University of Southern Denmark

High plasma microfibrillar-associated protein 4 is associated with reduced surgical repair in abdominal aortic aneurysms

Lindholt, Jes Sanddal; Madsen, Mathilde; Kirketerp-Møller, Katrine Lindequist; Schlosser, Anders; Kristensen, Katrine Lawaetz; Andersen, Carsten Behr; Sorensen, Grith Lykke

Published in: Journal of Vascular Surgery

DOI: 10.1016/j.jvs.2019.08.253

Publication date: 2020

Document version: Final published version

Document license: CC BY-NC-ND

Citation for pulished version (APA): Lindholt, J. S., Madsen, M., Kirketerp-Møller, K. L., Schlosser, A., Kristensen, K. L., Andersen, C. B., & Sorensen, G. L. (2020). High plasma microfibrillar-associated protein 4 is associated with reduced surgical repair in abdominal aortic aneurysms. Journal of Vascular Surgery, 71(6), 1921-1929. https://doi.org/10.1016/j.jvs.2019.08.253

Go to publication entry in University of Southern Denmark's Research Portal

Terms of use This work is brought to you by the University of Southern Denmark. Unless otherwise specified it has been shared according to the terms for self-archiving. If no other license is stated, these terms apply:

• You may download this work for personal use only. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying this open access version If you believe that this document breaches copyright please contact us providing details and we will investigate your claim. Please direct all enquiries to [email protected]

Download date: 28. Sep. 2021 High plasma microfibrillar-associated protein 4 is associated with reduced surgical repair in abdominal aortic aneurysms

Jes Sanddal Lindholt, PhD,a Mathilde Madsen, BSc,b Katrine Lindequist Kirketerp-Møller, PhD,b Anders Schlosser, PhD,b Katrine Lawaetz Kristensen, MD,c Carsten Behr Andersen, PhD,a and Grith Lykke Sorensen, PhD, DMSci,d,e Viborg and Odense, Denmark

ABSTRACT Objective: Identifying biomarkers for abdominal aortic aneurysms (AAA) could prove beneficial in prognosis of AAA and thus the selection for treatment. Microfibrillar-associated protein 4 (MFAP4) is an protein that is highly expressed in aorta. MFAP4 is involved in several tissue remodeling-related diseases. We aimed to investigate the potential role of plasma MFAP4 (pMFAP4) as a biomarker of AAA. Methods: Plasma samples and data were obtained for 504 male AAA patients and 188 controls in the Viborg Vascular (VIVA) screening trial. The pMFAP4 levels were measured by Alphalisa. The Mann-Whitney U test assessed differences in pMFAP4 levels between the presence and absence of different exposures of interest. The correlation between pMFAP4 and aorta growth rate were investigated through spearman’s correlation analysis. Immunohistochemistry and multiple logistic regression adjusted for potential confounders assessed the association between pMFAP4 and AAA. Multiple linear regression assessed the correlation between pMFAP4 and aorta growth rate. Cox regression and competing risk regression were used to investigate the correlation between AAA patients with upper tertile pMFAP4 and the risk of undergoing later surgical repair. Results: A significant negative correlation between pMFAP4 and aorta growth rate was observed using spearman’s correlation analysis (r ¼0.14; P ¼ .0074). However, this finding did not reach significance when applying multiple linear regression. A tendency of decreased pMFAP4 was observed in AAA using immunohistochemistry. Competing risk regression adjusted for potential confounders indicated that patients with upper tertile pMFAP4 had a hazard ratio of 0.51 (P ¼ .001) for risk of undergoing later surgical repair. Conclusions: High levels of pMFAP4 are associated with a decreased likelihood of receiving surgical repair in AAA. This observation warrants confirmation in an independent cohort. (J Vasc Surg 2019;-:1-9.) Keywords: MFAP4; Abdominal aorta aneurysms; Surgical repair; Growth rate; Extracellular matrix

An abdominal aortic aneurysm (AAA) is a severe cardio- prevalence of AAA is 3.3% in men ages 65 to 74, and vascular disease with a very high mortality rate. It is clin- 0.7% in woman ages 61 to 76.2,3 Patients rarely display ically defined as a permanent and localized dilation of symptoms before rupture, and once the aneurysm rup- the aorta of at least 50% of the normal vessel diameter tures the survival rate is only 10% to 15%.4 Diagnosis or being at least 3 cm in diameter.1 Population-based and prognosis currently depend on technologies such ultrasound screening for AAA has showed that the as ultrasound examination and computed tomography

From the Cardiovascular Research Unit, Viborg Hospital, Viborga; the Cancer Author conflict of interest: A.S. and G.L.S. are inventors of U.S. Patent No. and Inflammation Research, Department of Molecular Medicine, University 9,988,442 and EP17199552.5 owned by University of Southern Denmark. of Southern Denmark, Odenseb; the Department of Cardiac, Thoracic, and Correspondence: Grith Lykke Sorensen, PhD, DMSci, Department of Molecular Vascular Surgery,c and Centre of Individualized Medicine in Arterial Diseases Medicine, Cancer and Inflammation Research, University of Southern (CIMA), Department of Cardiothoracic and Vascular Surgery,d Odense Univer- Denmark, J.B. Winsløws Vej 25, 3rd floor 5000, Odense, Denmark (e-mail: sity Hospital, Odense; and the Vascular Research Unit, Department of [email protected]). Vascular Surgery, Viborg Regional Hospital, Viborg.e The editors and reviewers of this article have no relevant financial relationships to Funded by the Danish Research Council, The Lundbeck Foundation, The Novo disclose per the JVS policy that requires reviewers to decline review of any Nordisk Foundation, Brødrene Hartmann Foundation, Augustinus Founda- manuscript for which they may have a conflict of interest. tion, The A.P. Møller Foundation for the Advancement of Medical Science, 0741-5214 Direktør Kurt Bønnelycke og Hustru fru Grethe Bønnelyckes Fond, The Maj- Copyright Ó 2019 The Authors. Published by Elsevier Inc. on behalf of the So- gaards Eftf. Fru Lily Benthine Lunds Fond af 1.6. 1978, Henry og Astrid Møllers ciety for Vascular Surgery. This is an open access article under the CC BY- Fond, Torben og Alice Frimodts Fond, Handelsgartner Ove William Buhl Ole- NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). sen og ægtefælle fru Edith Buhl Olesen Mindelegat, Bagermester August https://doi.org/10.1016/j.jvs.2019.08.253 Jensen og Hustrus Legat. The funding bodies had no roles in the design of the study and collection, analysis, interpretation of data or in writing the manuscript.

