In-vitro Identification of Distinctive Metabolic Signatures of Intact ...

European Journal of Vascular and Endovascular Surgery 44 (2012) 442e450

Contents lists available at SciVerse ScienceDirect

European Journal of Vascular and Endovascular Surgery journal homepage: www.ejves.com

In-vitro Identification of Distinctive Metabolic Signatures of Intact Varicose Vein Tissue via Magic Angle Spinning Nuclear Magnetic Resonance Spectroscopy M.A. Anwar a, J. Shalhoub a, P.A. Vorkas b, C.S. Lim a, E.J. Want b, J.K. Nicholson b, E. Holmes b, A.H. Davies a, * a b

Academic Section of Vascular Surgery, Department of Surgery & Cancer, Imperial College London, UK Section of Biomolecular Medicine, Department of Surgery & Cancer, Imperial College London, UK

WHAT THIS PAPER ADDS Varicose veins affect one-third of adults in western population. Complications of the disease such as leg oedema and ulceration affect the patients’ health and lifestyle. Pathogenesis of primary varicose veins is unclear. This novel study has elucidated the metabolic profile of intact varicose vein as compared to non-varicose vein. Understanding the significance of differential metabolic profile in future may untangle the biological mechanism of disease initiation and progression.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 12 September 2011 Accepted 19 May 2012 Available online 20 July 2012

Objectives: Nuclear magnetic resonance (NMR) spectroscopy is an established tool for metabolic profiling of tissues or biofluids with utility in identifying disease biomarkers and changes in enzymatic or gene expression. This pilot study aims to compare the metabolic profiles of intact varicose and non-varicose vein tissue via magic angle spinning (MAS) NMR spectroscopy with a view to promoting the understanding of the pathogenesis of varicose vein formation. Methods: Varicose vein tissue (n ¼ 8) was collected from patients undergoing varicose veins surgery. Control non-varicose great saphenous vein samples were collected from patients undergoing lower limb amputation (n ¼ 3) and peripheral arterial bypass surgery (n ¼ 5). Intact tissue samples (average weight 10.33 0.8 mg) from each vein segment were analysed using 1D MAS 1H NMR (600 MHz) spectroscopy. For selected vein samples, two-dimensional (2D) NMR experiments were performed. Differences between spectra from varicose and non-varicose tissues were elucidated using a variety of multivariate statistical analyses. Results: The metabolic profiles of varicose veins samples were clearly differentiated from non-varicose veins samples. Lipid metabolites were present at a higher concentration in the non-varicose veins group whilst creatine, lactate and myo-inositol metabolites were more characteristic of the varicose veins group. Conclusion: We demonstrate differential metabolic profiles between varicose veins and non-varicose veins. Elucidating the metabolic signature underlying varicose veins can further improve our understanding of the biological mechanisms of disease initiation, progression, and aid in identifying putative therapeutic targets. Ó 2012 European Society for Vascular Surgery. Published by Elsevier Ltd. All rights reserved.

Keywords: Vein Varicose veins Metabolic profile Metabonomics NMR spectroscopy MAS NMR

Background

* Corresponding author. A.H. Davies, Academic Section of Vascular Surgery, Charing Cross Hospital, 4th Floor, Fulham Palace Road, London W6 8RF, UK. Tel.: þ44 (0) 20 3311 7320; fax: þ44 (0) 20 3311 7362. E-mail addresses: [email protected], [email protected] (A.H. Davies).

