Image created by Dr. Michael J. Miller
New research led by Aston University’s Dr Ghaniah Hassan-Smith, has found a new technique to diagnose infections of the central nervous system, such as meningitis and encephalitis, which can be difficult to determine using conventional methods.
The method developed by Dr Hassan-Smith, clinical senior lecturer at Aston Medical School, and honorary consultant neurologist at University Hospitals Birmingham (UHB) and her collaborators, is based on metagenomic sequencing, an existing and extremely sensitive genetic test, combined with a new and unique computational ‘filtering’ system to identify likely disease-causing bacteria and viruses from the complex information generated.
The research team included Dr Zaki Hassan Smith, clinical associate professor at Aston Medical School and consultant endocrinologist at UHB, and collaborators at the University of Birmingham – Professor Nicholas Loman, director of the Institute of Microbiology and Infection, Dr Joshua Quick, UKRI Future Leaders and Dr Nicola Cumley, clinical scientist.
Central nervous system infections can be serious and require rapid diagnosis and treatment. Usually, such infections are diagnosed by taking cerebrospinal fluid by lumbar puncture and combining knowledge of the patient’s symptoms with laboratory tests such as microscopy, microorganism culture and targeted polymerase chain reaction (PCR) tests for particular organisms. This requires the medical team to have some knowledge of what they are looking for.
In a substantial proportion of patients with a suspected central nervous system infection (over 50% in some settings, such as intensive care), no causative pathogenic organism is identified. This may be because the pathogen is unexpected, present only in small amounts or because antibiotics or antiviral treatment have already been started.
Rather than testing for individual pathogens, metagenomic sequencing analyses all the genetic material present in a sample, takes out all the unnecessary genomic data and identifies DNA and RNA and which microorganisms it is likely to have come from. This means medical staff can look for many different bacteria, viruses and other microorganisms simultaneously, including unusual or unexpected pathogens. This is particularly relevant for central nervous system infections, where there can be major diagnostic challenges.
The researchers tested cerebrospinal fluid samples from patients with central nervous system infections, sourced from UHB, and compared them with non-infectious neurological control samples.
Samples can also be contaminated with other microorganisms, for example those found on skin or in laboratory reagents. This can lead to signals which appear disproportionately important. Dr Ghaniah Hassan-Smith and the team developed a series of filters to remove this ‘background noise’ based on bioinformatics. This is a technique that uses computational methods to understand biological data, in this case, the DNA and RNA signals from the fluid samples.
The filters are able to separate potentially meaningful pathogen signals from this background noise and so determine what the infectious agent is. The method brings together several pieces of sequencing information rather than relying on a single result.
Dr Ghaniah Hassan-Smith said:
“Central nervous system infections such as meningitis and encephalitis can be extremely serious, but in many patients we never identify the organism responsible. Conventional tests are very good at finding the pathogens we suspect, but metagenomic sequencing gives us the exciting possibility of looking much more broadly, including for unusual or unexpected causes of infection.
“The difficulty is that looking for everything also means finding a lot of noise. Not every microbial signal detected in a sample represents a genuine infection, so interpreting these very complex datasets is one of the major challenges in bringing metagenomic sequencing into clinical practice.
“This isn't intended to replace conventional microbiology. Rather, we hope the work contributes to developing more standardised ways of interpreting metagenomic sequencing and ultimately helps move this powerful technology closer to routine clinical practice.”
Reference
Cumley, N., Quick, J., Brier, T. et al. Pathogen detection in central nervous system infections: moving metagenomic sequencing closer to clinical practice. BMC Infect Dis 26, 1517 (2026). https://doi.org/10.1186/s12879-026-13276-9
Abstract
Background
Central nervous system infections (CNSI) contribute significantly to global disability and mortality, but the causative agent is often undetected. Metagenomic sequencing offers the potential to enhance diagnostic sensitivity, particularly in cases of unusual or partially treated infections. However, caution is required in interpretation of metagenomics data due to technical artefacts from contamination or non-specific read mapping which can reveal a broad spectrum of biologically plausible but diagnostically unlikely organisms.
Methods
This study compares the performance of metagenomic sequencing with standard clinical microbiology methods using cerebrospinal fluid (CSF) from patients with CNSI and non-infected control samples. To evaluate sensitivity of different laboratory approaches, we sequenced DNA and RNA metagenomic libraries extracted from CSF, using both cell-free and cellular fractions. We then devised a set of simple, easily interpreted yet rigorous filters tailored for clinical metagenomics to generate a framework for result interpretation that can be readily applied by clinical scientists.
Results
We demonstrate that composite filtering strategies are essential to reduce misleading signals and support standardised workflows. Additionally, our results suggest that a cell-free sample preparation approach can improve confidence in identifying clinically relevant pathogens, highlighting the impact of sample preparation on results quality.
Conclusion
In this study we describe a reproducible method that can be incorporated into a practical framework for clinical application of metagenomic sequencing in CNSI diagnostics.