Assessment of a Serum Assay for Quantification ofAbdominal Aortic Calcification

Paula Clancy, Lisa Oliver, Rumal Jayalath, Petra Buttner, Jonathan Golledge

Intimal vascular calcification is an important marker ofatherosclerosis, and its severity is an independent risk

factor for cardiovascular events.1 Measurement of coronaryor aortic calcification requires CT-based imaging and elabo-rate data analysis to ensure accuracy.2 A blood assay whichcould quantify the severity of vascular calcification would beless expensive, avoid exposure to radiation, and be moreaccessible than present imaging based methods. Osteoprote-gerin (OPG) and osteopontin (OPN) are present in humanserum and have been implicated in vascular calcification.3,4 Inthis prospective study we assessed the value of a number ofserum assays for OPG and OPN in determining abdominalaortic calcification. Firstly, we investigated the assay charac-teristics and reproducibility of 5 ELISAs using a subset ofpatients. Based on the findings of these assessments, weselected 3 assays for full evaluation in the entire cohort.

MethodsOne hundred and nine patients with peripheral vascular disease hadfasting serum obtained the morning before measurement of infrare-nal abdominal aortic calcification. We assessed a variety of differentcommercial ELISAs (details are given in the supplemental Methods,available online at http://atvb.ahajournals.org).

ResultsAssessment of Assay CharacteristicsWe identified a large number of commercial ELISAs forOPG with varying reported assay characteristics (see supple-mental Methods). We initially compared the OPG ELISAs interms of recognized assay requirements.5 The calibrationcurves and linearity of the assays were excellent (supplemen-tal Table I). Concordance correlation coefficients for intra-and inter-assay reproducibility were good but most reproduc-ible for the DuoSet assay (supplemental Table II). Each OPGassay uses a different form of OPG as a reference standard(supplemental Figure I). As a measure of accuracy weassessed the ability of the different assays to recognize addedOPG spiked into serum. The mean recovery was 57�7.5%,83�5.0%, and 58�9.0% for the Duoset, Biovendor, andBiomedica ELISAs, respectively (n�8 for each assay). The

physiological ligand RANKL had little effect on the assayresults (see supplemental Results). Two assays for serumOPN were assessed (Quantikine R&D Systems and IBLImmuno-Biological Laboratories). We found similar assaycharacteristics for these kits (mean recovery 90�4.5% and88.3�9.1%, intra- and inter-assay concordance correlationcoefficient 0.990 to 0.994).

Comparison of Serum OPG and OPN inDetermining Aortic CalcificationWe assessed the ability of serum OPG assessed with bothBioVendor and DuoSet ELISAs and OPN measured with theR&D quantikine kit to predict abdominal aortic calcification.The characteristics of the patients entered into the study areshown in supplemental Table III. The mean volume ofinfrarenal abdominal aortic calcification was 153.7�142.5cm2 (range 4 to 5650 cm2). Serum OPG measured by DuoSet(r�0.26, P�0.006) and BioVendor assays (r�0.27,P�0.004) but not serum OPN (r�0.03, P�0.76) was corre-lated with aortic calcification (Figure, a). Patients weredivided into quartiles based on volume of aortic calcification.Serum OPG but not OPN was related to calcification quartile(Figure, b). Receiver operating characteristic curves demon-strated that a cut-off value of OPG of 72 pM (DuoSet;sensitivity 85%, specificity 55%, negative predictive value92% based on prevalence of 25%, area under the curve 0.68,95% CI 0.58 to 0.79, P�0.004) and 8.49 pM (Biovendor;sensitivity 56%, specificity 76%, negative predictive value84% based on prevalence of 25%, area under the curve 0.67,95% CI 0.56 to 0.79, P�0.007) were the most accurate indetermining the highest quartile of aortic calcification (Fig-ure, c).

