B-Type Natriuretic Peptide and Impedance Cardiography at the Time of Routine Echocardiography Predict Subsequent Heart Failure Events

Journal of Cardiac Failure Vol. 15 No. 1 2009

B-Type Natriuretic Peptide and Impedance Cardiography at the Time of Routine Echocardiography Predict Subsequent Heart Failure Events LUIS R. CASTELLANOS, MD, MPH,1,2 VIKAS BHALLA, MD,1 SUSAN ISAKSON, BS,1 LORI B. DANIELS, MD, MAS, FACC,1,3 MEENAKSHI A. BHALLA, MD,1 JEANNETTE P. LIN, MD,1,3 PAUL CLOPTON, MS,1 NANCY GARDETTO, NP MSN,1 MAX HOSHINO, BS,1 ALBERT CHIU, BS,1 ROBERT FITZGERALD, PhD,1 AND ALAN S. MAISEL, MD, FACC1,3 La Jolla, California; Sacremento, California; San Diego, California

ABSTRACT Background: Detection of heart failure (HF) in stable outpatients can be difficult until an overt event occurs. This study sought to determine whether the combination of B-type natriuretic peptide (BNP) and impedance cardiography (ICG) could be used in a nonacute clinical setting to risk stratify and predict HF-related events in stable outpatients. Methods and Results: Patients undergoing routine outpatient echocardiography underwent ICG and BNP testing and were followed for one year for HF-related events (Emergency Department [ED] visit or hospitalization due to HF or all-cause death). A total of 524 patients were analyzed, resulting in 57 HF-related events; 16 ED visits, 17 hospitalizations, and 24 all-cause deaths. Using Cox regression analyses, BNP and systolic time ratio index (STRI) by ICG proved to be the strongest predictors of future HF-related events. Patients with a BNP O100 pg/ml and STRI O0.45 sec-1 had a significantly lower event-free survival rate than those with a high BNP and low STRI (67% versus 89%, P5.001). In patients with LV dysfunction only, if both BNP and STRI values were high, the relative risk of a HF-related event increased by 12.5 (95 % C.I. 4.2-36.7), when compared with patients with a low BNP and low STRI (P!.001). Conclusions: In a nonacute clinical setting, both BNP and ICG testing can provide unique predictive power of long-term HF-related events in a stable cohort of patients with and without LV dysfunction. (J Cardiac Fail 2009;15:41e47) Key Words: B-type natriuretic peptide, impedance cardiography, ventricular dysfunction, echocardiography.

Heart failure (HF) is the most frequent cause of hospitalization among elderly people, with 900,000 hospitalizations and approximately 250,000 deaths each year.1 Prompt detection and risk stratification of patients with left ventricular (LV) dysfunction can allow early initiation of therapies that have been proven to reduce mortality, improve quality of life, and decrease the costs associated with the development of complications. Although echocardiography is the preferred method for the detection of LV dysfunction, its

routine use for diagnostic and prognostic evaluation is limited due to its cost, availability and complexity.2 B-type natriuretic peptide (BNP) is a cardiac neurohormone that is released from the ventricles in response to LV volume expansion and pressure overload.3 BNP levels are known to be elevated in patients with LV dysfunction and correlate with echocardiographic findings, New York Heart Association (NYHA) class and severity of heart failure.4,5 Impedance cardiography (ICG) is a noninvasive method to determine hemodynamic parameters and electromechanical timing intervals.6,7 Measurements from ICG, such as systolic time intervals, have also been shown to correlate with echocardiographic findings of LV dysfunction8 and have been used to identify patients with low ejection fraction and changes in ejection fraction.9,10 Packer et al11 published a prospective evaluation of HF patients in which ICG parameters were used to successfully identify patients at high short-term risk for developing worsening heart failure. In patients presenting to the ED with dyspnea, the combination of BNP and ICG parameters can help

From the 1Division of Cardiology and the Department of Medicine Veteran’s Affairs San Diego Healthcare System, La Jolla, CA; 2Division of Cardiovascular Medicine, University of California, Davis Medical Center, Sacramento, CA and 3Division of Cardiology, University of California at San Diego, San Diego, California. Manuscript received January 22, 2008; revised manuscript received September 2, 2008; revised manuscript accepted September 11, 2008. Reprint requests: Alan S. Maisel, MD, VAMC Cardiology 111-A, 3350 La Jolla Village Drive, San Diego, CA 92161. Dr. Maisel received research grant support from CardioDynamics. 1071-9164/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.cardfail.2008.09.003

41

42 Journal of Cardiac Failure Vol. 15 No. 1 February 2009 physicians diagnose patients with decompensated heart failure.12 However, it is unknown if the combination of these parameters provide diagnostic and prognostic value in clinically stable patients. Therefore, we sought to determine whether the combination of BNP and ICG could be used to risk stratify and predict HF-related adverse events in a stable cohort of patients in a nonacute clinical setting, irrespective of the presence or absence of LV dysfunction.

