Urinary proteome analysis in hypertensive patients with left ventricular diastolic dysfunction

Abstract

Aims

Despite the significant heart failure (HF) burden on society, easily applicable screening techniques, particularly for the early detection of asymptomatic left ventricular (LV) dysfunction, are lacking. The present study aimed to identify and test a set of urinary polypeptides that may indicate early LV diastolic dysfunction as defined by echocardiography in hypertensive patients in a cross-sectional case–control study nested within the FLEMish study on ENvironment, Genes and Health Outcome (FLEMENGHO).

Methods and results

To identify potentially discriminating urinary biomarkers for LV diastolic dysfunction, we applied capillary electrophoresis coupled to mass spectrometry. In the discovery set, we compared 19 hypertensive patients with asymptomatic LV diastolic dysfunction with 19 healthy controls. In the absence of adjustment for multiple testing, 85 urinary peptides were different between cases and controls at a P-value of <0.033. With adjustment for multiple testing, three potential biomarkers remained significantly different between cases and controls (P ≤ 0.02). We combined the 85 potential biomarkers in a high-dimensional model (classifier), which we applied in a blinded manner to an independent test set of 16 hypertensive patients with symptomatic HF and 16 healthy controls. Upon unblinding, the area under the receiver operating characteristic curve (AUC) of the HF classification was 0.84 (95% CI: 0.70–0.98; P = 0.001).

Conclusion

In asymptomatic hypertensive patients with LV diastolic dysfunction, we identified a set of urinary polypeptides specific for essential hypertension with LV diastolic dysfunction that subsequently distinguished hypertensive patients with overt HF from healthy controls.

Introduction

Heart failure (HF) is a progressive condition that begins with risk factors for left ventricular (LV) dysfunction (e.g. hypertension), proceeds to asymptomatic changes in cardiac structure (e.g. LV hypertrophy) and function (e.g. impaired relaxation), and then evolves into clinically overt HF, disability and death.¹ The 5-year mortality rate for symptomatic HF is ~60%. Heart failure can clinically manifest with predominantly systolic or diastolic LV dysfunction, or both. In a randomly selected population sample, the frequency of asymptomatic echocardiographically diagnosed diastolic LV dysfunction (early stage) is as high as 27%.² This constitutes a large pool of subjects at high risk for diastolic HF. Despite its significant burden on society, the diagnosis of LV diastolic dysfunction and diastolic HF remains difficult.

Easily applicable screening techniques for the early detection of asymptomatic stages of LV dysfunction are lacking. Clinical proteomics aims at defining novel biomarkers for the early diagnosis of the disease, their prognostic assessment, and the selection and monitoring of proper treatment modalities. In this study, we focused on the urinary proteome, because urine can be easily sampled and contains an array of low-molecular-weight proteins and peptides that can be analysed without additional manipulation, such as proteolysis.³ Capillary electrophoresis coupled to mass spectrometry (CE-MS)^4,5 is now an established method, which allowed identifying peptide markers specific for a variety of diseases, such as diabetes and diabetic nephropathy,⁶ graft rejection,⁷ and coronary artery disease.⁸

One of the core features of cardiac remodelling in hypertensive patients at risk of developing HF and patients with overt HF is related to increased interstitial deposition and cross-linking of type-I collagen. Thus, an impairment of the normal extracellular matrix (ECM) turnover might result in fibrosis of the myocardium, and therefore lead to increased LV stiffness. Capillary electrophoresis coupled to mass spectrometry might be used for the identification and validation of effective screening biomarkers including collagen polypeptides for the risk stratification of patients with an early stage of LV dysfunction. Achieving this goal would meet a pressing clinical need. Thus, the aim of the present study was to identify and validate a set of urinary polypeptides in hypertensive patients that may indicate LV diastolic dysfunction as defined by echocardiography in a cross-sectional case–control study nested within the FLEmish study on ENvironment, Genes and Health Outcome (FLEMENGHO).

Methods

Study population

This study is a sub-analysis of a large-scale family-based study on the genetic epidemiology of cardiovascular phenotypes (FLEMENGHO), for which recruitment started in 1985 and continued through 2008.⁹ At each contact, standardized and validated questionnaires were completed to collect detailed information about each participant’s personal and familial medical history, use of medication, smoking habits, intake of alcohol, physical activity, and lifestyle. High-fidelity phenotyping at the examination centre included clinical and anthropometric measurements, echocardiography, blood sampling, and 24-h urine collection. Biological samples were stored in a biobank for later analysis. The Ethics Committee of the University of Leuven approved the FLEMENGHO study. All participants gave their written inform consent.