1 2 Lindholt et al Journal of Vascular Surgery --- 2019 scanning. Biomarkers could be helpful, particularly in the prognosis of AAA, including the need for preventive ARTICLE HIGHLIGHTS repair and rupture and in understanding the individual d Type of Research: Prospective case control study pathogenesis of AAA development. d Key Findings: In this study of 504 male abdominal The extracellular matrix (ECM) protein microfibrillar- aortic aneurysm (AAA) patients and 188 controls associated protein 4 (MFAP4) is highly expressed in hu- AAA patients with upper tertile plasma microfi- man elastic tissues and localized in elastic fibers in blood brillar-associated protein 4 (pMFAP4), 49% had vessels.5 MFAP4 binds to ECM fibers including collagen, reduced likelihood of receiving later surgical repair tropoelastin, and fibrillin, and it has been shown to pro- (subhazard ratio, 0.51; 95% confidence interval, 0.35- mote elastic fiber formation in vitro.6-10 This role was 0.76; P ¼ .001). partially supported by observations of pulmonary airspace d Take Home Message: AAA patients with the highest enlargement and elastic defects in MFAP4-deficient levels of pMFAP4 have significantly reduced risk of mice.11 Furthermore, MFAP4 has been shown to bind receiving later surgical repair. RGD-dependent integrins and induce smooth muscle cell proliferation and migration, and to promote mono- cyte chemotaxis.12 These effects might be part of the screening by three mobile teams at 14 venues. Enrolment mechanism by which MFAP4 is involved in several tissue started October 2008 and stopped in January 2011. Over- remodeling-related diseases including liver fibrosis, pul- all, 18,749 men attended the screening.24 Participants monary airspace enlargement, bronchial hyperplasia in with positive test results were offered secondary pro- allergic asthma, and arterial neointimal formation.11-14 phylaxis and/or referred to their general practitioner. A Circulating MFAP4 may reflect a spillover from vascular major registry-based follow-up report including AAA wall EMC turnover and has been suggested as a potential repair after 5 years was recently published.2 biomarker for chronic obstructive pulmonary disease, liver Diagnostic criterium for AAA was aortic diameter of fibrosis and cirrhosis, and peripheral artery disease.5,13,15-19 least 30 mm.24 AAA was diagnosed in 615 cases (3.3%; Several studies have supported that MFAP4 levels may 95% confidence interval [CI], 3.0-3.6).25 Cases with an be influenced by the presence of vascular aneurysms; AAA of greater than 50 mm were referred for vascular however, the regulation of the protein has shown inconsis- surgical evaluation, and those less than 50 mm were tency in these studies. Two studies, both including prote- invited annually to a control scan to check for expansion. omic analyses of human AAA tissue, showed decreased Plasma samples were obtained from 711 subjects, but MFAP4 expression in AAA patients.20,21 Another study of data were not available for all included in the cohort. patients with , a disorder of connective Controls were chosen as age-matched screening at- tissue associated with AAA development, showed upre- tenders. Plasma and DNA samples were taken from pa- gulation of MFAP4 in aneurysmal tissue.22 These observa- tients with AAA and 188 controls at baseline (original tions were supported by a recent study of internal time of screening) for biobanking. Hypertension and mammary arteries from patients with AAA, which showed chronic obstructive pulmonary disease were self- MFAP4 to be upregulated.23 reported at baseline. The aim of this study was to test the hypothesis that The trial was approved by the scientific ethical commit- variation in plasma MFAP4 (pMFAP4) could serve as tee of the Mid Jutland Region (Danish: Region Midtjyl- biomarker of AAA and the natural history of AAA using land) (M20080018) and registered in the Clinical trials the Viborg Vascular (VIVA) screening trial.2 register (NCT00662480).

METHODS Ethics approval and consent to participate. Written Designs. We performed a case control study using informed consent was obtained from each patient cases of AAA and healthy controls in the VIVA screening included in the study. The study protocol conforms to trial to test performance of pMFAP4 as independent the ethical guidelines of the 1975 Declaration of Helsinki biomarker for AAA. We also undertook a prospective and the study protocol has previously been approved by cohort study testing pMFAP4 as a prognostic biomarker the Research Ethic Committee for Southern Denmark for aneurysm growth and for later surgical repair during (S-20080140, S-20100044 and S-20090082). surveillance in the VIVA screening trial. Measurement of pMFAP4. pMFAP4 was measured The VIVA screening trial. VIVA is a randomized, clini- using a modified AlphaLISA immunoassay (Perkin Elmer, cally controlled study designed to evaluate the benefits Waltham, Mass) as described elsewhere.5 Monoclonal of vascular screening and modern vascular prophylaxis anti-MFAP4 (HG-HYB 7-14) antibody was conjugated to in a population of 50,000 men aged 65 to 74 years. A AlphaLISA Acceptor beads (Perkin Elmer) at a concen- combined screening program for AAA, peripheral artery tration of 0.1 mg antibody/mg beads following the disease, and hypertension setup included decentralized manufacturer’s instructions. Monoclonal anti-MFAP4 Journal of Vascular Surgery Lindholt et al 3 Volume -, Number -