LV dysfunction had at least one of the following: a) systolic dysfunction defined by an ejection fraction lower than 50% or any wall-motion abnormality, or b) diastolic dysfunction. Diastolic dysfunction was determined using Doppler measurements of mitral inflow and Doppler tissue imaging (DTI) of the mitral annulus, as previously described by Redfield et al.15 Subjects whose measurements were suggestive but not definitive for diastolic dysfunction were classified as indeterminate. Measurement of BNP Levels

Methods Study Cohort Sample The University of California San Diego Institutional Review Board approved this prospective observational study and all patients provided written, informed consent. The study population was a convenience sample of 550 patients from the San Diego Veteran’s Affairs (VA) Healthcare System referred for an outpatient echocardiogram between July, 2003, and April, 2005. Patient referrals were made by clinic physicians, attending physicians, or nurse practitioners. Outpatients who reported for an echocardiography examination for either routine follow up for established LV dysfunction or for a suspicion of new or worsening LV dysfunction were approached for the study when research coordinators were present (approximately 6 hours a day, five days a week). Patients were excluded from the study if echocardiography was performed to either rule out endocarditis, a cardioembolic source of a stroke, severe aortic regurgitation, unstable arrhythmias, atrial fibrillation or flutter with rapid ventricular response, and presence of minute ventilation permanent pacemakers in subjects. All measured parameters (ICG, echocardiography, and blood pressure) and blood draws were performed within 1 hour of each other. After processing the samples, data was lacking for 26 individuals, and they were excluded from the study. No patients were lost to follow up, and no patient asked to be removed from the study. HF-Related Events All patients were followed for HF-related events for one year after the initial BNP level, echocardiogram and ICG parameters were taken. For the sake of consistency with similar studies,11,13 an HF-related event was prospectively defined as the composite endpoint of an ED visit or hospitalization because of HF, or allcause death. Two trained, licensed physicians independently reviewed all medical notes, including ED visit forms, hospital admission notes, hospital discharge notes and billing codes for any ED visit as well as hospital admission that included HF as the primary diagnosis. When the primary diagnosis of HF was identified, it was verified by cross-referencing it with the attending physician’s progress note. If a patient had more than one HFrelated event, only the earliest event was counted. Deaths were adjudicated based on a documented physician’s death note. To eliminate the possibility of bias, the reviewing physicians were blinded to BNP, ICG parameters, and echocardiogram results. Echocardiography Classifications Echocardiograms were performed by certified sonographers and interpreted by a cardiologist who was blinded to BNP levels and ICG results. Left ventricular systolic and diastolic volumes and ejection fraction were derived from biplane apical views using the modified Simpson’s rule algorithm.14 Patients classified with

All samples were collected by venipuncture into EDTA tubes at the time of echocardiography. The blood samples were kept at room temperature and analyzed within 4 hours of the draw time. Before analysis, each tube was inverted several times to ensure homogeneity. The whole blood was then analyzed for BNP (Bayer, ADVIA CentaurÒ, Tarrytown, NY) as described previously.16 Measurement of ICG All ICG measurements were obtained with the BioZ ICG Monitor (CardioDynamics, San Diego, CA) as previously described.17,18 Four dual ICG sensors were placed on the patient: above the base of the neck and under each ear, and one on either side of the thorax in the mid-axillary line at the level of the xiphoid. A cable with eight ICG lead wires was attached to the individual sensor sites. An integrated oscillometric blood pressure cuff was attached to the patient’s left arm, and the patient was placed in the supine position. After confirmation of a visible ICG waveform on the ICG monitor screen, the patient was instructed to remain relaxed while ICG hemodynamic data were collected. After approximately 3 minutes, the operator printed an ICG status report. ICG variables evaluated in this study have been described previously and include measurements of cardiac blood flow, vascular resistance, fluid status, and indices of electromechanical timing and contractility.17 Systolic time ratio index (STRI) was defined as STR divided by R-to-R interval. Statistical Analyses Cox regression backward elimination analysis was performed to determine the association between independent variables and the composite end point of all-cause death, hospitalization or ED visit due to heart failure. The log values of continuous variables were used when the probability density curve did not represent a Gaussian distribution. For continuous variables, the nonparametric Mann-Whitney U test was used for comparisons, and values were expressed as medians with inter-quartile ranges. The chisquare test was used for nominal variables expressed as frequency counts. Additionally, Cox regression models were created to assess the joint ability of BNP levels and ICG parameters to predict adverse HF-related events over a period of 365 days. Receiver operator characteristic (ROC) curves were constructed for the predictors that emerged in the Cox regression models. The optimal cut point for each parameter was determined from the ROC curve as the round number cut point nearest the upper left corner (using Pythagoras’s Theorem). Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and likelihood ratios were calculated19 based on these cut points. We calculated odds ratios (OR) for continuous predictors and HR for group comparisons. Patient cohort groups were stratified according to

