Articles Unbiased plasma proteomics discovery of biomarkers for improved detection of subclinical atherosclerosis Estefanía N ~ez,a,d Valentín Fuster,a,b María Go un mez-Serrano,a,d Jose Manuel Valdivielso,h Juan Miguel Ferna ndez-Alvira,a Diego Martínez-Lo pez, Jose Manuel Rodríguez, Elena Bonzon-Kulichenko, Enrique Calvo, Alvaro Alfayate,a c a a,d a,d Marcelino Bermudez-Lopez,h Joan Carles Escola-Gil,i,j Leticia Ferna ndez-Friera,a Isabel Cerro-Pardo,c Jose María Mendiguren,f Fatima Sanchez-Cabo, Javier Sanz, Jose María Ordovas, Luis Miguel Blanco-Colio,c,d Jose Manuel García-Ruiz,a   a a,b   e Borja Ib an~ez,a,c,d Enrique Lara-Pezzi,a,d Antonio Fern andez-Ortiz,a,d,g Jose Luis Martín-Ventura,c,d,* and Jes azquez a,d,** us V a Centro Nacional de Investigaciones Cardiovasculares Carlos III, Madrid 28029, Spain b Icahn School of Medicine at Mount Sinai, New York, USA c IIS-Fundacion Jimenez-Díaz, Madrid, Spain d CIBER de Enfermedades Cardiovasculares (CIBERCV), Madrid, Spain e US Department of Agriculture Human Nutrition Research Center on Aging, Tufts University, Boston, USA f Banco de Santander, Madrid, Spain g Hospital Clínico de San Carlos, IdISSC, Madrid, Spain h Institut de Recerca Biom edica, Lleida, Spain i  Biom Institut de Investigacio edica Sant Pau, Barcelona, Spain j CIBER de Diabetes y Enfermedades Metabo licas (CIBERDEM), Spain Summary Background Imaging of subclinical atherosclerosis improves cardiovascular risk prediction on top of traditional risk eBioMedicine 2022;76: factors. However, cardiovascular imaging is not universally available. This work aims to identify circulating proteins 103874 Published online 10 Feb- that could predict subclinical atherosclerosis. ruary 2022 Methods Hypothesis-free proteomics was used to analyze plasma from 444 subjects from PESA cohort study (222 https://doi.org/10.1016/j. with extensive atherosclerosis on imaging, and 222 matched controls) at two timepoints (three years apart) for dis- ebiom.2022.103874 covery, and from 350 subjects from AWHS cohort study (175 subjects with extensive atherosclerosis on imaging and 175 matched controls) for external validation. A selected three-protein panel was further validated by immunotur- bidimetry in the AWHS population and in 2999 subjects from ILERVAS cohort study. Findings PIGR, IGHA2, APOA, HPT and HEP2 were associated with subclinical atherosclerosis independently from traditional risk factors at both timepoints in the discovery and validation cohorts. Multivariate analysis ren- dered a potential three-protein biomarker panel, including IGHA2, APOA and HPT. Immunoturbidimetry con- firmed the independent associations of these three proteins with subclinical atherosclerosis in AWHS and ILERVAS. A machine-learning model with these three proteins was able to predict subclinical atherosclerosis in ILERVAS (AUC [95%CI]:0.73 [0.70 0.74], p < 1 £ 10 99), and also in the subpopulation of individuals with low car- diovascular risk according to FHS 10-year score (0.71 [0.69 0.73], p < 1 £ 10 69). Interpretation Plasma levels of IGHA2, APOA and HPT are associated with subclinical atherosclerosis indepen- dently of traditional risk factors and offers potential to predict this disease. The panel could improve primary preven- tion strategies in areas where imaging is not available. Funding This study was supported by competitive grants from the Spanish Ministry of Science, Innovation and Universi- ties (BIO2015-67580-P, PGC2018-097019-B-I00, PID2019-106814RB-I00 and SAF2016-80843-R), through the Carlos III Institute of Health-Fondo de Investigacion Sanitaria grant PRB3 (IPT17/0019 - ISCIII-SGEFI / ERDF, ProteoRed), Abbreviations: CV, cardiovascular; SA, subclinical atherosclerosis; PESA, Progression of Early Subclinical Atherosclerosis study; AWHS, Aragon Worker Health Study; FHS, Framingham Heart Study; CACS, coronary artery calcium score; MS, mass spectrome- try; PIGR, polymeric immunoglobulin receptor; IGHA1 and IGHA2, immunoglobulin heavy constant alpha 1 and 2; IGHG1, IGHG2, IGHG3 and IGHG4, immunoglobulin heavy constant gamma 1, 2, 3 and 4; IGKC, immunoglobulin kappa constant; IGLC2, immunoglobulin lambda constant 2; IGHD, immunoglobulin heavy constant delta; HPT, haptoglobin; HEP2, heparin cofactor 2; APOA, apolipoprotein(a); GELS, gelsolin; CD5L, CD5 antigen-like; Lp(a), lipoprotein(a) *Corresponding to: Cardiovascular Proteomics Laboratory, Centro Nacional de Investigaciones Cardiovasculares Carlos III, C/Mel- chor Fernandez Almagro 3, Madrid 28029, Spain. **Corresponding to: Vascular Research Lab., IIS-Fundacion Jimenez Dıaz, Autonoma University, Av. Reyes Catolicos 2, 28040 Madrid, Spain. E-mail addresses: jlmartin@fjd.es (J.L. Martın-Ventura), jvazquez@cnic.es (J. Vazquez). www.thelancet.com Vol 76 Month February, 2022 1 Articles CIBERCV and CIBERDEM, the Fundacio MaratoTV3 (grant 122/C/2015) and “la Caixa” Banking Foundation (project HR17-00247). The PESA study is co-funded equally by the Centro Nacional de Investigaciones Cardiovasculares (CNIC), Madrid, Spain, and Banco Santander, Madrid, Spain. The ILERVAS study was funded by the Diputacio de Lleida. The study also receives funding from the Instituto de Salud Carlos III (PI15/02019; PI18/00610; RD16/0009) and the FEDER funds. The CNIC is supported by the Instituto de Salud Carlos III (ISCIII), the Ministerio de Ciencia, Innovacion y Universidades (MCNU) and the Pro CNIC Foundation. Copyright Ó 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Keywords: Subclinical atherosclerosis; Proteomics; Biomarkers; IGHA2; APOA; HPT Research in context Evidence before this study Introduction Atherosclerosis is the leading cause of death worldwide, Atherosclerosis is the leading cause of death worldwide. and early prevention is the best approach to fight this pan- The natural history of atherosclerosis starts in child- demic disease. Cardiovascular risk assessment is based on hood, involves a protracted subclinical phase and diag- equations that use traditional risk factors. Identification of nosis usually occurs in advanced stage or following a atherosclerosis by non-invasive imaging has been shown to cardiovascular (CV) event. Early prevention is thus the improve risk stratification over classical equations. Clinical best approach to fight this pandemic disease. To guide practice guidelines recommend screening for atherosclero- sis by imaging and using these measures as risk modifiers, actions for prevention, clinical practice guidelines rec- especially in low-moderate risk individuals. However, cardio- ommend individual risk assessment by using algo- vascular imaging is not universally available. The identifica- rithms that are based on traditional CV risk factors.1 tion of plasma biomarkers closely associated with However, there is a substantial variation in the amount subclinical atherosclerosis could overcome this limitation of atherosclerosis among individuals belonging to the and make it possible to improve risk prediction in a wider same risk category,2,3 and interest has emerged in the scale. In this regard, existing biomarkers to date, such as C- use of non-invasive imaging techniques for screening reactive protein, provide limited additional value. atherosclerotic burden to improve CV risk assessment. Numerous works have shown that the detection of Added value of this study coronary calcification or carotid plaques using non-inva- This is the largest study to date exploring the association sive imaging tools improves risk prediction and reclassi- between plasma protein levels and subclinical atherosclero- fication compared with only conventional risk sis by using high-throughput, unbiased quantitative proteo- factors.4 13 Indeed, current guidelines (2021 ESC guide- mics. The discovery phase was undertaken in a population lines) consider coronary artery calcium score and carotid of 444 individuals from the PESA cohort study undergoing plaque detection as risk modifiers in CV risk assess- simultaneous imaging and blood testing at two different