Direct Detection of Shigella in Stool Specimens by Use of a Metagenomic Approach

The underestimation of Shigella species as a cause of childhood diarrhea disease has become increasingly apparent with quantitative PCR (qPCR)-based diagnostic methods versus culture. We sought to confirm qPCR-based detection of Shigella via a metagenomics approach. Three groups of samples were selected from diarrheal cases from the Global Enteric Multicenter Study: nine Shigella culture-positive and qPCR-positive (culture+ qPCR+) samples, nine culture-negative but qPCR-positive (culture- qPCR+) samples, and nine culture-negative and qPCR-negative (culture- qPCR-) samples. Fecal DNA was sequenced using paired-end Illumina HiSeq, whereby 3.26 × 108 ± 5.6 × 107 high-quality reads were generated for each sample. We used Kraken software to compare the read counts specific to "Shigella" among the three groups. The proportions of Shigella-specific nonhuman sequence reads between culture+ qPCR+ (0.65 ± 0.42%) and culture- qPCR+ (0.55 ± 0.31%) samples were similar (Mann-Whitney U test, P = 0.627) and distinct from the culture- qPCR- group (0.17 ± 0.15%, P < 0.05). The read counts of sequences previously targeted by Shigella/enteroinvasive Escherichia coli (EIEC) qPCR assays, namely, ipaH, virA, virG, ial, ShET2, and ipaH3, were also similar between the culture+ qPCR+ and culture- qPCR+ groups and distinct from the culture- qPCR- groups (P < 0.001). Kraken performed well versus other methods: its precision and recall of Shigella were excellent at the genus level but variable at the species level. In summary, metagenomic sequencing indicates that Shigella/EIEC qPCR-positive samples are similar to those of Shigella culture-positive samples in Shigella sequence composition, thus supporting qPCR as an accurate method for detecting Shigella.