IgA antibodies to phosphocholine associate with long-term cardiovascular disease risk

DSpace https://erepo.uef.fi

Rinnakkaistallenteet Terveystieteiden tiedekunta

2018

IgA antibodies to phosphocholine associate with long-term

cardiovascular disease risk

Kankaanpää, J

Elsevier BV

Tieteelliset aikakauslehtiartikkelit

© Elsevier B.V.

CC BY-NC-ND https://creativecommons.org/licenses/by-nc-nd/4.0/

http://dx.doi.org/10.1016/j.atherosclerosis.2017.12.010

https://erepo.uef.fi/handle/123456789/6284

Downloaded from University of Eastern Finland's eRepository

Accepted Manuscript

IgA antibodies to phosphocholine associate with long-term cardiovascular disease risk

Jari Kankaanpää, Maritta Sämpi, Risto Bloigu, Chunguang Wang, Ramin Akhi, Y.

Antero Kesäniemi, Anne M. Remes, Olavi Ukkola, Sohvi Hörkkö

PII: S0021-9150(17)31429-6

DOI: 10.1016/j.atherosclerosis.2017.12.010 Reference: ATH 15298

To appear in: Atherosclerosis Received Date: 19 May 2017 Revised Date: 25 November 2017 Accepted Date: 6 December 2017

Please cite this article as: Kankaanpää J, Sämpi M, Bloigu R, Wang C, Akhi R, Kesäniemi YA, Remes AM, Ukkola O, Hörkkö S, IgA antibodies to phosphocholine associate with long-term cardiovascular disease risk, Atherosclerosis (2018), doi: 10.1016/j.atherosclerosis.2017.12.010.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

M AN US CR IP T

AC CE PT ED

1

IgA antibodies to phosphocholine associate with long-term cardiovascular disease risk

Jari Kankaanpää^a,b, Maritta Sämpi^a, Risto Bloigu^c, Chunguang Wang^a,b, Ramin Akhi^a,b, Y. Antero Kesäniemi^d, Anne M. Remes^b,e,f,g, Olavi Ukkola^d, Sohvi Hörkkö^a,b

a) Medical Microbiology and Immunology, Research Unit of Biomedicine, University of Oulu.

POB 5000, University of Oulu FI-90014, Finland.

b) Medical Research Center and Nordlab Oulu, University Hospital and University of Oulu, Oulu, Finland. POB 5000, University of Oulu FI-90014, Finland.

c) Medical Informatics and Statistics Research Group, University of Oulu, Oulu, Finland. POB 5000, University of Oulu FI-90014, Finland.

d) Research Unit of Internal Medicine, Medical Research Center Oulu, Oulu University Hospital, and University of Oulu, Oulu, Finland. POB 5000,University of Oulu FI-90014, Finland.

e) Institute of Clinical Medicine- Neurology, University of Eastern Finland, Kuopio, Finland.

POB 1627, FI-70211 University of Eastern Finland, Finland

f) Department of Neurology, Kuopio University Hospital, Kuopio, Finland. POB 100, FI-70029, Kuopio, Finland.

g) Research Unit of Clinical Neuroscience, University of Oulu, Finland. POB 5000,University of Oulu FI-90014, Finland.

Address for correspondence:

University of Oulu, POB 5000, University of Oulu 90014, Finland.

Email address:sohvi.horkko@nordlab.fi Sohvi Hörkkö,

M AN US CR IP T

AC CE PT ED

2

ABSTRACT

Background and aims: Antibodies to phosphocholine and oxidized LDL (oxLDL) are proposed to modify progression of atherosclerosis. We investigated the prognostic value of antibodies to phosphocholine (PCho), Streptococcus pneumoniae cell wall polysaccharide (CWPS) and oxLDL in defining the long-term CVD survival.

Methods: CVD incidence was followed for 18 years and analyzed with baseline plasma IgM, IgG and IgA antibody levels to PCho, CWPS and oxLDL in 1044 subjects of Oulu Project Elucidating Risk of Atherosclerosis study (OPERA).

Results: During the follow-up period, 195 subjects (18.7%) had a CVD event. Cox model with ACC/AHA CVD adjustments (ASCVD) showed that IgA levels to PCho and IgA to CWPS were statistically significant factors predicting CVD risk. IgM and IgG antibodies to PCho, CWPS and oxLDL had no effect on CVD risk after adjusting for other risk factors. Net reclassification improvement (categories: 17-year risk <15%, 15 to 30%, > 30%), was 0.06 (-0.001 – 0.12, p<0.054), and IDI was 0.0124 (0.0036 – 0.0211, p<0.006). with IgA-PCho added to the ASCVD risk model. Seventeed (9.4%) study subjects with CVD events were correctly reclassified into higher risk category while 9 (5.0%) subjects were classified into lower risk category. Among the non-cases, 58 (8.7%) subjects were correctly reclassified into lower risk, and 46 (5.9%) were reclassified into higher risk category.

Conclusions: Plasma IgA antibodies to PCho and Streptococcus pneumoniae CWPS are significant predictors of long-term CVD risk. Additional studies on the role of IgA antibodies in atherogenesis and CVD are warranted.

