Developing a severity assessment test for Ebola infection

Posted on March 25, 2020   by Laura Cox

Researchers are looking for new ways to predict the disease outcomes of infection with Ebola virus. As part of her PhD project, Dr Jocelyn Pérez Lazo aimed to predict the course of infection – including how severe disease would be in the patient – by measuring host biomarkers.

iStock/Motortion

In December 2013, Ebola virus disease, the often fatal condition caused by infection with the Ebola virus, re-emerged for the first time since 2001. Killing as many as 90% of those infected, Ebola virus is one of the deadliest known human pathogens. When an outbreak occurs, quick action and huge amounts of resources are required to control the spread of disease and prevent loss of life.  

As with any infectious disease, predicting which patients are most at risk and those most vulnerable to severe disease can be vital in enabling healthcare workers to allocate resources appropriately and provide effective treatment. 

Currently, the most reliable way to predict outcomes of Ebola virus infection is by measuring viral load. Viral load is the amount of virus present in a sample collected from a patient  

– for example, of blood or saliva – and is measured by Ct value. A higher Ct value (22 or greater) is a low viral load and a lower Ct value (20 or lower) is a high viral load. Viral titre results in the mid-range cannot provide accurate predictions about whether a patient will survive.  

“Although the viral load is able to stratify well patients into risk groups (survivors or fatal outcome), especially when the Ct value is high or low, it is hard to predict the outcome of patients with mid-range Ct values” said Jocelyn. 

In an effort to assess the risk of a patient at diagnosis, Jocelyn hopes a new test can be developed that does not use viral load to predict disease outcomes. Instead, measuring naturally-occurring characteristics, such as genes called host biomarkers, could help researchers to predict how severe disease might be in patients infected with Ebola virus. Using machine learning, a previous study identified 10 genes that could be used as biomarkers for this purpose. Jocelyn tested their potential using samples collected during the 2014–15 New Guinea outbreak of Ebola virus: “Applying machine learning models based on the transcript abundance of these genes, along with the Ebola virus Ct value, could classify correctly the outcome of more than 90% of clinical samples collected by the European Mobile Laboratory during the outbreak in Guinea (2014–2015). The results of this work indicate that not only Ebola virus Ct values could be used as a predictor of the clinical outcome in Ebola virus disease patients, but host biomarkers are also strong predictors” she said. 

Jocelyn hopes her findings can be used to provide a more accurate way to predict disease outcomes, particularly in patients with mid-range Ct values, by using a combination of tests measuring viral load and the gene MS4A4A, which was identified as a biomarker during the research. “This study also suggests the high capability of Ebola virus Ct and one specific host biomarker to predict the risk of Ebola virus disease mortality and support the idea of using both for the prediction of the clinical outcome in situations where viral load is a poor predictor,” she said.