(HG-HYB 7-18) was modified by labeling with (þ)-biotin analyses were then performed with the resulting b-coef- N-hydroxysuccinimide ester (Sigma H1759; Sigma, St ficient being the growth rate. With this analysis, outliers Louis, Mo) to permit binding to AlphaLISA streptavidin- were regressed towards the mean. coated donor beads. The AlphaLISA procedure was per- As previously described, predefined logistic regression formed using 384-well microtiter plates (white opaque analysis was used to identify whether the upper tertile OptiPlate, Perkin Elmer) containing 5 mL of diluted serum pMFAP4 at baseline was independently associated with (final dilution, 1:100), 2 nmol/L biotinylated HG-HYB 7-18, AAA after adjustment for potential confounders identi- and 10 mg/mL HG-HYB 7-14 conjugated to acceptor fied in the univariate analyses by a P value below 0.10.27 beads in a total of 20 mL AlphaLISA HiBlock Buffer (Per- Multiple linear regression analysis was used to identify kinElmer). The reaction mixture was incubated at room whether the level of pMFAP4 at baseline was indepen- temperature for 60 minutes. Streptavidin donor beads dently associated with aneurysmal growth rate after were then added to reach a final concentration of 40 mg/ similar adjustment for potential confounders. We con- mL, and the plate was incubated at room temperature in structed a Cox proportional hazard model to study the dark for another 30 minutes, after which time it was whether AAA patients at the upper tertile at baseline read on an EnVision reader (PerkinElmer) using the had an altered likelihood of receiving later surgical repair. AlphaScreen protocol. Briefly, the AlphaScreen protocol The model was adjusted for potential confounders iden- used the AlphaScreen 570 emission filter, a flash/time tified in the univariate analyses by a P value of less than ratio of 0.55, a measurement height of 1 mm, an excita- .10. Patients entered the survival analysis at baseline tion time of 0.18 second, and an emission time of (time of screening) and were censored at end of follow- 0.37 second. The experiments were performed in dupli- up on December 31, 2014, or death. cate except for the standards and quality controls, which Finally, we estimated subhazard ratios (SHR) with 95% were performed in quadruplicate. The duplicate sample CI using competing risk regression. This method was covariance was accepted if it was 10% or less. Standards chosen to account for informative censoring during the were prepared by the serial dilution of recombinant long follow-up of this study. Time was measured from MFAP4 (rMFAP4) overexpressing CHO cell culture su- baseline screening until a participant underwent surgery, pernatant in AlphaLISA HiBlock Buffer. Standards died, or reached the end of follow-up, whichever came included serial dilutions from 8000 to 7.8 mU/mL. The first. We adjusted for the same confounders as in the concentration of rMFAP4 was determined using Amino main analysis. Acid Analysis by Alphalyse A/S (Odense, Denmark). When Categorical variables were treated as dummy variables. measured in serum, 1 U/mL equals 38 ng/mL. Quality The residuals of the regression analyses were plotted in controls were prepared from human serum pools and histograms and q-q plots to test for normal distribution. from rMFAP4 pools. One pool was prepared to contain a SPSS 21.0 (SPSS Inc, Chicago, Ill) and Stata Release 15.1 low content of MFAP4 (QLow), and two pools were (StataCorp, College Station, Tex) were used as statistical spiked with purified rMFAP4 (QMid) and (QHigh). The packages. three quality controls were included in each plate. Human tissue samples and histochemical analysis. Interplate variation was accepted if all quality control Nonaneurysmal aorta sections (n ¼ 6) were sampled from measurements were within two standard deviations the pararenal aorta from kidney donors with brain death. obtained from more than 10 consecutive runs. The tissue was fixed in formaldehyde and paraffin Statistical analyses. The statistical tests were defined a embedded from the deceased organ donors with the priori and in a similar way as in previous publica- authorization of the French Biomedicine Agency (PFS tions.19,26,27 Nonparametric statistics were used for 09-007) as previously described.28 Aneurysmal aortas descriptive and univariate analyses. The assumption of (n ¼ 22) were collected from the affected infrarenal region normality of continuous measurements was confirmed of patients undergoing surgical repair in the VIVA by visual inspection of residual log-log plots. This sup- screening trial (registration number NCT00662480). These ported the use of parametric statistics. Differences in the samples were collected from the anterior aorta from the level of pMFAP4 between cases and controls were most dilated part.24 The collection was approved by the assessed by the Mann-Whitney U test. Spearman’s cor- regional scientific ethics committee (M20080028) and by relation coefficient was used to study the univariate as- the Danish national data protection authorities. Aneu- sociation between the level of pMFAP4 and aneurysmal rysmal aortas were frozen before fixation and paraffin growth rate as well as potential numerical confounders. embedding. Human aorta data were analyzed anony- Growth rates were calculated based on ultrasound mously. Immunohistochemical staining was carried out scans from the baseline and annual follow-up visits. using the monoclonal anti-MFAP4 antibody as previously Regarding each case followed, their maximal aortic described.5 Histology slides were scanned at 20 magni- diameter was plotted using observation time as x-axis fication using a NanoZoomer-XR (Hamamatsu Photonics, and aortic diameter as the y-axis. Linear regression Hamamatsu City, Japan), and image acquisition was 4 Lindholt et al Journal of Vascular Surgery --- 2019

