We read with interest the study of Nolte et al. (6) evaluating a completely automated enteroviral real-time PCR assay (EV-PCR) with cerebrospinal fluid (CSF). Among 434 patients, 113 (26%) had enteroviral meningitis with no bacterial coinfection. Six patients had bacterial meningitis with EV-PCR-negative CSF. In accordance with the work of Nigrovic et al. (5), who found no bacterial coinfection among 735 patients with enteroviral meningitis, they conclude that EV-PCR-positive CSF may lead to rapid patient discharge without antibiotic treatment.
However, we wish to describe two recent cases, diagnosed in 2010 at Robert-Debré Hospital, Paris, France, which illustrate the fact that EV-PCR positivity does not rule out the possibility of bacterial coinfection.
The first case involved a previously healthy 5-year-old boy. On admission, he had a temperature of 36°C, with headache and nuchal rigidity. One week previously, he had received a 5-day course of amoxicillin (80 mg/kg of body weight/day) for fever and headache. The leukocyte count was normal, the C-reactive protein level was <10 mg/liter, and procalcitonin was at <0.1 ng/ml. CSF analysis showed 638 cells/mm3 (89% lymphocytes), 0.37 g/liter protein, and 4.2 mmol/liter glucose. Gram staining was negative. EV-PCR using the GeneXpert enterovirus assay (Cepheid, Maurens-Scopont, France) was positive. However, Streptococcus pneumoniae soluble antigens were detected (Binax Now), and S. pneumoniae PCR with CSF was positive (3). Blood and CSF cultures were negative. He was discharged after a 10-day course of cefotaxime (200 mg/kg/day) and vancomycin (60 mg/kg/day).
In the second case, a 14-year-old boy with renal graft-versus-host disease was admitted to the nephrology unit with a typical meningeal syndrome, including fever (38.3°C) and very high blood pressure. A ceftriaxone bolus (50 mg/kg) was immediately administered. Computed tomography (CT) scan was normal. The leukocyte count was 11,300/mm3, the C-reactive protein level was 20 mg/liter, and the procalcitonin level was 0.9 ng/ml. CSF analysis showed 710 cells/mm3 (91% neutrophils), 0.25 g/liter protein, and 3.7 mmol/liter glucose. Gram staining and culture of the CSF were negative. CSF was positive by both EV-PCR and Neisseria meningitidis serogroup B PCR (8). He was discharged after a 7-day course of cefotaxime (200 mg/kg/day).
In 1996, Dronkert et al. (4) noted that since the first description of concomitant viral and bacterial meningitis (9), only 14 cases had been published. However, Sferra et al. found that CSF samples from up to 2.8% of their 276 patients were positive by both viral and bacterial culture (7). Since 2007, in our hospital, 2 (1.3%) of 150 patients with EV-PCR-positive CSF were also positive for bacteria. Although representing a major advance, the completely automated PCR systems like the GeneXpert automat may paradoxically constitute a risk in the rare cases of bacterial coinfection, especially considering that up to 30 to 50% of children may have received antibiotics before CSF sampling (1, 2). Although rare, failure to diagnose mixed meningitis and thus to prescribe timely antibacterial chemotherapy may have dramatic consequences. Rapid sensitive techniques should be systematically used to search for at least S. pneumoniae and N. meningitis in CSF with pleocytosis, especially in antibiotic-pretreated children in industrialized countries.