Varicose veins affect one-third of adults in the Western world.1 Marked morphological differences clearly distinguish varicose veins from non-varicose veins. Varicose veins also exhibit expression of various transcripts and proteins linked to specific genes regulating cell growth and apoptosis, extracellular matrix metabolism and inflammatory processes.2e4 Studies have also looked at

1078-5884/$ e see front matter Ó 2012 European Society for Vascular Surgery. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejvs.2012.05.020

M.A. Anwar et al. / European Journal of Vascular and Endovascular Surgery 44 (2012) 442e450

the differential gene expression in varicose veins using DNA hybridisation techniques.5,6 Evidence has strongly suggested that primary vein wall pathology is a major contributor in the development of varicose veins.7,8 However, transcriptomic (study of individuals’ gene expression as RNA) and proteomic (study of translation of RNA into proteins) analyses have not yet allowed elucidation of the cellular and molecular changes described in the wall of varicose veins, or the mechanisms responsible for disease progression. Metabonomics is defined as “the quantitative measurement of the dynamic multi-parametric metabolic response of living systems to pathophysiological stimuli or genetic modification.”9 Metabolites are intermediate or end-products of cellular metabolism and hold key information regarding ongoing biological processes. When exposed to radiofrequency pulses in an external magnetic field, magnetically active atomic nuclei such as protons in a biological sample show slight differences in their transition frequency depending upon their immediate electronic and chemical environment. These differences are the basis of the spectral fingerprint containing discrete signals reflecting multiple metabolites and their concentrations.10 Various stimuli including disease, diet and drugs affect cellular metabolism and hence the metabolite pools in the extracellular tissues and fluids are affected as a result of this altered metabolism.11 High-resolution magic angle spinning (MAS)-nuclear magnetic resonance (NMR) spectroscopy is a non-destructive approach to analyse intact tissue and is found to be valuable in providing information regarding the metabolites’ concentration and compartmentalisation within the tissues.12,13 Several studies have demonstrated toxicological and diagnostic applications of high-resolution MAS-NMR spectroscopy of intact tissues including colonic, renal and hepatic tissues.12,14,15 The biochemical composition of tissue spectra collected by MAS-NMR is then analysed via mathematical modelling using chemometric software to extract and identify robust information on cellular biological events. This can lead to the elucidation of cellular pathways, provide information on gene or enzyme functions and can aid in the determination of the influence, efficacy and toxicity of targeting therapeutics.9,13,16 Identifying new cellular pathways using metabonomic approaches may address the undisclosed biological events, which have not been picked up by other systems biology approaches, including transcriptomics and proteomics. This preliminary study aims to assess the viability of using a metabonomic approach to investigate the biochemical composition of vein wall tissue and to explore the pathogenesis of varicose veins disease, which may help to improve our understanding of its aetiology and its management. Methods Research ethics committee approval was obtained from the Riverside Research Ethics Committee, London, UK. Varicose veins were collected from patients undergoing surgery for lower extremity varicose veins (n ¼ 8). The presence of truncal reflux in the great saphenous vein (GSV) was confirmed and vein marking was undertaken, both using duplex ultrasound prior to surgery. Retrieved GSV varicose samples were also inspected carefully to select for analysis only dilated and tortuous vein segments. Three control non-varicose GSV samples were collected from patients undergoing lower limb amputation and five non-varicose GSV samples from individuals having peripheral arterial bypass surgery. Each vein sample was collected fresh at the time of surgery, snap frozen in liquid nitrogen and stored at 80 C. The entire circumference of vein tissue was cut using a sterile scalpel and forceps; any peri-vascular fat was gently removed. To assess tissue homogeneity within an individual sample and to establish whether the metabolic

