Prospective identification of enteroviruses involved in meningitis in 2006 through direct genotyping in cerebrospinal fluid

Abstract

Enterovirus infections were investigated with special emphasis on performing rapid molecular identification of enterovirus serotypes responsible for aseptic meningitis directly in cerebrospinal fluid (CSF). Enterovirus genotyping was carried out directly with specimens tested for the diagnostic procedure, using two seminested PCR assays designed to amplify the complete and partial gene sequences encoding the VP1 and VP4/VP2 capsid proteins, respectively. The method was used for identifying the enterovirus serotypes involved in meningitis in 45 patients admitted in 2005. Enterovirus genotyping was achieved in 98% of the patients studied, and we obtained evidence of 10 of the most frequent serotypes identified earlier by genotyping of virus isolates. The method was applied for the prospective investigation of 54 patients with meningitis admitted consecutively in 2006. The enterovirus serotypes involved were identified with the cerebrospinal fluid (CSF) of 52 patients (96%) and comprised 13 serotypes within the human enterovirus B species and 1 within the human enterovirus A species. The three most common serotypes were echovirus 13 (E13; 24%), E6 (23%), and coxsackievirus B5 (11.5%), a pattern different from that observed in 2005. Genotyping of virus isolates was also performed in 35 patients in 2006 (meningitis, n = 31; other diseases, n = 4). By comparison, direct genotyping in CSF yielded a more complete pattern of enterovirus serotypes, thereby allowing the detection of rare serotypes: three less common serotypes (CB2, E21, and E27) were not detected by indirect genotyping alone. The study shows the feasibility of prospective enterovirus genotyping within 1 week in a laboratory setting.

FIG. 1.

Phylogenetic tree based on the complete VP4 and partial VP2 encoding sequences (420 nucleotides) of enteroviruses obtained from CSF specimens and isolates and all HEV prototype strains. The tree was constructed by the neighbor-joining method and validated with 1,000 bootstrap pseudoreplicates. Only bootstrap values of greater than 70% are shown. Genetic distances were calculated with Tamura-Nei's model of evolution, and branch length is drawn to the indicated scale (proportion of nucleotide substitutions per site).