rss
Arch Dis Child 98:388-390 doi:10.1136/archdischild-2013-303858
  • Archimedes

Question 2
Is there a role for MRI as an adjunct for diagnosing bacterial meningitis?

  1. Shankar Upadhyayula
  1. Correspondence to Dr Shankar Upadhyayula, Infectious Diseases, Children's Hospital of Pittsburgh, Pittsburgh, PA 15206, USA; shankar.upadhyayula{at}chp.edu
  • Received 11 February 2013
  • Accepted 15 February 2013

Scenario

You are asked to consult on a 3-week-old neonate who was admitted for a ‘full septic screen’. He had a lumbar puncture several hours after starting antibiotics. The lumbar puncture was traumatic and revealed pleocytosis (increased red and white cells). Cultures remained sterile after 72 h, but there was concern that the antibiotics had caused a false negative culture result. You wonder if in this scenario MRI would be useful to aid diagnosis of meningitis.

Structured clinical question

In a 3-week-old neonate with suspected meningitis [patient], does MRI [intervention] assist in the diagnosis of bacterial meningitis [outcome]?

Search strategy and outcome

PubMed was searched using the following keywords: (MRI) AND (diagnosis of) AND (infectious meningitis). This resulted in 260 articles. Seven of these articles were relevant to our question, of which two were review articles. We excluded studies examining tuberculous and/or fungal meningitis only.

The search was repeated independently by a librarian. An additional search via Ovid MEDLINE and Scopus databases did not identify any new papers. Related links and references in the selected articles were reviewed.

Commentary

Lumbar puncture with cerebrospinal fluid (CSF) evaluation is and will remain the gold standard for the diagnosis of meningitis. As the incidence of bacterial meningitis has reduced significantly due to vaccination, fewer lumbar punctures are now being performed. Previous studies have shown that there is an increased risk of unsuccessful or traumatic lumbar puncture when performed by physicians with less experience.1 It is therefore possible that trainee physicians having less experience would encounter a number of traumatic or dry taps. It is standard of care to commence intravenous antibiotics if the lumbar puncture cannot be performed or is unsuccessful. As a result we often encounter clinical situations where children have received intravenous antibiotics but also have later traumatic lumbar punctures. In these situations physicians have to make difficult decisions concerning diagnosis and treatment duration.

Recent reviews have suggested that it may not be possible to interpret CSF pleocytosis in traumatic lumbar punctures in neonates using conventional ratios.2 Physicians are then likely to treat these patients as previously administered antibiotics make it impossible to rule out bacterial meningitis in this setting. This decision cannot be taken lightly as it may have significant implications including repeated intravenous cannulation or peripherally inserted central catheter placement for antibiotic administration. In addition, the patient's family is inconvenienced, while an extended hospital stay of 2–3 weeks has its own iatrogenic risks. In the longer term, a diagnosis of meningitis has implications for follow-up and concerns about neurodevelopmental outcome.

Imaging of the central nervous system has been used in the past primarily to diagnose meningitis complications.3 ,4 However, more recent studies have suggested a role for MRI in the initial diagnosis of infectious meningitis.5–9 We would like to emphasise that the available data (number of studies and their strength of evidence) on the role of MRI for this indication are rather limited.

A review of the studies listed in table 2 showed that the specificity of MRI (ie, negative imaging findings in those who did not have meningitis) was high and ranged from 93%6 to 100%.7 ,9

Table 2

Is there a role for MRI as an adjunct for diagnosing bacterial meningitis?

The sensitivity of MRI was more variable, being 9%,8 85%,6 95%9 and 100%.7 Kamra et al5 observed that sensitivity may be higher for bacterial and fungal meningitis than for viral meningitis. A similar observation was made by Kamran et al.9 However, Splendiani et al7 noted equally high MRI sensitivity for both bacterial and viral meningitis. It is possible that MRI sensitivity may depend on the degree of inflammatory response in the meninges and brain, and hence may vary with the aetiology of meningitis.

It is also unclear which MRI sequence(s) have the highest yield. While a number of studies have indicated that contrast-enhanced FLAIR6 ,7 is the most useful, others have suggested a role for magnetisation transfer5 and some contrast-enhanced T1.9 Data from the study by Lee et al,8 which reported the lowest sensitivity among the studies included in this review, stress that subcortical low intensity on MRI is not reliable. It is also important to note that signal changes on MRI are caused by agents used for sedation and supplemental oxygen administration.3 ,10 ,11

Interpretation of MRI images (particularly the subtle changes) may be somewhat subjective. Therefore in ‘real-life’, readings may be less reliable than in study conditions.

Most of the above studies included children, but the vast majority of the study populations were adults. To our knowledge there are no studies exclusively examining children let alone infants, and prospective studies are needed. In conclusion, there are currently insufficient data to support the use of cranial MRI as a reliable method to rule out meningitis or guide management decisions.

Clinical bottom line

  • MRI cannot be recommended to rule out meningitis due to its poor sensitivity; it may be better for bacterial compared to viral meningitis, but sensitivity varies depending on MRI technique used and data are limited. (Grade B)

  • MRI appears to be more specific but cannot be recommended to rule in meningitis because data are very limited for children and absent for infants. (Grade D)

Footnotes

  • Competing interests None.

  • Provenance and peer review Commissioned; internally peer reviewed.

References