By Tom Jenkins
For August 2019, we have selected Wilson, Sample et al. Clinical metagenomic sequencing for diagnosis of meningitis and encephalitis. N Eng J Med 2019; 380:24;2327-2340.
This study assessed the ability of a metagenomic next generation sequencing approach to detect pathogens in cerebrospinal fluid (CSF) in a real-world setting, finding that it added value to routine microbiological testing (culture, serological antigen and antibody testing, polymerase chain reaction (PCR)) in patients with meningitis and encephalitis. Therapeutic implications for antimicrobial management were demonstrated.
The lead authors of this study are based at the University of California and eight USA sites participated in a one year prospective case series. Patients with meningitis, encephalitis and/or myelitis, who did not yet have a microbiological diagnosis were enrolled. They were identified by physician referral, or screening of electronic patient records. In addition to routine testing and clinical care, CSF samples were sent to a single laboratory at the University of California for next generation sequencing metagenomic analysis for microbial nucleic acids. A previously validated assay was applied, which tested to a depth of 5-10,000,000 single-end 140 base-pair reads. Species-specific taxonomic classification was performed. Mean testing time was 90 hours. Results were fed back to participating clinicians in real-time through teleconferences, and physician reports on influences on management recorded through standardised surveys. The primary outcome was positive and negative percentage agreement between metagenomic next generation sequencing and conventional results. Incidental findings and laboratory contaminants were judged by expert consensus at teleconference, taking into account the clinical picture. Any positive results obtained by metagenomic testing alone were confirmed with targeted testing, for example, Sanger sequencing.
Four hundred and eighty-two participants were screened, 285 met enrolment criteria, 214 were enrolled, and 204 completed the study. Fifty-six percent were male and mean age was 40 years. Sixty-four percent had encephalitis, 34% meningitis, and 2% myelitis. Forty-one percent were immunocompromised. Patients were seriously ill; 49% were admitted to intensive care and 30 day mortality was 11%.
An aetiological diagnosis was made in 51% of patients; 28% were infectious and 8% autoimmune. Thirty-three percent of infectious diagnoses were identified both on conventional and metagenomic testing, 45% by conventional testing only, and 22% by metagenomic testing only. Infections that were diagnosed solely by metagenomics included St Louis encephalitis virus, hepatitis E, and Streptococcus agalacticae. Other infections detected by metagenomics with a negative result on conventional tests included Neisseria, Nocardia farcinica, Candida tropicalis, Klebsiella aerogenes, Streptococcus mitis, Enterococcus faecalis and MW polyomavirus, the latter of unclear clinical significance. In cases that were negative on metagenomic testing but positive by other means, 11 infections were diagnosed by antibody testing alone with negative culture, PCR and antigen testing, 7 diagnosed from samples other than CSF (such as brain biopsy or abscess fluid) and 8 had borderline positive or discordant results, interpreted as low CSF titres (agents included Mycobacterium bovis and tuberculosis, Cryptococcus neoformans, Propionibacterium acnes, Fusobacterium, Staphylococcus aureus, cytomegalovirus and herpes simplex type II). Post-hoc, it was identified that these missed pathogens were frequently detected by metagenomics but did not reach pre-specified thresholds for reporting positivity. Nineteen viral infections were detected by metagenomics and judged incidental, and three tests judged false positive (Pantonea, Staphylococcus aureus and Strep agalacticae). Overall, the highest diagnostic yield was seen by combining conventional and metagenomic testing and, in this group, conventional testing was positive in 27 patients and metagenomic testing positive in 32. In 13 infections diagnosed only with metagenomic testing, treatment was adjusted in 7 patients. Physicians reported that metagenomic testing improved patient management and the capability to exclude co-infections was felt valuable.
A limitation is the relatively low prevalence of infectious aetiology identified in this cohort compared with previous studies. In 35% of patients, the first CSF sample was not available and testing was on a subsequent CSF sample, which generally followed empirical antimicrobial treatment, and may have contributed to the low prevalence. The authors concluded that metagenomic testing represents a potential step forward in the diagnosis of meningoencephalitis, and that the preferred timing and patient population to target remain to be defined.
Prof Johann Sellner, Paracelsus Medical University, Salzburg commented: “The real strength of this approach is that it tests for viral, bacterial, parasitic and fungal infections concurrently, without the need for a precise a priori hypothesis about the likelihood of a particular infectious agent. Results must be weighed and interpreted carefully in the clinical context.”