Comparison of the DuoSet and BioVendor ELISAThe serum OPG concentrations measured by the two OPGELISAs were noted to be very different with limited concor-dance between values (Table). To further investigate thesedifferences the 109 serum samples and opposing manufactur-er’s standards were assessed in each assay. These resultsallowed the calculation of the concentration of the serumsamples by reference to the two different standards in eachassay separately. Using the supplied standards to calculateresults the mean (�standard deviation), OPG concentrationsfor the samples were 78.4�32.4 and 7.59�2.75 pM in theDuoset and Biovendor ELISAs, respectively. The differencein values were markedly reduced by the use of similarstandards in the different ELISA plates; eg, using Biovendorstandards to read the Duoset and Biovendor ELISAs resultedin mean serum OPG concentrations of 9.73�5.86 and7.59�2.75, respectively (Table).

From the Vascular Biology Unit (P.C., L.O., R.J., J.G.), School ofMedicine, James Cook University, and the School of Public Health andTropical Medicine (P.B.), James Cook University, Townsville, Austra-lia.

Correspondence to Professor Jonathan Golledge, the Vascular Biol-ogy Unit, School of Medicine, James Cook University, Townsville,Qld, 4811, Australia. E-mail jonathan.golledge@jcu.edu.au

(Arterioscler Thromb Vasc Biol. 2006;26:2574-2576.)© 2006 American Heart Association, Inc.

Arterioscler Thromb Vasc Biol. is available at http://www.atvbaha.orgDOI: 10.1161/01.ATV.0000242799.81434.7d

2574

Letter to the Editor

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

by guest on A

ugust 19, 2016http://atvb.ahajournals.org/

Dow

nloaded from

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

by guest on A

ugust 19, 2016http://atvb.ahajournals.org/

Dow

nloaded from

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

by guest on A

ugust 19, 2016http://atvb.ahajournals.org/

Dow

nloaded from

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

by guest on A

ugust 19, 2016http://atvb.ahajournals.org/

Dow

nloaded from

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

by guest on A

ugust 19, 2016http://atvb.ahajournals.org/

Dow

nloaded from

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

by guest on A

ugust 19, 2016http://atvb.ahajournals.org/

Dow

nloaded from

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

by guest on A

ugust 19, 2016http://atvb.ahajournals.org/

Dow

nloaded from

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

Discussion

A number of previous studies have assessed the association ofserum calcification markers, such as osteoprotegerin andosteopontin, with abdominal aortic calcification.1,3,4,6–8 Inmany of these studies the reproducibility and accuracy of theassessment methods has not been quantitated, with a numberof investigators relying on subjective techniques such as plainX-ray.3 Other studies have only involved small numbers ofpatients4,6,8 or were reserved to those patients with renalfailure.7 In this study we used a carefully validated CTprotocol to accurately measure the volume of infrarenal aorticcalcification and, in addition, thoroughly assessed the assaycharacteristics for our calcification markers (supplementalTables I and II).2 We found an association between serumconcentrations of OPG but not OPN and severity of infrarenalabdominal aortic calcification (Figure). The association sug-gests the potential value of serum assays for OPG in predict-ing aortic calcification and cardiovascular risk. Indeed, serumconcentrations of OPG have been related to cardiovascularevents.9 However, our study also emphasized that at presentthe available commercial ELISAs for OPG have a number of

differences with marked variation in calculated serum con-centrations with different assays (Table). These variationsappear largely attributable to differences in the standards usedin the ELISAs (Table; supplemental Figure I). Which assaygives values closest to true serum OPG requires furtherassessment, but the greater similarity of the DuoSet standardto full-length OPG suggests this ELISA may be morerepresentative. Further improvements in OPG assays maylead to better sensitivity and specificity in detecting signifi-cant aortic calcification and allow the assay to becomeclinically useful. The relatively low cost of the DuoSetELISA supports the use of this assay in large screeningstudies.

Sources of FundingThis project is supported by grant numbers RO1 HL080010-01 fromthe National Institute of Health, USA, and 279408 from the NationalHealth and Medical Research Council Australia.

DisclosuresNone.

A, Scatter plot of relationship between serum OPG (measured with biovendor assay) and infrarenal abdominal aortic calcification sever-ity. B, Relationship between serum OPG concentration (measured with DuoSet assay) and quartile of abdominal aortic calcificationseverity. Shown are box plots representing median and interquartile range for patients with different quartiles of aortic calcification. C,Receiver operating characteristic curve for serum OPG (measured with both DuoSet and BioVendor assays) in determining the highestquartile of abdominal aortic calcification.