443

profile is relatively uniform across a length of vein tissue, one vein segment from each group (varicose and non-varicose veins control) was divided into three adjacent parts and analysed independently. In total, 10 samples from eight varicose veins patients and 10 samples from eight non-varicose veins control patients were analysed. Due to ethical limitations upon the study design, patients from whom varicose veins tissue samples were obtained were younger and had fewer co-morbidities as compared with those from whom non-varicose veins were collected (Table 1). Each sample was weighed (average weight 10.33 0.8 mg). Intact tissue samples were prepared as previously described13,14 and spun at 5 kHz in high-resolution MAS 600 MHz 1H-NMR spectroscope (Bruker BioSpin, Rheinstetten, Germany). NMR spectroscopy of vein tissue Standard one-dimensional spectra were used with suppression of the water signal to adjust the dynamic range to the metabolites of interest. The CarrePurcelleMeiboomeGill (CPMG) spin-echo pulse sequence was also applied in a second measurement to reduce effects of broader resonances from high-molecular-weight compounds, therefore enabling sharp resonances from lowmolecular-weight compounds to be easily identified. A total spinespin relaxation time of 120 ms was chosen (n ¼ 300, s ¼ 200 ms). For each sample, 256 scans were acquired into 64 K data points. Spectral width of d 20 ppm was used with relaxation delay of 2 s and acquisition time of 2.72 s. Each spectrum was phased, calibrated using the chemical shift of the alanine methyl group at d1.48 or the b-glucose (1H) doublet at d 4.644, and baseline was corrected using TOPSPIN 2.0a software (Bruker BioSpin, Rheinstetten, Germany). Chemometric analysis of NMR spectra Spectra were imported into MATLAB R2009b (MathworksÔ, 2009) using in-house software. The region containing the water resonances (from d4.68 to d5.24) was removed from each spectrum and all spectra were subsequently aligned and normalised using probabilistic quotient normalisation.17 Data were transferred to SIMCA-Pþ 11.5 statistical software (UMETRICSÔ, Sweden). Principal components analysis (PCA) and orthogonal partial least square regression (OPLS) analysis were employed using unit variance (UV) scaled data to identify the presence of inherent similarities in the data and investigate whether the varicose veins group systematically differed in biochemical composition from the control group. OPLS is a supervised linear regression approach which can model separately the co-varying variation from structured noise (orthogonal variation), while simultaneously

Table 1 Patient demographic and clinical features of varicose veins and non-varicose veins groups.

Age, mean (range) years Gender Smoker Ischaemic heart disease Hypertension Malignancy Peripheral arterial disease Deep venous thrombosis Diabetes Connective tissue disease Steroid use

Varicose veins (n ¼ 8)

Non-varicose veins (n ¼ 8)

42 (31e62) 6 Male/2 female 4 0 1 0 0 0 0 0 0

69 (32e94) 7 Male/1 female 2 2 7 0 7 0 3 0 0

444


maximising the X and Y covariance, that is, the differences between two groups.18 An in-house mathematical modelling algorithm in MATLAB (Mathworks, 2009) was employed to identify the strength of correlation of the influential metabolite peaks with a particular class. The colour-coded orthogonal 2 partial least square (O2PLS) regression coefficient plot identifies the strength of correlation of a metabolite to a particular tissue type, therefore highlighting significant differences between varicose and nonvaricose veins. We performed receiver operating characteristic (ROC) curve analysis to examine the diagnostic accuracy of partial least-squares driven class discrimination. An area between 0.9 and 1 under the ROC curve represents a strong diagnostic accuracy of the model.19 Individual metabolite identities were assigned by using data from other studies examining human tissue metabolic profiles via MAS NMR and using the human metabolome database.12,14,20e22 The assignments were confirmed by two-dimensional (2D) NMR experiments on selected varicose veins and non-varicose veins samples, and through the use of the Bruker Amix/SBase database. 2D NMR disperses the data into two spectra showing mainly correlation between adjacent protons (1He1H Correlation Spectroscopy or COSY) or proton and carbon atoms 1He13C heteronuclear single quantum coherence (HSQC) spectroscopy, facilitating elucidation of the structure of each metabolite.13,21 J-resolved spectra which display the coupling patterns and J-coupling values in a second orthogonal dimension were also used to identify metabolites and interpret overlapping signal patterns.23