Table I. Baseline characteristics of continuous variables for abdominal aortic aneurysm (AAA) patients and controls Baseline data Controls (n ¼ 187-188) AAA (n ¼ 503-504) P value Age, years 69.3 [66.9-72.1] 69.9 [67.5-72.4] .031 BMI, kg/m2 25.9 [24.2-27.8] 27.1 [24.8-29.5] <.001 Systolic BP, mm Hg 145 [136-160] 155 [142-168] <.001 Diastolic BP, mm Hg 80 [75-87] 87 [80-95] <.001 Lowest ABI, units 1.10 [1.02-1.19] 0.96 [0.88-1.04] <.001 Aorta diameter, mm 18.1 [16.7-20.1] 37.2 [32.8-44.3] <.001 pMFAP4, U/mL 9.7 [6.6-13.3] 8.9 [6.4-12.6] .171 ABI, Ankle-brachial index; BMI, body mass index; BP, blood pressure; pMFAP4, plasma microfibrillar-associated protein 4. Data are presented as median [interquartile range].

Table II. Baseline characteristics of categorical variables for controls and abdominal aortic aneurysm (AAA) patients Clinical outcomes Controls (n ¼ 186-188) AAA (n ¼ 503-504) P value Smoking 35 (18.6) 205 (40.7) <.001 First-degree relative with AAA 7 (3.8) 35 (7.0) .120 Diabetes mellitus 26 (14.0) 56 (11.1) .302 COPD 2 (1.1) 20 (4.0) .053 Hypertension 78 (42.2) 269 (53.5) .008 Previous peripheral artery disease 1 (0.5) 7 (1.4) .348 Previous acute myocardial infarct 3 (1.6) 30 (6.0) .017 Previous stroke 2 (1.1) 17 (3.4) .098 Atherosclerosisa 19 (10.1) 107 (21.2) .001 Glucocorticoid 13 (6.9) 44 (9.0) .386 Low-dose aspirin 47 (25.0) 240 (48.1) <.001 Statins 66 (35.1) 263 (52.8) <.001 AMI, Acute myocardial infarct; COPD, chronic obstructive pulmonary disease; PAD, peripheral artery disease. Data are presented as number (%). aComposite variable for former hospital submission owing to stroke, AMI, angina pectoris, and PAD.

obtained using NDP.view2 software (NanoZoomer Digital Associations between pMFAP4 and baseline vari- Pathology, Hamamatsu Photonics). Representative im- ables. Associations between levels of pMFAP4 and base- ages were automatically adjusted for contrast in Adobe line variables are shown in Table III together with the Photoshop CC 2017 (San Jose, Calif). Positive and negative differences in the level of pMFAP4 in the absence or controls for immunostaining of MFAP4 using antibody presence of different exposures or clinical outcomes. The HG-HYB 7-14 was previously shown using Mfap4þ/þ (wild spearman’s correlation showed a significant negative type) and Mfap4/ (gene-deficient mice). The specificity correlation between pMFAP4 and aneurysmal growth of HG-HYB 7-14 staining in human arteries was previously rate (coeff ¼0.14; P ¼ .0074) as well as between demonstrated in Wulf-Johansson et al.5 pMFAP4 and the confounder diastolic blood pressure, while a significant positive correlation was observed be- RESULTS tween pMFAP4 and the confounders age and body mass Baseline observations and pMFAP4. pMFAP4 was index (BMI). In addition, the Wilcoxon rank-sum test measured in blood samples drawn at initial screening showed a significant positive association between in 711 subjects from the VIVA trial, including both AAA pa- pMFAP4 and the following variables: low-dose aspirin tients and controls.2 Normal distribution was not intake, statins intake, hypertension, diabetes mellitus and observed for pMFAP4, and the median (interquartile atherosclerosis, whereas a significant negative associa- range [IQR]) was found to be 9.3 U/mL (6.5-12.9 U/mL) for tion was observed between pMFAP4 levels and smoking. all subjects. Baseline characteristics are shown in Table I and Table II, but were not available for all 711 subjects. In Association between pMFAP4 and AAA. The Wilcoxon addition, aneurysmal growth rate was calculated during rank-sum test did not show a significant difference in follow-up for AAA patients. The median growth rate was pMFAP4 levels between AAA patients (8.9 U/mL) and found to be 2.5 mm/year (IQR, 1.4-3.7 mm/year). controls (9.7 U/mL) (Table III). However, the upper tertile Journal of Vascular Surgery Lindholt et al 5 Volume -, Number -

Table III. Relationships between plasma microfibrillar-associated protein 4 (pMFAP4) levels and clinical variables measured at baseline pMFAP4 (U/mL) in pMFAP4 (U/mL) in Clinical variables persons with variable persons without variable P valuea Binary distribution AAA 9.7 [6.6-13.3] 8.9 [6.5-12.6] .1709 First-degree relative with AAA 9.3 [6.5-12.9] 9.2 [5.9-12.3] .9650 Smoking 9.8 [7.0-14.1] 7.9 [5.7-10.6] <.0001 Atherosclerosis 8.9 [6.3-12.3] 10.9 [6.9-15.7] .0029 COPD 9.3 [6.5-12.9] 7.9 [5.7-10.6] .1452 Diabetes mellitus 8.9 [6.3-12.5] 11.3 [7.5-16.3] .0002 Hypertension 8.7 [6.2-12.0] 9.7 [6.7-14.1] .0115 Peripheral artery disease 9.1 [6.4-12.8] 9.6 [6.7-13.7] .3698 Previous acute myocardial infarct 9.3 [6.5-12.8] 9.0 [6.5-13.4] .9164 Previous stroke 9.1 [6.5-12.8] 11.2 [6.5-15.9] .2382 Glucocorticoids 9.3 [6.5-12.9] 9.5 [7.0-12.8] .4987 Low-dose aspirin 8.8 [6.1-12.5] 9.6 [6.9-13.3] .0347 Statins 8.7 [5.9-12.2] 9.7 [6.8-13.8] .0069 r P valueb Continuous distribution Age 0.12 .0012 Weight 0.07 .0567 Height 0.01 .8110 BMI 0.08 .0392 Systolic BP 0.05 .2401 Diastolic BP 0.10 .0068 Aorta diameter 0.07 .0705 Lowest ABI 0.001 .9629 AAA, Abdominal aortic aneurysm; ABI, ankle-brachial index; BMI, body mass index; BP, blood pressure; COPD, chronic obstructive pulmonary disease. aWilcoxon rank-sum test of pMFAP4 variation between individuals with or without clinical variables with binary distribution listed in the leftmost column. pMFAP4 measurements are presented as median [interquartile range]. bSpearman rank correlation was used to investigate the relation between pMFAP4 and variables with continuous distribution. r [ Spearman’s rank correlation coefficient.

