Systemic lupus erythematosus (SLE) commonly damages the blood system and often manifests as blood cell abnormalities. The performance of biomarkers for predicting SLE activity still requires further improvement. This study aimed to analyze blood cell parameters to identify key indicators for a SLE activity prediction model. Clinical data of 138 patients with SLE (high activity, n = 40; moderate activity, n = 44; mild activity, n = 37; low activity, n = 17) and 100 healthy controls (HCs) were retrospectively analyzed. Data from 89 paired admission-discharge patients with SLE were collected. Differences and associations between blood cell parameters and disease indicators, as well as the relationship between the these parameters and organ damage, were examined. Machine-learning methods were employed to develop a prediction model for disease activity evaluation. Most blood cell parameters (22/26, 84.62%) differed significantly between patients with SLE and HCs. Analysis of 89 paired patients with SLE revealed significant changes in most blood cell parameters at discharge. The standard deviation of lymphocyte volume (SD-V-LY), red blood cell (RBC) count, lymphocyte percentage (LY%), hemoglobin(HGB), hematocrit(HCT), and neutrophil percentage(NE%) correlated with disease activity. By employing machine learning, an optimal model was established to predict active SLE using SD-V-LY, RBC count, and LY% (area under the curve [AUC] = 0.908, sensitivity = 0.811). External validation indicated impressive performance (AUC = 0.940, sensitivity = 0.833). Correlation analysis revealed that SD-V-LY was positively correlated with ESR, IgG, IgA, and IgM but was negatively correlated with C3 and C4. The RBC count was linked to renal and hematopoietic system impairments, whereas LY% was associated with joint/muscle involvement. In conclusion, SD-V-LY is associated with SLE disease activity. SD-V-LY combined with RBC count and LY% contributes to a prediction model, which can be utilized as an effective tool for assessing SLE activity.
Download full-text PDF |
Source |
---|---|
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC11439007 | PMC |
http://dx.doi.org/10.1038/s41598-024-72977-w | DOI Listing |
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!