For urinary biomarker discovery, we selected 19 hypertensive patients with echocardiographically diagnosed asymptomatic LV diastolic dysfunction and 19 age-matched healthy controls. The detailed echocardiographic protocol is available in the Supplementary material online, Methods. We defined cases as having hypertension (systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg or the use of antihypertensive drugs) and moderate diastolic dysfunction based on age-specific echocardiographic criteria (Supplementary material online, Methods).² Subjects with a previous history of coronary heart disease, valvular heart diseases, renal failure, or diabetes were excluded from the discovery phase.

For biomarker validation, an additional test set of 16 hypertensive patients with mild-to-moderate symptomatic HF (NYHA class II-III) and 16 healthy controls were selected from the FLEMENGHO cohort. We defined cases as having hypertension (systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg or the use of antihypertensive drugs) and moderate/severe LV diastolic dysfunction based on age-specific echocardiographic criteria (Supplementary material online, Methods) and symptoms that might be attributable to HF, such as dyspnoea, peripheral oedema, pulmonary congestion, and chest pain. To collect the information on the HF symptoms during the follow-up visit, we administered the standardized London School of Hygiene cardiovascular, dyspnoea, and respiratory questionnaires.¹⁰

Sample preparation and capillary electrophoresis coupled to mass spectrometry analysis

All participants collected 24-urine samples within 1 week of the echocardiographic examinations. Aliquots were stored at −80°C. Urine (0.7 mL) was thawed immediately before analysis and diluted with 0.7 mL of 2 M urea, 10 mM NH₄OH containing 0.02% SDS.¹¹ To remove higher molecular mass proteins, such as albumin and immunoglobulin G, the sample was ultrafiltered using Centrisart ultracentrifugation devices (20 kDa MWCO; Sartorius, Goettingen, Germany) at 3000 g relative centrifugal force until 1.1 mL of filtrate was obtained. This filtrate was then applied onto a PD-10 desalting column (GE Healthcare, Uppsala, Sweden) equilibrated in 0.01% NH₄OH in HPLC-grade H₂O (Roth, Germany) to decrease matrix effects by removing urea, electrolytes, and salts, and to enrich polypeptides. Finally, all samples were lyophilized, stored at 4°C, and suspended in HPLC-grade H₂O shortly before CE-MS analyses.¹²

Capillary electrophoresis coupled to mass spectrometry analyses were performed using a P/ACE MDQ capillary electrophoresis system (Beckman Coulter, Fullerton, USA) online coupled to a micrO-TOF MS (Bruker Daltonic, Bremen, Germany).¹² The ESI sprayer (Agilent Technologies, Palo Alto, CA, USA) was grounded, and the ion spray interface potential was set between −4 and −4.5 kV. Data acquisition and mass spectrometry acquisition methods were automatically controlled by the capillary electrophoresis via contact-close relays. Spectra were accumulated every 3 s, over a range of charge states (m/z) 350–3000. Accuracy, precision, selectivity, sensitivity, reproducibility, and stability of the CE-MS measurements have been published elsewhere.¹³ The quality control criteria were as follows:¹⁴ The sample must contain a minimum of 950 features, a minimum mass resolution of 8000, and a minimum migration time interval of 10 min. After calibration, migration time deviation must be <0.35 min.

Data processing

Mass spectra were processed using MosaiquesVisu software, including peak picking, deconvolution, and de-isotoping.¹⁵ Migration time and peak intensity were normalized using internal polypeptide standards.¹⁶ These fragments are believed to be the result of normal biological processes and appear to be unaffected by any disease state studied to date on the basis of 13 000 samples in our database.^17,18 The resulting peak list characterizes each polypeptide by its molecular mass, normalized capillary electrophoresis migration time, and normalized signal intensity. All detected polypeptides were deposited, matched, and annotated in a Microsoft SQL database, allowing further analysis and comparison of multiple patient groups.

Statistical methods and identification of biomarkers

We compared means and proportions of clinical and echocardiographic characteristics of the discovery and test samples by means of a t-test and the χ² statistics, respectively, using SAS software, version 9.1.3 (SAS Institute, Cary, NC, USA).