Comparison Between Serum OPG Measured by DuoSet and Biovendor ELISAs

CCC �95%-CI�* Bias** r (P value)***

Duoset/Biovendor using own standards (n�109) 0.014 70.796 r�0.99

�0.013, 0.014� (�31.278) P�0.001

Duoset/Biovendor using Biovendor standards (n�109) 0.044 44.110 r�1.00

�0.042, 0.045� (�15.253) P�0.001

Duoset/Biovendor using Duoset standards (n�109) 0.047 68.657 r�0.97

�0.045, 0.048� (�28.594) P�0.001

ELISA Plate Duoset Biovendor

Duoset standard, pmol/L 78.4�32.4 51.7�18.0

Biovendor standard, pmol/L 9.73�5.86 7.59�2.75

*CCC �95%-CI��Concordance correlation coefficient and 95%-confidence interval.**Bias�mean difference (standard deviation of differences).***r�Pearson correlation coefficient for correlation between average and mean differences of values and

respective P value.Values in bold are those obtained using the standards provided with the relevant assay while those in italics are

using the other manufacturers standards.

Clancy et al Quantification of Abdominal Aortic Calcification 2575

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

References1. Jayalath RW, Mangan SH, Golledge J. Aortic calcification. Eur J Vasc

Endovasc Surg. 2005;30:476–488.2. Jayalath RW, Jackson P, Golledge J. Quantification of abdominal aortic

calcification on CT. Arterioscler Thromb Vasc Biol. 2006;26:429–430.3. Rhee EJ, Oh KW, Jung CH, Lee WY, Oh ES, Yun EJ, Baek KH, Kang

MI, Kim SW. The relationship between four single nucleotide poly-morphisms in the promoter region of the osteoprotegerin gene and aorticcalcification or coronary artery disease in Koreans. Clin Endocrinol(Oxf). 2006;64:689–697.

4. Nitta K, Ishizuka T, Horita S, Hayashi T, Ajiro A, Uchida K, Honda K,Oba T, Kawashima A, Yumura W, Kabaya T, Akiba T, Nihei H. Solubleosteopontin and vascular calcification in hemodialysis patients. Nephron.2001;89:455–458.

5. Linnet K. Necessary sample size for method comparison studies based onregression analysis. Clin Chem. 1999;45:882–894.

6. Bezerra MC, Calomeni GD, Caparbo VF, Gebrim ES, Rocha MS, PereiraRM Low bone density and low serum levels of soluble RANK ligand are

associated with severe arterial calcification in patients with Takayasuarteritis. Rheumatology (Oxford). 2005;44:1503–1506.

7. Nitta K, Akiba T, Uchida K, Otsubo S, Takei T, Yumura W, Kabaya T,Nihei H. Serum osteoprotegerin levels and the extent of vascular calci-fication in haemodialysis patients. Nephrol Dial Transplant. 2004;19:1886–1889.

8. Nitta K, Akiba T, Uchida K, Otsubo S, Takei T, Yumura W, Kabaya T,Nihei H. The progression of vascular calcification and serum osteopro-tegerin levels in patients on long-term hemodialysis. Am J Kidney Dis.2003;42:303–309.

9. Kiechl S, Schett G, Wenning G, Redlich K, Oberhollenzer M, Mayr A,Santer P, Smolen J, Poewe W, Willeit J. Osteoprotegerin is a risk factorfor progressive atherosclerosis and cardiovascular disease. Circulation.2004;109:2175–2180.

KEY WORDS: aortic calcification � osteoprotegerin � osteopontin �sensitivity � specificity

2576 Arterioscler Thromb Vasc Biol. November 2006

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

Paula Clancy, Lisa Oliver, Rumal Jayalath, Petra Buttner and Jonathan GolledgeAssessment of a Serum Assay for Quantification of Abdominal Aortic Calcification

Print ISSN: 1079-5642. Online ISSN: 1524-4636 Copyright © 2006 American Heart Association, Inc. All rights reserved.

