Background:Systemic lupus erythematosus (SLE) is a multisystemic inflammatory disorder [1]. Given that immunosuppressive therapy is adopted as the predominant treatment option for SLE, up to half of SLE patients develop infections during their disease progress, and bacterial infection serves as the leading cause of morbidity and mortality in SLE patients [2]. Owing to the therapeutic regimen to bacterial infection and SLE flare are absolutely opposite, timely diagnosis and correct treatment are of vital importance, and improper treatment strategy may be fatal. No single biomarker, however, has exhibited sufficient sensitivity and specificity to serve as a standard tool for distinguishing bacterial infection from SLE flare.Objectives:To find a method by integrating cytokines, lymphocyte cells and routine examination biomarkers to observe its capacity for identifying bacterially infected SLE patients.Methods:Total 175 SLE patients (65 infected and 110 flare) were recruited into our study. The criteria of bacterial infection was positive isolation of bacteria, typical clinical symptoms and signs, imaging positive results and positive feedback on antibacterial treatment and lupus flare was regarded as three points higher than their previous SLEDAI. The disease activity of SLE patients was evaluated based on Systemic Lupus Erythematosus Disease Activity Index (SLEDAI). Lymphocyte cells (CD3+T, CD4+T, CD8+T, B, NK, Th1, Th2, Th17 and Treg) and cytokines [interleukin-2 (IL-2), IL-4, IL-6, IL-10, tumor necrosis factor-α (TNF-α), interferon-γ (IFN-γ) and IL-17] were measured by flow cytometry. Blood routine examination, erythrocyte sedimentation rate (ESR), C-Reactive Protein (CRP) Complement 3 (C3), C4, procalcitonin (PCT), immunoglobulin M (IgM), IgA and IgG were also evaluated. Partial least square discriminant analysis (PLS-DA) and supervised orthogonal PLS-DA (OPLS-DA) were applied to perform multivariate analysis of the data and further group the patients with bacterial infection. Receiver operating characteristic (ROC) curves were also plotted to investigate the ability of individual indicator and the combination of multiple indicators to identify bacterial infection.Results:The PLS-DA model showed a clear identification effect by the performance of R2Y=0.991 and Q2=0.970. The OPLS-DA model (R2Y=0.996 and Q2=0.991) exhibited a better separation of patients with bacterial infection. And the Observed vs. predicted plot of the OPLS-DA model demonstrated that all SLE patients were correctly separated into infected or flare groups, indicating that the model had a strong predictive ability for bacterial infection. For single indicator, infected patients had higher WBC, neutrophil (NEUT), ESR, CRP and PCT (P=0.002, 0.019, 0.002, <0.001, <0.001, respectively), and lower Treg cells (P=0.012). The levels of serum IL-6, IL-10, IFN-γ and TNF-α (P<0.001, =0.022, 0.014, 0.011, respectively) were significantly increased in infected group. ROC curves showed that the combination of the ten indicators showed the largest AUC and the highest accuracy, as well as balanced and relatively high sensitivity and specificity. Furthermore, the AUC of the combination was greatly higher than that of WBC, NEUT, ESR, CRP, PCT, Treg, IL-6, IL-10, IFN-γ and TNF-α (P<0.001).Conclusion:PLS-DA, OPLS-DA models including cytokines, lymphocyte cells and routine biomarkers and combination of WBC, NEUT, ESR, CRP, PCT, Treg, IL-6, IL-10, IFN-γ and TNF-α in ROC curve may be more predictive for finding bacterial infection in SLE and may prompt clinicians more promptly and accurately to help them make correct medication.References:[1]Illescas-Montes R, Corona-Castro CC, Melguizo-Rodríguez L, et al. Infectious processes and systemic lupus erythematosus. Immunology 2019;158:153-160.[2]Furst DE, Breedveld FC, Kalden JR, et al. Updated consensus statement on biological agents for the treatment of rheumatic diseases. Ann Rheum Dis 2002; 61: ii2–7.Disclosure of Interests:None declared
Recovery of members of the Pasteurella-Actinobacillus-group from guineapigs