M AN US CR IP T

AC CE PT ED

3

Keywords: Phosphocholine, streptococcus pneumoniae, antibody, cardiovascular disease, biomarker

ABBREVIATIONS

CVD, cardiovascular disease; LDL, low-density lipoprotein; OxLDL, oxidized LDL; PCho, phosphocholine;

CWPS, Streptococcus Pneumoniae cell wall polysaccharide; OPERA, Oulu Project Elucidating Risk of Atherosclerosis; ICD, International Classification of Diseases; CHD, coronary heart disease;

MDA-LDL, malondialdehyde modified LDL; CuOx-LDL, copper oxidized LDL;

IMT, intima media thickness; ASCVD, ACC/AHA atherosclerotic cardiovascular disease risk model;

HDL, high density lipoprotein; FRS, Framingham Risk Score; hsCRP, high sensitivity CRP; IDI, integrated discrimination index; NRI, net reclassification improvement;

MAA, malondialdehyde-acetaldehyde; PAF, platelet-activating factor.

INTRODUCTION

Atherosclerotic cardiovascular disease (CVD) develops gradually over lifetime and it is often difficult to observe before potentially hazardous manifestations. Risk-estimation systems have been developed to assist in everyday clinical patient care. It has been acknowledged that the effect of adding new risk factors to well-established risk factors is usually quite small, but still may aid to appropriately reclassify some patients (1). The total CVD risk is generally a sum of several interacting risk factors influencing final decisions in clinical management.

M AN US CR IP T

AC CE PT ED

4

Phosphocholine (PCho) is a small molecule universally present in nature. PCho is prominently found on a variety of different organisms including Haemophilus influenza, (2) Lactobacillus spp,(3) Streptococcus spp(2). Aspergillus spp(3), nematodes (3), and on some protozoa such as Plasmodium falciparum (4). PCho is synthesized from choline, which is a vital nutrient for eukaryotes, but it is not essential for most prokaryotes. PCho is the immunodominant epitope on Streptococcus pneumoniae cell wall polysaccharide (CWPS), and antibody responses to CWPS in humans have been studied for over 70 years(5).

Low density lipoprotein (LDL) is a causative risk factor in the development of atherosclerosis. Upon oxidative modification of LDL lipids, oxidative neo-epitopes are generated, which are recognized by receptors of the innate and adaptive immune system. One of the antigenic epitopes in oxLDL for the immune recognition is the phosphocholine (PCho) head group of oxidized phospholipids exposed on LDL surface (6). Autoantibodies binding to the PCho moiety on oxidized phospholipids have been demonstrated to be present in human plasma (7), and PCho epitopes are found in atherosclerotic plaques (8). Antibodies binding to oxidatively modified LDL are generally accepted to be modifying factors in atherogenesis. Low plasma IgM antibodies binding to oxidative neo- epitopes are linked to increased intima media thickness (9) and the risk of cardiovascular disease events (10). Less definite findings have been identified on IgG and IgA antibodies to oxidized epitopes in atherosclerosis (11).

Humoral immune response to PCho, the immunodominant epitope in many microbes and in oxLDL, has been proposed to be an independent risk factor for development of atherosclerosis and coronary artery disease (7). The present study aimed to investigate the long-term CVD survival and autoantibodies to PCho and CWPS and OxLDL in 1044 subjects of the OPERA-study (Oulu Project Elucidating Risk of Atherosclerosis). We have earlier reported that baseline IgM antibodies to oxLDL are inversely associated with carotic artery intima media thickness among the OPERA study subjects (9). Here, we show that baseline IgA antibody levels to PCho and CWPS are

M AN US CR IP T

AC CE PT ED

5

independently associated with increased risk of cardiovascular disease events during 18 years of follow-up.

M AN US CR IP T

AC CE PT ED

6

MATERIALS AND METHODS

Subjects

The OPERA study population consisted of subjects selected from the Social Insurance Institute Register of Finland in the early 1990s. The subjects were selected randomly among those eligible for hypertension medication reimbursement (n=600), with age and sex-matched controls (n=600) without eligibility for hypertension medication reimbursement. The study population represented approximately 3% of the age-group population of the City of Oulu region in Finland and it was described in detail previously (12). Of the 1200 selected and contacted subjects, 1044 participated in this study (87%). The study protocol was approved by the Ethical committee of Faculty of Medicine, University of Oulu, and an informed consent was obtained from each participant. The study followed the principles of the Declaration of Helsinki.

Cardiovascular endpoint

Baseline data of the study subjects was collected during 1990-1993.Causes of death and CVD events were obtained from the Finnish Causes-of-Death Register and the Hospital Discharge code register until the end of the year 2009. The mean follow-up time in the study population was 18 years. Endpoints were defined according to FINRISK criteria (13) and classified according to International Classification of Diseases revision 8 (ICD-8) or ICD-9 before 1994 and ICD-10 thereafter. Coronary heart disease (CHD) endpoint was defined according to the FINRISK criteria (I20-I22 [ICD10] and 410, 4110 [ICD8/9], coronary artery bypass or angioplasty or I20-I25, R96, R98 as causes of death). Common cardiovascular disease endpoint (CVD) included subjects with CHD event or stroke excluding subarachnoidal hemorrhage (I61, I63 (not I636) and I64 [ICD-10], and 431, 4330A, 4331A, 4339A, 4340A, 4341A, 4349A, 436 [ICD-9] and 431 (except 43101, 43191) 433, 434, 436 [ICD-8].

M AN US CR IP T

AC CE PT ED

7

Determination of antibody levels to PCho, CWPS and oxLDL

Isolated pooled human LDL was modified using malondialdehyde (MDA-LDL) and copper sulfate (CuOx-LDL) as previously described (14). Plasma IgM, IgG and IgA antibodies to MDA-LDL, CuOx-LDL, phosphocholine (PC-KLH, Biosearch Technologies, PC-1013-5, Novato, CA) and Streptococcus pneumoniae cell wall polysaccharide (Statens Serum Institute Diagnostica, 3459, Copenhagen, Denmark) were determined using chemiluminescent immunoassay. IgM and IgG antibody levels were expressed as relative units of standard human IgM or IgG. IgA antibodies and IgM or IgG antibodies to PCho or CWPS were measured as relative light units. Intra-assay coefficients of variation and specificity of antibodies to oxLDL have been reported previously (14).

