Author + information
- ↵⁎Department of Clinical Biochemistry, East Kent Hospitals NHS Trust, Canterbury, Kent, United Kingdom CT1 3NG
We were interested in the data of Anwaruddin et al. (1) demonstrating that in a group of patients presenting with dyspnea, of whom approximately one-third had chronic kidney disease (CKD), there was a strong relationship between amino-terminal pro-brain natriuretic peptide (NT-proBNP) and glomerular filtration rate (GFR) (r = −0.55), which remained independently highly significant in their multiple regression analysis. Despite this, Anwaruddin et al. (1) conclude that the diagnostic performance of NT-proBNP for congestive heart failure (CHF) is unaffected by the concomitant presence of kidney disease.
We have presented comparative data on the effect of BNP and NT-proBNP in a cohort of patients with CKD with a range of GFR between 5 and 60 ml/min/1.73 m2but not yet treated with dialysis, the majority of whom did not have CHF, thereby providing an opportunity to focus on the effect of diminishing GFR (2). The relationship between GFR and NT-proBNP was similar to that described by Anwaruddin et al. (rs= −0.53), whereas that for BNP was less strong (rs= −0.36). Anwaruddin and colleagues also report a weaker relationship between GFR and BNP (r = −0.18), although their BNP data were limited to those patients known to have CHF. Using a multiple regression approach we quantified the relationship between GFR and natriuretic peptide concentrations: for each 10 ml/min/1.73 m2decline in GFR, a 21% increase in BNP could be anticipated, compared to a 38% increase in NT-proBNP. We concluded that the effect of declining GFR on natriuretic peptide concentration was greater for NT-proBNP than for BNP. This is explicable from an understanding of the basic biology of these co-secreted peptides, with BNP having several known pre-renal clearance mechanisms in addition to renal elimination, whereas NT-proBNP is believed to be cleared by glomerular filtration alone.
How could 2 such similar datasets arrive at opposite conclusions? Anwaruddin et al. used NT-proBNP in a “rule-in” mode for heart failure, whereas in many studies and practical health service applications, lower thresholds have been proposed to enable use of the test in rule-out mode—a negative test result suggesting further investigation for heart failure (e.g., echocardiography) is probably unwarranted (3). The diagnostic thresholds they have selected (>450 pg/ml at <50 years and >900 pg/ml at >50 years) are therefore higher than the manufacturer’s usual decision thresholds. Despite this, however, 32% of patients with CKD (GFR <60 ml/min/1.73 m2) and no CHF had NT-proBNP concentrations in excess of these thresholds. An upward revision of the cut-point for CKD patients to 1200 pg/ml slightly reduced this nonspecificity such that 28% of CKD patients with NT-proBNP concentrations >1,200 pg/ml did not have CHF: however, this would generally not be considered acceptable performance for a diagnostic test used in clinical decision making. We would expect poor performance if NT-proBNP were to be used for rule-out decisions in patients with reduced GFR.
It is unfortunate that the investigators (1) did not undertake BNP testing in their entire cohort so that true comparative diagnostic performance could be evaluated. On the basis of their data, we would disagree with their conclusion that NT-proBNP is useful for diagnosing CHF across a wide range of renal functions.
↵1 Please note: Professor Christopher P. Price is a former employee of Bayer Diagnostics Plc.
The research cited in this letter was partly funded by Bayer Diagnostics Plc.
- American College of Cardiology Foundation
- Anwaruddin S.,
- Lloyd-Jones D.M.,
- Baggish A.,
- et al.
- Hobbs F.D.R.,
- Davis R.C.,
- Roalfe A.K.,
- Hare R.,
- Davies M.K.,
- Kenkre J.E.