Clinical features of central nervous system infections and experience in differential diagnosis from neuropsychiatric lupus erythematosus in a cohort of 8491 patients with systemic lupus erythematosus

Background In clinical practice, discrimination between central nervous system (CNS) infections in patients with systemic lupus erythematosus (SLE) and neuropsychiatric lupus erythematosus (NPSLE) could be urgent and critical yet extremely challenging. Given this, this study aimed to investigate the clinical features and outcomes of infections in the CNS in patients with SLE and to establish a simplified scoring system for guiding the discrimination of CNS infections from NPSLE. Methods A total of 95 patients who were identified as having CNS infections among 8491 SLE patients between January 1992 and January 2018 were included in this retrospective study. NPSLE patients admitted at the same period were randomly selected for comparison. Key factors either clinically valuable or statistically significant for discriminating CNS infections from NPSLE were integrated to build a simplified scoring system. Another group of 22 SLE patients complicated with suspected newly onset of CNS infections or NPSLE admitted after January 2018 was enrolled to verify the utility of the scoring system. Results Sixty-three positive pathogens were identified in 59 patients of the total 95 CNS infection cases. Compared with the NPSLE group, the CNS infections group had a longer disease duration (21.0 [3.0–50.0] vs. 1.0 [0–22.0] months, P < 0.05), exhibited more fever (96.8% vs. 23.2%, P < 0.001) and polymorphonuclear leukocyte leukocytosis in the cerebrospinal fluid (CSF) (45.6% vs. 0.5%, P < 0.05), and had significantly decreased CSF glucose (2.0 ± 1.3 vs. 3.3 ± 0.9 mmol/L, P < 0.01), whereas hypocomplementemia seemed to be a strong hint of NPSLE (44.6% vs. 77.4%, P < 0.001). A simplified scoring system integrated with 8 key factors was established for guiding clinical differential diagnosis. By setting the cutoff value at 4 and verifying in a group of SLE patients complicated with newly occurred suspected CNS infection or NPSLE, a sensitivity of 85.7% and specificity of 93.3% with the area under the curve (AUC) being 0.93 (95%CI 0.80–1.00) were obtained. Conclusions CNS infections are a fatal complication of SLE and can be difficult to discriminate from NPSLE. A simplified scoring system may help to make preliminary discrimination of CNS infections from NPSLE. Electronic supplementary material The online version of this article (10.1186/s13075-019-1971-2) contains supplementary material, which is available to authorized users.