Antibody measurements were performed during the years 2001-2002 using samples stored at -20°

C and cross-sectional data has been previously reported elsewhere (9, 14).

Clinical measurements

BMI was calculated dividing weight (kg) by height squared (m²). Waist hip ratio was measured dividing waist (cm) by hip (cm). Alcohol and smoking consumption was determined using validated questionnaires. Smoking was calculated as pack-years, where one pack-year equals 20 cigarettes smoked in a day for one year. Alcohol usage was measured in grams of absolute alcohol per week. Systolic and diastolic blood pressure at baseline was measured using an automated oscillometric blood pressure recorder (Dinamap model 18465X, Crition Ltd, Ascot, UK), after 10-15 min rest in a sitting position, from the right arm. Blood pressure was measured three times with one minute resting time, and the mean of the latter two measures was used in the analysis. Diagnostic criteria for type 2 diabetes included fasting glucose over 6.1 mmol/l and/or 2-h plasma glucose level over 10mmol/l in oral glucose tolerance test or if the study subject was on diabetes medication (WHO criteria).

M AN US CR IP T

AC CE PT ED

8 Ultrasonography measurements

The detailed IMT measurement procedure has been reported earlier (9). Intima media thickness (IMT) was defined as the distance between the media-adventitia interface and lumen-intima interface. The mean IMT measurement was defined as the mean of both internal carotid arteries, both bifurcations and the highest common carotid artery measurements from both sides. The measurement of liver adiposity was based on liver-kidney contrast measured with ultrasonography (14) and classifications for statistical analyses were made between subjects with normal liver brightness and those with fatty liver.

Laboratory measurements

Venous blood samples were drawn in EDTA tubes, after an overnight fast. Plasma was separated by 2000 rpm centrifugation at +4°C for 10 min. Lipoprotein fractions were separated by ultracentrifugation. Total cholesterol, HDL, LDL and triglyceride levels were determined by enzymatic colorimetric method (Boehringer Diagnostica, Mannheim GmbH, Germany) as described (12) and analyzed with a Kone Spesific analyser (Selective Chemisty Analyzer, Kone Intstruments, Espoo, Finland). High sensitivity C-reactive protein (hsCRP) was analyzed using a chemiluminescent kit (Diagnostic Systems Laboratories Inc., Webster, TX, USA).

Statistical analyses

Statistical analyses were performed using IBM SPSS statistics version 22.0 (IBM Corp, Armonk, NY, USA). Reclassification statistics, ROC-curve and AUC (Area under curve) were calculated using the R statistical software (R, version 3.2.5, 2016; R Development Core Team 2016, R foundation for Statistical Computing, Vienna, Austria). Continuous variables are presented as mean ± standard deviation (SD), and dichotomous as percentages. Statistical significances between groups at baseline were tested using the Mann-Whitney U-test for continuous variables and the Chi-squared test for categorical variables. Correlation of specific antibody levels to plasma

M AN US CR IP T

AC CE PT ED

9

lipoprotein levels was assessed using Spearman Rank Correlation coefficient. Cardiovascular survival was assessed using Kaplan-Meier survival curves with plasma antibody levels in tertiles.

Statistical significances of the Kaplan-Meier survival curves were calculated using Log-Rank test.

The association between baseline plasma antibody levels and risk of development of CVD events during follow-up time was estimated with Cox proportional hazards models. The models were calculated with plasma antibody levels both as categorical variables in tertiles and continuous variables, with risk calculated according to one standard deviation increase in plasma antibody levels. Model 1 included study group, age, sex and BMI as adjusting factors. Model 2 included covariates in model 1 with LDL cholesterol levels, intima media thickness, diabetes and smoking in packet years. Model 3 included covariates in model 2 and high sensitivity CRP (hsCRP), alcohol usage (g/week), fatty liver and HDL. We also assessed plasma IgA to PCho and CWPS using ACC/AHA atherosclerotic cardiovascular disease(ASCVD) risk variables in the model (15). ASCVD variables included study group, age, sex, diabetes, HDL- and total cholesterol levels, smoking status and systolic blood pressure. . All participants were assumed to be caucasian. Variables in this model correspond to the variables in the Framingham global CVD risk model (16). The study group in each model was represented by the original selection criteria, i.e. eligibility for hypertension medication reimbursement. Proportional Hazards assumption was tested with Schoenfield residuals and was satisfied in every model.

The improvement in CVD risk prediction was tested with logistic regression analysis by adding IgA to PCho and CWPS into the ASCVD variables. To ensure equal follow-up times in logistic regression models, the risk prediction statistics was calculated in events occurring up to 17 years of the follow-up time. Study subjects having events after 17 years were considered as non-event cases at the time-point of 17 years. Subjects censored before 17 years without an event were excluded from risk prediction statistics calculations. The goodness of fit was evaluated using Hosmer-Lemeshow test with the predicted probabilities divided in deciles. The predicted

M AN US CR IP T

AC CE PT ED

10

probabilities were entered to receiver operating characteristics (ROC) curve and the area under curve (AUC) was calculated. Statistical differences in AUC values were determined using the method by DeLong et al. (17). The strength of association was calculated with Nagelkerke’s pseudo R² –test.

Reclassification tables are reported with categorical and continuous net reclassification improvement (NRI) and integrated discrimination index (IDI) (18). Risk categories <15%, 15-30%

and >30% were used for categorical NRI in the analyses.

