Plasma levels of intermedin (adrenomedullin-2) in healthy human volunteers and patients with heart failure

Peptides 76 (2016) 19–29

Contents lists available at ScienceDirect

Peptides journal homepage: www.elsevier.com/locate/peptides

Plasma levels of intermedin (adrenomedullin-2) in healthy human volunteers and patients with heart failure David Bell a,∗ , Brian J. Gordon b , Anita Lavery c , Katie Megaw d , Michael O. Kinney e , Mark T. Harbinson a,e a

School of Medicine, Dentistry and Biomedical Sciences, The Queen’s University of Belfast, Northern Ireland, UK Peterborough City Hospital, Peterborough, England, UK c Hillingdon Hospitals NHS Trust, London, England, UK d Southeastern Health and Social Care Trust, Northern Ireland, UK e Belfast Health and Social Care Trust, Belfast City Hospital, Northern Ireland, UK b

a r t i c l e

i n f o

Article history: Received 18 September 2015 Received in revised form 23 November 2015 Accepted 15 December 2015 Available online 6 January 2016 Keywords: Intermedin/adrenomedullin-2 Adrenomedullin Heart failure Healthy subjects Human Renal function Cardiac resynchronization therapy Plasma levels

a b s t r a c t Intermedin/adrenomedullin-2 (IMD) is a member of the adrenomedullin/CGRP peptide family. Less is known about the distribution of IMD than for other family members within the mammalian cardiovascular system, particularly in humans. The aim was to evaluate plasma IMD levels in healthy subjects and patients with chronic heart failure. IMD and its precursor fragments, preproIMD25–56 and preproIMD57–92 , were measured by radioimmunoassay in 75 healthy subjects and levels of IMD were also compared to those of adrenomedullin (AM) and mid-region proadrenomedullin45–92 (MRproAM45–92 ) in 19 patients with systolic heart failure (LVEF < 45%). In healthy subjects, plasma levels (mean + SE) of IMD (6.3 + 0.6 pg ml−1 ) were lower than, but correlated with those of AM (25.8 + 1.8 pg ml−1 ; r = 0.49, p < 0.001). Plasma preproIMD25–56 (39.6 + 3.1 pg ml−1 ), preproIMD57–92 (25.9 + 3.8 pg ml−1 ) and MRproAM45–92 (200.2 + 6.7 pg ml−1 ) were greater than their respective bioactive peptides. IMD levels correlated positively with BMI but not age, and were elevated in heart failure (9.8 + 1.3 pg ml−1 , p < 0.05), similarly to MRproAM45–92 (329.5 + 41.9 pg ml−1 , p < 0.001) and AM (56.8 + 10.9 pg ml−1 , p < 0.01). IMD levels were greater in heart failure patients with concomitant renal impairment (11.3 + 1.8 pg ml−1 ) than those without (6.5 + 1.0 pg ml−1 ; p < 0.05). IMD and AM were greater in patients receiving submaximal compared with maximal heart failure drug therapy and were decreased after 6 months of cardiac resynchronization therapy. In conclusion, IMD is present in the plasma of healthy subjects less abundantly than AM, but is similarly correlated weakly with BMI. IMD levels are elevated in heart failure, especially with concomitant renal impairment, and tend to be reduced by high intensity drug or pacing therapy. © 2016 Elsevier Inc. All rights reserved.

1. Introduction There has been much recent interest in the measurement of biomarkers in cardiovascular disease [9]. In the field of heart failure, B type natriuretic peptide (BNP) and its precursor fragment N terminal pro B type natriuretic peptide (NT-proBNP) have become established diagnostic and prognostic markers, and their use has been encouraged in clinical guidelines [44,31]. A variety of alternative or additional markers have been proposed,

Abbreviations: CGRP, calcitonin gene-related peptide; CRT, cardiac resynchronization therapy; TFA, trifluroacetic acid. ∗ Corresponding author at: Therapeutics and Pharmacology, Centre for Medical Education, School of Medicine, Dentistry and Biomedical Sciences, QUB, Whitla Medical Building, 97 Lisburn Road, Belfast BT9 7BL, UK. Fax: +44 28 9097 2451. E-mail address: [email protected] (D. Bell). http://dx.doi.org/10.1016/j.peptides.2015.12.003 0196-9781/© 2016 Elsevier Inc. All rights reserved.

with previous interest in adrenomedullin (AM) [10,14,35,37] and now renewed focus on its precursor fragment, mid-region proadrenomedullin45–92 (MRproAM45–92 ) [20,8,26]. Intermedin (also known as adrenomedullin-2; IMD), a novel peptide independently discovered by 2 research groups in 2004 [38,41], is related to and shares a family of receptors with CGRP and AM [38,41,2]. Proteolytic processing of a larger prepro-IMD precursor yields a series of biologically active C-terminal peptides as well as the precursor fragments, preproIMD25–56 and preproIMD57–92 (Fig. 1). Exogenous IMD is a potent systemic vasodilator [32,49], influences regional blood flow and water-electrolyte homeostasis [43], augments cardiac contractility and protects against oxidative stress and ischaemic insult in rodents (reviewed in Ref. [2,15]). IMD is detected, but at lower levels than AM, in adult rodent cardiomyocytes [3–5]. Data in humans are scarce but in tissue obtained at autopsy from subjects without known cardiac or renal disease, IMD

20

D. Bell et al. / Peptides 76 (2016) 19–29

Fig. 1. Structure of the human IMD/AM2 prepro-peptide. Cleavage sites towards the C-terminus of the prepro-IMD molecule (indicated by arrows) give rise to 3 mature IMD/AM2 peptides (IMD1–53 , IMD1–47 ) (IMD long) and IMD8–47 (IMD short). The precursor fragment preproIMD25–92 is not biologically active and is further cleaved between Arg56-Pro57 to generate two fragments, preproIMD25–56 and preproIMD57–92 .

