Author + information
- A. Mark Richards, MD⁎ ( )()
- ↵⁎Reprint requests and correspondence:
Dr. A. Mark Richards, Department of Medicine, Christchurch Hospital, Riccarton Avenue, P.O. Box 4345, Christchurch, New Zealand. OR Cardiovascular Research Institute, National University Heart Centre, Singapore, 1E Kent Ridge Road, NUHS Tower Block, Level 9 (Cardiac Department), NUHCS, 1192998 Singapore
Life-time risks of heart failure (HF) and/or cardiovascular death (CVD) in the elderly are high, and strategies for accurate risk stratification are needed. In this issue of the Journal, deFilippi et al. (1) report from the CHS (Cardiovascular Health Study) on an assessment of N-terminal pro–B-type natriuretic peptide (NT-proBNP) and echocardiographic left ventricular ejection fraction (LVEF) in screening for risk of subsequent new-onset HF and CVD in ambulant elderly patients. In established cardiac disease, both indicators are independent predictors of adverse cardiovascular outcomes and, in combination, improve risk stratification (2). In contrast, the authors conclude that, in this initially HF-free population, LVEF did not reliably add to risk stratification beyond information provided from NT-proBNP levels.
The CHS recruited 5,888 independent adults (≥65 years of age) from 4 communities. Those with known HF were excluded, and analyses were conducted on 4,137 participants with left ventricular systolic function by using echocardiography and assay of NT-proBNP. The population was divided into those with NT-proBNP levels ≥190 and <190 pg/ml. LVEF was divided into 2 groups (normal ≥55%; abnormal <55%). Because NT-proBNP exceeded 190 pg/ml in 29.5% and LVEF was abnormal in 7.7%, most elevated NT-proBNP values were not associated with an abnormal LVEF. Over 10 years of follow-up, 1,112 new-onset HF events and 893 cardiovascular-related deaths occurred.
When indexed to the group with both low-peptide and normal LVEF (n = 2,783; 67.3%), adjusted hazard ratios were progressively higher (1.75, 2.75, and 5.73 for HF and 1.68, 1.92, and 2.95 for CVD) in the low-peptide/low ejection fraction, high-peptide/normal ejection fraction, and high-peptide/low ejection fraction groups, respectively. Hazard ratios for both endpoints differed significantly between the 2 high-peptide groups but not between the 2 groups with peptide levels <190 pg/ml. Both C-statistic and net reclassification improvement (NRI) tests showed no important additive value in adding LVEF to NT-proBNP. The authors comment that 14 participants with elevated NT-proBNP levels would need to be screened to detect 1 patient with subnormal LVEF (<45%). The “take-home message” offered is “… once adjusting for the multiple comorbidities often present in ambulatory older patients, we were unable to demonstrate that an assessment of LVEF could further stratify prognosis after measurement of NT-proBNP. In the presence of an elevated NT-proBNP level in this population, a tailored approach to cardiac imaging appears most appropriate.”
What can we gain from the risk stratification documented in the current report? The sensitivity and predictive values of screening tests depend on the threshold values adopted. Using 190 pg/ml to define elevation of NT-proBNP and <55% to mark abnormal LVEF in the current study, about one-half of both endpoints occurred in patients who did not cross either threshold, so sensitivity is poor. In those in whom both markers were adverse, 57% incurred HF over 10 years compared with 21% if neither was positive. The corresponding figures for CVD were 19.1% and 7.9%. Is this useful stratification, guiding allocation of patients to more or less aggressive surveillance and pharmacotherapy? At progressively higher threshold values of NT-proBNP, individual risk may become high enough to confidently mandate early and aggressive investigation and treatment.
Lack of prognostic contribution from LVEF reflects the inability of an infrequent marker to provide much overall predictive value. Multivariable analysis incorporating NT-proBNP as a continuous variable and LVEF as a semi-quantitative variable did support an independent association of LVEF with outcomes. This finding suggests that given sufficiently fine-grained data with an objective percent LVEF calculated for each individual rather than use of broad semiquantitative categories, LVEF may have retained some independent predictive power. However, retention of significance in multivariable models does not equate to meaningful clinical value. Assessment of the additive value of a marker has been strengthened in recent years by the introduction of NRI analyses to the biostatistical armamentarium, and this tool has been put to good effect by deFilippi et al. (1) to illustrate convincingly how little LVEF adds overall to prognostication in the ambulant elderly (3). Uncommon abnormalities can carry potential prognostic significance in only a few people.
