News

Instead of directly trying to find the specific bacteria or viruses causing diseases, a consortium of researchers from Europe and Australia explored how paediatric patients’ RNA levels respond to different types of infection as well as non-infectious inflammatory conditions. They used these molecular clues to pinpoint the cause of the illness, paving the way for effective, precise and rapid diagnosis.

University of Melbourne’s Professor Lachlan Coin, Laboratory Head at the Doherty Institute and part of this consortium, said that while this approach has already shown potential for distinguishing between specific conditions, it has only been used for simple distinctions between two outcomes, for example bacterial vs viral infection.

“To make a precise diagnosis when multiple diseases are possible, you need to consider them all at once,” he said.

In a study published in the journal Med, the researchers harnessed the power of RNA sequencing to study the level at which genes are ‘turned on’ (expressed) when the body fights an infection. They examined data from more than 1,600 children and discovered a 161-gene expression pattern that accurately differentiates a wide range of diseases at once.

“What we found is that these specific patterns can accurately predict whether the diseases are bacterial, viral, inflammatory, malaria, tuberculosis or Kawasaki disease (KD), while also pinpointing the type of pathogens causing them”, said Professor Coin, co-author of the study.

“One of the exciting aspects of this work is that our new approach requires just one blood sample from the paediatric patient. By analysing the expression levels of a panel of genes within that single sample, we can discriminate between 18 diseases and  predict the cause of the fever. It has never been done before and could revolutionise diagnosis in febrile children around the world.”

While rapid pathogen identification is crucial for optimal treatment, it is equally vital for clinical teams to know the correct broad disease category (such as whether it is viral, bacterial or inflammatory). Indeed, failing to diagnose a life-threatening bacterial infection is a very serious mistake, while missing a viral infection without a specific treatment may be less critical. To address this, the researchers used a specialised technique known as "cost-sensitive learning" to prioritise predictions of diseases for which an incorrect diagnosis could have the most harmful consequences.

This innovative method could lead to faster and more precise clinical diagnoses, reducing missed cases and unnecessary antibiotic use. It represents a significant stride toward improving healthcare for children with fevers around the world.

“We are moving towards being able to classify childhood febrile illness using a single blood sample. This opens the door to a new approach for clinical diagnosis,” said Professor Coin.

“It’s important to note that more work is needed before this test can be implemented in clinical practice. This includes optimising how the results are presented and the amount of detail given to healthcare teams, deciding whether the results should show the most likely cause and the likelihood of each potential cause, and if they should include recommendations for case management. Clinical studies will also be required to demonstrate the benefit of inclusion of host-based transcriptomics alongside traditional diagnostics in managing febrile illness in children.”

Peer review: Med (DOI: https://doi.org/10.1016/j.medj.2023.06.007)

Funding: European Union’s Seventh Framework Programme for Research (FP7), Horizon2020 (EU), Wellcome Trust, Medical Research Foundation (UK), NIHR Imperial Biomedical Research Centre (UK).

Collaborators: This work was done in collaboration with a number of international consortia – Pediatric Emergency Medicine Kawasaki Disease Research Group (PEMKDRG), the UK Kawasaki Genetics Consortium, the GENDRES consortium, the EUCLIDS consortium and the PERFORM consortium.