localized primarily to myocardium and to renal tubular cells [30]. IMD is also expressed in human ventricular cardiac fibroblasts, cardiomyocytes and cardiac microvascular and aortic endothelial cells [33,1,15] and in human renal mesangial and tubular cells (Bell and Metcalfe, unpublished observation) maintained in culture. At mRNA level, expression of IMD is negligible in leukocytes circulating in human blood obtained from healthy subjects but is upregulated markedly in heart failure [27]. Given the vasodilator, positively inotropic, natriuretic and diuretic actions [41,2,15] reported in laboratory studies so far, IMD may have an important role to play in the counter regulatory response to heart failure. Data in humans however are limited. Recently, a number of studies reported on augmentation of IMD levels in the plasma of patients following acute myocardial infarction, correlating with the extent of coronary stenosis and providing a potential prognostic marker for further adverse cardiovascular events [25,36,42,48]. Although such studies included carefully matched controls, relatively little is yet known about characteristics influencing plasma levels of IMD in healthy individuals, how IMD levels are affected by other disease states and clinical interventions, or relate to those of its own precursor fragments or other members of the CGRP/adrenomedullin peptide family. The aim therefore was to measure levels of IMD and the precursor fragments, preproIMD25–56 and preproIMD57–92 , in healthy control subjects, and explore anthropometric influences on these levels. IMD levels in healthy subjects were compared with those measured in a cohort of patients with stable systolic heart failure. Within the heart failure group, the influence of various parameters including intensity of drug therapy and cardiac resynchronization therapy (CRT) on plasma peptide levels was assessed. As comparator, IMD levels were compared with those of AM and mid-region proAM45–92 in healthy subjects, and in heart failure patients since these comparators have been found previously to be elevated in such patients [35,37,8]. 2. Methods 2.1. Subjects Two groups were studied. 75 healthy subjects aged 18 years and above were recruited from amongst our clinical and university colleagues. Subjects were excluded if they had cardiac symptoms, a known history of heart failure or any known cardiac disease, known risk factors for cardiac disease including diabetes and hypertension, or were taking any regular medication. 19 patients with heart failure were recruited from the Belfast Health and Social Care Trust heart failure service. All subjects had stable chronic heart failure, left ventricular systolic dysfunction (ejection fraction < 45%) and were receiving maximally tolerated medical therapy (including diuretics, angiotensin converting enzyme inhibitors or angiotensin receptor antagonists, aldosterone receptor antagonists and beta

adrenoceptor antagonists as tolerated). A subgroup of patients with heart failure and systolic dysfunction attending for cardiac resynchronization therapy (CRT) were studied within 24 h prior to (baseline) and for up to 12 months after this device-based intervention. Approval was obtained from the Queen’s University of Belfast School of Medicine, Dentistry and Biomedical Sciences Research Ethics Committee and Office of Research Ethics Committees for Northern Ireland (ORECNI). The study was performed in accordance with The Declaration of Helsinki and the Belfast Trust Research Governance Framework. All subjects provided informed written consent. Height and weight were obtained for each subject and body mass index (BMI) calculated. Clinical features, details of LVEF, intensity of medical therapy (those taking 4 or fewer standard therapies for heart failure as listed above) and biochemical parameters including renal function were recorded for the patients. NT-proBNP was determined using the commercially available and validated Elecsys proBNP assay (Roche Diagnostics Ltd.) and analysed on a Roche Modular Analyser (E module).

2.2. Peptide biomarker sampling and analysis Following venepuncture, 30 ml of blood was immediately collected into chilled bottles containing EDTA and the protease inhibitor aprotinin (0.6TIU (780KIU) ml−1 of blood, Apollo Scientific, UK). After immediate centrifugation at 3000 × g for 20 min at 4 ◦ C. the plasma was stored with aprotinin at −80 ◦ C pending analysis. Plasma was defrosted, acidified with an equal volume of trifluoroacetic acid (TFA, 1% v/v, Sigma–Aldrich, UK) and centrifuged at 12000 × g for 20 min at 4 ◦ C. The supernatant was applied onto pre-equilibrated C18 SEP columns (Waters Corporation, UK) eluted with 60% acetonitrile (Riedel de Haen, Germany) in 1% TFA and evaporated to dryness using a combination of centrifugal concentration (Hetero Vac, Scandinavia) and lyophilisation (Edwards, Modalyo). Reconstituted samples were assayed using commercially available radio-immunoassays (RIA; Phoenix Pharmaceuticals Inc., California, USA). Human AM, human MRproAM45–92 , and human IMD were assayed in all subjects; human preproIMD25–56 and human preproIMD57–92 were additionally measured in healthy control subjects. Each RIA Exhibits 100% cross-reactivity with its respective peptide antigen and no crossreactivity with any of the other peptides measured in this study. The AM RIA detects total AM (both glycated and mature forms) and also displays limited cross-reactivity with shorter fragments such as AM26–52 (12% cross-reactivity) which represent degradation products of the bioactive peptide. The IMD RIA Exhibits 100% cross reactivity with the 53, 47 and 40 amino acid versions of the human peptide. IMD was measured down to 1 pg ml−1 by the use of serial concentration steps. For each of the peptide assays, coefficients of intra-and inter-assay variation were <5% and 10%,

D. Bell et al. / Peptides 76 (2016) 19–29

respectively. Recovery of standard peptides from the columns was typically >95%.

[a]

21

80 70

2.3. Statistical analysis

AM [pg.ml -1]

60

Data were analysed with the assistance of the Department of Statistics, Centre for Public Health, Queen’s University Belfast. Data analysis and graphical representation were carried out using SSPS (Version 15.0, SSPS Inc., Chicago, IL) and GraphPad Prism software (Version 4.0, GraphPad Software Inc., San Diego, CA). Data are presented as mean values + SE. Statistical analyses between groups were performed by an unpaired Student’s t test or ANOVA for multiple points. Linear regression analysis was used to examine relationships between two variables. Values of p < 0.05 were considered significant.

50 40

r= 0.494 p< 0.001

30 20 10 0

3. Results

0

5

10

3.1. Healthy subjects

3.2. Heart failure population The characteristics of the heart failure patients are given in Table 1. All patients were in NYHA Class II or III and had left ventricular ejection fraction <45%; 13/19 patients had moderate renal impairment (GFR <60 ml/min/1.73 m2 ). Plasma levels of IMD (9.8 + 1.3 v 6.3 + 0.6 pg ml−1 ; p < 0.05), AM (56.8 + 10.9 v 25.8 + 1.8 pg ml−1 ; p < 0.01), and MRproAM45–92 (329.5 + 41.9 v 200.2 + 6.7 pg ml−1 ; p < 0.01), were each greater in the heart failure group than respective values obtained in the healthy control population (Table 2). Receiver operating characteristic (ROC) analysis generated AUCs in the region 0.65–0.70 (statistically nonsignificant, data not shown). Levels of all 3 peptides were less significantly elevated in the subgroup of heart failure patients receiving high dose therapy (all 4 of the classes: - beta adrenoceptor antagonists, diuretics, ACE inhibitors or AT1 receptor antagonists, and aldosterone receptor antagonists) compared with those receiving 3 or fewer of the agents (Fig. 5). Plasma levels of IMD were elevated in the 13 heart failure patients with renal impairment (GFR <60 ml min−1 1.73 m−2 ) but could not be distinguished in 6 heart failure patients with normal renal function (11.3 + 1.8 pg ml−1 v 6.5 + 1.0 pg ml−1 ; p = 0.036) from those of healthy subjects (Fig. 6(a) Plasma levels of AM and MRproAM45–92 did not discriminate heart

