Plasma N-terminal pro–B-type natriuretic peptide (NT–pro-BNP) levels improve preoperative cardiac risk stratification in vascular surgery patients. However, single preoperative measurements of NT–pro-BNP cannot take into account the hemodynamic stress caused by anesthesia and surgery. Therefore, the aim of the present study was to assess the incremental predictive value of changes in NT–pro-BNP during the perioperative period for long-term cardiac mortality. Detailed cardiac histories, rest left ventricular echocardiography, and NT–pro-BNP levels were obtained in 144 patients before vascular surgery and before discharge. The study end point was the occurrence of cardiovascular death during a median follow-up period of 13 months (interquartile range 5 to 20). Preoperatively, the median NT–pro-BNP level in the study population was 314 pg/ml (interquartile range 136 to 1,351), which increased to a median level of 1,505 pg/ml (interquartile range 404 to 6,453) before discharge. During the follow-up period, 29 patients (20%) died, 27 (93%) from cardiovascular causes. The median difference in NT–pro-BNP in the survivors was 665 pg/ml, compared to 5,336 pg/ml in the patients who died (p = 0.01). Multivariate Cox regression analyses, adjusted for cardiac history and cardiovascular risk factors (age, angina pectoris, myocardial infarction, stroke, diabetes mellitus, renal dysfunction, body mass index, type of surgery and the left ventricular ejection fraction), demonstrated that the difference in NT–pro-BNP level between pre- and postoperative measurement was the strongest independent predictor of cardiac outcome (hazard ratio 3.06, 95% confidence interval 1.36 to 6.91). In conclusion, the change in NT–pro-BNP, indicated by repeated measurements before surgery and before discharge is the strongest predictor of cardiac outcomes in patients who undergo vascular surgery.
Patients who undergo vascular surgery are at high risk for peri- and postoperative cardiac events due to underlying coronary artery disease. N-terminal pro–B-type natriuretic peptide (NT–pro-BNP) levels improve preoperative cardiac risk stratification for surgical patients. NT–pro-BNP is a cardiac neurohormone that is synthesized in the ventricular myocardium and is released in response to ventricular wall stretching and myocardial ischemia. However, a single preoperative measurement of NT–pro-BNP cannot reflect the hemodynamic changes caused by anesthesia and surgical stress. However, they might in fact be the consequence hemodynamic instabilities during the perioperative period, with subsequent episodes of prolonged subclinical myocardial ischemia, associated with adverse long-term cardiac outcomes. Previous studies have demonstrated that changes in NT–pro-BNP over brief periods are related to adverse outcomes in acute coronary syndromes and acute decompensated heart failure. However, data on the use of repeated perioperative NT–pro-BNP measurements in vascular surgery patients are lacking. In the present study, we evaluated the incremental predictive value of changes in NT–pro-BNP during the perioperative period for long-term cardiac mortality in patients who underwent vascular surgery.
Methods
The study population consisted of patients who underwent elective vascular surgery at the Erasmus Medical Center (Rotterdam, The Netherlands) from 2007 to 2010. Patients were identified in a prospectively maintained database including all patients who underwent vascular surgery at this institution. The medical ethics committee of the hospital was informed about the study, and all procedures of this retrospective study met the approval of the medical ethics committee of the Erasmus Medical Center.
Before surgery, a detailed cardiac history was obtained from each patient, including angina pectoris, myocardial infarction, percutaneous coronary intervention or coronary artery bypass grafting, heart failure (defined as the presence of heart failure symptoms according the New York Heart Association classification or previous hospital admission for decompensated heart failure), and stroke or transient ischemic attack. Furthermore, cardiovascular risk factors were recorded, and included age, smoking history, hypertension (blood pressure ≥140/90 mm Hg or medical therapy to control hypertension), diabetes mellitus (fasting glucose level ≥7.0 mmol/L or medication to control diabetes), hypercholesterolemia (plasma cholesterol level ≥5.5 mmol/L or treatment with lipid-lowering drugs), and renal dysfunction (defined as serum creatinine >2 mg/dl). Other data collected included history of chronic obstructive pulmonary disease (defined as a forced expiratory volume in 1 second <70% of age- and gender-predicted value), site of surgery (abdominal aortic, lower extremity, or carotid), and type of procedure (endovascular or open).
Finally, the use of the following medications was recorded at baseline and at the time of discharge: aspirin, statins, β-blocking agents, calcium antagonists, diuretics, angiotensin-converting enzyme inhibitors, and angiotensin receptor blockers. Treatment goals were defined according to current guidelines for patients with peripheral arterial disease and included low-dose aspirin (80 mg/day), statins, β-blocking agents (titrated to a perioperative heart rate of 50 to 70 beats/min) in patients with or at risk for ischemic heart disease, and angiotensin-converting enzyme inhibitors in patients with left ventricular (LV) ejection fractions (LVEFs) <40%.
