York Heart Association status and N-terminal pro–B-type natriuretic peptide level represented by median values (horizontal bars) , twenty-fifth and seventy-fifth percentiles (lower and upper limits of boxes) , and 1.5 times interquartile range (error bars) . Median N-terminal pro–B-type natriuretic peptide values for each group are 431.5, 1,100, and 2,711 pmol/ml for New York Heart Association classes II, III, and IV, respectively.

An initial serum NT–pro-BNP cut-off value of 1,256 pg/ml (95% CI 811 to 1,700) for prediction of death was identified by receiver operating characteristic curve analysis (area under the curve 0.747). A serum NT–pro-BNP concentration of 1,256 pg/ml showed 63.5% sensitivity, 77.5% specificity, and a negative predictive value of 88.5% for a fatal outcome. Cox regression modeling confirmed that baseline NT–pro-BNP was highly significantly associated with mortality in our PAH patient population (p = 0.010; Table 2 ) adjusting for gender, age, NYHA functional class, mRAP and baseline eGFR. The hazard ratio for 1-U increase in baseline log ₁₀ (NT–pro-BNP) was 1.92 (95% CI 1.01 to 3.64). Kaplan–Meier curves for survival in those with NT–pro-BNP levels >1,256 pg/ml and those with levels below this cutoff are shown in Figure 5 . Differences between the 2 groups were significant (p <0.0001, log-rank test).

Table 2

Cox regression models of single N-terminal pro–brain natriuretic peptide measurement

Baseline Measurement	Value	Hazard Ratio	SE	p Value
Female gender	0.82	2.27	0.71	0.2473
Age	0.04	1.04	0.01	0.0004
New York Heart Association class III	0.33	1.39	0.39	0.3980
New York Heart Association class IV	1.09	2.97	0.46	0.0189
Mean right atrial pressure	0.02	1.02	0.02	0.4062
Estimated glomerular filtration rate	−4.01	0.02	1.87	0.0324
Log 10 (baseline N-terminal pro–brain natriuretic peptide)	0.65	1.92	0.33	0.0469

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. Pulmonary Function in Patients With Acute Coronary Syndrome Treated With Ticagrelor or Clopidogrel (from the Platelet Inhibition and Patient Outcomes [PLATO] Pulmonary Function Substudy)
2. Predictors of Response to Cardiac Resynchronization Therapy in Patients With a Non-Left Bundle Branch Block Morphology
3. Comparison of Coronary Microcirculation in Female Nurses After Day-Time Versus Night-Time Shifts
4. Threshold for the Upper Normal Limit of Indexed Epicardial Fat Volume: Derivation in a Healthy Population and Validation in an Outcome-Based Study

Stay updated, free articles. Join our Telegram channel

Join