[b]

PreproIMD 57-92 [pg.ml -1]

The characteristics of the healthy subjects are given in Table 1. The study group was balanced in regard to gender and encompassed a wide range of ages (20–69 years) and BMI values (range 17.6–40.0 kg m−2 ). The plasma level of each of the peptides (mean + SE) is given in Table 2. IMD was detected at low levels in plasma (6.3 + 0.6 pg ml−1 ), less abundantly than AM (25.8 + 1.8 pg ml−1 ). There was a modest correlation between levels of the two peptides (r = 0.49; p < 0.001; Fig. 2(a). The IMD precursor fragments, preproIMD25–56 and preproIMD57–92 , were present in plasma at higher concentrations than IMD and were related to each other (r = 0.63; p < 0.001; Fig. 2(b), but neither correlated significantly with IMD. IMD was related to BMI (Fig. 3) with stepwise multivariate analysis showing a small 6.3% increase in IMD for each unit increase in BMI. Age and gender were not influences on IMD levels (Fig. 4). PreproIMD25–56 and preproIMD57–92 were not influenced by BMI, age (Figs. 3 and 4) or gender (data not shown). Similarly to the situation for IMD, the precursor molecule MRproAM45–92 was more abundant than AM itself, but was not related to it (p = 0.14). AM levels were correlated with BMI (Fig. 3) similarly to IMD, but in contrast both AM and MRproAM45-92 were weakly related to age (Fig. 4).

15

20

25

30

IMD [pg.ml-1]

250

r= 0.439 p<0.001

200

150

100

50

0

0

50

100

150

200

250

-1

PreproIMD25-56 [pg.ml ] Fig. 2. Relationships between (a) AM and IMD and (b) the IMD fragments preproIMD25–56 and preproIMD57–92. Values are given in pg ml−1 .

failure patients on the basis of renal status (p = 0.97, p = 0.80; data not shown). Plasma levels of MRproAM45–92 correlated positively with those of AM (p < 0.05) in the heart failure patients. The plasma levels of NT-proBNP were not correlated with those of any of the three peptides. There was an inverse relationship between plasma IMD levels and diastolic blood pressure in the heart failure group (p < 0.05, Fig. 6(b); no other significant correlations were detected between any of the three peptides and the haemodynamic, functional or biochemical characteristics of the patients given in Table 1. In the subgroup undergoing cardiac resynchronisation pacing, IMD tended to be reduced 6 months after the intervention (9.8 + 1.3 to 4.1 + 1.0 pg ml−1 , p = 0.07; Fig. 7); similar results were noted for AM (65.0 + 13.2 to 29.2 + 6.6 pg ml−1 , p<0.05; Fig. 7). The fall in AM tended to be progressive from a few weeks after implantation, though the decline in IMD was less marked early after intervention. MRproAM45–92 did not change significantly (275 + 30 to 334.2 + 33 pg ml−1 , p > 0.05). NYHA functional classification was also reduced from baseline (2.91 + 0.08 to 2.18 + 0.13 p < 0.05 at 6 months) although NT-proBNP levels were not significantly reduced after CRT in the group

22

D. Bell et al. / Peptides 76 (2016) 19–29

Table 1 Characteristics of the healthy control population (n = 75) and heart failure patients (n = 19). Control group n = 75 Characteristics

Male

Female

Total

Gender (n) Age (years) Weight (kg) Height (cm) BMI category (kg/m2 ) Underweight (n) < 18.5 Normal weight (n) 18.5–24.9 Overweight (n) 25.0–29.9 Obese (n) > 30.0 kg/m2 20–29 Years (n) 30–39 Years (n) 40–49 Years (n) 50–59 Years (n) 60–69 Years (n)

36 45.5 (22–67) 81.5 (54.5–123.6) 180.3 (148.0–195.8) 25.5 (19.9–38.0) 0 14 17 5 5 11 6 10 4

39 44.0 (20-63) 65.0 (45.0–99.5) 165.0 (152.0–180.3) 23.6 (17.6–40.0) 2 21 12 4 8 7 12 9 3

75 44.0 (20–67) 73.2 (45.0–123.6) 172.0 (148.0–195.6) 25.1 (17.6–40.0) 2 35 29 9 13 18 18 19 7

Heart failure patients n = 19 Age, years Gender BMI, kg m−2 Ejection fraction % NYHA Classification Systolic BP mmHg Diastolic BP mmHg Haemoglobin Sodium mmol/l Potassium mmol/l Urea mmol/l Creatinine micromol/L GFR ml min−1 1.73 m−2 NT-pro-BNP Beta adrenoceptor antagonist (n/19) ACE inhibitor or AT1 receptor antagonist (n/19) Aldosterone receptor antagonist (n/19) Diuretic (n/19)

75.0 (46–84) 14 male, 5 female 26.9 (24.5–41.5) 23.0 (10.0–38.0) Class II 8 /Class III 11 108.0 (90–182) 62.0 (48–78) 12.6 (10.8–14.2) 141.0 (134.0–144.0) 4.6 (3.8–5.4) 8.75 (5.0–28.8) 118.0 (41.0–205.0) 57.0 (29.0–188.0) 1698 (358–7355) 17 18 14 15

Data are given as median values (range).