Self-evidently, LVEF cannot predict HF in the presence of a preserved ejection fraction (HF-PEF), which is common in HF occurring in the elderly (4). In this regard, it is notable that a high proportion of heart failure events in the report from deFilippi et al. (1) occurred in the presence of a normal LVEF.
This paper reflects an interesting maturation in our understanding of information offered by the circulating B-type cardiac peptides. Early clinical studies focused on the utility or otherwise of plasma B peptides in assisting diagnosis of acute decompensated heart failure and on the ability to detect the presence of a depressed LVEF (5,6). With successive reports confirming the relationship of B-type peptides to both short- and long-term all-cause mortality and cardiovascular outcomes in both clinic and community, the focus has shifted to defining the best markers to accurately predict outcomes in specific settings (7).
Despite undisputed power to independently stratify patients for risk of important adverse outcomes, knowledge of elevated NT-proBNP (or B-type natriuretic peptide) levels in the asymptomatic patient may yet leave us destitute of any useful response. This contrasts with our response to knowledge of LVEF because evidence indicates that treatment of asymptomatic reduced LVEF will improve outcomes (8). There is no analogous evidence connecting awareness of asymptomatic elevated plasma NT-proBNP/B-type natriuretic peptide to any specific therapeutic response known to alter outcome. Even if such elevations heighten ability to predict HF-PEF, we have no therapy proven to improve outlook (9). Furthermore, the association of NT-proBNP with depressed LVEF remains indisputable (1,2) and, despite the conclusion of the current paper, this association will inevitably encourage progression to echocardiography when an elevated NT-proBNP is found in older ambulant patients. The current report tells us this approach will have low yield in terms of finding depressed LVEF, but it also tells us that in 1 in 14 cases when NT-proBNP exceeds 190 pg/ml, we will miss a reduced LVEF (<45%) by not referring for echocardiography. What is the clinician to do? Advice to adopt a “tailored approach” is enigmatic and does not resolve this conundrum.
Echocardiography is noninvasive and informative, revealing details of structure and function that can never be elucidated by B-type peptide measurements alone. It would be helpful to know what proportion of CHS echocardiograms revealed any previously undiagnosed cardiac pathology (including wall motion abnormalities, left ventricular hypertrophy, hypertrophic cardiomyopathy, asymptomatic valve dysfunction, and elevations in pulmonary artery pressure) and their associations with NT-proBNP levels. A proportion of such findings would warrant surveillance and/or institution of treatment. Rather than focusing solely on the ability of echocardiography to provide an LVEF that turns out to add little beyond NT-proBNP in long-term prognostication for HF or death, it seems important to examine the full picture to assess the utility of NT-proBNP in alerting clinicians to any actionable cardiac abnormality. Further inspection of the CHS data may provide some insight into this issue. It is likely NT-proBNP–triggered echocardiography will reveal a high prevalence of diastolic dysfunction, but the physician may still be left with no clear indication for introducing any specific therapy. Given the accumulating evidence, it is possible we should conduct controlled therapeutic trials to assess the benefits or otherwise of introducing treatment to those with predefined combinations of deranged peptide levels and echocardiographic abnormalities currently associated with poorer prognosis but not yet validated as indications for intervention.
Dr. Richards has received speaker honoraria, travel grants, and research grants from Roche Diagnostics.
↵⁎ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
- American College of Cardiology Foundation
- deFilippi C.R.,
- Christenson R.H.,
- Kop W.J.,
- Gottdiener J.S.,
- Zhan M.,
- Seliger S.L.
- Richards A.M.,
- Nicholls M.G.,
- Espiner E.A.,
- et al.
- Januzzi J.L.,
- Van Kimmenade R.,
- Lainchbury J.,
- et al.
- Di Angelantonio E.,
- Chowdhury R.,
- Sarwar N.,
- et al.