M AN US CR IP T

AC CE PT ED

11

RESULTS

Baseline data

Baseline measurements of study subjects are shown in Table 1. During the follow-up period, 195 subjects (18.7%) had a CVD event. Study subjects with CVD events were older and more frequently male, had higher levels of total cholesterol, LDL cholesterol and triglycerides, and lower levels of HDL cholesterol compared to subjects with no CVD events. The study subjects with CVD events also had a higher prevalence of diabetes and hypertension (Table 1). The variables not available for statistical analysis are listed in Supplemental Table 1.

Study subjects with CVD events during follow-up had higher plasma IgA antibody levels to PCho (p<0.00005) and CWPS (p<0.008), compared to study subjects with no CVD events (Supplemental Fig. 1 and Supplemental Table 2). In these unadjusted analyses, the study subjects with CVD events also had lower plasma IgM antibody levels to copper oxidized LDL (CuOx-LDL; p<0.027), phosphocholine (PCho; p<0.007) and Streptococcus pneumoniae cell wall polysaccharide (CWPS;

p<0.031). No correlation between specific IgA levels and lipoprotein or triglyceride values was observed.

M AN US CR IP T

AC CE PT ED

12

Baseline plasma antibody levels and the long-term risk of CVD

Kaplan-Meier estimate survival analysis

To estimate the probability of the study subjects’ survival without CVD events, the Kaplan-Meier estimate analysis was used (Fig. 1 and Supplemental Fig. 2). Subjects with high IgA levels to PCho had higher frequency of CVD events during follow-up compared to subjects with low IgA levels to PCho (Log-rank p<0.002, Fig. 1). Similarly, subjects with high IgA levels to CWPS had more CVD events compared to those with low IgA to CWPS (Log-rank p<0.011, Fig. 1). In contrast, subjects with high IgM to PCho had lower frequency of CVD events compared to subjects with low IgM to PCho (Log-rank p<0.02, Supplemental Fig. 2). There were no significant differences in CVD survival probabilities associated with plasma IgA levels to MDA-LDL and CuOx- LDL (Fig. 1), and with IgM levels to MDA-LDL (Supplemental Fig. 2). Plasma IgM to CWPS and CuOx-LDL and plasma IgG levels to oxLDL, PCho and CWPS were not associated with CVD survival probability (data not shown).

Cox regression model analysis

The long-term risk of developing CVD events was assessed using Cox proportional hazard survival models. Plasma IgA-PCho and IgA-CWPS were statistically significant predictors of risk of developing cardiovascular events during follow-up (Fig. 2 and Supplemental Table 3). This finding was tested in multivariate models adjusted with known risk factors (Fig. 2). The adjustments were also made with ASCVD variables and IgA-PCho (HR 1.194, 95% CI 1.049 – 1.360) and IgA-CWPS (HR 1.184, 95% CI 1.046 – 1.341) remained significant predictors of CVD events (Fig. 2). Plasma IgM and IgG to oxLDL, PCho and CWPS did not predict long-term risk of CVD events in Cox regression models (Supplemental Table 3). The highest tertile of IgM to PCho was associated with

M AN US CR IP T

AC CE PT ED

13

lower risk of CVD events, but was not statistically significant when adjusted with known CVD risk factors.

Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) analysis

The incremental prognostic value of the findings was assessed using logistic regression models with ASCVD variables in the presence and absence of IgA-PCho and IgA-CWPS. Calibration of the model was satisfactory when measured by Hosmer Lemeshow test (p>0.05 for all models, Table 2). The strength of association tested with Nagelkerkes pseudo R² increased from 0.198 to 0.213 with the addition of IgA-PCho, and from 0.198 to 0.211 with the addition of IgA-CWPS, suggesting improvements in the explained variability. There was also an increase in the area under the ROC curve when adding IgA-PCho (from 0.751 to 0.754) and IgA-CWPS (from 0.751 to 0.759), but differences were not statistically significant.

Table 3 shows reclassification of study subjects when IgA-PCho was added into the model with ASCVD variables. The addition of IgA-PCho into the model reclassified 17 (9.4%) study subjects with CVD events correctly into the higher category, and 9 (5.0%) subjects were classified intothe lower risk category. Among study subjects without CVD events, 58 (8.7%) were correctly reclassified into lower risk category, and 46 (5.9%) were reclassified into higher risk category.

NRI_event was 0.044 (-0.011 – 0.010) and NRI_nonevent was 0.015 (-0.010 – 0.041). Categorical NRI was 0.060 (95% CI -0.001 – 0.1207, p< 0.0537) with 15% and 30% risk categories, continuous was NRI 0.26 (95% CI 0.097 – 0.42, p<0.00173) and IDI was 0.0124 (0.0036 – 0.0211 p<0.0058) (Table 2). Reclassification of study subjects using IgA-CWPS was similar to IgA-PCho (Supplemental Table 4).

M AN US CR IP T

AC CE PT ED

14

DISCUSSION

The atherosclerotic process is complex and takes substantial time, during which different risk factors influence the progression rate. In clinical setting, treatment options are based on evaluation of several risk factors. The present study investigated the prognostic value of plasma antibodies to phosphocholine, Streptococcus pneumoniae cell wall polysaccharide and oxidized LDL in a population-based study cohort. Our data showed that IgA antibodies to PCho and CWPS independently predicted long-term risk of developing CVD events, and further improved patient risk assessment. Plasma antibody measurements are relatively easy to perform, inexpensive and do not require complex laboratory equipment, and therefore, can be considered to have potential for wide-spread use in everyday clinical risk assessment.