Table 2 Plasma concentrations of adrenomedullin (AM), intermedin (IMD) and mid region proadrenomedullin (MRproAM) in patients and healthy controls, and of IMD precursor fragments in healthy controls. Peptide

Normal controls

Heart failure

AM (pg ml−1 ) AM (pM) IMD (pg ml−1 ) IMD (pM) MRproAM45–92 (pg ml−1 ) MRproAM45–92 (pM) PreproIMD25–56 (pg ml−1 ) PreproIMD25–56 (pM) PreproIMD57–92 (pg ml−1 ) PreproIMD57–92 (pM)

22.9 (2.25–71.43) 3.82 (0.38–11.91) 4.99 (0–27.97) 0.95 (0–5.33) 198.8 (69.57–333.4) 38.8 (13.6–65.1) 33.99 (0.32–220.6) 9.8 (0.09–63.5) 14.6 (0–193.6) 3.8 (0–50.1)

42.46 (11.67–64.83) 7.03 (1.93–10.73* 8.41 (1.73–21.65) 1.63 (0.33–4.20)** 268.7 (101.9–739.6) 52.5 (19.9–144.6)***

Values given as pg ml−1 and pM. Data are presented as mean + SE. * p = 0.01. ** 0.01 < p < 0.05. *** p < 0.001 for heart failure versus control subjects.

as a whole despite overall significant clinical improvement and echocardiographic reverse modelling.

4. Discussion This study is the first to report on the plasma levels of IMD and its putative precursor fragments, preproIMD25–56 and preproIMD57–92 , measured together in the healthy general human population across a wide range of age and BMI values. IMD was detected at low levels in the plasma of healthy individuals, and

was weakly influenced by BMI but not by age or gender. In contrast, levels of the structurally similar AM were somewhat higher and were influenced by age in addition to BMI. Levels of the precursors, preproIMD25–56 , preproIMD57–92 and MRproAM45–92 were all substantially elevated in comparison to their respective active peptide. To our knowledge, this is also the first study of IMD plasma levels in patients with chronic heart failure and how these are influenced by medical or surgical intervention. In systolic heart failure, levels of IMD (and AM) were substantially elevated, with some suppression of levels in those receiving extensive medical

D. Bell et al. / Peptides 76 (2016) 19–29

23

Correlation vs. BMI [a]

[b] 80

350

(pre)proAM 45-92 [pg.ml -1]

r= 0.312 p=0.006

70 60

AM [pg.ml -1]

400

50 40 30 20

300 250 200 150 100

r= 0.333 p=0.004

50

10 0 15

20

25

30

35

0 15

40

20

BMI [kg.m ]

30

[d] r= 0.515 p<0.001

-1

IMD [pg.ml ]

25

preproIMD 25-56 [pg.ml -1]

[c]

20 15 10 5 0 15

20

25

30

35

40

BMI [kg.m-2]

-1

preproIMD 57-92 [pg.ml ]

[e]

25

30

35

40

BMI [kg.m-2]

-2

250

200

r= -0.077 p= ns

150

100

50

0 15

20

25

30

35

40

BMI [kg.m-2]

250

r= 0.159 p= ns

200

150

100

50

0 15

20

25

30

35

40

-2

BMI [kg.m ] Fig. 3. Relationships between plasma levels (pg ml−1 ) of (a) AM; (b) MRproAM45–52 ; (c) IMD; (d) preproIMD25–56 ; (e) preproIMD57–92 and BMI in a healthy control population (n = 75). Ns = not significant.

therapy. IMD, though not AM, was significantly elevated in heart failure patients with concurrent renal impairment, and in those with low diastolic blood pressure. Neither peptide was related to levels of NT-proBNP indicating that their measurement might provide independent information additional to that of NT-proBNP as part of a multi-biomarker panel strategy.

4.1. Levels in the control population IMD plasma levels were lower than those recorded in other recent studies, conducted in predominantly middle-aged Asian populations and measured using the same commercial RIA employed in the current study [25,36,42]. Levels of AM in the

24

D. Bell et al. / Peptides 76 (2016) 19–29

Correlation vs. Age [a]

80

400

[b]

70

r= 0.278 p=0.016

50 40 30 20

(pre)proAM 45-92 [pg.ml -1]

AM [pg.ml -1]

60

10 0 15

r= 0.293 p=0.011

350 300 250 200 150 100 50

25

35

45

55

65

0 15

75

25

age [years]

[c]

r= -0.133 p=ns

20 15 10 5

35

45

55

65

75

age [years]

preproIMD 57-92 [pg.ml -1]

[e]

[d] preproIMD 25-56 [pg.ml -1]

IMD [pg.ml -1]

25

25

45

55

65

75

age [years]

30

0 15

35

250

r= -0.148 p=ns

200

150

100

50

0 15

25

35

45

55

65

75

age [years]

250

r= -0.010 p=ns

200

150

100

50

0 15

25

35

45

55

65

75

age [years] Fig. 4. Relationships between plasma levels (pg ml−1 ) of (a) AM; (b) MRproAM45–52 ; (c) IMD; (d) preproIMD25–56 ; (e) preproIMD57–92 and age in the healthy control population (n = 75). Ns = not significant.

general population were broadly similar to those reported previously [19,23]. The higher plasma levels of AM may primarily reflect secretion from the vasculature and atria [18]. The origin of IMD

in human plasma is unclear though it is abundantly expressed in mammalian kidney, gastrointestinal tract, hypothalamus and pituitary [38,40]. In human tissue obtained at autopsy from subjects

D. Bell et al. / Peptides 76 (2016) 19–29

25

Fig. 5. Plasma levels (pg ml−1 ) of AM, IMD and mid regional pro-adrenomedullin45 92 (MR-proAM45–92 ) in patients with heart failure stratified by treatment intensity, and in control subjects. The 4 drug classes used in heart failure comprised beta blockers, ACE inhibitors or angiotensin receptor blockers, loop diuretics, and aldosterone antagonists. *p < 0.05, ***p < 0.001.

Fig. 6. (a) Plasma levels of IMD (pg ml−1 ) in heart failure patients (n = 19) stratified by normal (GFR > 60 ml/min/1.73 sqm) or abnormal (GFR < 60) renal function (**p < 0.05). IMD is elevated in those with reduced GFR (11.3 + 1.8 pg/ml) compared with normal renal function (6.5 + 1.0 pg/ml; p = 0.036). (b) Relationship between plasma IMD level (pg.ml−1 ) and diastolic blood pressure (mmHg) in patients with heart failure (r = −0.56; p = 0.01).