Greenville Avenue, Dallas, TX 75231is published by the American Heart Association, 7272Arteriosclerosis, Thrombosis, and Vascular Biology

doi: 10.1161/01.ATV.0000242799.81434.7d2006;26:2574-2576Arterioscler Thromb Vasc Biol. 

http://atvb.ahajournals.org/content/26/11/2574World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://atvb.ahajournals.org/content/suppl/2006/11/05/26.11.2574.DC1.htmlData Supplement (unedited) at:

  http://atvb.ahajournals.org//subscriptions/

at: is onlineArteriosclerosis, Thrombosis, and Vascular Biology Information about subscribing to Subscriptions:

  http://www.lww.com/reprints

Information about reprints can be found online at: Reprints: 

document. Question and AnswerPermissions and Rightspage under Services. Further information about this process is available in the

which permission is being requested is located, click Request Permissions in the middle column of the WebCopyright Clearance Center, not the Editorial Office. Once the online version of the published article for

can be obtained via RightsLink, a service of theArteriosclerosis, Thrombosis, and Vascular Biologyin Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:

by guest on August 19, 2016

http://atvb.ahajournals.org/D

ownloaded from

1

ATVB/2006/128314 R1 ON LINE SUPPLEMENT DATA FOR ASSESSMENT OF A SERUM ASSAY FOR

QUANTIFICATION OF ABDOMINAL AORTIC CALCIFICATION.

SUPPLEMENTAL METHODS

Patients: Ethical approval for the study was granted by the local Institutional Ethics

Committee. Written informed consent was obtained from all participants. Patients were

recruited prospectively and consecutively from the Vascular Outpatients clinic between

May 2004 and June 2006. The inclusion criteria for the study were:

1) A diagnosis of symptomatic peripheral vascular disease based on clinical assessment

or Abdominal Aortic Aneurysms (AAA) based on ultrasound;

2) Computer Tomography Angiogram (CTA) required to further assess peripheral

vascular disease or AAA according to the treating physician.

The exclusion criteria were:

1) previous surgical (endovascular or open) repair of AAA or aortic occlusive disease

(AOD);

2) patients deemed not suitable for a CT angiogram by the treating physician, including

contrast allergy, inability to lie flat, serum creatinine > 120µM.

A total of 109 patients who fulfilled the entry criteria participated in the study. No

patients who entered the study were subsequently excluded from analysis (thus no flow

diagram has been presented). Data sheets were completed regarding the patients’ medical

history and blood tests. The data was stored on a Microsoft Excel spreadsheet (Microsoft

Office 2003). Hypertension was defined by a history of high blood pressure or receiving

2

treatment to reduce blood pressure. Diabetes was defined by a fasting blood glucose ≥ 7.0

mmol/L, or history of, or treatment for hyperglycaemia. Ischaemic heart disease was

defined by a history of angina, myocardial infarction or previous treatment by coronary

medications or intervention. Smoking status was classified into current smokers (smoked

within the last month), ex-smokers (given up for more than one month) and never

smokers.

Imaging: Contrast enhanced CT images were obtained using a 4-slice multiscanner

(MX800, Philips). The imaging and workstation protocols utilised to measure infrarenal

abdominal aortic calcification have previously been validated in detail [1]. Briefly

selected images were reconstructed on a workstation utilising defined thresholds (Center

Hounsfield Unit (CH) level 1400 and Window width Hounsfield Unit (WH) 2000),

automated function setting and image magnification. The reproducibility of calcification

volumes measured with this protocol were excellent (intra and inter-observer coefficient

of variation approximately 1%) [1].

OPG and OPN ELISA: A large number of different commercial ELISAs for OPG are

available [2-9]. The DuoSet OPG ELISA (R & D Systems) is a developmental ELISA

economically suitable for assay of a large number of samples. No precision parameters

are provided by the manufacturers of this ELISA. The OPG ELISA available from

Biomedica [9] has a detection limit of 0.14pM and precision values (intra- and inter-assay

coefficient of variation of 4-10% (n=2) and 7-8% (n=2), respectively) supplied by the

manufacturer. The Biovendor OPG ELISA [3] is also provided with detailed assay

characteristics, including detection limit, 0.4pM, intra-assay coefficient of variation, 2.4-

7% (n=8), inter-assay coefficient of variation, 3.4-7.4% (n=3) and recovery values of 84-

3

105%. Each of these assays uses a different recombinant standard (Supplemental Figure

1). The standard used in the Duoset ELISA is the full length OPG protein expressed in

sf9 cells (insect cell line). The Biovendor standard is also expressed in sf9 cells, but

comprises the OPG protein minus the signal peptide fused to the Fc domain of human

IgG. The Biomedica standard encompasses the cysteine-rich TNFR domain and is

expressed in NSO cells (mammalian cell line). We assessed two OPN ELISAs available

from R & D Systems and IBL Immuno-Biological Laboratories, both of which are

provided with detailed assay characteristics.