BNP and ICG Predict Heart Failure Events combinations of BNP and STRI; these groups were then compared using age-adjusted event survival curves.

Results Patients’ Characteristics

The cohort study consisted of 524 patients who presented to the outpatient echocardiography station for routine echocardiogram. B-type natriuretic peptide levels, echocardiographic and ICG parameters were measured in all patients, and descriptive statistics were obtained: 511 (97%) were male and 413 (79%) were White with an average age of 67, plus or minus 13 years. Additional population prevalence of co-morbid diseases, patient characteristics and medications in those with a HF-related event versus no event are shown in Table 1. At 365 days of follow-up, a total of 57 patients with HFrelated adverse events including 16 ED visits, 17 hospital admissions, and 24 all-cause deaths, were recorded. Patients who experienced a HF-related event had a higher prevalence of preexisting HF (49% vs. 22%), myocardial Table 1. Baseline Cohort Characteristics HF-Related Event Characteristics (Total Number of Patients 5 524)

Patients (%)

Age (years) 67 6 13 Gender (% male) 97 African American 10 White 79 Hispanic 6 Medical history Arrhythmia 25 AV stenosis 9 Angina (stable) 9 Coronary artery disease 39 CABG 15 COPD 14 Hypertension 67 PCI 8 Myocardial infarct 23 Dyslipidemia 57 Diabetes mellitus 32 Chronic kidney disease 11 Heart failure 25 Systolic dysfunction 20 Systolic and diastolic dysfunction 11 Medications ACE-I 49 Beta blockers 46 Calcium channel blockers 26 Antiarrhythmics 4 Diuretics 42 Digoxin 13 Angiotensin II receptor blockers 6 Anticoagulants 23 Antihyperlipidemics 46 Hypoglycemic agents 20 Nitrates 17

No (%)

Yes (%)

P Value

97 10 79 6

100 9 70 10

NS NS NS NS

23 8 1 38 13 14 66 8 21 57 31 10 22 18 11

40 12 4 51 25 14 75 5 35 54 32 18 49 35 28

.005 NS NS NS .022 NS NS NS .018 NS NS NS .001 .002 .002

48 45 26 3 40 12 6 21 45 19 16

54 54 26 9 58 14 7 35 54 21 19

NS NS NS .04 .01 NS NS .023 .021 NS NS

Patient values are expressed as mean 6 SD or by frequency (%). CABG, coronary artery bypasss graft; COPD, chronic obstructive pulmonary disease; HF, heart failure; PCI, percutaneous intervention; ACEI, angiotensin-converting enzyme inhibitor; AV, aortic valve.



Castellanos et al

43

infarct (35% vs. 21%), coronary artery bypass graft surgery (25% versus 13%) and arrhythmia (40% vs.23%), compared with those without HF-related events during the follow up (P!.02 for all). Review of the echocardiograms revealed that there was a higher prevalence of systolic dysfunction (35% versus 18%), diastolic dysfunction (70% vs. 52%) or both (28% vs. 11%) in the cohort of patients with a HF-related event, in comparison with those without an event (P ! .03 for all). Among medications, diuretics (58% vs. 40%), anti-coagulants (35% vs. 21%), and anti-hyperlipidemic agents (54% vs. 45%) were noted to be significantly more frequently prescribed in the group with HF-related events (P ! 0.03 for all comparisons). Demographic data, ICG parameters and serum markers for patients with a HF-related event versus no event are shown in Table 2. Serum markers and hemodynamic parameters are presented as medians with interquartile ranges. Of the demographic information, only heart rate was significantly higher in the HF-related event group (72 min1 vs. 68 min1P 5 .02). Patients with a subsequent HF-related event had higher absolute BNP level (135 pg/ ml vs. 53 pg/ml) and creatinine (1.3 mg/dl vs. 1.1 mg/dl) when compared with those without an event (P 5 .001 for both). Individuals with a HF-related event during the follow-up period had a lower stroke index (36 ml/m2 vs. 42 ml/m2), and lower LV ejection time (288 msec vs. 308 msec), and higher pre-ejection period (117 msec vs. 109 msec), STR (0.42 vs. 0.36) and STRI (0.55 sec1 vs. 0.39 sec1), (P !.008 for all). Cox Regression Analysis