non-varicose veins group (Fig. 2). Creatine, myo-inositol, lactate and glutamate were found in relatively higher levels in varicose veins tissues (Fig. 3). These findings were supported by the results of 2D NMR on selected vein tissue. Analysis of the PCA scores plot of CPMG spectra of varicose and non-varicose veins (Fig. 4a) showed that the variance in the first two components was skewed by two outliers. These outliers were two samples out of a series of three biological replicates from the same vein. Outlier samples contained substantially higher concentrations of triglycerides compared to other samples. However, after exclusion of these two outlying samples, variation was still relatively higher in the non-varicose veins group compared to varicose veins samples (Fig. 4b). There was evidence for clustering of a subset of the samples from the varicose veins group and so OPLS analysis was used to identify systematic variation relating to disease class. To avoid potential bias in OPLS discriminatory analyses, only one out of three replicates from the same vein segment in each group was included for the OPLS discriminatory analysis. Although the number of samples in this pilot study was small, the OPLS scores plot (Fig. 5) clearly highlights the separation of the two groups, with tighter clustering of the varicose veins samples compared to non-varicose veins samples. Examination of the O2PLS coefficients plot showed that creatine, myo-inositol and lactate were highly correlated with the varicose veins group (Fig. 6). The ROC curve analysis confirmed that the diagnostic accuracy of our model was excellent with an area under the curve score of 0.92 (Fig. 7).

Results

Discussion

1 H NMR spectra from all samples with relevant varicose (n ¼ 10) and non-varicose veins (n ¼ 10) spectra are shown in Fig. 1. Assigned spectra from each group are shown in Figs. 2 and 3. Spectral similarities within each group and differences between the two groups of veins were evident. The large peaks of triglycerides (unsaturated and saturated) were found in abundance in the

We observed that the metabolic profiles of varicose veins vary significantly from non-varicose veins. The main differentiating biomarkers are creatine, myo-inositol and lactate, which were higher in varicose veins tissue than non-varicose veins tissue. On the other hand, triglycerides were found in abundance in nonvaricose veins.

Figure 1. Spectra from 10 non-varicose vein samples and 10 varicose vein samples. Similarities of spectral peaks within the varicose veins group representing homogeneity within the group. In comparison, 10 spectra from non-varicose vein samples are different to the varicose veins spectra. Spectra coloured in blue in each group represent the samples taken from the single vein segment of a patient of that group. There is clearly more intra-segmental versus inter-sample homogeneity.


445

Figure 2. Magic Angle Spinning-1H NMR (600 MHz) CarrePurcelleMeiboomeGill (CPMG) spectrum detailing the metabolic profiles of a non-varicose vein tissue.

Creatine is a nitrogenous organic acid originating from the diet, and also synthesised in the liver. It is stored in skeletal muscle where it acts as a cellular energy source.24 Intracellular adenosine triphosphate (ATP) transfers its phosphate group to creatine to

form phosphocreatine.25 Phosphocreatine is found in a concentration ranging between 20 and 35 mM in skeletal muscle cells and 5e10 mM in other excitable tissue including brain and smooth muscle.25 In our study, varicose veins samples were observed to be

446


Figure 3. Magic Angle Spinning-1H NMR (600 MHz) CarrePurcelleMeiboomeGill (CPMG) spectrum listing the metabolic profiles of a varicose vein tissue.

thick walled on inspection. Hypertrophied and hyperplastic smooth muscle mass in the varicose veins wall is therefore responsible for the creatine abundance observed in this group. However, varicose veins can also have smooth muscle atrophy and be thin walled; therefore, inclusion of a large sample including both thin and thick-walled varicose veins, with metabolite extraction for NMR and mass spectrometry analysis, is recommended for future studies. Myo-inositol, in addition to regulating intracellular osmolality in the brain,26 forms a structural component of the phosphatidylinositol and phosphatidylinositides, which function as second messengers. Phosphatidylinositides are involved in lipid signalling and cell signalling via intracellular pathways containing phosphatidylinositol kinases,27,28 protein kinase C and nuclear factor kappa B (NFkB).29,30 NFkB signalling plays a role in the regulation of adhesion molecules, cytokines, chemokines, growth factors and matrix metalloproteinases (MMPs) in vascular cells.31,32 Varicose veins have been found to be associated with upregulation of growth factors, including vascular endothelial growth factor, transforming