26

D. Bell et al. / Peptides 76 (2016) 19–29

Fig. 7. AM and IMD levels (pg ml−1 ) within 24 h prior to (baseline, t = 0) and at various time points after cardiac resynchronization therapy (CRT). AM levels fell progressively being significantly reduced by 6 months (p < 0.05; One way ANOVA with Post hoc-Dunnetts). IMD declined towards normal levels after 6 months but just failed to reach significance.

without known cardiac or renal disease, IMD was predominantly localised to cardiomyocytes rather than endocardium, pericardial adipose tissue or coronary artery smooth muscle [30,40]. In the kidney, IMD was localised to renal tubular cells but much less abundantly so to renal arteriolar smooth muscle. Overspill from such tissues could account for the presence of the peptide in plasma. Expression of IMD and AM is negligible in leukocytes circulating in the blood of healthy subjects [27] and it is unlikely that these cells would make a significant contribution to plasma levels in the absence of disease. Positive correlation between plasma levels of the two peptides could reflect a common stimulus for secretion, common transport mechanism, or common mechanism for clearance/degradation of IMD and AM. The plasma levels of preproIMD25–56 and preproIMD57–92 were greater than those of IMD. Similarly levels of MRproAM45–92 were greater than those of AM. The presence of these precursor fragments in human plasma is consistent with secretion of larger precursor molecules with final processing of AM and IMD each occurring largely extracellularly. Morganthaler and co-workers (2005) [8] also detected MRproAM45–92 levels much greater than those of AM reported in other studies [19,22]. The lack of correlation and of molar equivalence despite their common origin may reflect differences between the bioactive peptides and precursor fragments in regard to stability, biological activity and metabolic pathway. The plasma level of AM may be influenced by binding of biologically active AM to cell surface receptors, rapid enzymatic degradation, and the existence of a binding protein [34] which renders AM less accessible to detection by radioimmunoassay [29]. Similar considerations could in theory apply to the relationship between preproIMD25–56 and preproIMD57–92 and IMD. The levels of preproIMD25–56 and preproIMD57–92 positively correlated with each other, reflecting a shared origin from a common precursor, though lack of molar equivalence might reflect differing routes of elimination or stability.

4.2. Relationship to BMI Plasma levels of AM, IMD and MRproAM45–92 were correlated significantly with BMI in the general population. Few previous studies have reported on the relationship between BMI and plasma IMD levels although Qin et al. (2013) did not find a correlation between

these two parameters in a combined study population of healthy controls and patients with acute coronary syndrome, with mean BMI values in the normal range. Interestingly, plasma IMD levels reported in Yamac et al. (2014) study in obese middle-aged patients with normal coronary anatomy on angiography were greater than those reported for healthy controls with normal BMI values in Lv et al. (2013) study. An association with BMI has been noted previously for AM peptides [19,23] and is consistent with a causal association between BMI and expression and/or release of AM from a larger precursor rather than metabolism of AM. Such correlation however is lacking between BMI and the IMD precursors and cannot therefore account for the relationship between BMI and plasma IMD levels. Oxidised LDL [16] and hyperinsulinaemia [23] stimulate AM secretion from vascular endothelial and smooth muscle cells which may be important in obese subjects at risk of glucose intolerance and diabetes. However AM is also expressed in human adipocytes and this expression is enhanced in obese subjects [24]. IMD expression has been detected in pericardial adipocytes [30] indicative of a possible contribution of adipose tissue to circulating levels of IMD. Overall, raised AM (and IMD) levels in the plasma of obese subjects may represent a protective mechanism against various complications of obesity. The contribution of the vasculature to plasma IMD levels is less clear than for AM and spill-over of IMD produced by adipocytes into the blood may explain increased levels in obesity.

4.3. Relationship to age and gender The levels of the peptides and their precursor fragments did not differ significantly between males and females. This is in keeping with previous reports for AM [19] and MRproAM45–92 [29]. Yamac et al. (2014) did report slightly higher plasma IMD levels in males than females in their population of patients undergoing diagnostic coronary angiography, but the majority of those diagnosed with severe stenosis were male while those found to have normal coronary anatomy were predominantly female. Qin et al. (2013) noted that IMD levels were lower in younger subjects than in an older population, but the former had no known clinical problems while the latter were being admitted to hospital with suspected acute coronary syndrome. In contrast, the current study incorporated a larger number of healthy adults relatively

D. Bell et al. / Peptides 76 (2016) 19–29

evenly distributed across five decades. Plasma levels of AM and MRproAM45–92 correlated positively with age but those of IMD did not, possibly reflecting the paucity of expression of IMD within the vasculature of healthy individuals. Knowledge of the extent of the influence of BMI, gender and age on plasma levels of these biomarkers will supply useful reference values for comparison studies in patients with cardiovascular diseases, and reinforces the importance of appropriate matching of control subjects. 4.4. Levels in heart failure Levels of IMD, AM and MRproAM45–92 were elevated in systolic heart failure compared with controls. In contrast with the lower levels measured in control subjects, plasma levels of MRproAM45–95 correlated positively with, but were greater than those of AM, in patients with moderate systolic chronic heart failure. Recent publications have demonstrated that MRproAM45–92 is elevated in heart failure, is related to systolic dysfunction in patients with ischaemic heart disease, and carries prognostic information in patients with acute heart failure or myocardial infarction [20,26]. We have additionally found that the extent of peptide elevation was also influenced by intensity of medical therapy in agreement with the finding of [10]. To our knowledge, our study provides the first human data on IMD levels in human heart failure and indicates that IMD is elevated in patients with chronic heart failure, though less so than AM or MRproAM45–92 , most likely reflecting the more limited distribution and differing stimuli for release of IMD relative to AM (reviewed in Ref. [15]. Plasma IMD distinguished heart failure patients with renal impairment from those with normal renal function while AM and MRproAM45–92 did not; this might reflect overspill of synthesis of IMD locally within the kidney in response to renal hypoperfusion and ischemia [30]. Indeed, the peptide has been implicated specifically in mammalian renal physiology since IMD augments renal blood flow and directly attenuates renal tubular sodium reabsorption in rodents [12] and may therefore, in contrast to AM, provide a useful marker of cardio-renal syndrome. A negative correlation between plasma IMD level and diastolic blood pressure was noted; this could reflect secretion of IMD from the heart due to ischaemia arising as a consequence of reduced coronary perfusion [47,17]. In addition to falling in association with medical therapy, peptide levels fell after CRT. In a sub study of the CARE-HF study, Fruhwald et al. (2007) reported that NT-proBNP fell at 3 months after CRT and was associated with favourable reverse remodelling of the left ventricle. This suggests that the positive (reverse) cardiac remodelling resulting from CRT has reduced the stimulus for AM and IMD release; the differing time course of AM and IMD suppression may reflect differential effects over time on their respective stimuli for release. We are not aware of any data on the effect of CRT on levels of either peptide. Morales reported that high AM levels prior to CRT predicted remodelling after therapy [28], but did not report levels after therapy; no data exist for IMD. None of the 3 peptides was related to levels of NT-proBNP, and so it is anticipated that any clinical information they provide would therefore be independent of or additive to this assay. Others have also reported that AM or its’ precursor peptide can provide useful additional information, independent of BNP or its’ precursor peptide [37,46,21,45]. The difference likely reflects differing distribution, stimuli for secretion, and route and timelines for elimination of the two peptide biomarkers. The two markers are looking at different aspects of the pathophysiology of heart failure, supporting the rationale for screening strategies based on the development of multi-marker panels. One previous study reported wide variation in NT-proBNP results and a non-significant change in levels after 12 months of CRT [13]. NT-proBNP levels might only be reduced in those that display significant reverse remodelling