ICG parameters, clinical variables and serum biomarkers were analyzed using Cox proportional hazards analysis across the 365-day follow up to determine the strongest independent predictors of a future HF-related event for the cohort of patients (Table 3). After backward elimination, only BNP and STRI remained as significant predictors of a HF-related event (P ! .004 for both). When ICG parameters were analyzed without BNP in the Cox model, STRI remained the only significant predictor of a HFrelated adverse event (P ! .001). Furthermore, after repeating the analysis in Table 3, but accounting for age, diabetes, beta blocker treatment and BMI (!25, 25e35, and O35) in the model, both BNP and STRI remained significant predictors (P ! .001 andP 5 .002, respectively), but none of the other variables achieved statistical significance. The optimal prognostic BNP and STRI cut-point values, 100 pg/ml and 0.45 sec1 respectively, were determined from receiver operating characteristic curves. Based on these values, four groups were formed (high or low BNP and high or low STRI). Cox models were then used to test for event-free survival differences among these groups. A significant difference was found among groups as illustrated in Figure 1. Pairwise contrasts among the groups showed that a combination of BNP O100 pg/ml and STRI O.45 sec1 resulted in a significantly lower event-free

44 Journal of Cardiac Failure Vol. 15 No. 1 February 2009 Table 2. Clinical and ICG Variables for the Cohort study No HF-Related Event

Demographic Age BMI BSA HR SBP DBP Impedance cardiography CI SV SI SVRI ACI VI TFC LCWI PEP LVET STR STRI Serum markers BNP Creatinine BUN

HF-Related Event

N

Mean 6 SD

N

Mean 6 SD

P Value

467 467 467 465 464 464

67 6 13 29 6 6 2.1 6 .23 68 6 12 138 6 20 77 6 11

57 57 57 57 57 57

69 6 11 28 6 6.1 2.1 6 .24 72 6 14 133 6 24 75 6 13

NS NS NS .02 NS NS

N 465 465 465 464 464 464 464 464 464 464 464 464

Median (Quartiles) 2.8 (2.4e3.1) 85 (71e99) 42 (36e48) 2532 (2170e3050) 63 (49e78) 35 (27e44) 30 (27e33) 3.3 (2.8e3.9) 109 (95e125) 308 (281e329) .36 (.30e.45) .39 (.30e.53)

N 57 57 57 57 56 57 57 57 57 57 57 57

Median (Quartiles) 2.6 (2.1e3.1) 78 (56e93) 36 (29e46) 2570 (2164e3362) 57 (48e81) 31 (23e44) 31 (28e34) 3.0 (2.4e3.6) 117 (101e145) 288 (258e317) .42 (.32e.64) .55 (.34e76)

P Value NS .008 .007 NS NS .04 NS .02 .002 .001 .001 .001

N 464 460 460

Median (Quartiles) 53 (20e122) 1.1 (0.9e1.3) 16 (12e21)

N 57 55 55

Median (Quartiles) 135 (64e346) 1.3 (1.0e1.5) 18 (12e25)

P Value .001 .001 NS

BMI, body mass index (kg/m2); BSA, body surface area (m2); HR, heart rate (min1); SBP, systolic blood pressure (mm Hg); DBP, diastolic blood pressure (mm Hg); CI, cardiac index (L/min/m2); SV, stroke volume (ml); SI, stroke index (ml/m2); SVRI, systemic vascular resistance index (dyne  sec  cm5  m2); ACI, acceleration index (1/100/sec2); VI, velocity index (1/1000/sec); TFC, thoracic fluid content (kOhm); LCWI, left cardiac work index (kg  m/m2); PEP, pre-ejection period (msec); LVET, left ventricular ejection time (msec); STR, systolic time ratio (unit less); STRI, STR*HR/60 (1/sec); BNP, B-type natriuretic peptide; BUN, blood urea nitrogen (mg/dL).