growth factor-b1 and acidic fibroblast growth factors.4,33,34 Similarly, inflammatory cytokines and adhesions molecules have been considered to be involved in vein valve insufficiency.35 Alteration in levels and activity of MMPs and their inhibitors have been implicated in the pathogenesis of varicose veins.36 Excess lactate represents increased anaerobic metabolism of pyruvate by the lactate dehydrogenase (LDH) enzyme,37 which occurs in a number of states including hypoxia. Importantly, hypoxia and increased activation of hypoxia-inducible pathway have been recently implicated in the pathogenesis of varicose veins.38,39 Increased activity of the anaerobic fraction of LDH has also been reported in varicose veins suggesting increased anaerobic glycolysis in this context.40 Association of the metabolites lactate and myo-inositol with such key signalling pathways proposes these pathways for future research. To our knowledge, this is the first study to characterise the metabolic profile of varicose veins by NMR spectroscopy. From this study, it is apparent that metabonomic techniques are suitable for characterising vein pathology. A limitation of our study is that the


447

a

b

Figure 4. Principal components analysis (PCA) scores plot of varicose vein and non-varicose vein CPMG spectra, showing variance between samples and the direction of the largest represents a vein sample. Percentage of variation in the NMR data explained by first and second principal components were 79% and 8%, variance in the data. Each - or respectively, in the presence of all replicates (a), and 53% and 19%, respectively, when two out of three replicates were removed from each group (b). Varicose vein samples are grouped together in the unsupervised PCA analysis. Intra-segmental versus inter-sample homogeneity is also noticeable in the non-varicose veins group.

control group did not accurately match with the varicose veins group in terms of demographic features, past medical history and medications. Factors such as age, medications and past medical history may influence the metabolic profile of vein tissue. However, to our knowledge, there has been no demonstration of the effect of age on metabolic profile of vein tissue so far. Statins and most of the other cardiovascular medications have a short half-life (less than 24 h)41 and neither parent statins nor their metabolites were

Figure 5. Orthogonal partial least square (OPLS) analysis separating the two classes (varicose and non-varicose veins) based on their metabolic profiles detected using MAS-1H NMR. Probabilistic quotient normalisation and unit variance settings were applied for data analysis. R2Ycum was 0.788 and Q2cum was 0.406 for orthogonal component. Q2 indicates the fraction of variation predicted by the model for cross validation. One non-varicose vein sample is closer to the varicose veins group (spectrum 5 in Fig. 1).

detected in the NMR profiles. Although this does not evade the possibility that an endogenous metabolic response to chronic use of statins might be detected, in light of the nature of the metabolic changes it seems unlikely that statins would be responsible for any of the observed differences in metabolic profile between the varicose vein and control groups. Each vein segment was washed with deuterium oxide before the experiment. It is challenging to obtain non-varicose vein control tissue from age-matched (often young and otherwise healthy) individuals. Use of inferior epigastric vein retrieved at inguinal hernia surgery may facilitate patient matching. However, this will not reflect the relevant haemodynamic stresses to which a GSV is exposed. We further plan to examine control vein tissue from various anatomical regions, including facial vein at the time of carotid endarterectomy, arm vein during haemodialysis fistula formation, inferior epigastric veins from inguinal hernia repairs and inferior mesenteric vein from bowel resections, to document any metabolic differences between the veins from different anatomical regions and to assign a metabolic profile of non-varicose vein. This study uses 1H MAS NMR spectroscopy to rapidly obtain a picture of the metabolic profile of vein tissue. Furthermore, metabolites can also be extracted from tissue and analysed by NMR. Correlation of metabolic profiles with histopathological analysis of tissue samples may further advance our understanding of the pathophysiology of varicose veins. These approaches will form part of our ongoing investigative plan, to be applied in larger sample sets, for establishing differential metabolic signature of varicose veins as compared to non-varicose veins.

