Increasing natriuretic peptide (NP) levels are associated with worse heart failure (HF) outcomes. Predictors of NP nonresponse have not been studied. The aim of this study was to identify predictors of nondecreasing NP levels during episodes of acute HF. A retrospective analysis was conducted in patients prospectively included in a registry of acute HF, with the primary diagnosis of acute HF. The objective under analysis was B-type NP (BNP) response, defined as a >30% decrease in BNP during hospitalization. Percentage of BNP variation was calculated as: % BNP variation = [(admission BNP − discharge BNP)/admission BNP] × 100. A logistic regression analysis was performed to study potential predictors of NP nonresponse. A multivariate model was built. A total of 496 patients were studied: 28.2% were considered nonresponders to the implemented HF treatment strategy. Identified predictors of nonresponse were older age, chronic HF, lower admission systolic blood pressure, anemia, renal dysfunction, and lower sodium on admission, as well as lower admission albumin and lower admission total cholesterol. Admission BNP was not a predictor of response. The only identified independent predictor of nonresponse was a low admission total cholesterol level (cutoff 125 mg/dl), with an odds ratio of 2.55 (95% confidence interval 1.59 to 4.11). This remained valid when the analysis was stratified according to admission BNP (cutoff 2,000 pg/ml) and according to statin use. In conclusion, a low admission total cholesterol level was a strong and independent predictor of BNP nonresponse in patients admitted with acute HF. The ability of cholesterol to predict BNP nonresponse was valid for patients with intrinsically low cholesterol and in those with statin-induced low cholesterol.
Highlights
- •
Predictors of BNP nondecrease have not been studied.
- •
Low total cholesterol was the only independent predictor of BNP nonresponse.
- •
Low total cholesterol appears to be a marker of disease severity.
- •
It is possible that BNP has a cholesterol synthesis suppression effect.
- •
The importance of the concept of “natriuretostat” in BNP tailored therapy is addressed.
The use of plasma levels of natriuretic peptides (NP) to guide treatment of patients with heart failure (HF) has been investigated in a number of randomized controlled trials, but there is no consensus regarding the benefits of this approach. The Use of Peptides in Tailoring Heart Failure Project study was the first trial to emphasize the concept of “responder” and “nonresponder” patients. Responders were defined as patients whose NP levels decrease by >30% during a definite time period in response to the HF strategy in use. In a subgroup analysis, nonresponder patients had a worse prognosis. Knowledge of NP nonresponse predictors would allow the identification of patients at risk for worse outcome who might require more intensive treatment or novel therapeutic strategies. We aimed to characterize the profile of the nonresponder HF patients and to identify predictors of nondecreasing plasma B-type NP (BNP) levels in patients admitted for acute HF.
Methods
We conducted a retrospective analysis in patients prospectively included in a registry of acute HF in the internal medicine department of a tertiary care academic hospital. From January 2009 to December 2010, all patients admitted to our department with the primary diagnosis of HF were included in an acute HF registry. The diagnosis of HF was made according to the 2008 European Society of Cardiology guidelines. Patients with systolic dysfunction and those with HF with preserved ejection fraction were included. Echocardiography was performed in all eligible patients during hospitalization. Severe left ventricular systolic dysfunction corresponded to a left ventricular ejection fraction (LVEF) <30%, moderate dysfunction to an LVEF of 30% to 40%, and mild dysfunction to an LVEF of 40% to 50%. An LVEF >50% was considered to indicate normal systolic function.
Patients were excluded from the registry when acute coronary syndromes were the cause of acute HF and when symptoms were ultimately attributed to causes other than HF. Patients with no echocardiographic structural or functional abnormalities were also excluded from the registry. Treatment, timing of discharge, and discharge medications were at the discretion of the attending physicians. Physicians were aware of the ongoing HF registry.
As part of the registry, all patients were collected fasting venous blood samples <48 hours after admission and again on the day of discharge, if patients were discharged alive. BNP was measured by way of a chemiluminescent immunoassay using an Architect i2000 automated analyzer (Abbott Diagnostics, Santa Clara, California). Serum sodium, creatinine, urea, albumin, total cholesterol, and C-reactive protein were measured using conventional methods with an Olympus AU5400 automated clinical chemistry analyzer (Olympus, Center Valley, Pennsylvania). Hemoglobin was assessed using an automated blood counter.
Demographic characteristics, medications in use, and co-morbidities were recorded. Diabetes mellitus was defined as a known previous diagnosis, current prescription of either an oral hypoglycemic agent or insulin, a fasting venous blood glucose level >126 mg/dl, or a random glucose level >200 mg/dl. Arterial hypertension was defined as the presence of previous diagnosis, record of antihypertensive pharmacologic treatment, or blood pressure >140/90 mmHg. Anemia was considered if the hemoglobin level was <13 g/dl in men and <12 g/dl in women. Renal dysfunction was considered when plasma creatinine was >1.5 mg/dl. Hyponatremia was considered when serum sodium was <135 mEq/L.