A link between plasma IgA concentrations and atherosclerosis was proposed nearly 30 years ago (19). Later, serum IgA to dietary and microbial antigens has been associated with clinical atherosclerosis in a few small studies (20, 21). To date, there is only a limited number of reports on circulatory IgA antibodies to PCho or OxLDL on human atherosclerotic cardiovascular diseases.

Patients with coronary artery disease and acute myocardial infarction are reported to have high levels of IgA to malondialdehyde acetaldehyde, the immunodominant epitope in malondialdehyde modified proteins (22). Low levels of serum IgA antibodies to PCho have been associated with increased intima media thickness progression in 226 individuals from the Swedish study population followed for four years (23). We documented earlier, using cross-sectional study design, that plasma IgA to PCho was associated with fasting blood glucose levels among non-diabetic study subjects of the OPERA-cohort (14), but we found no significant association between IgA levels to PCho and intima media thickness. In humans, IgA production exceeds other isotypes and is about 70% of the total immunoglobulin production. The main site for IgA production is on mucosal surfaces where IgA participates in immune homeostasis collectively with commensal mucosal

M AN US CR IP T

AC CE PT ED

15

bacteria. Systemic IgA is generally thought to be produced mainly by bone marrow B lymphocytes and the production is observed to be separate from mucosal IgA responses (24). Serum IgA levels have been also linked to diabetes (25), and environmental factors such as heavy alcohol drinking (26). Experimental studies with ApoE^-/- mice have shown increase in total serum IgA levels when treated with Th2 stimulatory IL-33, yet the mice had reduced atherosclerosis (27). Monomeric circulatory IgA binds to different receptors, including FcαRI receptor (CD89) and Fcα/Fcµ receptor, on a variety of cells, yet the exact function of circulatory IgA is not fully known. FcαRI receptors in hepatic Kupffer cells have been proposed to provide a second line of defense against intestinal bacteria by removing circulating bacteria opsonized with IgA entering through the intestinal surfaces and via portal vein (28).

In humans, choline is needed for the synthesis of cell membrane phosphocholine-containing lipids, and proinflammatory lipid mediators such as plateled-activating factor (PAF). Hazen and colleagues showed that choline and trimethylamine N-oxide levels independently associated with atherosclerosis risks in humans, and they were elegantly able to link, for the first time, gut microbial bacteria and phospholipid metabolism to atherosclerosis (29). From the microbial perspective, many commensal and pathogenic bacteria contain PCho within cell wall structural components, but not all host modulating responses of PCho are known. PCho acts as a virulence factor when bacteria exploit PAF-receptor on endothelial cells for invasion (30). PCho moieties on bacterial surface are also targeted via the immune system by immunoglobulin and CRP opsonization.

Microbes may use molecular mimicry as a mechanism to escape human immune system recognition, e.g. by displaying host derived small molecules, such as PCho, on their surfaces to avoid bacterial opsonization. PCho moiety may be hidden by the host immune self-recognition, but during, e.g. lipid peroxidation of phospholipids on LDL surface, the choline-head group of phospholipids is exposed and becomes available for humoral immune recognition (6). Epitope mapping studies with monoclonal antibodies specific to PCho, namely EO6, suggest that immune recognition of PCho does not cross-react with choline alone (6). In mice, EO6 subsets of natural

M AN US CR IP T

AC CE PT ED

16

IgM antibodies to PCho have been characterized to recognize evolutionary stable molecular patterns in bacteria (6) and additionally participate in apoptotic cell clearance (31). There has not been a clear demonstration yet of natural antibodies against PCho epitopes in humans (32).

Current pneumococcal vaccines are designed to induce antibodies binding to pneumococcal capsule also containing PCho moieties. The 23-valent pneumococcal vaccine induces a T-cell independent immune response and has been shown to stimulate IgA and IgM isotype antibodies to CWPS detectable up to one year post vaccination (33). Experimental animal studies have shown that pneumococcal vaccination induces antibodies binding to PCho epitopes both on oxLDL and on pneumococcal cell wall (34). There is also a substantial amount of animal studies documenting that immunization with oxLDL reduces atherosclerosis by inducing the production of antibodies binding to oxLDL. The potential link between immune response to PCho epitopes and atheroprotection has been investigated in several observational studies using the 23-valent pneumococcal polysaccharide vaccine. A large meta-analysis pooling eight observational studies together reported significantly lower likelihood of atherosclerotic cardiovascular events in patients 65 years and older, vaccinated with pneumococcal polysaccharide vaccine, but only when younger patients were excluded from the analysis (35). The proposed pathogenic mechanisms include antibody mediated inhibition of oxLDL and apoptotic cell uptake by macrophages (31). We have no record on the study subjects’ pneumococcal vaccinations. When baseline blood samples were collected in the early 1990s, pneumococcal vaccination was relatively uncommon in Finland (36).

Study limitations

The OPERA-study population was selected from the Finnish Social Insurance Institute Register and the selection was based on subjects’ eligibility for hypertension medication reimbursement and their age and sex-matched controls. Although the population represented approximately 3% of the age-group population of the City of Oulu region in Finland, applying the findings to the general population should be done with consideration. To avoid selection bias, the study group was

M AN US CR IP T

AC CE PT ED

17

included as a covariate in all examined statistical models. The main rationale of this study was to assess the risk of antibodies to oxidized epitopes in regards to CVD. We further assessed the increase in risk prediction using methods recently introduced in the literature (18) and were able to show a minor increase in risk prediction. Interpreting the increased predictive ability pointed by such measurements should be cautionary as discussed before (37).

Identifying high-risk patients prior to cardiovascular disease symptoms requires awareness of risk factors. Several scoring systems have been developed to calculate risk scores for patients and, for example, the Framingham risk score system is clinically used to estimate cardiovascular risk.