ment, specially in individuals with calculated CV risks timepoints three years apart. Validation of these findings was also performed by proteomics in an external cohort based on the major conventional risk factors around the (350 individuals from the AWHS cohort study). These analy- decisional thresholds.14 However, there are important ses allowed the identification of a panel of three biomarker concerns regarding availability and cost-effectiveness,14 proteins that are significantly associated with the presence as well as expertise requirements and radiation expo- of subclinical atherosclerosis even after adjusting for tradi- sure for the routine screening of atherosclerosis with tional risk factors. Quantification of these proteins by immu- imaging modalities. In this context, the concept of using noturbidimetry, a fast and simple technology, was shown to circulating biomarkers to improve the assessment of CV predict subclinical atherosclerosis in a large sample com- risk is not new. The high-sensitivity C-reactive protein posed of 2999 individuals from a third cohort (ILERVAS). has shown consistency across large prospective studies Subclinical atherosclerosis could also be predicted by this as a risk factor integrating multiple metabolic and low- panel in individuals with low cardiovascular risk. grade inflammatory factors, but its contribution to the Implications of all the available evidence existing methods of CV risk assessment has limited additional value.14 Today, mass spectrometry (MS)- Measuring plasma levels of IGHA2, APOA and HPT by based proteomics offers the opportunity of detecting immunoturbidimetry provides valuable information on novel protein biomarkers directly related to the pres- the presence of subclinical atherosclerosis. This approach may thus provide an affordable alternative to ence, extension and/or progression of atherosclerosis. cardiovascular imaging for the identification of subclini- To the best of our knowledge, unbiased deep quantita- cal atherosclerosis. These measures can be used for car- tive proteomics has not been used previously to study diovascular risk reclassification in low to moderate-risk associations between circulating plasma protein levels individuals in primary cardiovascular prevention. 2 www.thelancet.com Vol 76 Month February, 2022 Articles and subclinical atherosclerosis (SA) in large enough The external validation set was designed within the populations (e.g., more than 100 individuals). AWHS cohort15,20 as a nested case-control study, also In this work, we performed a case-control study restricted to men. The AWHS cohort was composed by aimed to detect novel protein biomarkers of SA using workers of a large car assembly plant in Figueruelas (Zara- unbiased deep MS-based proteomics on plasma from goza, Spain).15 Similar to PESA, the presence of plaques in 444 asymptomatic, middle aged men with or without both carotid and femoral arteries was determined using SA from the PESA study.4 The results were internally linear high-frequency 2-dimensional ultrasound probes.20 validated by repeating the proteomics in the same popu- Plaques were defined as a focal structure protruding lation after a 3-year follow up. The identified biomarkers 0.5 mm into the lumen or reaching a thickness 50% were also externally validated by proteomics in an exter- of the surrounding intima. CACS was also analyzed fol- nal cohort of 350 individuals from the AWHS study.15 lowing the Agatston method. Cases (175) were selected Finally, three selected proteins were also validated using among individuals with 3 or more vascular territories a commercially available turbidimetric test in the affected, and controls (175) among participants with no AWHS cohort and in a large cohort of 2999 individuals vascular territory affected. Controls were individually from the ILERVAS study.16 matched with cases based on a hierarchical single sum- mary score including age (caliper: 3 years), diabetes, smok- ing, dyslipidemia and hypertension. The final validation was performed in a subpopula- Methods tion from the ongoing ILERVAS study.21,22 ILERVAS is based on a middle-aged population with low-to-moder- Study populations and assessment of subclinical ate cardiovascular risk, randomized to undergo or not atherosclerosis vascular ultrasound examination.21,22 This cohort was For the discovery phase, a nested case-control study composed by subjects from 30 primary health care cen- within the prospective PESA cohort17 was designed ters in Lleida (Spain).21 The presence of plaques was (PESA-V1). The PESA cohort was composed by healthy, explored in carotid and femoral arteries by two-dimen- middle-aged employees of the Banco de Santander sional ultrasound.21 Plaque was defined as a focal Headquarters in Madrid (Spain).4 In PESA, the pres- intima-media thickness  1.5 mm protruding into the ence of atherosclerotic plaques was assessed by two- lumen.21,22 All subjects (men and women) with vascular dimensional vascular ultrasound of carotids, infrarenal ultrasound examination that completed biochemical abdominal aorta, and iliofemoral arteries and by non- analysis and had either 0 (1404 controls) or  3 vascular contrast cardiac computed tomography.17,18 Plaques territories affected (1595 cases) were selected. were defined as a focal protrusion into the arterial lumen of thickness >0.5 mm or >50% of the surround- ing intima-media thickness (IMT), or a diffuse thick- Protein digestion for proteomics ness >1.5 mm measured between the media-adventitia Plasma samples were collected in tubes with K-EDTA, and intima-lumen interfaces. Plaque thickness was cal- subjected to centrifugation at 1000 g at 4 8C for 10 min, culated by summing plaque areas from the two thickest and stored at -80. 5 µl of plasma of each individual (corre- plaques measured in each vascular territory: carotids, sponding to about 300 µg of protein measured by Nano- abdominal aorta and iliofemoral arteries.4,19 Coronary Drop 1000, Thermo Fisher), were mixed with 5 µl of a artery calcification score (CACS) was measured using buffer containing 50 mM Tris, 2 % SDS and 100 mM the Agatston scoring method; any Agatston score 1 is DTT, and boiled for 5 min for protein denaturation. Pro- considered indicative of atherosclerosis.4 Cases (222) teins were then subjected to filter-aided digestion (Nano- were selected among participants with 3 or more vascu- sep Centrifugal Devices with Omega Membrane-10 K, lar territories affected, and controls (222) among partici- PALL) according to manufacter’s instructions. Briefly, pants with 1 vascular territory affected. Controls were 320 µl of urea were added to each sample to dilute SDS individually matched with cases based on a hierarchical according to manufacter’s instructions, and they were single summary score including, in order of impor- transferred to the filter and subjected to centrifugation at tance, age (caliper: 3 years), diabetes, smoking, dyslipi- 14,000 g for 10 min. Cysteine residues were blocked with demia and hypertension. Due to the low prevalence of 50 mM iodoacetamide for 1 h at room temperature under subclinical atherosclerosis in women in PESA,17 the darkness. After two washes with 100 µl urea followed by study was restricted to men. To optimize statistical two washes with 100 µl of 100 mM ammonium bicarbon- power to detect significant protein alterations in plasma, ate pH 8.8, proteins were digested with trypsin (1:30-tryp- the discovery phase of the study was restricted to men. sin:protein, Promega) overnight at 37 8C. After protein For internal validation, plasma samples from the same digestion, peptides were eluted from the filter in two dif- individuals were collected three years later (PESA-V2), ferent steps using 40 µl of ammonium bicarbonate and except two cases and their matched controls that did not 50 µl of 500 mM NaCl, respectively. Peptides were acidi- renew their consent for ‘omics’ analysis. fied with 25% trifluoroacetic acid (TFA) to a final www.thelancet.com Vol 76 