Predictors of postinfectious inflammatory response syndrome in HIV-negative immunocompetent cryptococcal meningitis

Cryptococcal meningitis (CM) is an important opportunistic infection worldwide. After initiation of highly active antiretroviral therapy (HAART), about 10%–20% of patients with HIV–CM developed paradoxical cryptococcal immune reconstitution inflammatory syndrome (IRIS).1 2 Many retrospective studies have described risk factors for the development of HIV-related cryptococcal IRIS. A similar exuberant inflammatory response called postinfectious inflammatory response syndrome (PIIRS)1 3 is rarely noticed in previously healthy individuals. Early identification and treatment of PIIRS in patients with CM are very important. However, it is not clear which factors could predict the occurrence of CM-PIIRS. Therefore, in this study, we aimed to fill this research gap and to explore predictive indicators related to HIV-negative immunocompetent CM-PIIRS. There were 113 previously healthy patients who were evaluated and treated for CM enrolled between January 2011 and December 2019 at the Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, China. The final diagnoses and reasons for exclusion of patients are shown in online supplemental figure S1). Demographic characteristics (age gender), clinical symptoms (headache, fever, vision and hearing impairment, mental status, nausea and vomiting) and treatment history, blood and cerebrospinal fluid (CSF) samples testing, modified Rankin Scale (mRS) scores and lumbar puncture were analysed online supplemental table S1). ### Supplementary data [jnnp-2020-324921supp001.pdf] We used the random forests to select the useful predictor variables and to predict the development of PIIRS. The random forests were first fit in the training set (75%, n=84) using the selected variables and then were used to predict the PIIRS status of the testing set (25%, n=29). We demonstrate a parsimonious prediction rule by fitting a decision tree using all the patients and the most useful predictor variables. We further analysed the time from starting fungicidal therapy to the appearance of PIIRS or end of follow-up, and the association between the reduction rate of CSF …