Peripheral venous blood samples were obtained for measurement of NT–pro-BNP levels in all patients during the preoperative outpatient clinic or at hospital admission and before discharge. NT–pro-BNP concentration was determined using an electrochemiluminescence assay on an Elecsys (Hoffman-La Roche, Basel, Switzerland). The method is a “sandwich”-type quantitative immunoassay, based on polyclonal antibodies against epitopes in the N-terminal part of pro-BNP. The lower detection limit was 5 pg/ml. Intra-assay coefficients of variance at 271 and 6,436 pg/ml were 1.9% and 0.9%, respectively. Assays were performed by a laboratory technician blinded to the patients’ clinical data. Importantly, pre- and postoperative NT–pro-BNP levels were unknown for the treating physician and were not used for clinical management.
Preoperatively, patients underwent 2-dimensional transthoracic echocardiography during rest. Cardiac evaluation was performed using a portable Acuson Cypress ultrasound system (Siemens Medical Solutions USA, Inc., Mountain View, California) with a 3V2C transducer (3.0, 3.5, 2.5, and 2.0 MHz). The LVEF was assessed in the apical, 4-chamber, or 2-chamber view with the patient in left lateral decubitus position. Quantification of LV volumes was performed using the modified Simpson’s rule, with inter- and intraobserver variability of 9% to 12% and 6%, respectively. The LVEF was calculated as (LV end-systolic volume − LV end-diastolic volume) × 100/LV end-diastolic volume. LV dysfunction was defined as impaired LV systolic function with a LVEF <40%. Of note, all echocardiographic studies performed were for research purposes and were not used for clinical management.
The end point of this study was the occurrence of cardiovascular death, defined as any death with a cardiovascular cause, including those deaths after cardiac procedures, cardiac arrest, myocardial infarction, pulmonary embolus, stroke, or sudden deaths not ascribed to other causes. Mortality was considered cardiovascular unless explicit proof of a noncardiac cause could be delivered. Long-term mortality was assessed by approaching the municipal civil registries.
Dichotomous data are described as numbers and percentages. The continuous variables age and body mass index are described as mean ± SD. Continuous data with a significant skewed distribution were compared using the Mann-Whitney U test and are expressed as median (interquartile range [IQR]). Receiver-operating characteristic curve analysis was used to assess the optimal cut-off value of NT–pro-BNP for the prediction of long-term cardiac mortality. The optimal value of preoperative NT–pro-BNP for predicting long-term cardiac mortality was defined as the concentration with the largest sum of sensitivity plus specificity. Changes in plasma NT–pro-BNP levels from preoperatively until the first 30 postoperative days were calculated as median differences with their interquartile range. We applied multivariate Cox regression analyses to evaluate the relation between plasma levels of NT–pro-BNP and the subsequent changes in relation to the study end point. In the regression analyses, NT–pro-BNP was entered as a continuous dependent variable and was log-transformed to obtain normality. Multivariate regression analyses were adjusted for cardiac risk factors and factors recognized to influence NT–pro-BNP levels : age, angina pectoris, myocardial infarction, stroke, diabetes mellitus, renal dysfunction, body mass index, type of surgery, and the LVEF. Interactions between renal dysfunction, intraoperative fluids administered, medication at discharge (diuretics, angiotensin-converting enzyme inhibitors, and angiotensin receptor blockers) and the change in NT–pro-BNP levels with the study end point were evaluated by forcing these interaction terms in the multivariate regression model. Interaction terms were included in the multivariate regression model only when significant. We report crude and adjusted hazard ratios and their 95% confidence intervals. For all tests, p values <0.05 (2 sided) were considered significant. All analyses were performed using SPSS version 15.0 (SPSS, Inc., Chicago, Illinois).
Results
The study population consisted of 144 patients with peripheral arterial disease referred for elective noncardiac vascular surgery. Most of the patients underwent abdominal aortic surgery (n = 74 [51%]). Baseline characteristics of the total study population are listed in Table 1 . The mean age of the patients was 68 ± 10 years, and 69% were men. Almost half of the patients (41%) had histories of myocardial infarction. Diabetes mellitus and renal dysfunction were present in 28% and 24% of patients, respectively. Hypertension and hypercholesterolemia were the most frequent cardiovascular risk factors and were observed in 69% and 52% of the patients, respectively.