In the discovery phase, we compared the natural logarithm-transformed signal amplitude of the CE-MS urinary polypeptide profile between patients and controls using the Wilcoxon rank sum test.¹⁹ This non-parametric test is suitable for skewed proteomic data. We tested the null hypothesis that patients and controls have the same continuous distribution of signal amplitude of the CE-MS urinary polypeptide profile. The signal amplitude represents the calibrated counts (intensity) recorded by the mass spectrometry device. Statistical adjustment for multiple testing was performed by applying Benjamini-Hochberg.²⁰ This correction allows controlling the false discovery rate at a level of 0.05.²⁰ We searched for a cluster of urinary polypeptides discriminating between cases and controls based on the distribution of biomarkers in individual subjects. For each case and each control, the selected polypeptides were combined into a single summary variable, using the support-vector machine-based MosaCluster software, version 1.6.5 (Supplementary material online, Methods). In the test set, researchers blinded to the clinical condition of the study participants measured the cluster polypeptides. After breaking the code, we calculated the sensitivity and the specificity based on tabulating the number of correctly classified samples in the test set, using receiver operating characteristic (ROC) plots. The area under the ROC curve (AUC) provides a single measure of overall accuracy that is independent of any particular threshold.²¹

Results

Tables 1 and 2 list the clinical and echocardiographic characteristics of cases and controls in the discovery and test sets. Among cases in the discovery set, 8 (42.1%), 10 (52.6%), and 2 (10.5%) were on treatment with diuretics, beta-blockers, or renin system inhibitors (angiotensin-converting enzyme inhibitors or angiotensin receptor blockers). Among the 16 patients in the test set, these numbers were 9 (56.2%), 8 (50.0%), and 7 (43.7%), respectively. All urine samples were successfully analysed with CE-MS and passed all quality control criteria.

Table 1. Clinical characteristics of participants.

Characteristic	Discovery set			Test set
	Control (n = 19)	Case (n = 19)	P-value	Control (n = 16)	Case (n = 16)	P-value
Age, years	63.2 ± 7.5	67.0 ± 7.9	0.13	62.9 ± 4.2	74.5 ± 8.0	< 0.0001
Women, n (%)	9 (47.4)	11 (57.9)	0.52	7 (43.7)	9 (56.3)	0.48
Body mass index, kg/m2	24.9 ± 2.4	29.5 ± 4.1	0.002	24.7 ± 2.8	31.8 ± 7.1	0.001
Systolic pressure, mmHg	120.5 ± 8.4	151.6 ± 12.0	<0.0001	123.5 ± 9.7	146.7 ± 21.6	0.001
Diastolic pressure, mmHg	74.9 ± 5.3	82.0 ± 11.5	0.02	74.3 ± 4.7	79.7 ± 9.9	0.06
Heart rate, b.p.m.	62.1 ± 5.9	60.7 ± 9.0	0.84	62.9 ± 10.2	69.1 ± 14.1	0.17
Questionnaire data
Current smoking, n (%)	6 (31.6)	4 (21.0)	0.47	5 (31.2)	1 (6.2)	0.07
Drinking alcohol, n (%)	9 (47.4)	5 (26.3)	0.18	8 (50.0)	6 (37.5)	0.49
Hypertensive, n (%)	0	19 (100)	—	0	16 (100)	—
Treated for hypertension, n (%)	0	13 (68.4)	—	0	14 (87.5)	—
NYHA class II-III, n (%)	0	0	—	0	16 (100)	—
Coronary heart disease, n (%)	0	0	—	0	7 (43.7)	—
Diabetes, n (%)	0	0	—	0	3 (18.7)	—
Biochemical data
Serum creatinine, μmol/L	84.5 ± 10.4	82.7 ± 10.2	0.63	87.610.9	104.1 ± 32.0	0.07
Glomerular filtration rate, mL/min	74.6 ± 9.7	73.4 ± 9.1	0.70	72.6 ± 11.6	60.9 ± 20.7	0.07
Blood glucose, mmol/L	4.86 ± 0.33	4.81 ± 0.55	0.72	4.94 ± 0.47	5.85 ± 1.17	0.06

Values are mean ( ± SD), or number of subjects (%). NYHA, The New York Heart Association classification.

Table 2. Echocardiographic characteristics of participants.