27

after CRT [11,50]. One small study showed a reduction in noradrenaline levels after CRT [39] but two larger studies did not shown any change in activation of RAAS or SNS [7,6]. Therefore natriuretic peptide levels may still remain elevated as a compensatory mechanism for other pathological processes that have not been influenced by CRT, even in some patients who have displayed reverse remodelling. 4.5. Study limitations This is one of the first systematic studies of IMD plasma levels in healthy subjects and in patients with systolic heart failure. As such some potential as yet unknown confounders could be operating in the cohorts studied. Further as a preliminary study these findings, and particularly the subgroup results, are limited by small numbers. In this small pilot study, it was not our intention to determine suitability of IMD as a diagnostic marker of heart failure. The healthy control group and heart failure group were not of equal size and were not appropriately matched to facilitate ROC analysis. Furthermore the patients were already taking three or more drugs to manage their heart failure. Confirmation and further assessment in a large cohort of heart failure patients is required. 5. Conclusions This study provides the first systematic report of the levels of IMD (adrenomedullin-2), relative to its precursor fragments and those of AM in a healthy control population, and for alteration in plasma levels of IMD in systolic heart failure. In healthy individuals IMD is present in the plasma in low levels and is weakly correlated to BMI. It is elevated in chronic systolic heart failure and in this setting appears to be influenced by renal function, blood pressure and intensity of medical therapy and may be reduced consequent to positive cardiac remodelling after CRT. Levels appear to be independent of NTproBNP and are only weakly related to the structurally similar AM peptide and its precursor fragment MR-proAM45–92 . Assessment of IMD in a large cohort of heart failure patients is now recommended. Future studies should also address the potential application of IMD as a plasma biomarker in the pathophysiology of various renal diseases. Authors contributions M.H. was involved in the overall conception and design of the study, research governance and the recruitment of healthy subjects and patients attending the Belfast Trust heart failure outpatients’ clinic. B.G. was responsible specifically for recruitment of patients selected to undergo device-based CRT, based on expertise in managing such patients. D.B. designed experimental protocols, was responsible for the day to day running of the project, analysis and writing of the paper. K.M., A.L. and M.K. were undergraduate students undertaking short-term research projects within the laboratory at the time of the study and who each contributed to the generation of some of the data. Funding This study was funded by Heart Research UK (Grant RG2578/09/12), and by Chest Heart and Stroke Association, Northern Ireland (Grant 200777) References [1] D. Bell, M. Campbell, M. Ferguson, L. Sayers, L. Donaghy, A. O’Regan, V. Jewhurst, M. Harbinson, AM1 receptor-dependent protection by intermedin

28

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9] [10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18] [19]

[20]

[21]

[22]

[23]

[24]

