Neuroimaging for prognosis of central nervous system infections: a systematic review and meta-analysis

Abstract

Background: Central nervous system (CNS) infections carry a severe prognosis and often require intensive care unit (ICU) admission. This study evaluated the prognostic value of neuroimaging in patients with all-type CNS infections.

Methods: Using a predefined strategy, we first conducted a systematic search of PubMed/MEDLINE, PubMed Central, Embase, Cochrane and Google Scholar. Eligible studies published between January 1st, 2000, and June 1st, 2023, were included. We considered randomized controlled trials, non-randomized trials, cohort studies, excluding abstracts, cost-effectiveness analyses, letters, conference proceedings, systematic reviews, and meta-analyses. Two authors independently screened publications and extracted data. The meta-analysis was performed using a random-effects model. The main outcomes were (1) unfavorable outcome, defined as severe functional disability or death, and (2) mortality. Pooled odds ratios (OR) and 95% confidence intervals (95%CI) were calculated for each neuroimaging feature. We performed prespecified subgroup analyses depending on type of CNS infection (bacterial meningitis, CNS tuberculosis, CNS cryptococcosis, viral encephalitis, and brain abscess), country income, and ICU admission status.

Results: Of 7,864 studies identified, 83 met the inclusion criteria, with 48 studies (6,434 patients) included in the meta-analysis. Abnormal MRI (OR: 3.55; 95%CI: 1.81-6.96; I²=0%), brain ischemia (OR: 4.65; 95%CI: 3.14-6.88; I²=28.5%), and hydrocephalus (OR: 4.56; 95%CI: 2.49-8.36; I²=61.5%) were significantly associated with unfavorable outcome. Hydrocephalus (OR, 3.99; 95%CI 1.83-8.70; I²=61%) and brain ischemia (OR, 3.51; 95%CI, 2.22-5.54; I²=16.4%) were associated with mortality. These associations remained consistent in patients with bacterial meningitis and in patients with CNS tuberculosis, but not in other CNS infections. Subgroup analyses depending on country income and ICU admission status revealed similar findings.

Conclusion: Neuroimaging provides essential prognostic information in patients with CNS infections. Abnormal MRI findings, cerebral ischemia, and hydrocephalus are associated with unfavorable outcome, particularly in bacterial meningitis and CNS tuberculosis. These neuroimaging features should be considered when discussing prognosis in affected patients.

Keywords: Autoimmune; CNS infection; Encephalitis; Functional outcome; Intensive care; Meta-analysis; Mortality; Neuroimaging; Outcome.

Fig. 1

PRISMA diagram of eligible studies

Fig. 2

Forest plot illustrating the relationship between abnormal MRI findings and unfavorable outcomes across included studies. “Quality index” refers to the quality scale adapted from NOS and RoBANS. Detailed description for each study is available in supplemental material. mRS, modified Rankin Score; GOS, Glasgow Outcome Scale; MBI, modified Barthel Index; CI, confidence interval; df, degrees of freedom; IV, inverse variance; Q, Q-statistic; Qm, Qm-statistic. Sonneville et al. 2023 [5], Sarton et al. 2021 [17], Jaquet et al. 2019 [16], Pichler et al. 2017 [27], Kim et al. 2016 [28], Kamei S. et al. 2005 [29], Zhong Y et al. 2017 [30]

Fig. 3

Forest plot illustrating the relationship between hydrocephalus and unfavorable outcomes across included studies. “Quality index” refers to the quality scale adapted from NOS and RoBANS. Detailed description for each study is available in supplemental material. mRS, modified Rankin Score; GOS, Glasgow Outcome Scale; MBI, modified Barthel Index; CI, confidence interval; df, degrees of freedom; IV, inverse variance; Q, Q-statistic; Qm, Qm-statistic. Wu X et al. 2021 [31], Gu et al. 2015 [32], Cao et al., 2022 [33], Verma et al., 2019 [34], Raut et al., 2013 [35], Chan et al. 2003 [36], Feng et al. 2021 [37], Huang et al. 2017 [38], Erdem et al. 2015 [39], Synmon et al. 2017 [40], Tsai et al. 2018 [41], Lu et al. 2006 [42], Kastenbauer et al. 2003 [43]

Fig. 4

Forest plot illustrating the relationship between brain ischemia and unfavorable outcomes across included studies. “Quality index” refers to the quality scale adapted from NOS and RoBANS. Detailed description for each study is available in supplemental materia. mRS, modified Rankin Score; GOS, Glasgow Outcome Scale; MBI, modified Barthel Index; CI, confidence interval; df, degrees of freedom; IV, inverse variance; Q, Q-statistic; Qm, Qm-statistic. Cao et al. 2022 [33], Wu et al. 2021 [31], Feng et al. 2021 [37], Zhang et al. 2019 [44], Verma et al. 2019 [34], Cantier et al. 2018 [45], Gu et al. 2015 [32], Wasay et al. 2014 [46], Tu et al. 2022 [47], Mishra et al. 2018 [48], Tsai et al. 2018 [41], Lu et al. 2006 [42], Kastenbauer et al. 2003 [43]