448


Figure 6. O2PLS co-efficient plot showing 1H NMR CPMG spectral data of veins using MAS technology. Spectra have been processed using probabilistic quotient normalisation and unit variance scaling. Red peaks represent highly correlated metabolites to the respective group and are responsible for the distinction between the two groups, yellow are less variable metabolites, and blue the least correlated. Highly correlated metabolites to the varicose veins group are creatine, lactate and myo-inositol. Statistical total correlated spectroscopy (STOCSY) of metabolic features at d3.26 confirms the correlation of myo-inositol peaks.


ROC curve, AUC = 0.92188 1 0.9 0.8

true positive rate

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

false positive rate Figure 7. Receiver operating characteristic (ROC) curve diagnostic analysis for predictive O2PLS model. Area under the curve score of 0.9218 shows excellent predictive accuracy of our model.

Conclusion This preliminary study demonstrates differential metabolic expression in varicose as compared with non-varicose veins. Elucidating the metabolic signature underlying varicose vein disease has the potential to further our understanding of biological mechanisms of disease initiation and progression, identify putative biomarkers and targets for prevention and treatment and merits further explanation. Acknowledgements The authors would like to acknowledge financial support from the Waters Corporation, Imperial College Healthcare Trust Fund, Graham-Dixon Charitable Trust, European Venous Forum, and Royal Society of Chemistry, UK. We acknowledge the support of Olaf Beckonert and Kirill Veselkov for their help in chemometric analysis and performing ROC curve analysis. Ethical Approval Ethical approval was obtained from the Riverside Research Ethics Committee, RREC 3092. Funding Acknowledgement section contains the names of research funding organizations who supported this project. Conflict of Interest None. References 1 Lim CS, Davies AH. Pathogenesis of primary varicose veins. Br J Surg 2009; 96(11):1231e42. 2 Michiels C, Bouaziz N, Remacle J. Role of the endothelium and blood stasis in the development of varicose veins. Int Angiol 2002;21:18e25.