Month February, 2022 3 Articles concentration of 1%, desalted with Oasis cartridges automatic switch between MS and MS/MS using a top- (Waters) following manufacture’s instructions, Speed-vac speed adquisition mode method. MS spectra were dried and stored at -20 8C. acquired in the Orbitrap analyser using full ion-scan mode with a mass range of 400 1500 mass-to-charge (m/z) and 70,000 FT resolution. The automatic gain con- TMT labeling trol target was set at 2 £ 105 with 50 ms maximun injec- Peptides were subjected to multiplexed isobaric labeling tion time. MS/MS was performed in the top-speed with 10-plex TMT reagents (Thermo Fisher Scientific) adquisition mode with 3 s cycle time. HCD fragmentation following manufacter’s instructions. Briefly, peptide was performed at 30% of normalized collision energy and concentration was measured using DirectDetect Infra- MS/MS spectra were analysed at a 60,000 resolution in red Spectrometer (Merck) and 50 µg of peptides from the Orbitrap. each sample were labeled.23,24 After labeling, peptides from all the samples in each TMT experiment were mixed in the same tube, acidified with 25%TFA to a Protein identification final concentration of 1% and desalted with Oasis car- For peptide identification MS/MS spectra were searched tridges (Waters). Aliquots of 1/10 (in volume) were with the SEQUEST HT algorithm implemented in Pro- saved for direct MS analysis (without peptide fraction- teome Discoverer 2.1 (Thermo Scientific) against a Uni- ation). Labeled peptide aliquots were Speed-vac dried prot database comprised human protein sequences and stored at -20 8C. Samples were processed in batches (July 2014), using trypsin digestion with a maximum of containing a maximum of 15 TMT experiments (120 2 missed cleavages, using as fixed modifications Cys samples). Each TMT experiment contained the samples carbamidomethylation (57.021464 Da) and TMT label- from 8 individuals and two channels were reserved for ing at N-terminal end and Lys (229.162932 Da), and reference internal standards constructed by pooling all Met oxidation (15.994915) as dynamic modification. the samples from each batch. Labeling and subsequent Precursor mass tolerance was set at 800 ppm and frag- LC-MS analysis was performed in a blind manner. ment mass tolerance at 0.03 Da; precursor charge range was set to 2 4. Results were analyzed using the proba- bility ratio method25 and the false discovery rate (FDR) Peptide fractionation was calculated based on the search of results against the Peptides were fractionated using the high pH reversed- corresponding decoy database using the refined phase peptide fractionation kit (Thermo Fisher Scien- method,26 with an additional filter for precursor mass tific) according to manufacter’s instructions. Briefly, tolerance of 15 ppm.27 1% FDR was used as criterion for cartridges were washed with 50% and 100% acetonitrile peptide identification. The three biomarkers identified (ACN), and equilibrated with 0.1% of TFA. 100 µg of in this work (HPT, IGHA2 and APOA) were unambigu- peptides were resuspended in TFA 0.1 % and loaded ously identified by unique peptides (Tables S1 S3). into the cartridges. Peptides were then eluted into five fractions with increasing amounts of ACN: Fr1 (12.5% ACN), Fr2 (15% ACN), Fr3 (17.5% ACN), Fr4 (20% Protein quantification ACN) and Fr5 (50% ACN). Fractions were Speed-vac Protein quantification and statistical and systems biol- dried and stored -20 8C until MS analysis. ogy analysis were performed using the models previ- ously developed in our laboratory28,29 with the SanXoT software package.30 Quantitative information was LC-MS analysis extracted from the MS/MS spectra of TMT-labeled pep- Each fraction of labeled peptide samples was analyzed tides. Peptide quantification was analyzed using the using an Easy-nLC 1200 system (Thermo Fisher Scien- WSPP model, which uses raw quantifications as input tific) coupled via a nanoelectrospray ion source (Thermo data and computes the protein log2-fold changes for Fisher Scientific, Bremen, Germany) to an Orbitrap each individual with respect to the average of the values Fusion mass spectrometer (Thermo Fisher Scientific) for of the two reference internal standard samples. In this PESA-V1 cohort, and to a Q-exactive HF Hybrid Quadru- model protein log2-ratios are expressed as standardized pole-Orbitrap (Thermo Fisher Scientific) for PESA-V2 and variables in units of standard deviation according to AWHS cohorts. C18-based reverse phase separation was their estimated variances (Zq values). performed using a PepMap 100 C18 5 mm 0.3 £ 5 mm as trapping column (Thermo Fisher Scientific) and a Pep- Map RSLC C18 EASY-Spray column 50 cm £ 75 mm ID Immunoturbidimetry as analytical column (Thermo Fisher Scientific). Peptides Plasma levels of IGHA2, HPT and APOA were mea- were loaded in buffer A (0.1% of formic acid in water (v/ sured by immunoturbidimetric assays (LK088.OPT, v)) and eluted with a 338 min linear gradient of buffer B NK058.OPT and LK098.OPT, respectively, from The (100% ACN, 0.1% formic acid (v/v)) at 200 nl/min. Mass Binding Site) using the Binding Site Optilite analyzer spectra were acquired in data-dependent manner, with an in a blinded manner. 4 www.thelancet.com Vol 76 Month February, 2022 Articles Machine learning Unbiased discovery of plasma proteins associated to For classification of individuals with subclinical atheroscle- subclinical atherosclerosis rosis, we used a distributed random forest (RF) model, an Quantitative proteomics was performed on 444 plasma ensemble method well established in the diagnostic pre- samples from PESA-V1 and 440 samples from PESA-V2. diction.32 The RandomForestClassifier method from the The complete analysis required more than 800 LC-MS ensemble Scikit-Learn module was used to implement the runs of 5.6 h each. A mean of 1093 proteins were quanti- RF model. Optimal values for RF hyperparameters were fied per sample, and 470 proteins could be quantified in obtained using 10-fold cross-validation for AUC optimiza- more than 80% of the individuals in each cohort. MS data tion on the test datasets using the Python library Scikit- were processed using automated statistical workflows28,31 Learn library.33 Hyperparameter tuning was performed constructed with the SanXoT package.30 sequentially using the RandomizedSearchCV module, to The correlation of protein levels with plaque thickness find an initial naÿve range of values for the different RF found at baseline (PESA-V1) were in general well-repro- hyperparameters,34 and the GridSearchCV module, to duced in the second visit (PESA-V2) (Figure 1a and Tables obtain the optimal combination of specific values to maxi- S4, S5). The behavior of proteins related to the humoral mize performance. AUC calculation for the training and immune response suggested that increased plaque thick- test sets was done applying the roc_auc_score function. ness was associated with an immunoglobulin isotype The class imbalance problem was avoided using the Strati- switch (Figure 1a, red circles). Protein levels in PESA-V1 fiedKFold method, which preserves the percentage of sam- individuals highly correlated with those of the same indi- ples for each class in all the folds. The primary outcome of viduals in PESA-V2 (Figure 1b), suggesting that the switch the RF model was a continuous variable between 0 and 1 remained stable along time. describing the probability of having SA. To obtain a curated list of proteins that were inde- pendently associated with SA, we firstly selected a sub- set of proteins that yielded a significant correlation with Statistics plaque thickness in PESA-V1 (FDR < 5%) and showed Correlation of protein Zq values with plaque thickness or correlation at 3-year follow-up (PESA-V2) (FDR < 15%) with CACS was analyzed by Pearson’s method. Adjustment (Table S6). Among these, we selected IGHA2, PIGR, for multiple hypothesis testing was performed by control- HEP2, HPT, APOA, GELS and IGKC as the proteins ling for the False