pMFAP4 was associated with 30% lower risk (crude odds adjusting for the potential confounders, cases in the upper ratio [OR], 0.70; 95% CI, 0.49-0.99; P ¼ .042) (Table IV). tertile showed 37% reduced risk of later surgical repair However, the association did not reach significance after (adjusted HR, 0.63; 95% CI,0 [.40 to 0.99; P ¼ .044). adjustment for age, BMI, diastolic blood pressure, Next, a competing risk analysis was performed smoking, hypertension, diabetes mellitus, atherosclerosis, (Table VI) and showed that AAA patients with upper ter- low-dose aspirin, and statins without changing the tile pMFAP4 had a significant SHR of .53 (95% CI, 0.37 calculated OR suggesting lack of power (adjusted OR, to 0.75; P < .001) for later surgical repair, and when 0.71; IQR, 0.47-1.08; P ¼ .110) (Table IV). Likewise, in uni- adjusting for the potential confounders using this statis- variate analysis, the level of pMFAP4 was significantly tical model, cases in the upper tertile showed 49% inversely associated with annual aneurysmal growth rate reduced risk of later surgical repair (adjusted SHR, 0.51; (rho ¼0.14; P ¼ .0074). However, in a multivariate linear 95% CI, 0.35 to 0.76; P ¼ .001). regression analysis with the above-mentioned potential Localization of MFAP4 in human aortic tissues. Immu- confounders, significance was not maintained nohistochemical staining indicated a trend for (coefficient ¼0.02; IQR, 0.06 to 0.02; P ¼ .297). decreased MFAP4 immunodetection in aortic tissues Association between pMFAP4 and later surgical repair from AAA patients compared to healthy controls (Fig). in AAA. Univariate Cox regression analysis (Table V) In total, 6 tissue sections were obtained from control showed that AAA patients with upper tertile pMFAP4 had aortas and 22 tissue sections was obtained from AAA a significant hazard ratio of 0.57 (95% CI, 0.38 to 0.85; aortas. Huge variation in MFAP4 staining intensity and P ¼ .005) for undergoing later surgical repair, and when distribution between samples both for controls and AAA 6 Lindholt et al Journal of Vascular Surgery --- 2019

Table IV. Logistic regression showing the association be- Table V. Adjusted Cox regression analysis of the need for tween presence of abdominal aortic aneurysm (AAA) and later surgical repair in patients with an abdominal aortic plasma microfibrillar-associated protein 4 (pMFAP4) and aneurysm (AAA) additional risk factors HR 95% CI P value OR 95% CI P value Univariate analysis Univariate analysis pMFAP4 upper tertile 0.57 0.38-0.85 .005 pMFAP4 upper tertile 0.70 0.49-0.99 .042 Multivariate analysis Multivariate analysis pMFAP4 upper tertile 0.63 0.40-0.99 .044 pMFAP4 upper tertile 0.71 0.47-1.08 .110 Age 0.98 0.92-1.04 .419 Age 1.08 1.00-1.16 .038 BMI 1.10 0.84-1.45 .478 BMI 1.11 1.05-1.18 .001 Smoking 1.11 0.77-1.59 .570 < Diastolic BP 1.07 1.05-1.09 .001 Hypertension 1.05 0.72-1.53 .795 < Smoking 4.34 2.68-7.01 .001 Diastolic blood pressure 1.01 0.99-1.02 .444 Hypertension 1.28 0.85-1.93 .241 Diabetes mellitus 0.65 0.33-1.28 .213 Diabetes mellitus 0.57 0.30-1.06 .075 Atherosclerosis 0.87 0.51-1.48 .601 Atherosclerosis 1.80 0.96-3.38 .068 Low-dose aspirin 0.68 0.44-1.06 .088 < Low-dose aspirin 2.67 1.65-4.32 .001 Statins 1.40 0.89-2.18 .143 Statins 1.52 0.94-2.46 .090 AAA diameter 1.07 1.06-1.09 <.001 fi BMI, Body mass index; BP, blood pressure; CI, con dence interval; OR, BMI, Body mass index; CI, confidence interval; HR, hazard ratio; odds ratio. pMFAP4, plasma microfibrillar-associated protein 4. All presented variables were included as covariates into one logistic All presented variables were included as covariates into one Cox regression analysis. regression analysis. was observed. The Fig shows examples of the varying intensities and distribution of MFAP4 staining in normal control (N1-3) and AAA tissue (AAA1-4). Despite over- Table VI. Competing risk analysis of the need for later surgical repair in patients with an abdominal aortic lapping staining intensities, there was a clear tendency aneurysm (AAA) for higher MFAP4 staining in healthy tissues compared to AAA tissues. Three of six tissue sections from healthy SHR 95% CI P value donors had N3-type MFAP4 staining intensity, whereas Univariate analysis only 1 of 22 tissue sections from AAA had AAA4-type pMFAP4 upper tertile 0.53 0.37-0.75 <.001 MFAP4 staining intensity and the majority of these Multivariate analysis were of the AAA1/2 types. pMFAP4 upper tertile 0.51 0.35-0.76 .001 DISCUSSION Age 0.98 0.93-1.04 .584 In the present study, we analyzed the association be- BMI 1.11 0.88-1.40 .364 tween pMFAP4 and the risk of AAA, aneurysmal growth Smoking 1.08 0.79-1.49 .628 rate, and later surgical repair, respectively, through a Hypertension 1.43 1.04-1.96 .028 combination of statistical analyses and immunohisto- Diastolic blood pressure 1.00 0.98-1.01 .756 chemistry. We found an inverse correlation between Diabetes mellitus 0.89 0.54-1.48 .665 pMFAP4 the intermediate clinical endpoints aneurysm Atherosclerosis 0.69 0.46-1.04 .078 size and growth rate. The correlation was only significant Low-dose aspirin 0.81 0.58-1.15 .247 fi for growth rate and this signi cance was lost after correc- Statins 1.15 0.80-1.64 .460 tion for confounders. However, significant association AAA diameter 1.02 1.01-1.03 <.001 was observed for the end point of later surgical repair, BMI, body mass index; CI, confidence interval; pMFAP4, plasma which is a variable that incorporates aneurysm size, microfibrillar-associated protein 4; SHR, subhazard ratio. growth rate, and the presence of symptoms. We All presented variables were included as covariates into the competing risk analysis. observed an association between the highest pMFAP4 levels and a 49% reduced risk of later surgical repair. Moreover, we observed an overlap in pMFAP4 levels be- Biomarkers may be helpful in the prognosis of AAA by tween cases and controls, with a tendency for decreased indicating the immediate risk of rupture and need for pMFAP4 in AAA tissue when observed using surgical repair. Currently, only a small number of circu- immunohistochemistry. lating biomarkers have convincingly been linked to No relevant guidelines exist for this study of AAA but AAA and none have been translated into clinical prac- the patients were offered operation in accord with rec- tice.30,31 Studies of new biomarker candidates are thus ommendations from Wanhainen.29 warranted. Journal of Vascular Surgery Lindholt et al 7 Volume -, Number -

