Abstracts

Rationale: To address the affection of large dose corticosteroids as the first line therapy on in-hospital infection and neurologic outcome in Anti-N-methyl-D-aspartate receptor (anti-NMDAR) encephalitis patients.

Methods: We retrospectively reviewed the clinical data of 262 anti-NMDAR encephalitis patients at West China Hospital between October 2011 and August 2020. The cases were grouped as large-dose corticosteroid treatment (n = 177) and non-corticosteroid treatment (n = 85). The development of infections proportion, length of hospital stay, neurologic outcome at discharge and risk factors of in-hospital infection were assessed. Demographic and laboratory variables obtained pre-treatment in 177 patients receiving large-dose corticosteroid therapy were screened using Least Absolute Shrinkage and Selection Operator (LASSO) regression and logistic regression to construct a diagnosis score. The receiver operating characteristic (ROC) and decision curve analyses were performed to assess the accuracy of the model.

Results: Comparing with non-corticosteroid group, patients receiving large-dose corticosteroids treatment had a higher infection rate (62.2% vs 48.2%, p = 0.033), a higher incidence of complications (74.58% vs 61.18%; P = 0.026), and higher mRS score of at discharge from hospital (3 [interquartile range, 1 to 3] vs 2 [interquartile range, 2 to 4], P < 0.001). Lab markers associated with inflammation including WBC, neutrophil-to-lymphocyte ratio (NLR) and the systemic immune inflammation index (SII) (9.69 vs 5.68; P < 0.001; 7.01 vs 3.51; P< 0.001; 1.54 vs 0.7; P< 0.001), more use of antibiotics (59.3% vs 41.25, p=0.007). Moreover, infection (OR=0.29), antibiotic (OR=5.86), LOS(OR=1.03), ICU admission (OR=0.18), gastrointestinal bleeding (OR=4.44) mRS score at discharge (OR=3.44) and WBC count before or after received therapy (OR=1.58, 0.67) were independently associated with corticosteroid use. From 89 indicators, seven variables were selected and integrated into the prediction model, including mRS scores on admission, seizure, body temperature, the uric acid (URIC), cerebrospinal fluid immunoglobulin G (CSF IgG), the neutrophil-to-lymphocyte ratio (NLR) and the lymphocyte-to-monocyte ratio (LMR). The prediction model exhibited an area under the curve (AUC) of 0.885 for the training cohort. The decision curve analysis (DCA) plot displayed a good net benefit for this model, and external validation confirmed its reliability.