Predictive Properties of BNP, STRI, and Both

The sensitivity, specificity, PPV, NPV, accuracy, and likelihood ratios (LR) for BNP and STRI based on the optimal cut-points (100 pg/ml and .45 sec1 respectively) are shown Table 3. Cox Regression Analysis of Individual Predicting Variables (BNP and STRI in Log Form) for the Entire Cohort at 365 Days Variable

B

SE

Wald

Only ICG variables in the Cox model 3.03 0.693 19.2 Log10 STRI BNP and ICG variables in the Cox model 2.151 0.713 9.10 Log10 STRI 1.217 0.218 31.3 Log10 BNP

OR (95% CI)

P Value

20.7 (5.3e80.5)

! .001

8.59 (2.1e34.7) 3.38 (2.2e5.2)

.003 !.001

SE, standard error; OR, odds ratio; CI, confidence interval; ICG, impedance cardiography; STRI, systolic time ratio index; BNP, B-type natriuretic peptide.

in Table 5. At 365 days, BNP had a sensitivity of 67%, specificity of 69%, and NPV of 94%. Combining BNP with STRI increased the specificity to 88% with a NPV of 94% while increasing the positive LR to 4.0. 1.00

Cumulative Event Free Survival

survival than any other group difference (P 5 .001), compared to a BNP O100 pg/ml and STRI ! .45 sec1. In patients with preserved LV ejection fraction (EF $ 50%), 67% had a STRI ! .45 sec1 and 47.8% had a BNP level !100 pg/ml, Table 4. Conversely, among individuals with a depressed LV ejection fraction (EF ! 50%), 63% of patients had a STRI O .45 sec, 1 and 42.7% had a BNP level O100 pg/ml.

0.95 0.90

*

0.85 0.80 0.75 0.70

0.60 0.55

**

BNP<100 & STRI<0.45 BNP<100 & STRI>0.45 BNP>100 & STRI<0.45 BNP>100 & STRI>0.45

0.65

0

50

100

150

200

250

300

350

Days

Fig. 1. Age-adjusted event-free survival curves for the combinations of BNP (100 pg/ml) and STRI (.45 sec1) for the entire cohort. *P 5 NS for comparison with group BNP!100 pg/ml & STRI!.45 sec1 and BNP!100 pg/ml & STRIO.45 sec1 . **P 5 .001 for comparison with group BNPO100 pg/ml & STRI!.45 sec1; P !.001 for comparison with group BNP!100 pg/ml & STRIO.45 sec1 and BNP!100 pg/ml & STRI!.45 sec1

BNP and ICG Predict Heart Failure Events Table 4. Percentage of Patients with the Combinations of STRI (sec1) and BNP (pg/mL) With Preserved and Depressed LV Ejection Fraction

Discussion

EF ! 50%

EF $ 50%

20.3% 42.7%

22.4% 10.7%

14.4% 22.6%

47.8% 19.2%

STRI O 0.45 sec1 and BNP ! 100 pg/ml BNP O 100 pg/ml STRI ! 0.45 sec1 and BNP ! 100 pg/ml BNP O 100 pg/ml

45

Castellanos et al

12.5 times, (95% CI 4.2e36.7, P ! .001) when compared with patients with a low BNP and low STRI.

LV Function Variables



STRI, systolic time ratio index; BNP, B-type natriuretic peptide; LV, left ventricular; EF, ejection fraction.

Figure 1 shows age-adjusted event-free survival curves of the patient population stratified according to the four possible combinations of BNP and STRI at their optimal cut points. A clear separation existed between the event-free survival curves of high BNP plus high STRI and the other three combination curves. Patients with both high BNP and STRI, (BNP O 100 pg/ml and STRI O .45 sec, 1 respectively), had a significantly lower event-free survival than those with a high BNP and low STRI (67% vs. 89%,P 5 .001). There was no significant difference in survival when comparing the groups with low BNP plus low STRI vs. low BNP plus high STRI. In the cohort of patients with LV dysfunction only, individuals with a high BNP and high STRI had a lower eventfree survival rate than those with a high BNP and low STRI (60% vs. 86%, P 5 .002) Figure 2. In addition, a high BNP and low STRI had a significantly lower event-free survival rate compared to a low BNP and low STRI (86% vs. 96%, P 5 .044). Patients with a low BNP and low STRI had the highest event-free survival (96%) in both the group with LV dysfunction only and the entire cohort. Age-adjusted HR were generated to further quantify the association between BNP and STRI combinations and HF-related adverse events, Table 6. The risk of having a HF-related event was significantly increased for the group with high BNP and high STRI (HR 6.9 [95% CI 3.4e13.7], P ! .001). For patients with LV dysfunction only, a high BNP and low STRI more than tripled the risk of having a HF-related event when compared to patients with low BNP and low STRI. If both BNP and STRI values were high, the risk of having a HF-related adverse event increased

