Data Availability.The data collected for this study was from 2 randomized controlled clinical trials. Requests to access the dataset from qualified researchers trained in human subject confidentiality protocols should be sent to the Steering Committees of the parent trials. Study Populations. IgG N-glycans were evaluated at baseline and after one year of randomized high-intensity stain interventions in 2 sub-studies of randomized trials. The discovery population was a sub-study of the Justification for the Use of Statins in Prevention: an Intervention Trial Evaluating Rosuvastatin (JUPITER, NCT00239681) trial with 239 participants with IgG N-glycan measurements at baseline and year-1. JUPITER (URL: https://www.clinicaltrials.gov) was a randomized, double-blind, placebo-controlled trial that tested high-intensity 20 mg/day rosuvastatin versus placebo for primary CVD prevention (median follow-up 1.9, maximum 5 years) in participants with elevated high-sensitivity C-reactive protein (hs-CRP 2 mg/L or higher) and average to low levels of LDL cholesterol (<130 mg/dL). The validation cohort was a sub-study of the Treating to New Targets (TNT, NCT00327691) trial among 711 participants with IgG N-glycan measurements at baseline and year-1. TNT (URL: https://www.clinicaltrials.gov) was a randomized, double-blind, controlled trial comparing the efficacy of high-dose (80 mg/day) versus low-dose (10 mg/day) atorvastatin for the secondary prevention of CVD events (median follow-up 4.9 years), in patients with clinically evident coronary heart disease. In both studies, participants provided written informed consent at the time of enrollment, and the study was approved by the local research ethics committee or institutional review board at each center and by the Mass General Brigham institutional review board(Boston, MA). The first and senior authors had full access to all data in the study and take responsibility for their integrity and data analysis. The primary study design consisted of two nested CVD case-control studies with matched pairs based on age and sex from participants in two randomized statin trials. When assessing the effect of statins on IgG N-glycans, the case and control data were pooled and analyzed by randomized statin assignment. When examining the association between IgG N-glycans and CVD, the data was analyzed as a paired case-control study. IgG N-Glycan Measurements. Plasma samples at baseline and one year from each study were placed in random order throughout 96-well plates. Additionally, for quality control (QC) and to avoid experimental biases, each plate contained 5 wells with duplicated samples from the same plate, 5 wells with duplicates from other plates, and 5 wells with aliquots of standard plasma sample (pooled plasma from healthy volunteers) to further control the repeatability of the procedure. Laboratory personnel were blinded to case or timepoint status. IgG N-Glycan profiling and IgG N-glycan data preparation are reviewed in detail elsewhere6,7. Briefly, IgG was isolated from individual plasma samples using CIM r-Protein G LLD 0.2 mL Monolithic 96-well plate8, while IgG N-glycans were released by peptide: N-glycosidase F6,9. Prepared samples were sent to the processing laboratory where they were stored at −20°C until ultrahigh-performance liquid chromatography analysis was performed. All chromatograms were separated in the same manner into 24 distinct biantennary complex IgG N-glycans. The amount of IgG N-glycans at each peak was expressed as the percentage of the total integrated area. In addition to these 24 directly measured IgG N-glycan peaks, 8 IgG N-glycosylation traits were calculated by summing the relative areas of IgG N-glycans with shared structural features, representing the percentage of IgG N-glycans with those features in the total IgG N-glycome7: agalactosylation, monogalactosylation, digalactosylation, asialylation, monosialylation, disialylation, bisecting N-acetylglucosamine, and core fucosylation. CVD Outcomes.In both studies, CVD outcomes were prospectively ascertained and confirmed through medical review by the respective clinical trial endpoint committees10,11.JUPITER cases were defined as incident myocardial infarction, stroke, coronary revascularization, unstable angina requiring hospitalization, or death. TNT cases were defined as nonfatal non–procedure-related myocardial infarction, resuscitation after cardiac arrest, fatal or nonfatal stroke, and coronary heart disease death. Clinical and Biomarker Risk Factors.Baseline questionnaires were used to collect sex, age, ethnicity, use of non-randomized medications, hypertension, smoking, and other relevant aspects of health history. LDL cholesterol concentrations were calculated by the Friedewald equation when triglycerides were <400 mg/dL and measured by ultracentrifugation when ≥400 mg/dL12-14. Statistical Methods.All analyses were performed using JUPITER as the discovery cohort, with a selection criterion of Benjamin-Hochman false discovery rate (FDR) < 0.05. The findings were validated in the validation cohort, TNT, using a p-value < 0.05. All individual IgG N-glycan peeks at baseline and year-1 were log-transformed, winsorized (to reduce the effect of possibly spurious outliers), and standardized to mean = 0 and scaled to standard deviation (SD) = 1 to allow for comparison of the effect estimates. The winsorization function automatically calculated the 5th and 95th percentiles and replaced values outside this range with the corresponding thresholds. This method limits the impact of outliers while preserving the overall distribution of the data. The effect of one year of randomized statintreatment versus control on individual levels of IgG N-glycans.In the discovery cohort, JUPITER, we fitted a linear regression model of year-1 IgG N-glycans on high-intensity statin treatment versus placebo, adjusting for baseline IgG N-glycan levels, and covariates sex, age, race, batch/plate, non-randomized use of statin, and the occurrence of CVD during the year of follow-up. We selected significant effects based on an FDR threshold <0.05. We replicated the findings in the TNT cohort, fitting a linear regression of year-1 IgG N-glycans on high- versus low-intensity statin treatment adjusting for baseline IgG N-glycans and covariates sex, age, race, and the occurrence of CVD. We considered findings with a p-value < 0.05 as validated. We also included the interaction term sex and statins as well as the interaction term ethnicity and statins in the model to assess ethnicity and sex interactions with statins. For comparison, we estimated the effect of using statin for a year on LDL cholesterol by fitting a linear regression model of year-1 LDL on statin treatment adjusting for baseline LDL levels and covariates sex, age, race, non-randomized use of statin, and the occurrence of CVD during follow-up. The architecture of IgG N-glycan connectivity.We adjusted the levels of IgG N-glycans at baseline and year-1 for the effect of covariates (sex, age, race, CVD occurrence during follow-up, and plate) and identified a data-driven Bayesian network of baseline and year-1 IgG N-glycans each separately at level 0.001 to reveal the IgG N-glycan connectivity. For constructing the networks, we used an order-independent implementation of the conditional independence structure, learning PC-algorithm15. For the comparison of Bayesian networks, we examined node connectivity. This was applicable because we had only 24 nodes and the Bayesian network was undirected. Association of the IgG N-glycans altered by statin therapy with CVD events.To evaluate if the IgG N-glycans that were altered by statins were also associated with CVD events, we fit conditional logistic regression models for matched case-control data. In the discovery cohort, JUPTER, CVD cases, and controls were matched for sex and age (±2years). In the validation cohort, TNT, CVD cases, and controls were matched on a disease risk score12 and statin randomization (low versus high statin dose). We first fit a model adjusted for covariates age and race. In the second model, we also considered additional covariates: LDL cholesterol, HDL (high-density-lipoprotein) cholesterol, hypertension, and smoking. The IgG N-glycan CVD associations in the discovery cohort JUPITER (p-value <= 0.05) were then validated in TNT. We also conducted a joint analysis of IgG N-glycans with CVD. Based on the IgG N-glycan architecture revealed by Bayesian network analysis, we assessed the association between directly connected IgG N-glycans and CVD in a single model.