| Variable | Value |
|---|---|
| Demographics | |
| Age (years) | 68 ± 10 |
| Men | 99 (69%) |
| Angina pectoris | 32 (22%) |
| Myocardial infarction | 59 (41%) |
| Coronary revascularization | 41 (29%) |
| Heart failure | 24 (17%) |
| Stroke | 38 (26%) |
| Smokers | 52 (26%) |
| Hypertension ⁎ | 100 (69%) |
| Diabetes mellitus | 40 (28%) |
| Hypercholesterolemia † | 75 (52%) |
| Renal dysfunction | 35 (24%) |
| Chronic obstructive pulmonary disease | 26 (18%) |
| Site and type of surgery | |
| Abdominal aortic | 74 (51%) |
| Lower extremity | 56 (39%) |
| Carotid | 14 (10%) |
| Open surgery ‡ | 106 (74%) |
| Measurements | |
| Body mass index (kg/m 2 ) | 26 ± 3.8 |
| LVEF (%) | |
| <30 | 10 (7%) |
| 30–40 | 30 (21%) |
| 40–50 | 59 (41%) |
| >50 | 45 (31%) |
| Serum creatinine (μmol/L) | 90 (72–113) |
| NT–pro-BNP (pg/ml) | |
| Preoperative | 314 (136–1,351) |
| Postoperative | 1,505 (404–6,453) |
⁎ Blood pressure ≥140/90 mm Hg or medical therapy to control hypertension.
† Plasma cholesterol ≥5.5 mmol/L or treatment with lipid-lowering drugs
At baseline, 99 patients (69%) had LVEFs <40%. Symptomatic heart failure was based on the presence of signs and symptoms according to New York Heart Association classification and was diagnosed in 24 patients (17%) of the total study population. Of these patients, 17 (71%) had systolic or combined heart failure, while 7 (29%) were diagnosed with heart failure with preserved ejection fractions. In the remaining 75 patients (52%) with LVEFs <40%, but without clinical symptoms of heart failure, systolic or combined LV dysfunction was observed in 40 patients (53%), and 35 patients (47%) had isolated diastolic dysfunction. The high prevalence of LV dysfunction was also reflected by the use of diuretics and angiotensin-converting enzyme inhibitors, which were prescribed in 1/3 of these patients. Furthermore, cardioprotective medications in the population, such as aspirin, statins, and β blockers, were prescribed in ≥2 of 3 patients. With the exception of angiotensin receptor blockers (p = 0.02), no significant differences were found between medications at screening and at discharge ( Table 2 ).
| Medication | At Screening | At Discharge | p Value |
|---|---|---|---|
| Aspirin | 98 (68%) | 95 (66%) | NS |
| Statins | 110 (77%) | 115 (80%) | NS |
| β-blocking agents | 137 (95%) | 141 (98%) | NS |
| Diuretics | 43 (30%) | 43 (30%) | NS |
| Angiotensin-converting enzyme inhibitors | 47 (33%) | 48 (33%) | NS |
| Calcium antagonists | 29 (20%) | 27 (19%) | NS |
| Angiotensin receptor blockers | 17 (12%) | 35 (24%) | 0.002 |
The median preoperative NT–pro-BNP level in our high-risk vascular surgery population was 314 pg/ml (IQR 136 to 1,351). During the first 30 postoperative days, repeated measurements of NT–pro-BNP were performed, and the median postoperative NT–pro-BNP level increased to 1,505 pg/ml (IQR 404 to 6,453). The median difference for the total population between pre- and postoperative NT–pro-BNP levels was 969 pg/ml (IQR 139 to 4,337). In all patients, the first postoperative NT–pro-BNP measurement was performed before hospital discharge. The mean length of hospital stay was 6.8 ± 3.1 days. After a median follow-up period of 13 months (IQR 5 to 20), the mortality end point was reached in 29 patients (20%), 27 (93%) of whom died secondary to cardiovascular causes. The association between preoperative NT–pro-BNP level and long-term cardiovascular mortality was assessed using a receiver-operating characteristic curve ( Figure 1 ). For preoperative NT–pro-BNP, the area under the curve was 0.668 (95% confidence interval 0.619 to 0.716), and the optimum discriminate threshold was 350 pg/ml. The changes in NT–pro-BNP levels between the pre- and postoperative periods were compared between survivors and patients who died ( Figure 2 ). The median preoperative NT–pro-BNP level in patients who died during the follow-up period was significantly higher compared to the survivors (795 vs 269 pg/ml, p = 0.002). In addition, the median difference between pre- and postoperative NT–pro-BNP level was significantly higher in patients who died compared to the survivors (5,336 vs 665 pg/ml, p = 0.010). We found no significant interaction between renal dysfunction, intraoperative fluids administered, medications at discharge, and the change in NT–pro-BNP levels with respect to the study end point. Using multivariate Cox regression analyses with adjustment for demographics and cardiac risk factors, preoperative NT–pro-BNP level as a log-transformed variable was an independent predictor of long-term cardiac mortality (hazard ratio 2.57 95% confidence interval 1.16 to 5.70). Importantly, the change in NT–pro-BNP level between pre- and postoperative measurements was the strongest independent predictor of cardiac outcome (hazard ratio 3.06 95% confidence interval 1.36 to 6.91; Table 3 ).