Several new biomarkers have been suggested to improve risk classification including antibodies to oxLDL, yet new additions to the traditional scoring systems often provide only small effects on the final risk scoring. The present study highlights the potential of IgA antibody measurements to phosphocholine as a new marker in CVD risk assessment and calls for additional studies on larger well-defined study populations.

M AN US CR IP T

AC CE PT ED

18

Conflict of interest

YAK has received research funding from Merck Sharpe and Dohme and has ownership of Orion Pharma Stocks. The other authors declare no conflict of interest. Funders had no role in the design or conduct of the study.

Financial support

This work is supported by The Finnish Foundation of Cardiovascular Research; University of Oulu Graduate School [J.K.]; Finnish Alzheimer’s disease research society [J.K.] Finnish Alzheimer’s disease foundation [J.K]; The Finnish Brain Foundation [J.K]; Finnish Foundation for Cardiovascular Research [M.S] and Kyllikki and Uolevi Lehikoinen Foundation [M.S].

Acknowledgments

We thank Sirpa Rannikko, Leena Ukkola, Saija Kortetjärvi, Heidi Häikiö, and Liisa Mannermaa for their expert technical assistance.

M AN US CR IP T

AC CE PT ED

19 References

1. Cooney MT, Dudina AL, Graham IM. Value and Limitations of Existing Scores for the Assessment of Cardiovascular Risk: A Review for Clinicians. J Am Coll Cardiol 2009;54:1209-1227

2. Kolberg J, Høiby EA, Jantzen E. Detection of the phosphorylcholine epitope in streptococci, Haemophilus and pathogenic Neisseriae by immunoblotting. Microb Pathog 1997;22:321-329

3. Potter M. Antigen-binding myeloma proteins in mice. Ann N Y Acad Sci 1971;190:306-321

4. Grabitzki J,Lochnit G. Immunomodulation by phosphocholine—Biosynthesis, structures and immunological implications of parasitic PC-epitopes. Mol Immunol 2009;47:149-163

5. MacLeod CM, Hodges RG, Heidelberger M, Bernhard WG. Prevention of pneumococcal pneumonia by immunization with spesific capsular polysaccharides. J Exp Med 1945;82:445-465

6. Friedman P, Hörkkö S, Steinberg D, Witztum JL, Dennis EA. Correlation of Antiphospholipid Antibody Recognition with the Structure of Synthetic Oxidized Phospholipids: Importance of schiff base formation and aldol condensation. J Biol Chem 2002;277:7010-7020

7. Frostegård J. Low level natural antibodies against phosphorylcholine: A novel risk marker and potential mechanism in atherosclerosis and cardiovascular disease. Clin Immunol 2010;134:47-54

8. Palinski W, Hörkkö S, Miller E, et al. Cloning of monoclonal autoantibodies to epitopes of oxidized lipoproteins from apolipoprotein E-deficient mice. Demonstration of epitopes of oxidized low density lipoprotein in human plasma. J Clin Invest 1996;98:800-814

9. Karvonen J, Päivänsalo M, Kesäniemi YA, Hörkkö S. Immunoglobulin M Type of Autoantibodies to Oxidized Low-Density Lipoprotein Has an Inverse Relation to Carotid Artery Atherosclerosis.

Circulation 2003;108:2107-2112

M AN US CR IP T

AC CE PT ED

20

10. Tsimikas S, Willeit P, Willeit J, et al. Oxidation-Specific Biomarkers, Prospective 15-Year Cardiovascular and Stroke Outcomes, and Net Reclassification of Cardiovascular Events. J Am Coll Cardiol 2012;60:2218-2229

11. Tsiantoulas D, Diehl CJ, Witztum JL, Binder CJ. B Cells and Humoral Immunity in Atherosclerosis. Circ Res 2014;114:1743-1756

12. Rantala AO, Kauma H, Lilja M, Savolainen MJ, Reunanen A, Kesäniemi YA. Prevalence of the metabolic syndrome in drug-treated hypertensive patients and control subjects. J Intern Med 1999;245:163-174

13. Pajunen P, Jousilahti P, Borodulin K, Harald K, Tuomilehto J, Salomaa V. Body Fat Measured by a Near-Infrared Interactance Device as a Predictor of Cardiovascular Events: The FINRISK'92 Cohort. Obesity 2011;19:848-852

14. Sämpi M, Veneskoski M, Ukkola O, Kesäniemi YA, Hörkkö S. High plasma immunoglobulin (Ig) A and low IgG antibody titers to oxidized low-density lipoprotein are associated with markers of glucose metabolism. J Clin Endocrinol Metab 2010;95:2467-2475

15. Goff DC, Lloyd-Jones D, Bennett G, et al. 2013 ACC/AHA Guideline on the Assessment of Cardiovascular Risk. Circulation 2013;

16. D’Agostino RB, Vasan RS, Pencina MJ, et al. General Cardiovascular Risk Profile for Use in Primary Care: The Framingham Heart Study. Circulation 2008;117:743-753

17. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837-845

M AN US CR IP T

AC CE PT ED

21

18. Pencina MJ, D'Agostino RB, Steyerberg EW. Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers. Stat Med 2011;30:11-21

19. Muscari A, Bozzoli C, Gerratana C, et al. Association of serum IgA and C4 with severe atherosclerosis. Atherosclerosis 1988;74:179-186

20. Kervinen H, Huittinen T, Vaarala O, et al. Antibodies to human heat shock protein 60, hypertension and dyslipidemia. A study of joint effects on coronary risk. Atherosclerosis 2003;169:339-344

