Association of proteomic signatures and immunophenotyping profiles with subclinical cardiovascular disease characterised by multimodal imaging in South African systemic lupus erythematosus patients

Abstract

Abstract Background Systemic lupus erythematosus (SLE) is a multisystem autoimmune disease associated with autoantibodies (aAb). Cardiovascular disease (CVD) is a leading cause of mortality later in the disease course. Protein signatures may have utility in predicting SLE patients at risk of CVD. Purpose To determine the association of aAb protein signatures and plasma markers of immune activation with subclinical CVD in South African SLE patients characterised by multimodal imaging. Methods SLE patients, without known CVD, were recruited from rheumatology clinics. Early cardiovascular (CV) involvement was characterised using carotid ultrasound, echocardiography and cardiovascular magnetic resonance (CMR). Commercially available ELISA and Luminex multibead platforms profiled plasma immunophenotyping and CV biomarkers including IL1B, IL6, TNFa, IFNg, MMP9, TIMP1, ICAM1, VCAM1, adiponectin, galectin3, ST2, NTproBNP, hsTropT and hsCRP. We investigated IgG and IgA aAb responses on 2 protein microarrays, first analysing pooled samples on the i-Ome array of over 1600 proteins and second, individual patient samples were analysed on an autoimmune disease array of 50 autoantigens. Raw data were imported and analysed using R. First, loess normalisation of proteomic data was conducted to remove any systematic experimental variation, followed by exploratory analyses on normalised data, then permutational multivariate analysis of variance (PERMANOVA) assessed which imaging variables were most influential on aAb profiles and these results were then further analysed using linear modelling. Results 18 SLE patients and 6 controls were included in this exploratory analysis. Mean age was 34±9 years, 89% were women and the most frequent CV risk factors were hypertension, smoking, and dyslipidaemia in 28%, 28% and 17%, respectively. On CMR, 72% of SLE patients were found to have subclinical CVD, predominantly myocardial (72%) and pericardial (22%) involvement. Compared to controls, SLE patients had significantly increased TNFa (p=0.046), VCAM1 (p=0.006), adiponectin (p=0.046) and NTproBNP (p=0.04). On IgG assay proteomic analyses, PERMANOVA results showed that impaired peak longitudinal strain (p=0.03) and the presence of any regional myocardial oedema (T2 ratio >1,9) (p=0.025) were the most influential variables related to peptide aAb profiles, and late gadolinium enhancement (LGE) was the most influential variable related to protein aAb profiles (p=0.075). Linear modelling considering peak longitudinal strain, T2 ratio (Figure 1) and LGE identified a list of 5, 9 and 13 candidate biomarkers respectively for these CMR parameters. Conclusions In South African SLE patients with a high burden of subclinical CVD associated with elevated immunophenotyping markers, we identified imaging parameters that were influential on protein signatures. Linear modelling identified a list of candidate biomarkers associated with CMR indicators of myocardial strain, oedema and fibrosis.Candidate biomarkers