Shotgun metagenomics to investigate unknown viral etiologies of pediatric meningoencephalitis

Abstract

Introduction Meningoencephalitis in children poses a diagnostic challenge, as etiology remains unknown for most of patients. Viral metagenomics by shotgun sequencing represents a powerful tool for investigating unknown viral infections related to these cases. Patients and methods In a two-year, reference-centre, retrospective study, we investigated the usefulness of viral metagenomics of cerebrospinal fluid (CSF) for the diagnosis of viral infectious meningoencephalitis in forty seven pediatric patients, forty of them previously tested negative with a routine neurologic panel of viral targets that included herpesvirus 1–3 and enterovirus. We enhanced the detection by targeting viral sequences by hybrid capture. Raw sequence data was analysed using three bioinformatics pipelines. Results Out of forty remaining children with meningoencephalitis of unknown viral etiology, a significant detection of viral nucleic acid by shotgun sequencing was found in twenty one, which was confirmed in ten of them by specific PCR: seven human endogenous retrovirus K113 (HER K113), one parechovirus 3, one human herpesvirus 5 (HHV5); one enterovirus B (Echovirus 9). The remaining eleven CSF were not confirmed by PCR: three rotavirus, one human herpesvirus 7 (HHV7), one influenza A, one mastadenovirus C, one sindbis virus, one torque teno virus, one human immunodeficiency virus 1 (HIV-1), one human alphaherpesvirus 3 (HHV3), one human alphaherpesvirus 2 (HHV2). Conclusions Underutilization of currently available meningitis-encephalitis diagnostic techniques such as BioFire® FilmArray® is the main cause of undiagnosed cases of meningoencephalitis. However, in this study we detected uncommon viruses that should be considered, including virus, rotavirus, sindbis virus, influenza A virus and HHV7. No other viral sequences that could be readily linked to CNS inflammation were detected. Some findings may stem from reagent or sample contamination, as seen with papillomavirus; for others, the clinical relevance of the virus remains uncertain and should be substantiated by further studies, as is the case with endogenous retrovirus K113 virus. Online bioinformatics pipeline CZID represents a valuable tool for analysing shotgun sequencing data in cases of neurological conditions with unknown etiology. Altogether, this study highlights the potential of shotgun sequencing in identifying previously unknown viral neuropathogens and sheds light on the interpretation issues related to its application in clinical microbiology.