To participate in this registry, all patients provided written informed consent, and the study protocol conformed to the ethical guidelines of the Declaration of Helsinki. Patients included in the registry were followed for 6 months; hospital registries and telephone contact with patients or their relatives were used to assess patients’ vital status. The objective under analysis was BNP response, defined as a >30% decrease in BNP during hospital admission. To further conduct this retrospective subanalysis, we excluded all patients who died in the hospital and hence had no discharge BNP measurements and no possible calculation of BNP variation.
Statistical Analysis
Percentage of BNP variation during admission was calculated as: % BNP variation = [(admission BNP − discharge BNP)/admission BNP] × 100. Responders were those patients in whom BNP decreased by >30%; the remaining patients were considered nonresponders. Responders and nonresponders were compared using chi-square tests for categorical variables and Student’s t tests and Mann-Whitney U tests for continuous variables normally and non-normally distributed, respectively. We calculated Spearman’s correlation coefficient between admission total cholesterol and BNP and between total cholesterol and BNP variation. Considering all-cause death as the event under analysis, a Cox regression analysis was performed to study the association of BNP nonresponse with prognosis. We built a multivariate model to assess the independent association of BNP response with better outcome. Patients were further cross-classified according to response (30% decrease during hospitalization) and admission BNP (cutoff 2,000 pg/ml) for additional prognostic information. Patients’ distribution was as follows: responders with low admission BNP, 204 patients; responders with high admission BNP, 152 patients; nonresponders with low admission BNP, 83 patients; and nonresponders with high admission BNP, 57 patients. The Kaplan-Meier method was used to display and compare survival curves according to the 4 groups created. A logistic regression analysis was performed to study potential predictors of NP nonresponse. A cutoff of 125 mg/dl was chosen for admission total cholesterol. A multivariate model was built. Analysis was then stratified according to admission BNP categories (cutoff 2,000 pg/ml) and also according to previous statin use. Interaction between statin use and admission cholesterol was formally tested.
Results
We studied 496 patients. Patients’ characteristics are listed in Table 1 . During hospitalization, BNP levels decreased by >30% in 356 patients (71.8%), and the reminder 140 (28.2%) were considered nonresponders to HF treatment. Responders and nonresponders were compared (see Table 1 ). Nonresponders were older and more often had chronic HF; they were also admitted with lower systolic blood pressure. They had lower admission sodium, albumin, and hemoglobin levels and higher creatinine levels. Admission total cholesterol levels were also lower in nonresponders. Groups were similar concerning previous medication, admission BNP, and length of hospital stay.
| Variable | All patients (496) | BNP responders (n=356) | BNP non-responders (n=140) | p-value |
|---|---|---|---|---|
| Men | 210 (42.3%) | 146 (41.0%) | 64 (45.7%) | 0.34 |
| Age (years), median (IQR) | 78 (72-84) | 77 (70-84) | 80 (74-86) | 0.005 |
| Hypertension | 369 (74.4%) | 264 (74.1%) | 105 (75.0%) | 0.78 |
| Diabetes mellitus | 254 (51.0%) | 180 (50.6%) | 74 (52.9%) | 0.64 |
| Atrial fibrillation | 224 (45.2%) | 154 (43.2%) | 70 (50.0%) | 0.11 |
| Chronic Heart Failure | 435 (87.7%) | 304 (85.4%) | 131 (93.6%) | 0.02 |
| Heart Failure hospitalization previous year | 116 (23.4%) | 76 (21.3%) | 40 (28.6%) | 0.09 |
| Ischemic aetiology | 203 (40.9%) | 145 (40.7%) | 58 (41.4%) | 0.89 |
| NYHA at admission IV (vs II/III) | 291 (58.7%) | 212 (59.6%) | 79 (56.4%) | 0.54 |
| Admission SBP (mmHg), median (IQR) | 132 (115-154) | 134 (117-156) | 124 (110-146) | 0.001 |
| LVSD (vs HFpEF) | 269 (54.2%) | 191 (53.6%) | 78 (55.7%) | 0.52 |
| Ejection fraction (%), median (IQR) | 40 (25-56) | 40 (26-56) | 40 (23-56) | 1.00 |