Assessment of ELISAs: The assays were assessed in terms of standard and sample

linearity, imprecision, recovery and interference. Samples and standards for analysis were

prediluted, aliquoted and stored at –20oC prior to assessment in the assays, following

manufacturer’s instructions. The ELISAs were performed on two separate days by the

same experienced scientist. To assess sample linearity two sets of pooled serum were

serially diluted (1 in 2, 1 in 4, 1 in 8 and 1 in 16) in buffer, starting with the

recommended sample volume or dilution for each specific ELISA kit. Accuracy was

assessed by measuring recovery of two different concentrations of OPG (DuoSet 8.33

and 16.67 pM, BioVendor 3.3 and 10 pM and BioMedica 3.3 and 10 pM) added to four

serum samples. The concentration, presentation (buffer used) and form of OPG was

based on manufacturer recommendations for the standards used in each assay and linear

range of the assays. Receptor activator of NF-Kb ligand (RANKL) is the main circulating

ligand for OPG and present in the serum at 0.019-7.6pM [10-15]. We assessed the

interference of adding 10pM of RANKL to different concentrations of OPG standards in

the three different ELISAs. Every sample (n=40) and standards (n=8, 7, 5 for Duoset,

4

Biovendor and Biomedica, respectively) were assayed in duplicate to assess intra-assay

variability on two separate days. Samples (n=35) were assayed on two different days to

assess inter-assay variability.

Data analysis: Clinical and laboratory data were prospectively entered into a spreadsheet

(Microsoft Excel) and later transferred to SPSS (Version 13.0) for Windows for further

analysis. Assay precision were assessed using concordance correlation coefficients

(CCC) with 95%-confidence intervals (95%-CI), scatter plots comparing the observed

values against the line of equality, Pearson’s correlation coefficient (r), standard

deviation and Bland-Altman residual plots [16,17]. The relationships between measured

serum OPG and OPN and aortic calcification were assessed by an independent researcher

by means of scatter plots, Spearman rank order correlation, box plots and Kruskal Wallis

test. In order to determine the cut-off OPG concentration with most accuracy (combined

sensitivity and specificity) in distinguishing the patients with aortic calcification of the

upper quartile, receiver operating characteristic (ROC) curves were created and area

under the curve (AUC) calculated.

SUPPLEMENTAL RESULTS

Supplemental Table I illustrates the standard and sample curve parameters for the OPG

assays. The intra and inter-assay reproducibility of the OPG assays is shown in

Supplemental Table II and Supplemental Figure 2. The patient characteristics are shown

in Supplemental Table III. The mean recovery of OPG in the presence of 10 pM RANKL

was 85±13%, 104±19% and 82±21% for DuoSet, BioMedica and Biovendor ELISAs

(n=4)

5

SUPPLEMENTAL FIGURE LEGENDS

Supplemental figure 1: Comparison of the structure of the standards used in the three

OPG ELISA kits.

Supplemental figure 2: Representative results of intra- and inter-assay reproducibility for

OPG ELISAs. Values outside the linear range of the own standard curve have been

excluded. Shown are sample scatterplots (upper) and Bland-Altman plots (lower) for

intra- (left) and inter-assay (right) repeats for Duoset and Biovendor ELISAs [17].

6

REFERENCES

1. Jayalath RW, Jackson P, Golledge J. Quantification of abdominal aortic calcification

on CT. Arterioscler Thromb Vasc Biol. 2006;26:429-30.

2. Moschen AR, Kaser A, Enrich B, Ludwiczek O, Gabriel M, Obrist P et al. The

RANKL/OPG system is activated in inflammatory bowel disease and relates to the state

of bone loss. Gut 2005;54:479-487.