D. Bell et al. / Peptides 76 (2016) 19–29 of human vascular and cardiac non-vascular cells from ischaemia-reperfusion injury, J. Physiol. 590 (2012) 1181–1197. D. Bell, B.J. McDermott, Intermedin (adrenomedullin-2): a novel counter-regulatory peptide in the cardiovascular and renal systems, Br. J. Pharmacol. 153 (Suppl. 1) (2008) S247–S262. D. Bell, Y. Zhao, F.P.G. McCoy, A.B. Devine, B.J. McDermott, Differential effects of an anti-oxidant intervention on cardiomyocyte expression of adrenomedullin and intermedin and their receptor components in chronic nitric oxide deficiency, Cell Physiol. Biochem. 20 (2007) 269–282. D. Bell, Y. Zhao, F.P.G. McCoy, A. Devine, B.J. McDermott, Expression of the counter-regulatory peptide Intermedin is augmented in the presence of oxidative stress in hypertrophied cardiomyocytes, Cell. Physiol. Biochem. 21 (2008) 409–420. D. Bell, Y.Y. Zhao, A.B. Devine, B.J. McDermott, Influence of atenolol and nifedipine on nitric-oxide deficient cardiomyocyte hypertrophy and expression of intermedin and its receptor components, Cell. Physiol. Biochem. 21 (2008) 203–214. G. Boriani, F. Regoli, D. Saporito, C. Martignani, T. Toselli, M. Biffi, G. Francolini, I. Diemberger, L. Bacchi, C. Rapezzi, R. Ferrari, A. Branzi, Neurohormones and inflammatory mediators in patients with heart failure undergoing cardiac resynchronisation therapy: time courses and prediction of response, Peptides 27 (2006) 1776–1786. M.U. Braun, T. Rauwolf, T. Zerm, M. Schulze, A. Schnabel, R.H. Strasser, Long-term biventricular resynchronisation therapy in advanced heart failure: effect on neurohormones, Heart 91 (2005) 601–605. E. Elmas, S. Lang, C.E. Dempfle, T. Kälsch, J. Papassotiriou, N.G. Morganthaler, M. Borggrefe, M. Brueckmann, Diagnostic performance of mid-regional pro-adrenomedullin as an analyte for the exclusion of left ventricular dysfunction, Int. J. Cardiol. 128 (2008) 107–111. M. Emdin, S. Vittorini, C. Passino, A. Clerico, Old and new biomarkers of heart failure, Eur. J. Heart Fail. 11 (2009) 331–335. T. Etoh, J. Kato, M. Takenaga, Differential hormonal profiles of adrenomedullin and proadrenomedullin N-terminal 20 peptide in patients with heart failure and effect of treatment on plasma levels, Clin. Cardiol. 22 (1999) 113–117. F.M. Fruhwald, A. Fahrleitner-Pammer, R. Berger, F. Leyva, N. Freemantle, E. Erdmann, D. Gras, L. Kappenberger, L. Tavazzi, J.-C. Daubert, J.G.F. Cleland, Early and sustained effects of cardiac resynchronization therapy on N-terminal pro-B-type natriuretic peptide in patients with moderate to severe heart failure and cardiac dyssynchrony, Eur. Heart J. 28 (13) (2007) 1592–1597. Y. Fujisawa, Y. Nagai, A. Miyatake, Y. Takei, K. Miura, T. Shoukouji, A. Nishiyama, S. Kimura, Y. Abe, Renal effects of a new member of adrenomedullin family, adrenomedullin2, in rats, Eur. J. Pharmacol. 497 (1) (2004) 75–80. I. Garcia-Bolao, A. Macias, B. Lopez, A. Gonzalez, J.J. Garvira, P. Azcarate, E. Alegria, J. Diez, A biomarker of myocardial fibrosis predicts long-term response to cardiac resynchronisation therapy, J. Am. Coll. Cardiol. 47 (2006) 2335–2337. N. Hirayama, K. Kitamura, T. Imamura, J. Kato, Y. Koiwaya, T. Tsuji, K. Kangawa, T. Eto, Molecular forms of circulating adrenomedullin in patients with congestive heart failure, J. Endocrinol. 160 (1999) 297–303. D. Holmes, M. Campbell, M. Harbinson, D. Bell, Protective effects of intermedin on cardiovascular, pulmonary and renal diseases: comparison with adrenomedullin and CGRP, Curr. Protein Pept. Sci. 14 (2013) 294–329. Y. Isumi, H. Shoji, S. Sugo, T. Tochimoto, M. Yoshioka, K. Kangawa, H. Matsuo, N. Minamino, Regulation of adrenomedullin production in rat endothelial cells, Endocrinology 139 (1998) 838–846. Y.X. Jia, J.H. Yang, C.S. Pan, B. Geng, J. Zhang, Y. Xiao, J. Zhao, H. Gerns, J. Yang, J.-W. Chang, J.K. Wen, C.-S. Tang, Y.-F. Qi, Intermedin 1-53 protects the heart against isoproterenol-induced ischemic injury in rats, Eur. J. Pharmacol. 549 (2006) 117–123. M. Jougasaki, J.C. Burnett, Adrenomedullin: potential in physiology and pathophysiology, Life Sci. 66 (2000) 855–872. J. Kato, K. Kitamura, T. Uemura, K. Kuwasako, T. Kita, K. Kangawa, T. Eto, Plasma levels of adrenomedullin and atrial and brain natriuretic peptides in the general population: their relations to age and pulse pressure, Hypertens. Res. 25 (2002) 887–892. S.Q. Khan, R.J. O’Brien, J. Struck, P. Quinn, N. Morgenthaler, I. Squire, J. Davies, A. Bergmann, L.L. Ng, Prognostic value of midregional pro-adrenomedullin in patients with acute myocardial infarction, J. Am. Coll. Cardiol. 49 (2007) 1525–1532. I.T. Klip, A.A. Voors, S.D. Anker, H.L. Hillege, J. Struck, I. Squire, D.J. van Veldhuisen, K. Dickstein, for the OPTIMAAL Investigators: MR-proAM45-92 was a stronger predictor of mortality in patients with heart failure after an acute myocardial infarction, Heart 97 (2011) 892–898. C. Letizia, G. Lacobellis, C. Caliumi, F. Leonetti, D. Cotesta, M.C. Ribaudo, L. Petramala, R. Cianci, M. Celi, E. D’erasmo, U. Di Mario, Acute hyperinsulinemia is associated with increased plasma adrenomedullin concentrations in uncomplicated obesity, Exp. Clin. Endocrinol. Diabetes 113 (2005) 171–175. C. Letizia, S. Subioli, S. Cerci, C. Caliumi, C. Verrelli, E. Delfini, M. Celi, L. Scuro, E. D’Erasmo, High Plasma Adrenomedullin concentrations in patients with high-renin essential hypertension, J. Renin Angiotensin Aldosterone Syst. 3 (2002) 126–129. Y. Li, C. Jiang, X. Wang, Y. Zhang, S. Shibahara, K. Takahashi, Adrenomedullin is a novel adipokine: adrenomedullin in adipocytes and adipose tissues, Peptides 28 (2007) 1129–1143.