Discovery Rate (FDR).35 Linear and logistic whose correlation with plaque thickness was more con- regression models were tested using SPSS software (IBM, sistently independent from individual risk factors (Table Armonk, New York). Associations were expressed as stan- S7). We also analyzed the correlation with coronary dardized odds ratios (ORs) with 95% confidence intervals artery calcium score (CACS), finding that HPT, GELS (CI). The C-statistic or area under the receiver operating and CD5L significantly correlated in PESA-V1 and V2 characteristic (ROC) curve (AUC) was used as a measure of (Table S8 and Fig. S1a), independently of individual risk predictive power. Comparison of AUC for different models factors (Table S9). Interestingly, although HPT, consid- was performed according to the method of DeLong.36 ered an acute-phase protein, correlated with both plaque thickness and CACS, none of the typical acute-phase Ethics proteins showed any significant correlation (Table S10). Ethical committee advice and patient informed consent Secondly, among the eight proteins selected, we ana- were obtained (Instituto de Salud Carlos III Ethics Com- lyzed which ones were independently associated with the mittee (PESA), the Central Institutional Review Board presence of SA by logistic regression analysis (Table 2). of Aragon (CEICA) (AWHS) and the Ethics Committee We selected PIGR, IGHA2, APOA, HPT and HEP2 since of The Catalan Health Service (Ref. CEIC-1410 Hospital their levels were significantly associated with an increased Arnau de Vilanova, Lleida, Spain) (ILERVAS)). likelihood of SA, and the associations were maintained after adjustment by FHS 10-year score37 or by the REGI- COR score38,39 (Table 2A). Finally, in multivariate analysis Role of funders we found that these five proteins could be combined into Funding sources played no role in study design; collec- three-protein panels, where PIGR could be replaced by tion, analysis or interpretation of the data; writing of the IGHA2 and HEP2 by HPT (Table 2B). report, or in submission of this paper for publication. Validation of protein associations by proteomics and Results turbidimetry in the AWHS cohort Clinical characteristics of cases and controls from the Quantitative proteomics using the same high-through- PESA, AWHS and ILERVAS cohorts are depicted in put approach was performed on 175 plasma samples Table 1. Overall, the three cohorts were constituted by from AWHS male participants with extensive SA and low or low-to-moderate risk participants according to 175 from matched controls. The correlations of plasma conventional risk scales. proteins with plaque thickness found in PESA-V1 were www.thelancet.com Vol 76 Month February, 2022 5 Articles Controls Cases p value PESA population at baseline (n = 222) (n = 222) Age, y, mean (SD) 48 (4) 49 (4) 0.023 SBP, mm Hg, mean (SD) 120 (12) 124 (12) 0.001 DBP, mm Hg, mean (SD) 75 (9) 77 (9) 0.004 Fasting glucose, mg/dl, mean (SD) 94 (11) 96 (15) 0.132 Total cholesterol, mg/dl, mean (SD) 201 (33) 210 (36) 0.012 LDL-C (mg/dl), mean (SD) 136 (30) 143 (33) 0.024 HDL-C, mg/dl, mean (SD) 44 (10) 43 (10) 0.139 Triglycerides, mg/dl, mean (SD) 106 (58) 121 (65) 0.009 BMI, kg/m2, mean (SD) 27.24 (3.1) 27.5 (3.2) 0.385 Current smoking, No. (%) 55 (25%) 91 (41%) <0.001 Hypertension, No. (%) 20 (9%) 24 (11%) 0.526 Obesity, No. (%) 39 (17.5%) 41 (18.4%) 0.788 Dislypemia, No. (%) 12 (5.4%) 27 (12.2%) 0.011 History of CVD, No. (%) 28 (12.6%) 47 (21%) 0.016 AWHS population (n = 175) (n = 175) Age, y, mean (SD) 49.6 (4.1) 51.0 (3.6) 0.001 SBP, mm Hg, mean (SD) 122.1 (12.0) 125.8 (13.9) 0.01 DBP, mm Hg, mean (SD) 81.4 (8.7) 82.5 (8.8) 0.243 Fasting glucose, mg/dl, mean (SD) 98.2 (17.9) 98.3 (17.2) 0.971 Total cholesterol, mg/dl, mean (SD) 215.4 (35.5) 221.6 (37.3) 0.111 HDL-C, mg/dl, mean (SD) 54.0 (11.1) 50.4 (10.4) 0.002 BMI, kg/m2, mean (SD) 27.21 (3) 27.6 (3.2) 0.369 Current smoking, No. (%) 30 (17.1%) 77 (44%) <0.001 Hypertension, No. (%) 42 (24%) 50 (28.6%) 0.379 Obesity, No. (%) 20 (11.4%) 19 (11%) 1 Dislypemia, No. (%) 49 (28%) 62 (35.4%) 0.111 ILERVAS population (n = 1404) (n = 1595) Male, No. (%) 489 (34.8%) 992 (62.2%) <0.001 Age, y, mean (SD) 56.31 (6.11) 58.90 (6.09) <0.001 SBP, mmHg, mean (SD) 127.91 (16.22) 135.90 (16.69) <0.001 DBP, mmHg, mean (SD) 80.79 (9.67) 83.43 (9.60) <0.001 Fasting glucose, mg/dl, mean (SD) 95.54 (14.86) 98.81 (18.35) <0.001 Total cholesterol, mg/dl, mean (SD) 218.02 (37.21) 226.34 (39.96) <0.001 LDL-C, mg/dl, mean (SD) 123.03 (30.54) 128.18 (33.80) <0.001 HDL-C, mg/dl, mean (SD) 66.70 (19.42) 64.58 (17.66) 0.002 Triglycerides, mg/dl, mean (SD) 142.73 (85.80) 169.59 (112.84) <0.001 BMI, kg/m2, mean (SD) 29.05 (5.12) 29.16 (4.8) 0.525 Current smoking, No. (%) 279 (19.8%) 626 (39.3%) <0.001 Hypertension, No. (%) 480 (34.2%) 768 (48.2%) <0.001 Obesity, No. (%) 426 (30.4%) 497 (31.2%) 0.63 Dislypemia, No. (%) 660 (47%) 899 (56.4%) <0.001 History of CVD, No. (%) 126 (9%) 186 (11.6%) 0.026 Table 1: Characteristics of the PESA, AWHS and ILERVAS populations. SI conversion factors: To convert glucose to mmol/L, multiply values by 0.0555; to convert cholesterol, cholesterol-low density and cholesterol-high-density to mmol/L, multiply values by 0.0259; to convert tryglicerides to mmol/L, multiply values by 0.0113. reproduced in the AWHS cohort (Fig. S1b, see also IGHA2, HPT and HEP2 was also confirmed in AWHS Table S11), except for GELS. The immunoglobulin iso- (Figure 1c). We also observed that the associations type switch was also reproduced in AWHS (Fig. S1b). remained significant after adjusting by CV risk scores The correlation of HPT and CD5L with CACS was lower (Figure 1c). APOA did not achieve statistical significance in AWHS than in PESA but maintained the same trend in AWHS but showed the same tendency than in PESA (Fig. S1c, Table S11). The association with SA of PIGR, (Figure 1c). 6 www.thelancet.com Vol 76 Month February, 2022 Articles Figure 1. Selection of protein biomarkers. (a) Comparison of Pearson’s correlations of relative plasma proteins levels with plaque thickness in PESA-V1 and PESA-V2 cohorts. Dot sizes are indicative of protein abundances (in number of peptides per protein). Inset: behavior of proteins related to humoral immune response (red) or yielding a significant correlation (blue). The immunoglobulin iso- type switch is reflected in the increase of PIGR, IGHA2 and IGHD, and in the decrease of IGLC2, IGKC, IGHG1 and IGHG2, while, in con- trast, the related isotypes IGHA1, IGHG3 and IGHG4 did not change. (b) Representative correlations between relative abundance values (expressed as standardized log2-ratios, Zq) of four representative proteins in PESA-V1 and PESA-V2 for the same individuals. A similar trend was found with the other proteins cited in the text. R indicates Pearson’s correlation coefficient and p, the statistical significance of the correlation. (c and d) Validation of PESA results in the AWHS cohort. Forest plots show OR of subclinical athero- sclerosis (cases vs controls) in AWHS, obtained by either proteomics (c) or turbidimetry (d). OR refer to protein values expressed in units of standard deviation, using univariate logistic regression models, or multivariate models adjusted by common Risk Scores, as indicated. Error bars indicate 95% confidence intervals of OR values. To validate the results by independent techniques, we Validation of protein associations by turbidimetry in selected IGHA2, APOA and HPT since these proteins the ILERVAS cohort formed a panel (Table 2B) that could be measured using To further validate the association of IGHA2, APOA commercially available antibody-based turbidimetry kits and HPT with SA using the turbidimetry kits, we ana- suitable to be used in clinical practice. Multivariate logistic lyzed blood samples from a larger cohort obtained from regression revealed that increased plasma concentration of the ILERVAS study.21,22 This validation cohort con- each one of these three proteins measured by turbidimetry tained