Natriuretic peptides have been recognized for their ability to rule out as well as assist in confirming the diagnosis of congestive HF in the acute setting.20e22 Although several groups have shown that using BNP alone or in combination with echocardiography helps identify high-risk patients,23,24 no large prospective studies have been completed that involve BNP and noninvasive hemodynamic parameters to risk stratify and predict HF outcomes in a stable cohort of patients, irrespective of LV function. Consequently, this study sought to answer the question of whether the combination of BNP along with ICG parameters allow the clinician to risk stratify and predict HF-related adverse events in a stable outpatient population regardless of the presence or absence of LV dysfunction. In this sample population, our results indicate that among many historical and clinical variables and noninvasive hemodynamic parameters, BNP and STRI are significant independent predictors of future HF-related adverse events at one year of follow-up. The prognostic ability of BNP was strengthened when combined with STRI. In the group of patients with an elevated BNP level, STRI was able to identify those individuals likely to have the lowest event-free survival rate. Patients having both a high BNP level and high STRI had the lowest event-free survival (67%) and were approximately 3 times more likely to have a HF-related adverse event when compared to those patients with a high BNP and low STRI. In the group with LV dysfunction only, individuals with a high BNP level and high STRI had the lowest event-free survival rate (60%) and were approximately 7 times more likely to have a HF-related event when compared with patients with a high BNP and low STRI. This suggests that STRI complemented rather than replicated the information provided by BNP to successfully identify patients at risk of developing a HF-related adverse event. ICG parameters have been previously used in dyspneic patients to aid in differential diagnosis and to help determine therapeutic options.25 Results from the Prospective Evaluation of Decompensation by Impedance Cardiography Test (PREDICT) trial demonstrated that ICG parameters (velocity index, thoracic fluid content index and LV ejection time)

Table 5. BNP and STRI at Optimal Cut-Point Values, 100 pg/mL and 0.45 sec1, Respectively, Used to Predict a HF-Related Event in the Cohort of Patients at 365 Days Marker(s) 0 to 365 days BNP (100) STRI (.45) BNP and STRI

Sensitivity (%)

Specificity (%)

Positive Predictive Value (%)

Negative Predictive Value (%)

Accuracy

LRþ

LR

67 (53e79) 58 (44e71) 47 (34e61)

69 (64e73) 64 (60e69) 88 (85e91)

21 (16e28) 17 (12e23) 33 (23e45)

94 (91e96) 93 (89e95) 94 (90e95)

68 64 84

2.2 1.6 4.0

0.48 0.65 0.60

BNP, B-type natriuretic peptide; STRI, systolic time ratio index; HF, heart failure; LR, likelihood error.

46 Journal of Cardiac Failure Vol. 15 No. 1 February 2009

Cumulative Event Free Survival

1.00 0.95 0.90

*

0.85 0.80 0.75 0.70 BNP<100 & STRI<0.45 BNP<100 & STRI>0.45 BNP>100 & STRI<0.45 BNP>100 & STRI>0.45

0.65 0.60 0.55

0

50

100

150

200

** 250

300

350

Days

Fig. 2. Age-adjusted event-free survival curves for the combinations of BNP (100 pg/ml) and STRI (.45 sec1) for the subgroup of patients with LV dysfunction only. *P 5 .044 for comparison with group BNP!100 pg/ml & STRI!.45 sec1 and P 5 NS for comparison with group BNP!100 pg/ml & STRIO.45 sec1. ** P 5 .002 for comparison with group BNPO100 pg/ ml & STRI!.45 sec1; P !.001 for comparison with group BNP!100 pg/ml & STRIO.45 sec1 and BNP!100 pg/ml & STRI! .45 sec1. BNP, B-type natriuretic peptide;

have independent predictive power to identify short-term (14-day) risk for adverse HF-related events among chronic HF patients.13 The results from our study not only support previously reported evidence that ICG parameters can be used to predict HF-related events,26 but more important, in prospective fashion, we showed that the ICG STRI, when combined with BNP, provides a significant increase in the long-term stratification of a stable cohort of patients at risk of having a HF-related adverse event. Based on this analysis and previous reports, one could postulate that in future studies these two modalities could be used together to further risk stratify patients and evaluate the efficacy of therapeutic interventions in the outpatient setting. Limitations