[25] Z. Lv, K. Wu, X. Chen, X. Zhang, B. Hong, Plasma intermedin levels in patients with acute myocardial infarction, Peptides 43 (2013) 121–125. [26] A. Maisel, C. Mueller, R. Nowak, W.F. Peacock, J.W. Landsberg, P. Ponikowski, M. Mockel, C. Hogan, A.H.B. Wu, M. Richards, L. Clopton P, G.S. Filippatos, S. Di Somma, I. Anand, L. Ng, L.B. Daniels, S.-X. Neath, R. Christenson, M. Potocki, J. McCord, G. Terracciano, D. Kremastinos, O. Hartmann, S. von Haehling, A. Bergmann, N.G. Morgenthaler, S.D. Anker, Mid-region pro-hormone markers for diagnosis and prognosis in acute dyspnea results from the BACH (Biomarkers in Acute Heart Failure) trial, J. Am. Coll. Cardiol. 55 (2010) 2062–2076. [27] C. Manuela, L. Sabatino, B. Svezia, R. Caruso, A. Verde, C. Caselli, T. Prescimone, D. Giannessi, D.R. Silvia, Adrenomedullin and intermedin gene transcription is increased in leukocytes of patients with chronic heart failure at different stages in the disease, Peptides 55 (2014) 13–16. [28] M.-A. Morales, M. Maltinti, M. Piacenti, S. Turchi, D. Giannessi, S. Del Ry, Adrenomedullin plasma levels predict left ventricular reverse remodeling after cardiac resynchronization therapy, Pacing Clin. Electrophysiol. 33 (2010) 865–872. [29] N.G. Morgenthaler, J. Struck, C. Alonso, A. Bergmann, Measurement of mid-regional proadrenomedullin in plasma with an immunoluminometric assay, Clin. Chem. 51 (2005) 1823–1829. [30] R. Morimoto, F. Satoh, O. Murakami, K. Totsune, T. Suzuki, H. Sasano, S. Ito, K. Takahashi, Expression of adrenomedullin-2/intermedin in human brain, heart and kidney, Peptides 28 (2007) 1095–1103. [31] National Institute for Health and Clinical Excellence. Chronic heart failure: Management of chronic heart failure in adults in primary and secondary care. August 2010. http://www.nice.org.uk/nicemedia/live/13099/50526/50526. pdf (20 June 2012). [32] C.S. Pan, J.H. Yang, D.Y. Cai, J. Zhao, H. Gerns, J. Yang, J.W. Chang, C.S. Tang, Y.F. Qi, Cardiovascular effects of newly discovered peptide intermedin/adrenomedullin-2, Peptides 26 (2005) 1640–1646. [33] L.J. Pearson, T.G. Yandle, M.G. Nicholls, J.J. Evans, Intermedin (adrenomedullin-2): a potential protective role in human aortic endothelial cell, Cell. Physiol. Biochem. 23 (2009) 97–108. [34] R. Pio, A. Martinez, E.J. Unsworth, J.A. Kowalak, J.A. Bengoechea, P.F. Zipfel, T.H. Elsasser, F. Cuttitta, Complement Factor H is a serum-binding protein for adrenomedullin, and the resulting complex modulates the bioactivities of both partners, J. Biol. Chem. 276 (2001) 12292–12300. [35] F. Pousset, F. Masson, O. Chavirovskaia, Plasma adrenomedullin, a new independent predictor of prognosis in patients with chronic heart failure, Eur. Heart. J. 21 (2000) 1009–1014. [36] Y.-W. Qin, X. Teng, J.-Q. He, J. Du, C.-S. Tang, Y.-F. Qi, Increased plasma levels of intermedin and brain natriuretic peptide associated with severity of coronary stenosis in acute coronary syndrom, Peptides 42 (2013) 84–88. [37] A.M. Richards, R. Doughty, M.G. Nicholls, Plasma NT-pro BNP and adrenomedullin: prognostic utility and prediction of benefit from carvedilol in chronic ischemic left ventricular dysfunction, J. Am. Coll. Cardiol. 37 (2001) 781–1787. [38] J. Roh, C.L. Chang, A. Bhalla, C. Klein, S.Y. Hsu, Intermedin is a calcitonin/calcitonin gene-related peptide family peptide acting through the calcitonin receptor-like receptor/receptor activity-modifying protein receptor complexes, J. Biol. Chem. 279 (2004) 7264–7274. [39] M. Seifert, M. Schlegl, W. Hoersch, E. Fleck, A. Doelger, M. Stockburger, C. Butter, Functional capacity and changes in the neurohormonal and cytokine status after long-term CRT in heart failure patients, Int. J. Cardiol. 121 (2007) 68–73. [40] K. Takahashi, K. Kikuchi, Y. Maruyama, T. Urabe, K. Nakajima, H. Sasano, Y. Imai, O. Murakami, K. Totsune, Immunocytochemical localization of Adrenomedullin 2/Intermedin-like immunoreactivity in human hypothalamus, heart and kidney, Peptides 27 (2006) 1383–1389. [41] Y. Takei, K. Inoue, M. Ogoshi, K. Kawahura, H. Bannai, S. Miyano, Identification of a novel adrenomedullin in mammals: a potent cardiovascular & renal regulator, FEBS Lett. 556 (2004) 53–58. [42] B. Tang, Z. Zhong, H.-W. Shen, H.-P. Wu, P. Xiang, B. Hu, Intermedin as a prognostic factor for major adverse cardiovascular events in patients with ST-segment elevation acute myocardial infaction, Peptides 58 (2014) 98–102. [43] M.M. Taylor, S.L. Bagley, W.K. Samson, Intermedin/AM-2 acts within the central nervous system to elevate blood pressure and inhibit food and water intake, Am. J. Physiol. 288 (2005) R919–R927. [44] The Task Force for the Diagnosis and Treatment of Acute and Chronic Heart Failure 2012 of the European Society of Cardiology. Developed in collaboration with the Heart Failure Association (HFA) of the ESC. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure 2012. European Heart Journal 10.1093/eurheartj/ehs104 (20 June 2012). [45] Y. Xue, P. Taub, N. Iqbal, A. Fard, P. Clopton, A. Maisel, MR-proAM45–92 predictive value for long term mortality in stable heart failure was independent of BNP, Eur. J. Heart Fail. 15 (2013) 1343–1349. [46] S. van Haehling, G.S. Fillippatos, J. Papassotiriou, M. Cicoira, E.A. Jankowska, W. Doehner, P. Rozentryt, C. Vassanelli, J. Struck, W. Banasiak, P. Ponikowski, D. Kremastinos, A. Bergmann, N.G. Morgenthaler, S.D. Anker, MRproAM45-92 is an independent predictor of mortality in CHF patients which adds prognostic information to NT-proBNP, Eur. J. Heart Fail. 12 (2010) 484–491. [47] H. Yamagishi, M. Yoshiyama, N. Shirai, K. Akioka, K. Takeuchi, J. Yoshikawa, Protective effect of high diastolic blood pressure during exercise against exercise-induced myocardial ischemia, Am. Heart J. 150 (2005) 790–795.

D. Bell et al. / Peptides 76 (2016) 19–29 [48] A.H. Yamac, A. Bacaksiz, Z. Ismailoglu, S. Kucukbuzcu, E. Sevgili, E. Asoglu, M. Nasifov, P. Jafarov, E. Erdogan, O. Goktekin, Implication of plasma intermedin levels in patients who underwent first-time diagnostic coronary angiography: a single centre, cross-sectional study, BMC Cardiovasc. Disord. 14 (2014) 182. [49] J.H. Yang, C.S. Pan, Y.X. Jia, J. Zhang, J. Zhao, Y.Z. Pang, J. Yang, Y.F. Qi, Intermedin 1-53 activates L-arginine/NOS/NO pathway in rat aortas, Biochem. Biophys. Res. Commun. 341 (2006) 567–572.

29

[50] C. Yu, J.W. Fung, Q. Zhang, C.K. Chan, I. Chan, Y.S. Chan, S.L. Kong, J.E. Sanderson, C.W. Lam, Improvement of serum NT-proBNP predicts improvement in cardiac function and favourable prognosis after cardiac resynchronisation therapy for heart failure, J. Card. Fail. 11 (Suppl. 5) (2005) S42–S46.