Diagnosis of viral infections of the central nervous system: clinical interpretation of PCR results.

Background: Standard laboratory techniques, such as viral culture and serology, provide only circumstantial or retrospective evidence of viral infections of the central nervous system (CNS). We assessed the diagnostic accuracy of PCR of cerebrospinal fluid (CSF) in the diagnosis of viral infections of the CNS.

Methods: We examined all the CSF samples that were received at our diagnostic virology laboratory between May, 1994, and May, 1996, by nested PCR for viruses associated with CNS infections in the UK. We collected clinical and laboratory data for 410 patients from Oxford city hospitals (the Oxford cohort) whose CSF was examined between May, 1994, and May, 1995. These patients were classified according to the likelihood of a viral infection of the CNS. We used stratified logistic regression analysis to identify the clinical factors independently associated with a positive PCR result. We calculated likelihood ratios to estimate the clinical usefulness of PCR amplification of CSF.

Findings: We tested 2233 consecutive CSF samples from 2162 patients. A positive PCR result was obtained in 143 patients, including 22 from the Oxford cohort. Logistic regression analysis of the Oxford cohort showed that fever, a virus-specific rash, and a CSF white-cell count of 5/microL or more were independent predictors of a positive PCR result. The likelihood ratio for a definite diagnosis of viral infection of the CNS in a patient with a positive PCR result, relative to a negative PCR result, was 88.2 (95% CI 20.6-378). The likelihood ratio for a possible diagnosis of viral infection of the CNS in a patient with a negative PCR result, relative to a positive PCR result, was 0.10 (0.03-0.39).

Interpretation: A patient with a positive PCR result was 88 times as likely to have a definite diagnosis of viral infection of the CNS as a patient with a negative PCR result. A negative PCR result can be used with moderate confidence to rule out a diagnosis of viral infection of the CNS. We believe that PCR will become the first-line diagnostic test for viral meningitis and encephalitis.