individuals that had extensive SA (1.595 cases) in AWHS was significantly associated with an increased and individuals without SA (1.404 controls), and, likelihood of SA (IGHA2: OR[95%CI]:1.37[1.08 1.74], importantly, 51% of the subjects from this cohort were p < .01; APOA: 1.27[1.02 1.57], p < .05; HPT: 2.09 women (see Table 1). Multivariate logistic regression [1.62 2.70], p = 2 £ 10 8, logistic regression analysis) confirmed that the plasma concentrations of these three (Figure 1d). We also observed that these associations proteins were significantly associated with SA after remained significant after adjusting by FHS 10-year adjusting by gender and conventional risk factors score37 or by the REGICOR score38,39 (Figure 1d). (IGHA2: 1.21[1.11 1.33], p = 3 £ 10 5; APOA: 1.18 www.thelancet.com Vol 76 Month February, 2022 7 8 Articles A: Individual proteins Univariate Adj. by FHS 10-year Adj. by Regicor Proteins p-val OR 95% CI p-val OR 95% CI p-val OR 95% CI Increased Polymeric immunoglobulin receptor (PIGR) 0.001 1.554 1.256 1.922 0.005 1.374 1.102 1.711 0.006 1.36 1.091 1.695 Ig alpha-2 chain C region (IGHA2) 0.008 1.295 1.069 1.567 0.024 1.259 1.031 1.538 0.021 1.265 1.037 1.544 Apolipoprotein(a) (APOA) 0.018 1.261 1.041 1.527 0.015 1.279 1.049 1.56 0.011 1.293 1.062 1.575 Haptoglobin (HPT) 0.001 1.412 1.157 1.723 0.05 1.225 0.994 1.51 0.05 1.228 0.998 1.512 Heparin cofactor 2 (HEP2) 0.001 1.414 1.195 1.673 0.005 1.285 1.078 1.531 0.007 1.275 1.069 1.52 Decreased Gelsolin (GELS) 0.008 0.772 0.638 0.936 0.107 0.849 0.695 1.036 0.169 0.869 0.712 1.061 Ig kappa chain C region (IGKC) 0.049 0.833 0.694 0.999 0.181 0.879 0.727 1.062 0.197 0.883 0.73 1.067 CD5 antigen-like (CD5L) 0.226 0.947 0.868 1.034 0.408 0.962 0.879 1.054 0.42 0.963 0.879 1.055 B: Protein panels Proteins p-val OR 95% CI Chi-square df Adj. p-val Model 1 Polymeric immunoglobulin receptor (PIGR) 0.003 1.4 1.12 1.76 44.305 5 2.01E-08 Ig alpha-2 chain C region(IGHA2) 0.016 1.29 1.05 1.6 Apolipoprotein(a) (APOA) 0.004 1.34 1.1 1.64 Haptoglobin (HPT) 0.434 1.1 0.87 1.38 Heparin cofactor 2 (HEP2) 0.006 1.33 1.09 1.64 Model 2 Polymeric immunoglobulin receptor (PIGR) 0.001 1.49 1.19 1.85 37.238 3 2.05E-07 Apolipoprotein(a) (APOA) 0.004 1.34 1.1 1.64 Heparin cofactor 2 (HEP2) 0.001 1.34 1.12 1.6 Model 3 Polymeric immunoglobulin receptor (PIGR) 0.001 1.52 1.22 1.89 33.49 3 1.27E-06 Apolipoprotein(a) (APOA) 0.006 1.32 1.08 1.61 Haptoglobin (HPT) 0.016 1.28 1.05 1.58 www.thelancet.com Vol 76 Month February, 2022 Model 4 Ig alpha-2 chain C region(IGHA2) 0.001 1.39 1.14 1.71 33.267 3 1.41E-06 Apolipoprotein(a) (APOA) 0.005 1.32 1.09 1.61 Heparin cofactor 2 (HEP2) 0.001 1.49 1.25 1.77 Model 5 Ig alpha-2 chain C region(IGHA2) 0.009 1.3 1.07 1.58 24.317 3 1.05E-04 Apolipoprotein(a) (APOA) 0.015 1.27 1.05 1.55 Haptoglobin (HPT) 0.001 1.39 1.14 1.7 Table 2: Logistic regression analysis of association with the presence of subclinical atherosclerosis in PESA-V1. Odds ratios (OR) refer to relative protein values determined by proteomics and expressed in units of standard deviation, using logistic regression models (univariate or bivariate models in A, or multivariate in B). To analyze potential biomarker panels among the five proteins, several multivariate models containing different protein combinations were tested in B. Model 1 included the five proteins. Models 2 5 showed several combinations of three proteins. HPT maintained its association with SA when HEP2 was not included in the models (Models 3 and 5), so that in the practice HPT and HEP2 could be interchanged. PIGR and IGHA2 also showed a similar behav- ior, so that three-protein models could be constructed with either PIGR (Models 2 and 3) or IGHA2 (Models 4 and 5). P-values were adjusted by Bonferroni. Articles [1.08 1.29], p = 1 £ 10 4; HPT: 1.24[1.13 1.37], Potential to predict the presence of subclinical p = 3 £ 10 5, logistic regression analysis) (Figure 2a). atherosclerosis The independence from risk factors was further con- To test the potential of the three-protein panel to predict firmed checking that the associations were maintained the presence of SA, we trained a machine-learning after stratifying the cohort into subpopulations having model that used as predictors the concentration of the or not each one of the main risk factors (Table 3). three proteins (model 3P). Ten-fold cross validation con- We then analyzed the behavior of these associations firmed the stability of AUC values in the testing popula- according to the CV risk of the individuals. Noteworthy, tions with a minimal overfitting (Figure 2b). ROC the associations of the three proteins with SA were analysis showed that 3P had a good performance to pre- maintained in the subpopulation of individuals with dict the presence of SA in the complete Ilervas popula- low CV risk (FHS 10-year score < 10%) (IGHA2: 1.24 tion (AUC [95%CI]: 0.73 [0.70 0.74], p < 1 £ 10 99 vs [1.12 1.38], p = 2 £ 10 5; APOA: 1.19[1.09 1.31], AUC= 0.5, Mann-Whitney statistic) (Figure 2c and d). p = 2 £ 10 4; HPT: 1.22[1.11 1.35], p = 7 £ 10 4, logistic Since APOA is a protein that is already known to associ- regression analysis) (Figure 2a). The associations were ate with atherosclerosis, we also constructed a model lost in the medium/high-risk group (FHS 10-year score that only considered the concentrations of IGHA2 and > 10%), except for HPT, which remained independently HPT (model 2P). While 2P maintained a reasonably associated to SA (Figure 2a). good performance to predict SA (AUC [95%CI]: 0.70 [0.68 0.72], p < 1 £ 10 80 vs AUC= 0.5, Mann-Whit- ney statistic), a comparative analysis showed that 3P Figure 2. Validation of biomarkers in the ILERVAS cohort. (a) Forest plots showing OR of subclinical atherosclerosis (cases vs con- trols) per each protein, obtained by turbidimetry in the complete ILERVAS population, or after stratifying it into low-risk (FHS 10- year score < 10%) or medium/high-risk (FHS 10-year score  10%) individuals. OR refer to protein values expressed in units of stan- dard deviation, using multivariate logistic regression models including the three proteins, gender, smoking, obesity, hypertension, dyslipidemia,history of CV disease and body mass index. (b) 10-fold cross validation of AUC values provided by the 3P model to detect the presence of subclinical atherosclerosis in train and test populations. Data are expressed as mean § SD. (c, d) Improve- ment in AUC values and in the ROC curves to detect subclinical atherosclerosis obtained by the 2P and 3P models in the complete population, or in the low-risk population (FHS 10-year score < 10%). Horizontal error bars in (c) represent 95% CI. P-values above asterisks indicate statistical significance in relation to the null hypothesis (AUC=0), calculated using the Mann-Whitney statistic; p- values from the comparative analysis between models 2P and 3P were calculated using DeLong’s test. www.thelancet.com Vol 76 Month February, 2022 9 Articles Multivariate Adj. by Gender and all Risk Factors (RFs) p-val OR 95% CI p-val OR 95% CI Smoking Non-smokers (N = 2026) Smokers (N = 891) IGHA2 0.002 1.197 1.07 1.338 0.057 1.187 0.995 1.418 APOA 0.001 1.188 1.075 1.313 0.084 1.182 0.978 1.428 HPT 0.358 1.053 0.944 1.174 2.206E-08 1.695 1.409 2.039 Hypertension Non-Hypertensive (N = 1705) Hypertensive (N = 1212) IGHA2 0.000378 1.247 1.104 1.408 0.083 1.14 0.983 1.323 APOA 0.024 1.139 1.017 1.276 0.002 1.254 1.086 1.446 HPT 0.000019 1.299 1.152 1.464 0.309 1.078 0.932 1.247 Obesity Non-Obese (N = 2021) Obese (N = 896) IGHA2 0.00001 1.288 1.151 1.442 0.707 1.034 0.869 1.23 APOA 0.001 1.191 1.073 1.323 0.037 1.189 1.01 1.399 HPT 0.000008 1.29 1.154 1.443 0.413 1.072 0.908 1.265 Dyslipemia No Dyslipemia (N = 1403) Dyslipemia (N = 1514) IGHA2 0.003 1.223 1.07 1.399 0.018 1.172 1.028 1.337 APOA 0.01 1.202 1.045 1.382 0.006 1.172 1.046 1.313 HPT 0.000005 1.359 1.191 1.55 0.366 1.062 0.932 1.21 CVDH No CVDH (N = 2607) CVDH (N = 392) IGHA2 0.000368 1.197 1.084 1.321 0.057 1.31 0.992 1.728 APOA 0.000286 