Fig. Varying intensities and distribution of microfibrillar-associated protein 4 (MFAP4) in aortic tissue obtained from controls and patients with an abdominal aortic aneurysm (AAA). Immunohistochemical staining showed variation of intensity and distribution of MFAP4 in aortic tissue obtained from normal controls (N1-3) and AAA patients (AAA1-4). Tissues from controls were obtained from the pararenal aorta, and tissues from AAA patients were obtained from the infrarenal anterior aorta. Tissues were stained using monoclonal anti-MFAP4 antibody and counterstained with Mayers hematoxylin. Scale bars: 250 mm.

Previous studies of circulating MFAP4 have demon- tertile pMFAP4 is associated with 30% lower risk of AAA. strated associations with other diseases including cardio- The relationship between pMFAP4 and AAA may thus vascular disease. To date, the strongest association not be linear and only the highest levels of pMFAP4 observed between circulating MFAP4 and clinical dis- seemed to be associated with reduced AAA outcomes. ease is in liver fibrosis/cirrhosis, where serum MFAP4 is Moreover, when adjusting for risk factors age, BMI, to- suggested to serve as an efficient biomarker by signifi- bacco smoking, diastolic blood pressure, hypertension, cant differentiation between stages of fibrosis.15,17 diabetes mellitus, atherosclerosis, low-dose aspirin, and Whereas the Wilcoxon rank-sum test did not show any statins the OR stayed the same, although the significance significant difference in pMFAP4 levels between AAA pa- was lost, indicating that that our analyses are at the tients and controls, univariate analyses showed that upper borderline of sufficient statistical power. We used current 8 Lindholt et al Journal of Vascular Surgery --- 2019 smoking, and not pack-years as a confounder in the anal- owing to law and reimbursement.33,34 Consequently, ysis because we have previously demonstrated that pack- all had follow-up without missing data. The analyses years do not influence circulating MFAP4 variation.32 were adjusted for known potential risk factors for We observed a substantial variation in the immunode- AAA, and the progression of AAA, however, by nature tection of MFAP4 for both AAA and controls. All tissue therewillalwaysbeariskofresidualconfounding sections were obtained from the same aorta locations meaning that a causal effect of pMFAP4 directly or in both groups, so differences in MFAP4 immunodetec- indirectly reducing AAA growth cannot be inferred tion caused by variation in tissue sampling seem unlikely. from these data. There was a clear indication of higher levels of MFAP4 immunodetection in control tissue compared to AAA tis- CONCLUSIONS sue. However, the AAA and control tissues were prepared The study shows that patients with the highest levels of at two different hospitals and AAA tissue sections were pMFAP4 have significantly reduced risk of later surgical fi frozen before xation, which may have resulted in a repair. Owing to overlap in distribution of pMFAP4 be- loss of antigen retrieval. Moreover, the samples were tween patients and controls, the highest pMFAP4 levels anonymous and we do not know the patient ages. have potential to serve as biomarker in a biomarker Nevertheless, the distribution of vascular MFAP4 was panel and future studies may help inform a suitable fl markedly altered in AAA, both re ecting a thinner cut off. Moreover, validating analyses in larger indepen- MFAP4 positive tunica media layer and loss of immuno- dent cohorts are needed to confirm the associations fl detection in highly amorph or in amed areas. observed in this study. One reason for lower MFAP4 protein expression in AAA fl may re ect the increased expression of proteases or the The authors thank the staff at the Institute of Pathology, loss of contractile vascular smooth muscle cells, which Odense University Hospital, for performing the paraffin are the site of synthesis of MFAP4 and a character feature sections and immunohistochemical staining of MFAP4, 12 of the pathogenesis of AAA. In contrast, MFAP4 has Prof. Jean Baptiste Michel, INSERM, CHU Xavier Bichat, fi been shown to bind elastic bers as well as tropoelastin France for kindly donating healthy human aorta sections and actively promote tropoelastin coacervation and and Vicki Nielsen, Institute of Molecular Medicine, Univer- thereby MFAP4 seem to have an important role in elastic sity of Southern Denmark for performing pMFAP4 fi 6 ber formation. The depletion of MFAP4 in AAA could measurements. thus partially explain the thinning of the elastic media. The cause of the pMFAP4 variation within both control AUTHOR CONTRIBUTIONS and AAA patients is uncertain but might reflect that in- Conception and design: JL, GS flammatory processes in AAA as well as non-AAA tissue Analysis and interpretation: JL, MM, KKM, AS, KK, GS of the elderly causes a reduction of MFAP4 and thus a Data collection: JL, CA reduced turnover of the protein with a reduced spillover Writing the article: JL, MM, KKM, KK, GS to the circulation. Disease- or age-induced processes are Critical revision of the article: JL, MM, KKM, AS, CA, GS the most likely causes of the observed variance of circu- Final approval of the article: JL, MM, KKM, AS, CA, KK, GS 18 lating MFAP4 since the heritability is relatively low. A Statistical analysis: JL, KK contributing factor to the variation within AAA could also Obtained funding: KKM, GS be these changes in ECM composition after disease pro- Overall responsibility: GS gression and the degradation of the tunica media layer, causing a decrease in the expression of MFAP4. According to this notion, the fraction of AAA patients who have sus- tained high levels of MFAP4 within the vascular wall would REFERENCES 1. Kent KC. Clinical practice. Abdominal aortic aneurysms. be those with decreased risk later surgical repair. N Engl J Med 2014;371:2101-8. A limitation of the present study is that it is a prelim- 2. Lindholt JS, Sogaard R. Population screening and interven- inary finding, and that observations need validation in a tion for vascular disease in Danish men (VIVA): a randomised separate sample. To our present knowledge, such vali- controlled trial. Lancet 2017;390:2256-65. dating cohort does not yet exist. Selection bias seems 3. Dahl M, Frost L, Sogaard R, Klausen IC, Lorentzen V, Lindholt J. A population-based screening study for cardio- unlikelyinourstudy,asthestudygroupisattenders vascular diseases and diabetes in Danish postmenopausal in a population-based screening trial with a relative women: acceptability and prevalence. BMC Cardiovasc Dis- high attendance rate (74%).2 Information bias seems ord 2018;18:20. also unlikely, as the ultrasound based measurement 4. Powell JT, Greenhalgh RM. Clinical practice. Small abdom- of aorta was performed by a validated methods inal aortic aneurysms. N Engl J Med 2003;348:1895-901. 5. Wulf-Johansson H, Lock Johansson S, Schlosser A, Trom- showing high position, and information of later AAA melholt Holm A, Rasmussen LM, Mickley H, et al. Localiza- surgical repair was based upon nation-wide registry tion of microfibrillar-associated protein 4 (MFAP4) in human data, where all AAA repair procedures are recorded tissues: clinical evaluation of serum MFAP4 and its Journal of Vascular Surgery Lindholt et al 9 Volume -, Number -