| Hemoglobin (g/dL), mean (SD) | 11.8 (2.2) | 12.0 (2.2) | 11.2 (1.9) | <0.001 |
| Sodium (mEq/L), median (IQR) | 139 (136-142) | 140 (137-142) | 138 (135-142) | <0.001 |
| Creatinine (mg/dL), median (IQR) | 1.33 (1.09-1.80) | 1.30 (1.08-1.69) | 1.50 (1.16-2.02) | 0.004 |
| C-reactive protein (mg/L) median (IQR) | 20.3 (9.5-55.1) | 20.5 (9.6-55.5) | 18.5 (9.3-54.6) | 0.74 |
| Troponin I (ng/mL), median (IQR) | 0.060 (0.030-0.143) | 0.060 (0.030-0.129) | 0.060 (0.030-0.188) | 0.25 |
| Albumin (g/L), mean (SD) | 35.4 (4.8) | 35.7 (4.8) | 34.7 (4.8) | 0.04 |
| BNP (pg/mL), median (IQR) | 1637.8 (915.0-2762.8) | 1743.4 (959.5-2785.8) | 1363.8 (796.9-2747.5) | 0.14 |
| Total cholesterol (mg/dL), median (IQR) | 150 (126-186) | 156 (131-194) | 130 (110-163) | <0.001 |
| Previous ACEi or ARB | 311 (62.7%) | 218 (61.2%) | 93 (66.4%) | 0.30 |
| Previous Beta blocker, | 248 (50.0%) | 183 (51.4%) | 65 (46.4%) | 0.30 |
| Previous Statin | 236 (47.6%) | 166 (46.6%) | 70 (50.0%) | 0.52 |
| Length of hospital stay (days), median (IQR) | 8 (6-12) | 8 (6-11) | 8 (6-13) | 0.15 |
| Deaths at 6 months | 93 (18.8) | 48 (13.5) | 45 (32.1) | <0.001 |
Six-month all-cause death occurred in 93 patients. Nonresponders had a multivariate-adjusted hazard ratio of 6-month death of 2.31 (95% confidence interval [CI] 1.48 to 3.61, p <0.001; adjustments to admission total cholesterol, hemoglobin, plasma creatinine, and albumin; admission BNP and admission systolic blood pressure; age and previous prognostic modifying therapy use). Nonresponse and higher admission BNP were the strongest independent predictors of worse outcome. The hazard ratio of death for admission BNP >2,000 pg/dl was 2.67 (95% CI 1.70 to 4.20). A nonsignificant trend toward higher all-cause mortality was seen in patients with admission cholesterol <125 mg/dl, with a hazard ratio of 1.42 (95% CI 0.92 to 2.22, p = 0.12).
When patients were cross-classified according to response and admission BNP, prognostic information was much increased. Among patients admitted with BNP levels <2,000 pg/ml, those who responded (n = 204) had a 7.8% death rate and those not responding a 19.3% death rate. Among patients with higher admission BNP levels, nonresponders had a 50.9% death rate and responders a 21.1% death rate. Figure 1 shows the Kaplan-Meier survival curves according to these groups.
Table 2 lists possible nonresponse predictors in a univariate approach. Admission BNP was not a predictor of response. The multivariate logistic regression model built for the identification of independent predictors of nonresponse in listed in Table 3 . The only identified independent predictor of nonresponse was a low admission total cholesterol level (cutoff 125 mg/dl).
| Variable | OR (95% Confidence Interval) | p-value |
|---|---|---|
| Male sex | 1.21 (0.82-1.80) | 0.34 |
| Age ≥ 75 years | 1.72 (1.12-2.65) | 0.01 |
| Hypertension | 0.94 (0.59-1.48) | 0.78 |
| Diabetes mellitus | 1.05 (0.86-1.27) | 0.64 |
| Atrial fibrillation | 1.38 (0.93-2.05) | 0.11 |
| Chronic Heart failure | 2.65 (1.16-6.04) | 0.02 |
| Heart Failure hospitalization previous year | 1.47 (0.94-2.30) | 0.09 |
| Ischemic aetiology | 1.03 (0.69-1.53) | 0.89 |
| NYHA at admission IV (vs II/III) | 0.88 (0.59-1.31) | 0.54 |
| Admission systolic blood pressure ≤ 115mmHg | 1.75 (1.12-2.74) | 0.01 |
| LVSD (vs HFpEF) | 1.14 (0.76-1.70) | 0.52 |
| Ejection fraction (per %) | 1.00 (0.99-1.01) | 0.92 |
| Anaemia | 2.13 (1.38-3.28) | 0.001 |
| Hyponatremia | 1.59 (0.97-2.63) | 0.07 |
| Renal dysfunction | 1.82 (1.22-2.71) | 0.003 |
| C-reactive protein (per mg/L) | 1.00 (1.00-1.00) | 0.85 |
| Troponin I (per ng/mL) | 1.00 (0.91-1.11) | 0.93 |
| Albumin (per g/L) | 0.96 (0.92-1.00) | 0.04 |
| B-type natriuretic peptide (per 100pg/mL) | 1.00 (0.99-1.01) | 0.61 |
| B-type natriuretic peptide >2000pg/mL | 0.92 (0.62-1.37) | 0.69 |
| Total cholesterol < 125mg/dL | 3.32 (2.15-5.13) | <0.001 |
| Previous ACEi or ARB | 0.67 (0.42-1.06) | 0.09 |
| Previous Beta blocker | 0.80 (0.50-1.25) | 0.32 |
| Previous Statin | 1.14 (0.77-1.68) | 0.52 |
| Length of hospital stay (per day) | 1.02 (0.99-1.05) | 0.22 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