21. Kovanen PT, Manttari M, Palosuo T, Manninen V, Aho K. Prediction of myocardial infarction in dyslipidemic men by elevated levels of immunoglobulin classes A, E, and G, but not M. Arch Intern Med 1998;158:1434-1439

22. Anderson DR, Duryee MJ, Shurmur SW, et al. Unique Antibody Responses to Malondialdehyde-Acetaldehyde (MAA)-Protein Adducts Predict Coronary Artery Disease. PLoS ONE 2014;9:e107440

23. Fiskesund R, Su J, Bulatovic I, Vikström M, de Faire U, Frostegård J. IgM phosphorylcholine antibodies inhibit cell death and constitute a strong protection marker for atherosclerosis development, particularly in combination with other auto-antibodies against modified LDL. Results in Immunology 2012;2:13-18

24. Macpherson AJ, McCoy KD, Johansen F, Brandtzaeg P. The immune geography of IgA induction and function. Mucosal Immunol 0000;1:11-22

25. Rodriguez-Segade S, Camiña MF, Carnero A, et al. High serum IgA concentrations in patients with diabetes mellitus: agewise distribution and relation to chronic complications. Clin Chem 1996;42:1064-1067

M AN US CR IP T

AC CE PT ED

22

26. Gonzalez-Quintela A, Alende R, Gude F, et al. Serum levels of immunoglobulins (IgG, IgA, IgM) in a general adult population and their relationship with alcohol consumption, smoking and common metabolic abnormalities. Clin Exp Immunol 2008;151:42-50

27. Miller AM, Xu D, Asquith DL, et al. IL-33 reduces the development of atherosclerosis. J Exp Med 2008;205:339-346

28. van Egmond M, van Garderen E, van Spriel AB, et al. FcalphaRI-positive liver Kupffer cells:

reappraisal of the function of immunoglobulin A in immunity. Nat Med 2000;6:680-685

29. Wang Z, Klipfell E, Bennett BJ, et al. Gut flora metabolism of phosphatidylcholine promotes cardiovascular disease. Nature 2011;472:57-63

30. Schenkein HA, Barbour SE, Berry CR, Kipps B, Tew JG. Invasion of Human Vascular Endothelial Cells byActinobacillus actinomycetemcomitans via the Receptor for Platelet-Activating Factor. Infect Immun 2000;68:5416-5419

31. Chang M, Bergmark C, Laurila A, et al. Monoclonal antibodies against oxidized low-density lipoprotein bind to apoptotic cells and inhibit their phagocytosis by elicited macrophages: Evidence that oxidation-specific epitopes mediate macrophage recognition. Proc Natl Acad Sci U S A 1999;96:6353-6358

32. Fiskesund R, Steen J, Amara K, et al. Naturally Occurring Human Phosphorylcholine Antibodies Are Predominantly Products of Affinity-Matured B Cells in the Adult. J Immunol 2014;192:4551-4559

33. Schütz K, Hughes RG, Parker A, et al. Kinetics of IgM and IgA Antibody Response to 23-Valent Pneumococcal Polysaccharide Vaccination in Healthy Subjects. J Clin Immunol 2012;33:288-296

M AN US CR IP T

AC CE PT ED

23

34. Binder CJ, Hörkkö S, Dewan A, et al. Pneumococcal vaccination decreases atherosclerotic lesion formation: molecular mimicry between Streptococcus pneumoniae and oxidized LDL. Nat Med 2003;9:736-743

35. Ren S, Newby D, Li SC, et al. Effect of the adult pneumococcal polysaccharide vaccine on cardiovascular disease: a systematic review and meta-analysis. Open Heart 2015;2:

36. Fedson DS. Pneumococcal Vaccination in the United States and 20 Other Developed Countries, 1981–1996. Clin Infect Dis 1998;26:1117-1123

37. Kerr KF, Wang Z, Janes H, McClelland RL, Psaty BM, Pepe MS. Net Reclassification Indices for Evaluating Risk Prediction Instruments: A Critical Review. Epidemiology 2014;25:

M AN US CR IP T

AC CE PT ED

24 Figure legends:

Figure 1.Kaplan Meier survival estimates for cardiovascular disease events in tertiles of IgA to oxidation specific epitopes.

The total number of study subjects included in the analysis was 1038 and the statistical significance between the separation curves was measured using Log-rank test. CuOx-LDL, copper-oxidized LDL;CWPS, Streptococcus pneumoniae cell wall polysaccharide; MDA-LDL, malondialdehyde-modified LDL; PCho, phosphocholine.

Figure 2. Plasma antibody levels and long-term risk of developing cardiovascular disease (CVD) events in Cox proportional hazards model analysis.

Hazards ratios are calculated for 1-SD unit increase in antibody levels (left side columns) and for tertiles of antibody levels (right side columns). Squares represent hazard ratios and whiskers represent 95% confidence intervals. Model 1: adjusted for study group, age, sex and BMI. Model 2:

adjusted for variables in model 1 and LDL cholesterol level, smoking in packet years, mean intima media thickness, hypertension medication use (yes/no), type 2 diabetes (yes/no). Model 3:

adjusted for variables in model 2 and hsCRP level, alcohol usage (g/week), fatty liver (yes/no) and HDL cholesterol level. ASCVD: adjusted for variables in ACC/AHA atherosclerotic cardiovascular disease risk score factors: age, sex, type 2 diabetes (yes/no) hypertension medication use (yes/no) smoking status (yes/no), total cholesterol, HDL cholesterol and systolic blood pressure. 95% CI,

M AN US CR IP T

AC CE PT ED

25

95% confidence interval; BMI, Body mass index; CWPS, Streptococcus pneumoniae cell wall polysaccharide; ASCVD: ACC/AHA atherosclerosis cardiovascular disease rick model; hsCRP, High sensitivity C-reactive protein; PCho, phosphocholine.