3. Kyrtsonis MC, Vassilakopoulos TP, Siakantaris MP, Kokoris SI, Gribabis DA,

Dimopoulou MN, et al. Serum syndecan-1, basic fibroblast growth factor and

osteoprotegerin in myeloma patients at diagnosis and during the course of the disease.

Eur J Haematol 2004;72:252-8.

4. Jono S, Ikari Y, Shioi A, Mori K, Miki T, Hara K, et al. Serum osteoprotegerin levels

are associated with the presence and severity of coronary artery disease. Circulation

2002;106:1192-1194.

5. Schoppet M, Sattler AM, Schaefer JR, Herzum M, Maisch B, Hofbauer LC. Increased

osteoprotegerin serum levels in men with coronary artery disease. J Clin Endocrinol

Metab 2003;88:1024-1028.

6. Kiechl S, Schett G, Wenning G, Redlich K, Oberhollenzer M, Mayr A, et al.

Osteoprotegerin is a risk factor for progressive atherosclerosis and cardiovascular

disease. Circulation 2004;109:2175-2180.

7. Rhee EJ, Lee WY, Kim SY, Kim BJ, Sung KC, Kim BS, et al. Relationship of serum

osteoprotegerin levels with coronary artery disease severity, left ventricular hypertrophy

and C-reactive protein. Clin Sci (Lond) 2005;108:237-243.

7

8. Pennisi P, Signorelli SS, Riccobene S, Celotta G, Di Pino L, La Malfa T, et al. Low

bone density and abnormal bone turnover in patients with atherosclerosis of peripheral

vessels. Osteoporos Int 2004;15:389-395.

9. Ziegler S, Kudlacek S, Luger A, Minar E. Osteoprotegerin plasma concentrations

correlate with severity of peripheral artery disease. Atherosclerosis 2005;182:175-180.

10. Taranta A, Fortunati D, Longo M, Rucci N, Iacomino E, Aliberti F, et al. Imbalance

of Osteoclastogenesis-Regulating Factors in Patients with Celiac Disease. J Bone Miner

Res 2004;19:1112-1121.

11. Fabrega E, Orive A, Garcia-Suarez C, Garcia-Unzueta M, Amado JA, Pons-Romero

F. Osteoprotegerin and RANKL in alcoholic liver cirrhosis. Liver Int 2005;25:305-310.

12. Kraj M, Owczarska K, Sokolowska U, Centkowski P, Poglod R, Kruk B. Correlation

of osteoprotegerin and sRANKL concentrations in serum and bone marrow of multiple

myeloma patients. Arch Immunol Ther Exp (Warsz) 2005;53:454-64.

13. Liu JM, Zhao HY, Ning G, Zhao YJ, Chen Y, Zhang Zh, et al. Relationships Between

the Changes of Serum Levels of OPG and RANKL with Age, Menopause, Bone

Biochemical Markers and Bone Mineral Density in Chinese Women Aged 20-75. Calcif

Tissue Int 2005;76:1-6.

14. Olmos JM, Amado JA, Valero C, Hernandez JL, Gonzalez-Macias J. Factors other

than glucocorticoids are involved in the osteoblast activity decrease caused by tissue

injury. Clin Endocrinol 2006;64:280-283.

15. Martini G, Valenti R, Giovani S, Gennari L, Salvadori S, Galli B, et al. Serum

Osteoprotegerin and serum RANKL in Paget’s Disease. Calcif Tissue Int 2003;72:780.

8

16. Lin LI. A concordance correlation coefficient to evaluate reproducibility. Biometrics

1989;45:255-268.

17. Bland JM, Altman DG. Statistical methods for assessing agreement between two

methods of clinical measurement. Lancet 1986;1:307-310.

9

Supplemental Table I: Assay parameters for the OPG ELISAs.