449

3 Gillespie DL, Patel A, Fileta B, Chang A, Barnes S, Flagg A, et al. Varicose veins possess greater quantities of mmp-1 than normal veins and demonstrate regional variation in mmp-1 and mmp-13. J Surg Res. 2002;106:233e8. 4 Kowalewski R, Malkowski A, Sobolewski K, Gacko M. Evaluation of afgf/bfgf and fgf signaling pathway in the wall of varicose veins. J Surg Res. 2009;155:165e72. 5 Cario-Toumaniantz C, Boularan C, Schurgers LJ, Heymann MF, Le Cunff M, Leger J, et al. Identification of differentially expressed genes in human varicose veins: Involvement of matrix gla protein in extracellular matrix remodeling. J Vasc Res. 2007;44:444e59. 6 Lee S, Lee W, Choe Y, Kim D, Na G, Im S, et al. Gene expression profiles in varicose veins using complementary DNA microarray. Dermatol Surg 2005;31:391e5. 7 Labropoulos N, Leon L, Kwon S, Tassiopoulos A, Gonzalez-Fajardo JA, Kang SS, et al. Study of the venous reflux progression. J Vasc Surg 2005;41:291e5. 8 Labropoulos N, Tassiopoulos AK, Kang SS, Mansour MA, Littooy FN, Baker WH. Prevalence of deep venous reflux in patients with primary superficial vein incompetence. J Vasc Surg 2000;32:663e8. 9 Nicholson JK, Lindon JC, Holmes E. ‘Metabonomics’: Understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological nmr spectroscopic data. Xenobiotica 1999; 29:1181e9. 10 Mitchell TN, Costisella B. Nmr - from spectra to structures: An experimental approach. NY: Springer-Verlag; 2004. 11 Nicholson JK, Connelly J, Lindon JC, Holmes E. Metabonomics: A platform for studying drug toxicity and gene function. Nat Rev Drug Discov 2002;1:153e61. 12 Bollard ME, Xu J, Purcell W, Griffin JL, Quirk C, Holmes E, et al. Metabolic profiling of the effects of d-galactosamine in liver spheroids using (1)h nmr and mas-nmr spectroscopy. Chem Res Toxicol 2002;15:1351e9. 13 Beckonert O, Coen M, Keun HC, Wang Y, Ebbels TM, Holmes E, et al. Highresolution magic-angle-spinning nmr spectroscopy for metabolic profiling of intact tissues. Nat Protoc 2010;5:1019e32. 14 Chan EC, Koh PK, Mal M, Cheah PY, Eu KW, Backshall A, et al. Metabolic profiling of human colorectal cancer using high-resolution magic angle spinning nuclear magnetic resonance (hr-mas nmr) spectroscopy and gas chromatography mass spectrometry (gc/ms). J Proteome Res. 2009;8:352e61. 15 Moka D, Vorreuther R, Schicha H, Spraul M, Humpfer E, Lipinski M, et al. Biochemical classification of kidney carcinoma biopsy samples using magicangle-spinning 1h nuclear magnetic resonance spectroscopy. J Pharm Biomed Anal 1998;17:125e32. 16 Lindon JC, Holmes E, Nicholson JK. Metabonomics techniques and applications to pharmaceutical research & development. Pharm Res. 2006;23:1075e88. 17 Dieterle F, Ross A, Schlotterbeck G, Senn H. Probabilistic quotient normalization as robust method to account for dilution of complex biological mixtures. Application in 1 h nmr metabonomics. Anal Chem. 2006;78:4281e90. 18 Ebbels TM, Lindon JC, Coen M. Processing and modeling of nuclear magnetic resonance (nmr) metabolic profiles. Methods Mol Biol. 2011;708:365e88. 19 Misra S, Fu AA, Anderson JL, Sethi S, Glockner JF, McKusick MA, et al. The rat femoral arteriovenous fistula model: Increased expression of matrix metalloproteinase-2 and -9 at the venous stenosis. J Vasc Interv Radiol 2008;19:587e94. 20 Yap IK, Clayton TA, Tang H, Everett JR, Hanton G, Provost JP, et al. An integrated metabonomic approach to describe temporal metabolic disregulation induced in the rat by the model hepatotoxin allyl formate. J Proteome Res. 2006;5: 2675e84. 21 Coen M, Hong YS, Cloarec O, Rhode CM, Reily MD, Robertson DG, et al. Heteronuclear 1h-31p statistical total correlation nmr spectroscopy of intact liver for metabolic biomarker assignment: Application to galactosamine-induced hepatotoxicity. Anal Chem. 2007;79:8956e66. 22 Wishart DS, Knox C, Guo AC, Eisner R, Young N, Gautam B, et al. Hmdb: A knowledgebase for the human metabolome. Nucleic Acids Res. 2009;37:D603e10. 23 Fonville JM, Maher AD, Coen M, Holmes E, Lindon JC, Nicholson JK. Evaluation of full-resolution j-resolved 1h nmr projections of biofluids for metabonomics information retrieval and biomarker identification. Anal Chem. 2010;82: 1811e21. 24 Spillane M, Schoch R, Cooke M, Harvey T, Greenwood M, Kreider R, et al. The effects of creatine ethyl ester supplementation combined with heavy resistance training on body composition, muscle performance, and serum and muscle creatine levels. J Int Soc Sports Nutr 2009;6:6. 25 Wallimann T, Wyss M, Brdiczka D, Nicolay K, Eppenberger HM. Intracellular compartmentation, structure and function of creatine kinase isoenzymes in tissues with high and fluctuating energy demands: The ‘phosphocreatine circuit’ for cellular energy homeostasis. Biochem J 1992;281(Pt 1):21e40. 26 Kukuljan M, Vergara L, Stojilkovic SS. Modulation of the kinetics of inositol 1,4,5-trisphosphate-induced [ca2+]i oscillations by calcium entry in pituitary gonadotrophs. Biophys J 1997;72:698e707. 27 Zou KH, O’Malley AJ, Mauri L. Receiver-operating characteristic analysis for evaluating diagnostic tests and predictive models. Circulation 2007;115:654e7. 28 Badier-Commander C, Couvelard A, Henin D, Verbeuren T, Michel JB, Jacob MP. Smooth muscle cell modulation and cytokine overproduction in varicose veins. An in situ study. J Pathol 2001;193:398e407. 29 Rapiejko PJ, Northup JK, Evans T, Brown JE, Malbon CC. G-proteins of fat-cells. Role in hormonal regulation of intracellular inositol 1,4,5-trisphosphate. Biochem J 1986;240:35e40. 30 Sansilvestri-Morel P, Rupin A, Jullien ND, Lembrez N, Mestries-Dubois P, Fabiani JN, et al. Decreased production of collagen type iii in cultured smooth muscle cells from varicose vein patients is due to a degradation by mmps: Possible implication of mmp-3. J Vasc Res. 2005;42:388e98.