Table 1. Baseline characteristics of the study subjects.

CVD during follow-up (18 ± 1 years)

No (n= 849) Yes (n=195) p-value

Demographics Age (years) Female (%)

50.9 ± 6.0

54.8%

52.8 ± 5.8

30.3%

< 0.0001

< 0.0001 Body composition

BMI (kg/cm²)

Waist-hip ratio (cm/cm) Alcohol (g/week)

Intima media bifurcation thickness (mm)

27.5 ± 4.6 0.86 ± 0.09

57 ± 83 0.86 ± 0.16

28.5 ±4.5 0.91 ± 0.08

79 ± 116 0.95 ± 0.24

< 0.01

< 0.0001

< 0.05

< 0.0001 Vascular risk factors

Smoker at baseline^a Smoking packet years Hypertension (%)

Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Diabetes (%)

27.6%

8.42 ± 12.85 49.5%

147 ± 22 88 ± 12

8.4%

35.9%

16.30 ± 17.91 61.0%

156 ± 22 93 ± 12

17.9%

< 0.02

< 0.0001

< 0.01

< 0.0001

< 0.0001

< 0.0001

M AN US CR IP T

AC CE PT ED

26 Metabolic measurements

Serum creatine (µmol/l) Blood fasting glucose (mmol/l) Fatty liver (%)

81 ± 15 4.6 ± 1.2

23.8%

89 ± 66 5.3 ± 2.2

42.2%

< 0.01

< 0.0001

< 0.0001 Lipid profile

Total Cholesterol (mmol/l) Triglycerides (mmol/l) HDL cholesterol (mmol/l) LDL cholesterol (mmol/l)

5.6 ± 1.0 1.48 ± 0.87

1.4 ± 0.4 3.5 ± 0.9

6.0 ± 1.0 2.02 ± 1.43

1.2 ± 0.4 3.8 ± 1.1

< 0.001

<0.0001

<0.0001

< 0.001 Inflammatory markers

hsCRP (µg/ml)

3553 ± 7594

4881 ± 6741

< 0.0001

Values are reported as mean ± SD or percentage.

LDL, low-density lipoprotein; HDL, high-density lipoprotein; hsCRP, high sensitivity C-reactive protein.

Statistical significance was tested with Mann-Whitney (continuous variables) and h Χ² test (categorical variables).

a Ex-smokers (n=249) and random smokers (n=13) are not listed as smokers.

M AN US CR IP T

AC CE PT ED

27

Table 2. Comparison of cardiovascular disease (CVD) event risk prediction statistics in ACC/AHA atherosclerotic cardiovascular disease risk (ASCVD) model with the addition of IgA antibodies to phosphocholine (PCho) and Streptococcus pneumoniae cell wall polysaccharide (CWPS).

ASCVD model ASCVD model and IgA to CWPS ASCVD model and IgA to PCho

Logistic models p-value p-value

Hosmer-Lemeshow p<0.262 p<0.063 p<0.445

Nagelkerke’s Pseudo R² 0.198 0.213 0.211

Categorial NRI (95% CI) 0.058 (-0.0027 – 0.121) < 0.061 0.060 (-0.001 – 0.121) < 0.054

NRI_event 0.022 ( -0.033 – 0.078) 0.044 (-0.011 – 0.010)

NRI_nonevent 0.036 (0.011 – 0.061) 0.015 (-0.010 – 0.041)

IDI 0.014 (0.0044 - 0.023) < 0.0040 0.012 (0.0036 – 0.021) < 0.0058

M AN US CR IP T

AC CE PT ED

28

Continuous NRI (95% CI) 0.29 (0.13 - 0.45) < 0.00048 0.26 (0.097 – 0.42) < 0.00173

AUC (95% CI) 0.751 (0.715 - 0.788) 0.759 (0.723 - 0.795) 0.754 (0.718 - 0.791)

∆AUC 0.008 0.1296 0.003 0.557

NRI, Net reclassification improvement; IDI, integrated dicrimination improvement; AUC, area under curve

M AN US CR IP T

AC CE PT ED

29

Table 3. Risk reclassification analysis using ACC/AHA ASCVD risk score variables and plasma IgA to phosphocholine (PCho).

ASCVD risk variables and

IgA to PCho

No CVD events <15% 15-30% >30%

Total

ASCVD riskvariables variables only

<15% 403 27 ↑ 1 ↑ 431

15-30% 37 ↓ 185 18 ↑ 240

>30% 0 21 ↓ 86 107

Total 440 233 105 778

M AN US CR IP T

AC CE PT ED

30

↑ Number of study subjects reclassified into higher risk category.

↓ Number of study subjects reclassified into lower risk category.

CVD, cardiovascular disease.

CVD events

<15% 33 6 ↑ 0 39

15-30% 5 ↓ 57 11 ↑ 73

>30% 0 4 ↓ 64 68

Total 38 67 75 180

M AN US CR IP T

AC CE PT ED

M AN US CR IP T

AC CE PT ED

M AN US CR IP T

AC CE PT ED

Highlights Highlights Highlights Highlights ^:

Antibodies to OxLDL modify the progression of atherosclerosis IgA class antibodies to OxLDL have received less interest IgA to phosphocholine increased the long-term risk of CVD.

This association was maintained regardless of known CVD risk factors.

Manuscript title: “IgA antibodies to phosphocholine associate to long term cardiovascular disease risk” by Kankaanpää et al.