Characteristic	Discovery set			Test set
	Control (n = 19)	Case (n = 19)	P-value	Control (n = 16)	Case (n = 16)	P-value
Conventional echocardiography
Left atrium diameter, cm	3.84 ± 0.44	4.34 ± 0.48	0.002	3.96 ± 0.51	4.70 ± 0.73	0.003
LV internal diameter, cm	4.83 ± 0.41	4.96 ± 0.50	0.43	5.06 ± 0.57	5.38 ± 0.49	0.13
Interventricular septum, cm	0.96 ± 0.12	1.17 ± 0.20	0.001	0.99 ± 0.16	1.12 ± 0.15	0.03
Posterior wall, cm	0.89 ± 0.10	1.05 ± 0.16	0.001	0.91 ± 0.15	1.03 ± 0.09	0.01
LV mass index, g/m2	87.7 ± 12.7	114.6 ± 26.4	0.0009	95.8 ± 21.9	120.6 ± 22.8	0.008
LV hypertrophy, n (%)	0	7 (41.2)	—	0	8 (50.0)	—
Ejection fraction, %	71.8 ± 5.7	71.7 ± 9.5	0.96	68.3 ± 6.1	55.7 ± 11.5	0.003
Ejection fraction <50%, n (%)	0	0	—	0	7 (43.8)	—
Transmitral Doppler data
E peak, cm/s	71.6 ± 12.6	73.3 ± 15.6	0.73	69.2 ± 8.4	65.9 ± 24.9	0.62
A peak, cm/s	67.2 ± 11.3	87.3 ± 17.7	0.0003	69.0 ± 10.0	79.9 ± 23.8	0.10
E/A ratio	1.09 ± 0.25	0.88 ± 0.27	0.02	1.02 ± 0.19	1.02 ± 0.87	0.99
Tissue Doppler velocities
E′ peak, cm/s	10.5 ± 2.11	6.77 ± 1.72	< 0.0001	10.5 ± 1.79	5.35 ± 1.60	< 0.0001
A′ peak, cm/s	11.3 ± 2.03	10.8 ± 2.01	0.46	11.3 ± 1.71	8.87 ± 2.79	0.01
E′/A′ ratio	0.98 ± 0.33	0.66 ± 0.25	0.002	0.96 ± 0.25	0.72 ± 0.51	0.14
E/E′ ratio	6.97 ± 1.27	11.1 ± 1.73	< 0.0001	6.66 ± 0.72	13.2 ± 4.63	< 0.0001
Diastolic dysfunction, n (%)	0	19 (100)	—	0	16 (100)	—

Values are mean ( ± SD), or number of subjects (%).

Discovery set

The compiled urinary proteomics data from cases and controls in the discovery set are shown in Figure 1A and B. In the absence of adjustment for multiple testing, 85 peptides were found different between cases and controls at a nominal P-value of <0.033. The distribution of these 85 biomarkers in the 19 cases and 19 controls is shown in Figure 1C and D. To classify urinary polypeptides according to whether they were down- or up-regulated in cases vs. controls, we calculated a ratio (R) of ∑ (ln signal amplitude × frequency/number of participants) in controls to ∑ (ln signal amplitude × frequency/number of participants) in cases. Supplementary material online, Table S1 lists further details on these 85 polypeptides ordered by ascending R. The information provided includes polypeptide identifier (SQL number), mass, CE-migration time, percentage of samples, in which the polypeptide could be detected, signal amplitude of the polypeptides, R and P-values for comparisons of the polypeptide signal amplitude distribution between cases and controls. With adjustment for multiple testing applied, three potential biomarkers (IDs 8725, 40737, and 61984) remained significantly different between cases and controls (P ≤ 0.02).

Figure 1.

(A and B) Compiled polypeptide patterns of the 19 hypertensive patients with LV diastolic dysfunction and 19 controls examined in the validation set. The molecular mass (0.8–25 kDa, on a logarithmic scale) is plotted against normalized CE migration time (18–45 min). The signal intensity is encoded by the peak height and colour. (C and D) Distribution of the 85 selected polypeptides for LV diastolic dysfunction in the same patients and controls. The signal intensity is magnified five-fold in comparison with (A) and (B).