1.189 1.083 1.305 0.09 1.225 0.969 1.549 HPT 0.000117 1.212 1.099 1.337 0.41 1.109 0.866 1.42 Table 3: Multivariate logistic regression analysis of association with the presence of subclinical atherosclerosis in ILERVAS subpopulations stratified according to main CV risk factors. Multivariate logistic regression was used to determine association of protein values with the presence of subclinical atherosclerosis (SA). All the models were adjusted by risk factors including: Age, Hypertension, Obesity, Dyslipemia, Smoking, History of Cardiovascular Disease (CVDH), Systolic Blood Pressure (SBP) and Diastolic Blood Pressure (DB). was significantly better than 2P (p = 3 £ 10 5, DeLong’s Fig. S2, inclusion of the three proteins into a machine test) (Figure 2c), suggesting that the inclusion of APOA learning model constructed using risk factors signifi- significantly improved the performance of the bio- cantly improved AUC values for prediction of SA as marker panel. We further analyzed whether the perfor- compared with a model constructed with the risk factors mance of the panel to predict SA in the population of alone. The improvement in performance was also main- individuals with low CV risk. The model 3P maintained tained in the subpopulation of individuals with low CV a similar performance in this subpopulation (AUC risk (Fig. S2). Taken together, these data support the [95%CI]: 0.71 [0.69 0.73], p < 1 £ 10 69 vs AUC= 0.5, notion that the three-protein biomarker panel provides Mann-Whitney statistic) (Figure 2c and d). 2P also useful information to predict the presence of SA, partic- showed a reasonable performance (AUC [95%CI]: 0.68 ularly in individuals that have a low risk of having CVD [0.66 0.70], p < 1 £ 10 52 vs AUC= 0.5, Mann-Whit- according to traditional risk scales. ney statistic), but again 3P was still significantly better (p = .007, DeLong’s test), suggesting that APOA also contributed to SA prediction in the low risk population. Discussion In the low CV risk population, the 3P model provided a To our knowledge, this study is the deepest and largest means to stratify the risk of having SA (Figure 3). Thus, unbiased mass spectrometry-based proteomics analysis more than 63% of the individuals having SA in 7 or to date in the search for plasma proteins related to ath- more territories were allocated to the two quintiles with erosclerosis. This was possible by using well-pheno- highest risk according to 3P score (>0.48), increasing typed cohorts for SA and by combining the quantitative this proportion to 81% when individuals had more than accuracy and robustness provided by multiplexed iso- 8 territories (Table S12). 54 to 59% of individuals with baric labeling, peptide fractionation and well-validated 3 6 affected territories, but only 27% of the controls, and fully automated quantitative statistical were also allocated to the two highest risk quintiles workflows.28,30,31 Internal validation in the 3-year fol- (Table S12). low-up PESA visit was essential to reduce error sources Finally, we also explored whether the protein panel in the discovery phase, to discard proteins with marked would improve a tentative model to predict SA con- biological variability, and to concentrate our efforts on a structed by using conventional risk factors. As shown in set of proteins having robust associations with SA. The 10 www.thelancet.com Vol 76 Month February, 2022 Articles and our results consistently indicate that the combina- tion of these three proteins, APOA, IGAH2 and HPT, may be a useful tool to improve prediction of atheroscle- rosis presence and extension in the subclinical phase, complementing the information provided by conven- tional risk factors. APOA is the main component of the Lp(a) particle, which has been identified as an independent predictor of coronary artery calcification, and epidemiological studies support a strong association between elevated Lp(a) and atherosclerotic CV disease outcomes,42,43 sug- gesting that Lp(a) plays a causal role in the disease.44 Current guidelines on CV prevention45 recommend the assessment of Lp(a) for individuals at high risk, however our results provide novel information that suggest that Lp(a) might be also determinant for prediction of SA in individuals considered at low risk according to conven- tional risk factors. Our results also indicate that APOA plasma levels provide significant information, in combi- nation with those of IGHA2 and HPT and eventually Figure 3. Risk stratification of subclinical atherosclerosis pre- other plasma proteins, to predict the presence of SA. dicted by the 3P model in the low-risk ILERVAS population. Bar Increased HPT levels have been observed in CAD heights are proportional to the number of individuals in each patients46 and were predictive of CV events.47 In this category. In the right panel, the population was separated respect, HPT could increase together with other acute into two groups according to FHS 10-year CV risk score. phase proteins under inflammatory conditions48 and The left panel represents inviduals with low CV risk (FHS combination of HPT with other inflammatory markers 10-year score < 10%), stratified in quintiles according to (white blood cell count, CRP) has been associated to CV the prediction given by 3P score. The categories in each events.49 In contrast, we could not find any evidence of bar represent the number of individuals according to the association of CRP or other inflammatory proteins with number of affected territories. SA in our study, highlighting a promising role for HPT in the detection of the disease in its subclinical stage. Elevated levels of total serum IgA have been reported in selected proteins showed alterations in their plasma lev- patients with severe atherosclerosis or with previous els that were stable over time and were also externally myocardial infarction or other major ischemic events validated using proteomics in an independent cohort and were found to correlate with myocardial infarction (AWHS). Furthermore, antibody-based turbidimetry and cardiac death in dyslipidemic men.50 However, the kits confirmed the association with SA of the three specific association between the IgA isoform IGHA2 selected proteins in two cohorts (AWHS and ILERVAS). and atherosclerosis has never been reported before. The Several studies have assessed potential plasma pro- differential behavior with plaque thickness of several tein biomarkers of SA in large cohorts of asymptomatic kinds of immunoglobulins in PESA and AWHS cohorts individuals.40,41 In contrast with these previous analy- support a role for the humoral immune response in the ses, which were targeted studies performed using asymptomatic phases of atherosclerosis51 and may sug- immunoassays, here we followed a hypothesis-free gest the existence of Ig class switching triggered by the approach that allowed testing the association with SA of differential activation of specific B-cell subsets proposed all the plasma proteins detectable by our LC-MS setup. to take place during atherosclerosis.51 Further support- From more than 1.000 proteins quantified per individ- ing this idea, we have very recently demonstrated that ual, we selected five proteins (PIGR, IGHA2, APOA, atherosclerotic mice undergo a germinal center anti- HPT and HEP2) that had a significant correlation with body immune response featuring increased somatic either plaque thickness or CACS and maintained their hypermutation load, producing a skewed distribution of association with an increased likelihood of SA after switched antibodies with altered proportions of Ig adjustment by risk scores in two independent cohorts isotypes.52 (PESA and AWHS). For translational purposes, we fur- Imaging studies have revealed discordances between ther selected a panel composed by three proteins conventional risk scores and the presence of (APOA, HPT and IGHA2) that can be measured using atherosclerosis,53,54 since a high percentage of low-risk indi- standard commercial kits