DuoSet (R&D Systems) Biovendor Biomedica

Regression parameters Day 1 Day 2 Day 1 Day 2 Day 1 Day 2

Gradient 0.0308 0.0323 0.0963 0.1313 0.0716 0.0856 y intercept 0.3016 0.2702 0.0032 0.0726 0.1153 0.0878

Standard linearity

R2 0.9909 0.9954 0.9983 0.9996 0.9992 1.000 Gradient 0.4133 0.3877 0.6996 0.9914 0.2821 0.3366

y intercept 0.1913 0.2067 0.0258 0.0465 0.1703 0.1960Serum 1

R2 0.9779 0.9785 0.9998 0.9990 0.9988 0.9793Gradient 0.5654 0.4624 0.7887 1.1949 0.1377 0.1819

y intercept 0.1957 0.1994 0.0117 0.0500 0.1540 0.1608

Serum linearity (doubling dilution) Serum

2 R2 0.9973 0.9904 0.9993 0.9994 0.9570 0.9936

10

Supplemental Table II: OPG assay reproducibility.

Intra-assay Day 1 Day 2 CCC

[95%-CI]* Bias** r p-value***

CCC

[95%-CI]* Bias** r p-value***

Duoset (n=48) 0.997

[0.997, 0.997]

0.001 (∀0.08)

r=0.11 p=0.485

0.997 [0.997, 0.997]

-0.019 (∀0.074)

r=-0.01 p=0.957

Biovendor (n=47)

0.997 [0.996, 0.997]

-0.012 (∀0.062)

r=-0.40 p=0.005

0.996 [0.995, 0.996]

-0.021 (∀0.090)

r=-0.13 p=0.372

Biomedica (n=45)

0.992 [0.991, 0.992]

-0.009 (∀0.049)

r=-0.25 p=0.082

0.988 [0.987, 0.989]

-0.017 (∀0.064)

r=-0.41 p=0.004

Inter-assay (comparison between days) CCC

[95%-CI]* Bias** r p-value***

Duoset (n=35) 0.929

[0.916, 0.940]

0.496 (∀1.983)

r=0.16 p=0.463

Biovendor (n=35)

0.977 [0.974, 0.979]

-0.632 (∀0.783)

r=0.05 p=0.791

Biomedica (n=35)

0.937 [0.930, 0.943]

-0.444

(∀0.731)

r=-0.31 p=0.075

* CCC [95%-CI] = Concordance correlation coefficient and 95%-confidence interval [16]; **Bias = mean difference (∀ standard deviation of differences) [17]; ***r = Pearson’s correlation coefficient for correlation between average and mean differences of values and respective p-value [17].

11

Supplemental Table III: Presenting characteristics of study patients. Characteristic (units) Number or mean (% or standard

deviation) Age (years) 68.2±8.5 Presentation: Intermittent claudication (IC) 62 (57%) Abdominal aortic aneurysm (AAA) 40 (37%) Both AAA and IC 7 (6%) Diabetes mellitus 27 (25%) Hypertension 75 (69%) Ischaemic heart disease 56 (51%) Current smoker 46 (42%) Ex-smoker 16 (15%) Non-smoker 47 (43%) Haemoglobin (g/dl) 13.9±1.8 Cholesterol (mM) 4.96±1.84 LDL (mM) 2.86±1.78 HDL (mM) 1.28±0.35 LDL= Low density lipoprotein; HDL= High density lipoprotein.

12

Supplemental Figure 1

H2N COOH SP Cys Cys Cys Cys DD DD HD DuoSet standard -46kD predicted, ~60kD actual

Biovendor standard -71kD predicted, ~77kD actual H2N Cys Cys Cys Cys DD DD HD COOH Fc frag of IgG

H2N Cys Cys Cys Cys COOH Biomedica standard -19.6kD predicted, ~19.9kD actual

13

Supplementary Figure 2

0

22

0 220

22

0 22Day 1

Day 2

-8

-4

0

4

0 22-8

-4

0

4

0 22

Diffe

renc

e of v

alues

Duoset

0

22

0 220

22

0 22

-8

-4

0

4

0 22-8

-4

0

4

0 22

Day 2

Day 1

Biovendor

Inter-assay variation

Mean of calculated values

Intra-assay variation

Mean of optical density readings

0

4

0 40

4

0 4

Biovendor -0.3

0

0.3

0 4-0.3

0

0.3

0 4

0

4

0 40

4

0 4

Diffe

renc

e of r

eadi

ngs

Duoset -0.3

0

0.3

0 4-0.3

0

0.3

0 4