Test set

We performed additional CE-MS analysis using an independent test set of 16 hypertensive patients with symptomatic HF and 16 healthy controls in a blinded manner and the data were subjected to the classifier. Upon unblinding, 11 out of 16 HF samples and 15 out of 16 healthy control samples (Figure 2A) were predicted correctly using a classification threshold of the d-score = −0.38 with a sensitivity of 69% (95% CI: 41–89) and a specificity of 94% (95% CI: 70–100). The AUC of the HF classification was 0.84 (95% CI: 0.70–0.98; P = 0.001; Figure 2B).

Figure 2.

(A) Mean ± 95% CI of the d-score used for the classification of the test set. (B) Receiver operator curve (ROC) analysis for urinary biomarker detection of LV dysfunction. The observed area under the curve for left ventricular dysfunction using a panel of 85 biomarkers = 0.84 (95% CI: 0.70–0.98; P = 0.0010).

Sequencing

Table 3 lists information on 70 of 85 polypeptides from the classifier, in which the sequence data (Supplementary material online, Methods) were determined. First, the three polypeptides, which remained statistically significant after adjustment for multiple testing, were collagen alpha-1(V) (ID 8725), WW domain-binding protein 11 (WBP11) (ID 61984), and isoform 1 of collagen alpha-1(XXVI) (ID 40737). All three polypeptides were significantly down-regulated in patients with LV dysfunction compared with the controls (Table 3). Secondly, we noted in patients with LV dysfunction a down-regulation of five other collagen alpha-1(V) fragments and the majority of collagen alpha-1(I) fragments (9 of 14; Table 3). In contrast, 11 of 13 fragments of collagen alpha-1(III) were found to be up-regulated in hypertensive patients with LV diastolic dysfunction (Table 3).

Table 3. List of polypeptides with available information on sequencing.