routinely used in clinical prac- viduals still present SA.17 Detection of SA by imaging tice. These proteins showed independent association approaches is relevant in the primary prevention field, given with SA in two separate cohorts (AWHS and ILERVAS), that it has been shown to improve risk stratification4,5 and www.thelancet.com Vol 76 Month February, 2022 11 Articles has been related to CV events. Imaging techniques, how- quantitative analysis of plasma proteomics and verified ever, require specialized personnel and are not widely avail- the data. DML, ICP and JLMV conducted the antibody- able. Thus, plasma proteins whose levels associate with SA based analyses and verified the data. AA and JV devel- have the potential of stratifying subjects at higher risk oped the machine-learning model. EN, JLMV, JV and before imaging analysis, and hence may have social and AFO drafted the manuscript. All authors provided economic impact. The protein levels measured using the important intellectual revisions to the manuscript and kits confirmed the independent association of APOA, read and approved the final manuscript IGAH2 and HPT with SA in the large ILERVAS cohort and were useful to predict SA in this cohort. In a previous study it was shown that the majority of individuals classified Acknowledgments at high risk by traditional scales had SA; however SA was We thank Simon Bartlett (CNIC) for English editing, also present in nearly 60% of participants at low risk, sug- Ignacio Mahillo Fernandez for support in statistics and gesting an association of atherosclerosis with characteristics Cesar Vegas-Dominguez and Lucia Ortega-Villanueva not considered in standard risk scales.17 In this regard, our for technical support. This study was supported by com- results showed that the three-protein panel was able to pre- petitive grants from the Spanish Ministry of Science, dict SA in the population of ILERVAS individuals consid- Innovation and Universities (BIO2015-67580-P, ered at low-risk of CV events, which could facilitate targeted PGC2018-097019-B-I00, PID2019-106814RB-I00 and preventive/therapeutic approaches in this group of subjects. SAF2016-80843-R), through the Carlos III Institute of We should comment here that, being an observa- Health-Fondo de Investigaci on Sanitaria grant PRB3 tional study, our inferences did not reflect direct causal- (IPT17/0019 - ISCIII-SGEFI / ERDF, ProteoRed), ity. We must recognize the potential for residual, CIBERCV and CIBERDEM, the Fundaci o MaratoTV3 uncontrolled confounding that might partly explain the (grant 122/C/2015) and “la Caixa” Banking Foundation associations. Another limitation of this study is that the (project HR17-00247). The PESA study is co-funded discovery phase was performed in a population formed equally by the Centro Nacional de Investigaciones Car- by men only. Although there is a possibility that some diovasculares (CNIC), Madrid, Spain, and Banco protein factors specific of women could have been Santander, Madrid, Spain. The ILERVAS study was missed, the discovered proteins were validated in ILER- funded by the Diputaci o de Lleida. The study also VAS, a cohort composed also by females, suggesting receives funding from the Instituto de Salud Carlos III that these factors are also valid for women. (PI15/02019; PI18/00610; RD16/0009) and the In conclusion, this high-throughput, unbiased FEDER funds. The CNIC is supported by the Instituto plasma proteomics study, performed in two large de Salud Carlos III (ISCIII), the Ministerio de Ciencia, cohorts, allowed the identification of a panel of bio- Innovacion y Universidades (MCNU) and the Pro CNIC marker proteins that associate with subclinical athero- Foundation. sclerosis independently of known cardiovascular risk factors. The biomarker panel was able to predict SA in a third cohort, and the performance of the prediction was Data sharing statement maintained in a subpopulation of individuals with low All MS raw data have been deposited in Peptide Atlas cardiovascular risk. The information provided by the (http://www.peptideatlas.org/PASS/PASS01382 and three biomarkers may be useful for evaluation of the http://www.peptideatlas.org/PASS/PASS01522). risk of having subclinical atherosclerosis in individuals that otherwise would be considered at low-risk in pri- mary cardiovascular prevention. Supplementary materials Supplementary material associated with this article can be found in the online version at doi:10.1016/j. Declaration of interests ebiom.2022.103874. The authors have no relationship to disclose. References Contributors 1 Khot UN, Khot MB, Bajzer CT, et al. Prevalence of conventional VF, JLMV and JV conceived and designed the study. VF, risk factors in patients with coronary heart disease. JAMA. 2003;290(7):898–904. LFF, JMM, JS, JMO, MBL, JCEG, BI, ELP and AFO 2 Silverman MG, Blaha MJ, Krumholz HM, et al. Impact of coronary developed the clinical cohorts, including recruitment of artery calcium on coronary heart disease events in individuals at the extremes of traditional risk factor burden: the multi-ethnic patients, collection of clinical and demographic data, study of atherosclerosis. Eur Heart J. 2014;35(33):2232–2241. and storing of samples. JMFA, FSC and JMGR designed 3 Yusuf S, Rangarajan S, Teo K, et al. Cardiovascular risk and events the subcohorts and supported the epidemiological anal- in 17 low-, middle-, and high-income countries. N Engl J Med. 2014;371(9):818–827. ysis. EN, MGS, EBK and EC conducted the plasma pro- 4 Fernandez-Ortiz A, Jimenez-Borreguero LJ, Penalvo JL, et al. The teomics analyses. EN, JMR and JV performed the progression and early detection of subclinical atherosclerosis 12 www.thelancet.com Vol 76 Month February, 2022 Articles (PESA) study: rationale and design. Am Heart J. 2013;166(6):990– 25 Martinez-Bartolome S, Navarro P, Martin-Maroto F, et al. Proper- 998. ties of average score distributions of SEQUEST: the probability 5 Peters SA, den Ruijter HM, Bots ML, Moons KG. Improvements in ratio method. Mol Cell Proteom. 2008;7(6):1135–1145. risk stratification for the occurrence of cardiovascular disease by 26 Navarro P, Vazquez J. A refined method to calculate false discovery imaging subclinical atherosclerosis: a systematic review. Heart. rates for peptide identification using decoy databases. J Proteome 2012;98(3):177–184. Res. 2009;8(4):1792–1796. 6 Baber U, Mehran R, Sartori S, et al. Prevalence, impact, and predic- 27 Bonzon-Kulichenko E, Garcia-Marques F, Trevisan-Herraz M, Vaz- tive value of detecting subclinical coronary and carotid atheroscle- quez J. Revisiting peptide identification by high-accuracy mass rosis in asymptomatic adults: the bioimage study. J Am Coll spectrometry: problems associated with the use of narrow mass Cardiol. 2015;65(11):1065–1074. precursor windows. J Proteome Res. 2015;14(2):700–710. 7 Martin SS, Blaha MJ, Blankstein R, et al. Dyslipidemia, coronary 28 Garcia-Marques F, Trevisan-Herraz M, Martinez-Martinez S, et al. artery calcium, and incident atherosclerotic cardiovascular disease: A novel systems-biology algorithm for the analysis of coordinated implications for statin therapy from the multi-ethnic study of ath- protein responses using quantitative proteomics. Mol Cell Proteom erosclerosis. Circulation. 2014;129(1):77–86. MCP. 2016;15(5):1740–1760. 8 Zavodni AE, Wasserman BA, McClelland RL, et al. Carotid artery 29 Navarro P, Trevisan-Herraz M, Bonzon-Kulichenko E, et al. Gen- plaque morphology and composition in relation to incident cardio- eral statistical framework for quantitative proteomics by stable iso- vascular events: the multi-ethnic study of atherosclerosis (MESA). tope labeling. J Proteome Res. 2014;13(3):1234–1247. Radiology. 2014;271(2):381–389. 30 Trevisan-Herraz M, Bagwan N, Garcia-Marques F, et al. SanXoT: a 9 Gibson AO, Blaha MJ, Arnan MK, et al. Coronary artery calcium modular and versatile package for the quantitative analysis of high- and incident cerebrovascular events in an asymptomatic cohort. throughput proteomics experiments. Bioinformatics. 