association with various cardiovascular conditions. PLoS One energetic metabolism-related proteins at human abdom- 2013;8:e82243. inal aortic aneurysm sites. J Vasc Surg 2012;55:1124-33. 6. Pilecki B, Holm AT, Schlosser A, Moeller JB, Wohl AP, Zuk AV, 22. Pilop C, Aregger F, Gorman RC, Brunisholz R, Gerrits B, et al. Characterization of microfibrillar-associated protein 4 Schaffner T, et al. Proteomic analysis in aortic media of (MFAP4) as a tropoelastin- and fibrillin-binding protein patients with Marfan syndrome reveals increased activity involved in elastic fiber formation. J Biol Chem 2016;291: of calpain 2 in aortic aneurysms. Circulation 2009;120: 1103-14. 983-91. 7. Lausen M, Lynch N, Schlosser A, Tornoe I, Saekmose SG, 23. Kidholm CL, Beck HC, Madsen JB, Palstrom NB, Lindholt JS, Teisner B, et al. Microfibril-associated protein 4 is present in Rasmussen LM. Preliminary analysis of proteome alterations lung washings and binds to the collagen region of lung in non-aneurysmal, internal mammary artery tissue from surfactant protein D. J Biol Chem 1999;274:32234-40. patients with abdominal aortic aneurysms. PLoS One 8. Toyoshima T, Nishi N, Kusama H, Kobayashi R, Itano T. 36- 2018;13:e0192957. kDa microfibril-associated glycoprotein (MAGP-36) is an 24. Grondal N, Sogaard R, Henneberg EW, Lindholt JS. The -binding protein increased in chick aortae during Viborg Vascular (VIVA) screening trial of 65-74 year old men development and growth. Exp Cell Res 2005;307:224-30. in the central region of Denmark: study protocol. Trials 9. Kasamatsu S, Hachiya A, Fujimura T, Sriwiriyanont P, 2010;11:67. Haketa K, Visscher MO, et al. Essential role of microfibrillar- 25. Grondal N, Sogaard R, Lindholt JS. Baseline prevalence of associated protein 4 in human cutaneous homeostasis abdominal aortic aneurysm, peripheral arterial disease and in its photoprotection. Sci Rep 2011;1:164. and hypertension in men aged 65-74 years from a pop- 10. Brandsma CA, van den Berge M, Postma DS, Jonker MR, ulation screening study (VIVA trial). Br J Surg 2015;102: Brouwer S, Pare PD, et al. A large lung gene expression study 902-6. identifying fibulin-5 as a novel player in tissue repair in 26. Logstrup BB, Masic D, Laurbjerg TB, Blegvad J, Herly M, COPD. Thorax 2015;70:21-32. Kristensen LD, et al. Left ventricular function at two-year 11. Holm AT, Wulf-Johansson H, Hvidsten S, Jorgensen PT, follow-up in treatment-naive rheumatoid arthritis pa- Schlosser A, Pilecki B, et al. Characterization of spontaneous tients is associated with anti-cyclic citrullinated peptide air space enlargement in mice lacking microfibrillar- antibody status: a cohort study. Scand J Rheumatol associated protein 4. Am J Physiol Lung Cell Mol Physiol 2017;46:1-9. 2015;308:L1114-24. 27. Lindholt JS, Kristensen KL, Burillo E, Martinez-Lopez D, 12. Schlosser A, Pilecki B, Hemstra LE, Kejling K, Calvo C, Ros E, et al. Arachidonic acid, but not omega-3 in- Kristmannsdottir GB, Wulf-Johansson H, et al. MFAP4 pro- dex, relates to the prevalence and progression of abdominal motes vascular smooth muscle migration, proliferation and aortic aneurysm in a population-based study of Danish men. accelerates neointima formation. Arterioscler Thromb Vasc J Am Heart Assoc 2018;7:e007790. Biol 2016;36:122-33. 