Document resource
Objectives Neutrophil extracellular traps (NETs) are the autoantigenic drivers of autoimmunity in systemic lupus erythematosus (SLE). Their persistence is driven by excessive NET formation and impaired NET degradation. We aimed to validate a novel NET load quantification method that integrates formation and degradation as a biomarker of immunological activity in SLE, with a focus on lupus nephritis (LN).Methods SLE serum consistently induces neutrophil clustering during NET formation, hampering regular quantification. We previously developed an automated confocal imaging assay, now enhanced with a Detectron2-based deep learning model that accurately segments extracellular DNA patterns to quantify NET clusters. In addition, we introduced a NET degradation assay, allowing us to integrate both formation and clearance into a composite NET load ( figure 1A). We implemented NET load quantification in 90 LN patients with a median of 2 years of longitudinal sampling (N=725 samples) and compared with 26 healthy controls. Validity and clinical utility of NET load were assessed using Spearman correlations, mixed-effects logistic regression, and canonical test characteristics.Results The NET formation and degradation assays showed high technical reproducibility. NET clusters and non-degraded NETs were virtually absent in healthy controls but markedly increased in patients during flare and pre-flare states, robustly distinguishing healthy individuals from those with disease. NET clusters and non-degraded NETs exhibited distinct correlation profiles; associations with commonly used biomarker SLEDAI were at most modest and showed no meaningful correlations with proteinuria. The composite NET load correlated with immunological components including anti-dsDNA (r=0.63) and complement consumption (C3 r=minus 0.42; C4 r=minus 0.44). Moreover, using binomial GLMMs with a patient-level random intercept, NET load achieved an excellent accuracy (AUC 0.986, sensitivity 0.944, specificity 0.968, odds ratio 3.0-59.4) in discriminating flaring disease from a state of low lupus disease activity (LLDAS), surpassing classic biomarkers like anti-dsDNA and SLEDAI. Mixed-effects logistic regression yielded data-driven cut-offs that maximally discriminated LLDAS from flare ( figure 1B). Longitudinally, NET load declined in long-term remission and peaked prior to or during flares (figure 1C).Abstract PT1:03 Figure 1Conclusions NET load offers a biologically grounded and technically robust biomarker for immunological activity in SLE, particularly in LN, enabling early flare detection and precise therapeutic monitoring.