ID	Sequence of polypeptide	Protein name	Cases		Controls		R
			%	MA	%	MA
8725	GDAGSKGpmV	Collagen alpha-1(V)	0.05	1.94	0.63	2.28	0.067
123106	RDVEEEEEEEGLEEDAELLTELQEVLG	Coiled-coil and C2  domain-containing protein 1B	0.05	1.98	0.47	2.63	0.080
1577	KGDTGPpGP	Collagen alpha-1(III)	0.05	1.65	0.47	1.85	0.095
103493	DEAGSEADHEGTHSTKRGHAKSRPV	Fibrinogen alpha	0.05	3.36	0.47	3.29	0.109
44146	DDFDAHKALEDDE	Isoform 1 of Histone-lysine  N-methyltransferase MLL2	0.11	1.91	0.58	2.49	0.146
4845	GGSGAmGSmD	Immunoglobulin-like and fibronectin  type III domain-containing  protein 1	0.16	1.55	0.63	2.44	0.161
37610	GPpGpPGSpGEQGPSG	Collagen alpha-1(I)	0.11	1.73	0.53	1.87	0.192
83441	GAVGEKGEPGEAGEpGLpGEGGPpG	Collagen alpha-1(V)	0.11	3.45	0.53	3.56	0.201
74703	KSSSHQDSSRmSSVGDYNT	Bone morphogenetic protein 5	0.11	2.64	0.53	2.7	0.203
101157	GPpGADGQpGAKGEpGDAGAKGDAGpPGPA	Collagen alpha-1(I)	0.11	1.97	0.53	1.98	0.207
103022	FNINNLDNNWLKMHFWFYYA	Dermatan-sulfate epimerase-like  protein	0.16	2.52	0.68	2.56	0.232
46091	KGETGDVGQMGppGPP	Collagen alpha-1(V)	0.16	2.08	0.53	2.24	0.280
32022	TYFPHFDLSHG	Haemoglobin subunit alpha	0.21	1.99	0.58	2.21	0.326
82708	GRTGDAGPVGPPGPpGppGpPGPPS	Collagen alpha-1(I)	0.32	2.57	0.84	2.68	0.365
98089	DEAGSEADHEGTHSTKRGHAKSRP	Fibrinogen alpha	0.32	2.97	0.84	3	0.377
61984	DQDKHDDSTDDSDTDK	WW domain-binding protein 11	0.53	2.64	1	3.12	0.448
46369	GPpGEKGGQGPPGpQGp	Collagen alpha-1(V)	0.32	2.78	0.68	2.84	0.461
143947	DQGPVGRTGEVGAVGPpGFAGEKGPSGEAGTAGPpGTpGPQG	Collagen alpha-2(I)	0.37	2.26	0.79	2.24	0.472
39275	DGVGQpGLPGpPGPpG	Collagen alpha-1(XVIII)	0.47	2.59	0.79	2.96	0.520
56493	KGDEGEAGDPGDDNNDI	Collagen alpha-1(VI)	0.47	2.56	0.79	2.74	0.556
41972	EQGLpGAAGQDGPpGP	Isoform D pre-pro-protein of  collagen alpha-1 (XI)	0.53	2.75	0.84	2.95	0.588
24168	GPpGPPGPSSNQG	Collagen alpha-6(IV)	0.58	2.8	0.84	3.26	0.593
107858	VSESSIHIIGVSLGAHVGGmVGQLFGGQ	Isoform 2 of phospholipase A1  member A	0.63	2.36	0.89	2.69	0.621
23356	GPpGPpGPSSNQG	Collagen alpha-6(IV)	0.58	2.63	0.84	2.9	0.626
97599	LGSHSQDEEDEDTEYFDAMEDS	101 kDa protein	0.58	2.59	0.89	2.66	0.634
23697	DDGEAGKpGRpG	Collagen alpha-1(I)	0.68	2.8	1	2.88	0.661
36566	EEEDSSDSSSDSE	Isoform 1 of AP-3 complex subunit  beta-1	0.58	2.77	0.74	3.27	0.664
26670	GQDGRpGPpGPpG	Collagen alpha-1(I)	0.63	3.02	0.84	3.3	0.686
58050	GPpGEAGKpGEQGVPGDLG	Collagen alpha-1(I)	0.63	2.57	0.84	2.79	0.691
28005	TYFPHFDLSHG	Haemoglobin subunit alpha	0.68	3.08	0.84	3.4	0.733
69979	KGSpGSDGpKGEKGDPGpEGP	Isoform 2C2A’ of collagen  alpha-2(VI) chain	0.79	2.86	0.95	3.17	0.750
40737	GPpGPAGNpGpSpNSP	Isoform 1 of collagen alpha-1(XXVI)	0.84	3.33	1	3.68	0.760
65368	WIDAPDDVFYMATEET	Metastasis-associated protein MTA1  79 kDa protein	0.79	3.17	0.89	3.61	0.779
73434	ADGSDLDAVSHGSmDSGHGTH	C-myc promoter-binding protein  isoform 1	0.84	3.1	1	3.28	0.794
