Characterization of cerebrospinal fluid (CSF) microbiota from patients with CSF shunt infection and reinfection using high throughput sequencing of 16S ribosomal RNAgenes

BACKGROUND: Nearly 20% of patients with cerebrospinal fluid (CSF) shunt infection develop reinfection. It is unclear whether reinfections are caused by an organism previously present or are independent infection events. OBJECTIVE: We used bacterial culture and high throughput sequencing (HTS) of 16S ribosomal RNA (rRNA) genes to identify bacteria present in serial CSF samples obtained from children who failed CSF shunt infection treatment. We hypothesized that organisms that persist in CSF despite treatment would be detected upon reinfection. DESIGN/METHODS: Serial CSF samples were obtained from 6 patients, 5 with 2 infections and 1 with 3 infections; the study was limited to those for which CSF samples were available from the end of infection and beginning of reinfection. Amplicons of the 16S rRNA gene V4 region were sequenced. Taxonomic assignments of V4 sequences were compared with bacterial species identified in culture. RESULTS: Seven infection dyads averaging 13.5 samples per infection were analyzed. A median of 8 taxa [interquartile range (IQR) 5–10] were observed in the first samples from reinfection using HTS. Conventional culture correlated with high abundance of an organism by HTS in all but 1 infection. In 6 of 7 infection dyads, organisms identified by culture at reinfection were detected by HTS of culture-negative samples at the end of the previous infection. The median Chao-Jaccard abundance-based similarity index for matched infection pairs at end of infection and beginning of reinfection was 0.57 (IQR 0.07–0.87) compared to that for unmatched pairs of 0.40 (IQR 0.10–0.60) [p = 0.46]. CONCLUSION(S): HTS results were generally consistent with culture-based methods in CSF shunt infection and reinfection, and may detect organisms missed by culture at the end of infection treatment but detected by culture at reinfection. However, the CSF microbiota did not correlate more closely within patients at the end of infection and beginning of reinfection than between any two unrelated infections. We cannot reject the hypothesis that sequential infections were independent.