2019;35 The MESA study. JACC Cardiovasc Imaging. 2014;7(11):1108–1115. (9):1594–1596. 10 Davidsson L, Fagerberg B, Bergstrom G, Schmidt C. Ultrasound- 31 Navarro P, Trevisan-Herraz M, Bonzon-Kulichenko E, et al. Gen- assessed plaque occurrence in the carotid and femoral arteries are eral statistical framework for quantitative proteomics by stable iso- independent predictors of cardiovascular events in middle-aged tope labeling. J Proteome Res. 2014;13(3):1234–1247. men during 10 years of follow-up. Atherosclerosis. 2010;209 32 Yang L, Wu H, Jin X, et al. Study of cardiovascular disease predic- (2):469–473. tion model based on random forest in eastern China. Sci Rep. 11 Lamina C, Meisinger C, Heid IM, et al. Association of ankle-bra- 2020;10(1):5245. chial index and plaques in the carotid and femoral arteries with car- 33 Pedregosa F, Varoquaux G, Gramfort A, et al. Scikit-learn: machine diovascular events and total mortality in a population-based study learning in python. J Mach Learn Res. 2011;12(null):2825–2830. with 13 years of follow-up. Eur Heart J. 2006;27(21):2580–2587. 34 Probst P, Wright MN, Boulesteix A. Hyperparameters and tuning 12 van der Meer IM, Bots ML, Hofman A, del Sol AI, van der Kuip strategies for random forest. Wiley Interdiscip Rev Data Min Knowl DA, Witteman JC. Predictive value of noninvasive measures of ath- Discov. 2019;9(3):e1301. erosclerosis for incident myocardial infarction: the rotterdam 35 Benjamini Y, Hochberg Y. Controlling the false discovery rate: a study. Circulation. 2004;109(9):1089–1094. practical and powerful approach to multiple testing. J R Stat Soc. 13 Polonsky TS, McClelland RL, Jorgensen NW, et al. Coronary artery 1995;57(1):289–300. calcium score and risk classification for coronary heart disease pre- 36 DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas diction. JAMA. 2010;303(16):1610–1616. under two or more correlated receiver operating characteristic 14 Visseren FLJ, Mach F, Smulders YM, et al. 2021 ESC guidelines on curves: a nonparametric approach. Biometrics. 1988;44(3):837–845. cardiovascular disease prevention in clinical practice. Eur Heart J. 37 Ford ES, Giles WH, Mokdad AH. The distribution of 10-year risk 2021;42(34):3227–3237. for coronary heart disease among US adults: findings from the 15 Casasnovas JA, Alcaide V, Civeira F, et al. Aragon workers' health national health and nutrition examination survey III. J Am Coll study-design and cohort description. BMC Cardiovasc Disord. Cardiol. 2004;43(10):1791–1796. 2012;12:45. 38 Marrugat J, D’Agostino R, Sullivan L, et al. An adaptation of the 16 Betriu A, Farras C, Abajo M, et al. Randomised intervention study Framingham coronary heart disease risk function to European to assess the prevalence of subclinical vascular disease and hidden mediterranean areas. J Epidemiol Commun Health. 2003;57(8):634– kidney disease and its impact on morbidity and mortality: the 638. ILERVAS project. Nefrologia. 2016;36(4):389–396. 39 Marrugat J, Solanas P, D'Agostino R, et al. [Coronary risk estima- 17 Fernandez-Friera L, Penalvo JL, Fernandez-Ortiz A, et al. Preva- tion in Spain using a calibrated Framingham function]. Rev Esp lence, vascular distribution, and multiterritorial extent of subclini- Cardiol. 2003;56(3):253–261. cal atherosclerosis in a middle-aged cohort: the PESA (progression 40 Saarikoski LA, Huupponen RK, Viikari JS, et al. Adiponectin is of early subclinical atherosclerosis) study. Circulation. 2015;131 related with carotid artery intima-media thickness and brachial (24):2104–2113. flow-mediated dilatation in young adults-the cardiovascular risk in 18 Fernandez-Friera L, Fuster V, Lopez-Melgar B, et al. Normal LDL- young finns study. Ann Med. 2010;42(8):603–611. cholesterol levels are associated with subclinical atherosclerosis in 41 Oikonen M, Wendelin-Saarenhovi M, Siitonen N, et al. Tissue the absence of risk factors. J Am Coll Cardiol. 2017;70(24):2979– inhibitor of matrix metalloproteinases 4 (TIMP4) in a population 2991. of young adults: relations to cardiovascular risk markers and 19 Fernandez-Alvira JM, Fuster V, Dorado B, et al. Short telomere carotid artery intima-media thickness. The cardiovascular risk in load, telomere length, and subclinical atherosclerosis: the PESA young finns study. Scand J Clin Lab Investig. 2012;72(7):540–546. study. J Am Coll Cardiol. 2016;67(21):2467–2476. 42 Ellis KL, Boffa MB, Sahebkar A, Koschinsky ML, Watts GF. The 20 Laclaustra M, Casasnovas JA, Fernandez-Ortiz A, et al. Femoral renaissance of lipoprotein(a): brave new world for preventive cardi- and carotid subclinical atherosclerosis association with risk factors ology? Prog Lipid Res. 2017;68:57–82. and coronary calcium: the AWHS study. J Am Coll Cardiol. 43 Alonso R, Andres E, Mata N, et al. Lipoprotein(a) levels in familial 2016;67(11):1263–1274. hypercholesterolemia: an important predictor of cardiovascular dis- 21 Sanchez E, Betriu A, Lopez-Cano C, et al. Characteristics of athero- ease independent of the type of LDL receptor mutation. J Am Coll matosis in the prediabetes stage: a cross-sectional investigation of Cardiol. 2014;63(19):1982–1989. the ILERVAS project. Cardiovasc Diabetol. 2019;18(1):154. 44 Ellis KL, Watts GF. Is lipoprotein(a) ready for prime-time use in the 22 Bermudez-Lopez M, Martinez-Alonso M, Castro-Boque E, et al. clinic? Cardiol Clin. 2018;36(2):287–298. Subclinical atheromatosis localization and burden in a low-to-mod- 45 Mach F, Baigent C, Catapano AL, et al. 2019 ESC/EAS guidelines erate cardiovascular risk population: the ILERVAS study. Rev Esp for the management of dyslipidaemias: lipid modification to reduce Cardiol (Engl Ed). 2021;74(12):1042–1053. cardiovascular risk. Eur Heart J. 2020;41(1):111–188. 23 Baldan-Martin M, Lopez JA, Corbacho-Alonso N, et al. Potential 46 Lee CW, Cheng TM, Lin CP, Pan JP. Plasma haptoglobin concen- role of new molecular plasma signatures on cardiovascular risk trations are elevated in patients with coronary artery disease. PLoS stratification in asymptomatic individuals. Sci Rep. 2018;8(1):4802. One. 2013;8(10):e76817. 24 Bagwan N, Bonzon-Kulichenko E, Calvo E, et al. Comprehensive 47 Holme I, Aastveit AH, Hammar N, Jungner I, Walldius G. Hapto- quantification of the modified proteome reveals oxidative heart globin and risk of myocardial infarction, stroke, and congestive damage in mitochondrial heteroplasmy. Cell Rep. 2018;23 heart failure in 342,125 men and women in the apolipoprotein mor- (12):3685–3697.e4. tality risk study (AMORIS). Ann Med. 2009;41(7):522–532. www.thelancet.com Vol 76 Month February, 2022 13 Articles 48 Graves KL, Vigerust DJ. Hp: an inflammatory indicator in cardio- 52 Lorenzo C, Delgado P, Busse CE, et al. ALDH4A1 is an atheroscle- vascular disease. Future Cardiol. 2016;12(4):471–481. rosis auto-antigen targeted by protective antibodies. Nature. 49 Holme I, Aastveit AH, Hammar N, Jungner I, Walldius G. Inflam- 2021;589(7841):287–292. matory markers, lipoprotein components and risk of major cardio- 53 Naqvi TZ, Mendoza F, Rafii F, et al. High prevalence of ultrasound vascular events in 65,005 men and women in the apolipoprotein detected carotid atherosclerosis in subjects with low Framingham mortality risk study (AMORIS). Atherosclerosis. 2010;213(1):299–305. risk score: potential implications for screening for subclinical ath- 50 Kovanen PT, Manttari M, Palosuo T, Manninen V, Aho K. Predic- erosclerosis. J Am Soc Echocardiogr. 2010;23(8):809–815. tion of myocardial infarction in dyslipidemic men by elevated levels 54 Pen A, Yam Y, Chen L, Dennie C, McPherson R, Chow BJ. Discor- of immunoglobulin classes A, E, and G, but not M. Arch Intern dance between Framingham risk score and atherosclerotic plaque Med. 1998;158(13):1434–1439. burden. Eur Heart J. 2013;34(14):1075–1082. 51 Tsiantoulas D, Diehl CJ, Witztum JL, Binder CJ. B cells and humoral immunity in atherosclerosis. Circ Res. 2014;114(11):1743– 1756. 14 www.thelancet.com Vol 76 Month February, 2022