Several limitations to the clinical investigation presented exist. This study was done at a single VA hospital, with Table 6. Age-Adjusted HR and Confidence Intervals for Combinations of BNP and STRI When Compared to Low BNP (!100 pg/ml) and Low STRI (!.45 sec1) in the Entire Cohort and in Patients with LV Dysfunction Only Entire Cohort Variables

HR (95% CI)

Only LVD

P P Value HR (95% CI) Value

Low BNP and high 0.96 (0.36e2.55) NS 2.12 (0.57e7.92) NS STRI High BNP and low 1.97 (0.84e4.60) .12 3.39 (1.03e11.2) .04 STRI High BNP and high 6.85 (3.42e13.7) !.001 12.5 (4.24e36.7) !.001 STRI HR, hazard ratio; BNP, B-type natriuretic peptide; LVD, left ventricular dysfunction; CI, confidence interval.

a group of patients who were recruited at the Heart Station (echocardiography suite); thus, the results may be difficult to generalize to a broader population. Both the cohort characteristics and area under the receiver-operated curves are dependent on the relatively homogenous patient population that was selected. Our cohort sample represents generally older, White men and war veterans with a high prevalence of cardiac risk factors. The cohort included individuals with a prior history of heart disease, as well as those patients who underwent an initial cardiac evaluation. This explains in part the relatively low percentage of patients taking evidence-based medical therapy. In addition, HF-related events were identified and verified by direct evaluation of the patient’s medical records from the San Diego Veteran’s Affairs Healthcare System. Therefore, it is plausible that a number of HF-related adverse events might have been missed if a patient received acute medical care at a different hospital and was not transferred back to the San Diego VA hospital as specified by VA guidelines. In patients with LV dysfunction only, breakdown by ejection fraction categories (EF O50%, 50%e30%, and !30%) could not be performed as the number of HF-related events became too small for statistical analysis in each subgroup. Finally, the number of events limited the number of predictors used in the survival analysis models. In a larger study with a greater number of events, more complex models could be tested to confirm our findings. Conclusions The results of this study suggest that BNP and ICG parameters, when used alone or in combination, are useful to risk stratify and prognosticate a HF-related event in stable patients with or without LV dysfunction. A BNP level below 100 pg/ml and STRI !.45 sec1 predicts a favorable event-free survival with a high NPV for a HF-related complication. The predictive property of both BNP and STRI increases in patients with LV dysfunction. Given the ease of use and point-of-care availability, these two modalities can facilitate more frequent noninvasive hemodynamic and neurohormonal monitoring of LV function in highrisk patients as a means to gauge disease status.

References 1. Krumholtz HM, Chen YT, Wan Y, Vaccarino V, Radford MJ, Horwitz RI. Predictors of readmission among elderly survivors of admission with heart failure. Am Heart J 2000;139:72e7. 2. Devereux RB, Liebson PR, Horan MJ. Recommendations concerning use of echocardiography in hypertension and general population research. Hypertension 1987;9(2 Pt 2):II97e104. 3. Chen HH, Burnett JC. The natriuretic peptides in the pathophysiology of congestive heart failure. Current Cardiology Reports 2000;3: 198e205. 4. Maisel AS, Koon J, Krishnaswamy P, Kazanegra R, Clopton P, Gardetto N, et al. Utility of B-natriuretic peptide as a rapid, pointof-care test for screening patients undergoing echocardiography to determine left ventricular dysfunction. Am Heart J 2001;141(3):367e74.