108574	DmGPpGPQGPpGKDGPPGVKGENGHPGSP	Isoform 2 of collagen alpha-1 (XIII)	0.79	3.56	0.89	3.85	0.821
90344	GKNGDDGEAGKpGRpGERGPpGPQ	Collagen alpha-1(I)	0.89	3.12	0.95	3.46	0.845
36759	PpGPpGFPGDpGPpG	Collagen alpha-3(V)	0.89	2.94	0.95	3.18	0.866
28561	SpGPDGKTGPpGPA	Collagen alpha-1(I)	0.89	3.36	0.89	3.79	0.886
107460	KNGETGPQGPPGPTGPGGDKGDTGPpGpQG	Collagen alpha-1(III)	0.95	2.91	1	3.11	0.889
32171	ApGDRGEpGPpGPA	Collagen alpha-1(I)	0.95	4.07	1	4.27	0.905
39322	GPpGPpGFPGDPGPpG	Collagen alpha-3(V)	1	3.2	1	3.49	0.917
35339	ApGEDGRpGPpGPQ	Collagen alpha-1(II)	1	3.36	1	3.53	0.952
81196	NGApGNDGAKGDAGApGApGSQGApG	Collagen alpha-1(I)	1	3.72	1	3.59	1.036
41601	DGQPGAKGEpGDAGAK	Collagen alpha-1(I)	1	3.72	1	3.56	1.045
62866	SGpQGppGSEGFTGPPGPQG	Collagen alpha-2(IV)	1	3.89	1	3.71	1.048
99021	QQEQLQQQQFQQQQEQLQQQ	Zinc finger protein 853	1	3.88	1	3.7	1.049
79136	AGPpGEAGKpGEQGVpGDLGApGP	Collagen alpha-1(I)	1	3.74	1	3.49	1.072
50840	DGApGKNGERGGpGGpGP	Collagen alpha-1(III)	0.95	4.17	0.95	3.86	1.080
72533	PpGEAGKpGEQGVpGDLGApGP	Collagen alpha-1(I)	0.95	3.49	0.95	3.21	1.087
57537	NDGApGKNGERGGpGGpGP	Collagen alpha-1(III)	1	4.02	0.95	3.82	1.108
50212	VGGGEQPPPAPAPRRE	Xylosyltransferase 1	0.89	2.7	0.89	2.43	1.111
60149	GNDGApGKNGERGGpGGpGP	Collagen alpha-1(III)	1	3.72	0.95	3.47	1.128
103198	ERGEAGIpGVpGAKGEDGKDGSpGEpGA	Collagen alpha-1(III)	0.89	2.94	0.89	2.47	1.190
104786	NRGERGSEGSPGHpGQpGppGpPGAPGP	Collagen alpha-1(III)	1	3.58	0.89	3.34	1.204
33135	GAPGPRGRDGEpGT	Isoform 1 of collagen alpha-1(II)	1	2.86	0.89	2.65	1.213
73291	DDKDEEDSPRPRSPPGGPD	Zinc finger and BTB  domain-containing protein 46	0.84	2.75	0.79	2.37	1.234
45021	RDGEPGTPGNpGPpGP	Isoform 1 of collagen alpha-1(II)	1	2.82	0.89	2.55	1.243
99475	DDILASPPRLPEPQPYPGAPHHSS	Collagen alpha-1(XVIII)	0.95	2.78	0.89	2.38	1.246
40294	DEPPQSPWDRVK	Apolipoprotein A-I	1	2.85	0.84	2.62	1.295
35424	AMFGPKGFGRGGAE	Cysteine-rich protein 1	0.95	2.79	0.79	2.56	1.311
131294	PGEDGEpGRNGNPGEVGFAGSpGARGFPGAPGLPGL	Collagen alpha-2(V)	1	2.87	0.79	2.71	1.341
111564	ERGEAGIpGVpGAkGEDGKDGSpGEpGANG	Collagen alpha-1(III)	0.89	3.21	0.79	2.67	1.354
104195	NRGERGSEGSPGHPGQPGPpGppGApGP	Collagen alpha-1(III)	0.89	2.61	0.74	2.29	1.371
28747	SpGERGETGPpGP	Collagen alpha-1(III)	1	3.44	0.74	3.32	1.400
44802	GGAGEpGKNGAKGEpGp	Isoform 1 of collagen alpha-1(III)	0.79	2.51	0.63	2.1	1.499
113452	NGEAGSAGPpGppGLRGSpGSRGLPGADGRAG	Collagen alpha-2(I)	0.89	2.47	0.58	2.29	1.655
69681	SNGNpGPpGPSGSpGKDGPpGP	Collagen alpha-1(III)	0.84	2.43	0.42	2.51	1.936
55516	RGSGGGGGGGGQGSTNYGKS	Isoform 3 of heterogeneous nuclear  ribonucleoprotein A/B	0.79	2.54	0.42	2.39	1.999
80360	ISVPGPMGPSGPRGLpGPpGApGP	Collagen alpha-1(I)	0.68	2.74	0.26	2.73	2.625
82784	ADGQpGAkGEpGDAGAKGDAGPpGP	Collagen alpha-1(III)	0.63	2.28	0.21	2.31	2.961