450


31 Sprague AH, Khalil RA. Inflammatory cytokines in vascular dysfunction and vascular disease. Biochem Pharmacol 2009;78:539e52. 32 Bond M, Chase AJ, Baker AH, Newby AC. Inhibition of transcription factor nfkappab reduces matrix metalloproteinase-1, -3 and -9 production by vascular smooth muscle cells. Cardiovasc Res. 2001;50:556e65. 33 Hollingsworth SJ, Powell G, Barker SG, Cooper DG. Primary varicose veins: Altered transcription of vegf and its receptors (kdr, flt-1, soluble flt-1) with sapheno-femoral junction incompetence. Eur J Vasc Endovasc Surg 2004; 27:259e68. 34 Jacob T, Hingorani A, Ascher E. Overexpression of transforming growth factorbeta1 correlates with increased synthesis of nitric oxide synthase in varicose veins. J Vasc Surg 2005;41:523e30. 35 Satokowa HHS, Igari T, Iwaya F, Midorikawa H. The appearance of cytokines and adhesion molecules in saphenous vein valves in chronic venous insufficiency. Phlebology 2002;16:106e10.

36 Aravind B, Saunders B, Navin T, Sandison A, Monaco C, Paleolog EM, et al. Inhibitory effect of timp influences the morphology of varicose veins. Eur J Vasc Endovasc Surg 2010;40:754e65. 37 Takase S, Pascarella L, Bergan JJ, Schmid-Schonbein GW. Hypertension-induced venous valve remodeling. J Vasc Surg 2004;39:1329e34. 38 Michiels C, Bouaziz N, Remacle J. Role of the endothelium and blood stasis in the appearance of varicose veins. Int Angiol 2002;21:1e8. 39 Lim CS, Kiriakidis S, Paleolog EM, Davies AH. Increased activation of the hypoxia-inducible factor pathway in varicose veins. J Vasc Surg 2012. 40 Pascarella L, Lulic D, Penn AH, Alsaigh T, Lee J, Shin H, et al. Mechanisms in experimental venous valve failure and their modification by daflon 500 mg. Eur J Vasc Endovasc Surg 2008;35:102e10. 41 Raffetto JD, Qiao X, Beauregard KG, Tanbe AF, Kumar A, Mam V, et al. Functional adaptation of venous smooth muscle response to vasoconstriction in proximal, distal, and varix segments of varicose veins. J Vasc Surg 2010;51:962e71.