BNP and ICG Predict Heart Failure Events 5. Krishnaswamy P, Lubien E, Clopton P, Koon J, Kazanegra R, Wanner E, et al. Utility of B-type natriuretic peptide (BNP) in elucidating left ventricular dysfunction(systolic and diastolic) in patients with and without symptoms of congestive heart failure at a Veterans Hospital. Am J Med 2001;111:274e9. 6. Labadidi Z, Ehmke DA, Durnin RE, Leaverton PE, Lauer RM. The first derivative thoracic impedance cardiogram. Circulation 1970;41: 651e8. 7. Strobeck J, Silver M. Beyond the four quadrants: The critical and emerging role of impedance cardiography in heart failure. Congest Heart Fail 2004;10(2 Suppl 2):1e6. 8. Bhalla V, Isakson S, Bhalla MA, Lin JP, Clopton P, Gardetto N, et al. Diagnostic ability of B-type natriuretic peptide and impedance cardiography testing to identify left ventricular dysfunction in hypertensive patients. Am J Hypertens 2005;18:S73e81. 9. Thompson B, Drazner MH, Dries DL, Yancy CW. Is impedance cardiography-derived systolic time ratio a useful method to determine left ventricular systolic dysfunction in heart failure? J Card Fail 2004;10(suppl):S38. 10. Parrott C, Burnham K, Quale C, Lewis DL. Comparison of changes in ejection fraction to changes in impedance cardiography cardiac index and systolic time ratio. Congest Heart Fail 2004;10(2 Suppl. 2):11e3. 11. Packer M, Abraham WT, Mehra MR, Yancy CW, Lawless CE, Mitchell JE, et al. Utility of impedance cardiography for the identification of short-term risk of clinical decompensation in stable patients with chronic heart failure. J Am Coll Cardiol 2006;47:2245e52. 12. Barcarse E, Kazanegra R, Chen A, Chiu A, Clopton P, Maisel A. Combination of B-type natriuretic peptide levels and non-invasive hemodynamic parameters in diagnosing congestive heart failure in the emergency department. Congest Heart Fail 2004;10:171e6. 13. Abraham WT, Trupp RJ, Mehra MR, Yancy CW, Lawless CE, Mitchell JE, et al. A multivariate impedance cardiography composite score can predict the occurrence of major heart failure events: the PREDICT study. Circulation 2004;110(Suppl III):597. 14. Schiller NB, Acquatella H, Ports TA, Drew D, Goerke J, Ringertz H, et al. Left ventricular volume from paired biplane two-dimensional echocardiography. Circulation 1979;60:547e50. 15. Redfield MM, Jacobsen SJ, Burnett JC, Mahoney DW, Bailey KR, Rodeheffer RJ. Burden of systolic and diastolic ventricular dysfunction in the community: appreciating the scope of the heart failure epidemic. JAMA 2003;289:194e202.



Castellanos et al

47

16. Wu AH, Packer M, Smith A, Bijou R, Fink D, Mair J, et al. Analytical and clinical evaluation of the Bayer ADVIA Centaur automated Btype natriuretic peptide assay in patients with heart failure: a multisite study. Clin Chem 2004;50:867e73. 17. Yancy C, Abraham W. Noninvasive hemodynamic monitoring in heart failure: utilization of impedance cardiography. Congest Heart Fail 2003;9:241e50. 18. Greenberg BH, Hermann DD, Pranulis MF, Lazio L, Cloutier D. Reproducibility of impedance cardiography hemohynamic measures in clinical stable heart failure patients. Congest Heart Fail 2000;6: 74e80. 19. Dao Q, Krishnaswamy P, Kazanegra R, Harrison A, Amirnovin R, Lenert L, et al. Utility of B-type natriuretic peptide (BNP) in the diagnosis of heart failure in an urgent care setting. J Am Coll Cardiol 2001;37:379e85. 20. Maisel AS, Krishnaswamy P, Nowak RM, McCord J, Hollander JE, Duc P, et al. Breathing Not Properly Multinational Study Investigators. Rapid measurement of B-type natriuretic peptide in the emergency diagnosis of heart failure. N Engl J Med 2002;347:161e7. 21. Lainchbury JG, Campbell E, Frampton CM, Yandle TG, Nicholls MG, Richards AM. Brain natriuretic peptide and n-terminal brain natriuretic peptide in the diagnosis of heart failure in patients with acute shortness of breath. J AM Coll Cardiol 2003;42:728e35. 22. Januzzi JL, Camargo CA, Anwaruddin S, Baggish AL, Chen AA, Krauser DG, et al. The N-terminal Pro-BNP investigation of dyspnea in the emergency department (PRIDE) study. Am J Cardiol 2005;95: 948e54. 23. Steg PG, Joubin L, McCord J, Abraham WT, Hollander JE, Omland T, et al. B-type natriuretic peptide and echocardiographic determination of ejection fraction in the diagnosis of congestive heart failure in patients with acute dyspnea. Chest 2005;128:21e9. 24. Peacock F, Summers R, Emerman C. Emergent dyspnea impedance cardiography-aided assessment changes therapy: The ED IMPACT trial. J Card Fail 2004;10:S119. 25. Springfield C, Sebat F, Johnson D, Lengle S, Sebat C. Utility of impedance cardiography to determine cardiac vs. noncardiac cause of dyspnea in the emergency department. Congest Heart Fail 2004;10(2 Suppl. 2):14e6. 26. Zewail A, Broom C, Eastwood C, Scroggins N, Kar B, Wadia Y, et al. Use of systolic time ratio and B-type natriuretic peptide to predict mortality in patients with heart failure. J Card Fail 2003;9:S105.