ID, polypeptide identifier (SQL number); %, percentage of samples, in which the polypeptide could be detected; MA, mean signal amplitude of the polypeptides. R was calculated as ∑ (ln signal amplitude × frequency/number of participants) in controls divided by ∑ (ln signal amplitude × frequency/number of participants) in cases. The polypeptides were ordered by ascending R.

Discussion

The key finding of our study was that by applying the CE-MS methodology to easily obtainable urine samples, it was possible to derive sets of urinary polypeptides specific for LV diastolic dysfunction in hypertensive patients. These pilot findings suggest that the urinary proteome might provide a clinically useful tool for screening and subsequent monitoring of subclinical LV dysfunction in hypertensive patients.

Under physiological conditions, the urinary protein content originates ~70% from the kidney and the urinary tract, while 30% is derived from plasma.²² ~60% of the total mass of urinary peptides and proteins represent fragments of collagen.¹⁷ Indeed, in our current study, most of the marker polypeptides originated from the metabolism of collagen (Table 3).

The collagen in the ECM is maintained by an equilibrium of synthesis and degradation, which is disturbed in chronic HF.²³ The cardiac ECM is predominantly composed of fibrillar collagen type I (85%) and type III (11%). Furthermore, small amounts of collagen types IV and V co-distribute with collagen I. The prime location of collagen V is at the fibril core and is important in nucleating collagen I-containing fibrils.²⁴ Left ventricular diastolic dysfunction and HF associated with hypertension are characterized by increased interstitial deposition and cross-linking of type I collagen. The modified ECM proteins lead to LV stiffness.²³ Animal models of the ageing heart revealed reduced proteolytic activity and degradation of collagen I.²⁵ Moreover, elevated levels of serum markers of collagen synthesis (amino-terminal peptide of procollagen type III, PIIINP) and degradation (carboxyl-terminal telopeptide of collagen type I, CITP) independently predicted mortality and cardiovascular disease in population studies.^26,27 Thus, an impairment of the normal ECM turnover might result in fibrosis of the myocardium, which is a core feature of cardiac remodelling in patients at risk of developing HF.

Our study showed in hypertensive patients with asymptomatic LV diastolic dysfunction a down-regulation of several fragments of collagens type I and type V as well as an up-regulation of fragments of collagen type III. Along similar lines, Zürbig et al.²⁸ reported a decreased urinary excretion of collagen alpha-1(I) fragments with ageing in humans. Down-regulation of collagen-alpha1(I) and fibrinogen fragments also occurs in chronic kidney disease.^6,29 However, the later fragments are partially different from those in the current study, indicating differential pathways in collagen breakdown in kidney and heart disease. In addition, the down-regulation of collagen V and the up-regulation of collagen III fragments are not a prominent feature in chronic kidney disease.^6,29

We also noted a down-regulation of polypeptide (ID 61984) associated with WBP11 in LV dysfunction hypertensive patients, which remained statistically significant after adjustment for multiple testing. There are two major functions of WBP11: a splicing factor and a modulator (inhibitor) of protein phosphatase-1 (PP-1).³⁰ In cardomyocytes, PP-1 plays a crucial role for calcium handling and relaxation via dephosphorylation of phospholamban.³¹ The activity of PP-1 is enhanced in HF. Therefore, reduced levels of WBP11, which we observed in patients with LV dysfunction in our study, might result in increased activity of PP-1. In turn, this would delay LV relaxation and results in the progression of LV dysfunction towards symptomatic HF in hypertensive patients.

The present study must be interpreted within the context of its limitations. We combined 85 potential biomarkers in a high-dimensional classifier, since the accuracy and robustness of a high-dimensional model improves with the number of biomarkers.¹⁹ In addition, it is by now accepted that a single biomarker may generally not be able to display a complex disease such as HF.⁴ However, the current study had a small sample size. Only three biomarkers remained significant upon the implementation of rigorous statistical testing. Nevertheless, conforming to the recommendations for clinical proteomics,³² we have tested the classification of independent samples from cases and controls, resulting in an ROC ~0.84. These observations suggest that urinary biomarkers might accurately identify LV dysfunction in hypertensive patients, once replicated in a substantially large population. Another limitation of the study is that the identified set of polypeptides might be specific only for essential hypertension with LV diastolic dysfunction. Further investigation is required to validate this set of polypetides in patients with HF caused by other pathologies, for example in patients with aortic valve stenosis. On the other hand, hypertension is the major risk factor for the development of HF and contributes a large proportion of HF cases in the general population. In the Framingham Heart Study,³³ the population-attributable risk for congestive HF associated with hypertension was 39% in men and 59% in women.

Future research should expand those preliminary findings by applying urinary proteome analysis to test the prognostic value of the derived set of urinary polypeptides against well-established clinical and laboratory prognostic indices in a large cohort. The long-term goal should be to develop the urinary proteomics approach into a clinically useful tool for the early detection of subclinical LV dysfunction and to monitor treatment. Insight in the urinary proteome will also provide clues to the pathogenetic mechanisms driving LV dysfunction and possibly identify targets for treatment.

In conclusion, from the discovery set in asymptomatic hypertensive patients, we derived a panel of urinary polypeptides that may be specific for essential hypertension with LV diastolic dysfunction. Furthermore, this set distinguished hypertensive patients with clinically overt HF from healthy controls in the test (validation) set. Thus, the urinary proteome might be useful as a tool for screening and subsequent monitoring of subclinical LV dysfunction in hypertensive patients.