28. Dutertre CA, Clement M, Morvan M, Schakel K, Castier Y, 13. Molleken C, Sitek B, Henkel C, Poschmann G, Sipos B, Alsac JM, et al. Deciphering the stromal and hemato- Wiese S, et al. Detection of novel biomarkers of liver cirrhosis poietic cell network of the adventitia from non- by proteomic analysis. Hepatology 2009;49:1257-66. aneurysmal and aneurysmal human aorta. PLoS One 14. Pilecki B, Schlosser A, Wulf-Johansson H, Trian T, Moeller JB, 2014;9:e89983. Marcussen N, et al. Microfibrillar-associated protein 4 mod- 29. Wanhainen A, Verzini F, Van Herzeele I, Allaire E, Bown M, ulates airway smooth muscle cell phenotype in experi- Cohnert T, et al. European Society for Vascular Surgery mental asthma. Thorax 2015;70:862-72. (ESVS) 2019 clinical practice guidelines on the management 15. Bracht T, Molleken C, Ahrens M, Poschmann G, Schlosser A, of abdominal aorto-iliac artery aneurysms. Eur J Vasc Eisenacher M, et al. Evaluation of the biomarker candidate Endovasc Surg 2018;57:8-93. MFAP4 for non-invasive assessment of hepatic fibrosis in 30. Moris D, Mantonakis E, Avgerinos E, Makris M, Bakoyiannis C, hepatitis C patients. J Transl Med 2016;14:201. Pikoulis E, et al. Novel biomarkers of abdominal aortic 16. Molleken C, Poschmann G, Bonella F, Costabel U, Sitek B, aneurysm disease: identifying gaps and dispelling mis- Stuhler K, et al. MFAP4: a candidate biomarker for hepatic perceptions. Biomed Res Int 2014;2014:925840. and pulmonary fibrosis? Sarcoidosis Vasc Diffuse Lung Dis 31. Chaikof EL, Dalman RL, Eskandari MK, Jackson BM, Lee WA, 2016;33:41-50. Mansour MA, et al. The Society for Vascular Surgery practice 17. Saekmose SG, Mossner B, Christensen PB, Lindvig K, guidelines on the care of patients with an abdominal aortic Schlosser A, Holst R, et al. Microfibrillar-associated protein 4: aneurysm. J Vasc Surg 2018;67:2-77.e2. a potential biomarker for screening for liver fibrosis in a 32. Johansson SL, Roberts NB, Schlosser A, Andersen CB, mixed patient cohort. PLoS One 2015;10:e0140418. Carlsen J, Wulf-Johansson H, et al. Microfibrillar-associated 18. Saekmose SG, Schlosser A, Holst R, Johansson SL, Wulf- protein 4: a potential biomarker of chronic obstructive pul- Johansson H, Tornoe I, et al. Enzyme-linked immunosorbent monary disease. Respir Med 2014;108:1336-44. assay characterization of basal variation and heritability of 33. Grondal N, Bramsen MB, Thomsen MD, Rasmussen CB, systemic microfibrillar-associated protein 4. PLoS One Lindholt JS. The cardiac cycle is a major contributor to 2013;8:e82383. variability in size measurements of abdominal aortic 19. Hemstra LE, Schlosser A, Lindholt JS, Sorensen GL. Microfi- aneurysms by ultrasound. Eur J Vasc Endovasc Surg 2012;43: brillar-associated protein 4 variation in symptomatic pe- 30-3. ripheral artery disease. J Transl Med 2018;16:159. 34. Laustsen J, Jensen LP, Hansen AK, Danish National 20. Didangelos A, Yin X, Mandal K, Saje A, Smith A, Xu Q, et al. Vascular R. Accuracy of clinical data in a population based Extracellular matrix composition and remodeling in human vascular registry. Eur J Vasc Endovasc Surg 2004;27:216-9. abdominal aortic aneurysms: a proteomics approach. Mol Cell Proteomics 2011;10. M111 008128. 21. Modrego J, Lopez-Farre AJ, Martinez-Lopez I, Muela M, Macaya C, Serrano J, et al. Expression of cytoskeleton and Submitted Jan 16, 2019; accepted Aug 15, 2019.