ABSTRACT
BACKGROUND: Polymerase chain reaction (PCR) is used to detect viruses in the cerebrospinal fluid (CSF) of patients with neurological disease. However, data to assist its use or interpretation are limited. OBJECTIVE: We investigated factors possibly influencing viral detection in CSF by PCR, which will also help clinicians interpret positive and negative results. METHODS: CSF from patients with was tested for human herpesviruses types 1-6, JC virus, enteroviruses, and Toxoplasma gondii. The likelihood of central nervous system (CNS) infection was classified as likely, possible, or unlikely. PCR findings in these categories were compared using single variable and logistic regression analysis. RESULTS: Of 787 samples tested, 97 (12%) were PCR positive for one or more viruses. Of episodes likely to be CNS viral infections, 30% were PCR positive compared to 5% categorised as unlikely. The most frequent positive findings were Epstein Barr virus (EBV), enteroviruses, and herpes simplex virus (HSV). Enteroviruses and HSV were found predominantly in the likely CNS viral infection group, whereas EBV was found mainly in the unlikely group. Positive PCR results were more likely when there were 3-14 days between symptom onset and lumbar puncture, and when CSF white cell count was abnormal, although a normal CSF did not exclude a viral infection. CONCLUSIONS: The diagnostic yield of PCR can be maximised by using sensitive assays to detect a range of pathogens in appropriately timed CSF samples. PCR results, in particular EBV, should be interpreted cautiously when symptoms cannot readily be attributed to the virus detected.
