Machine-learning: The Hunt for the Cause of Acute Flaccid Myelitis (AFM)

A “mysterious” disease has been affecting many dozens of people (at least in the U.S.), and scientists are trying to understand what the “signal” (the environment) is, and what genetic architecture (underlying genetic basis of a trait, and its variational properties, at the most basic level) it takes — to respond “favorably” vs “unfavorably” to the environmental agent. The disorder has been given the name acute flaccid myelitis (AFM), which means “sudden onset of an infection causing muscle weakness and paralysi” (which actually resembles the symptoms of polio). The Centers for Disease Control and Prevention (CDC) has confirmed 134 cases of AFM in the U.S. during 2018; it is fatal to many of those who develop the illness. Most of the evidence suggests that an enterovirus called EV-D68 is causing the illness, but researchers have not been able to find the virus in the spinal fluid of children with the disease [see the attached editorial].

Scientists are urgently trying to identify the cause by using a combination of host-response diagnostics — which look at how each patient’s immune system responds to pathogens — and machine-learning analysis (a new buzzword that is becoming common in research). This approach could lead to better diagnostics and provide insight into novel treatment of AFM. Host-response diagnostic tests have not been used in the clinic yet. However, researchers are developing novel tests to help identify other conditions that can be problematic to diagnose, including tuberculosis and bacterial meningitis.

This outbreak of AFM in 2018 started in October — and is the third in a series of outbreaks in the U.S. that have mysteriously occurred every other year since 2014; researchers have yet to find a definitive explanation for the pattern. It has taken an unusually long period of time for scientists to determine the cause of the illness. Blood samples taken from many patients with AFM contain EV-D68; however, many people who never develop AFM symptoms also have the virus in their blood. The prevailing hypothesis is that EV-D68 causes AFM, the virus damages the spinal cord rapidly, and then levels of the virus quickly become undetectable in spinal fluid and blood.

The composition of the immune system’s defense mechanisms differs — depending on which pathogens are present in the body. Hence, instead of looking for the agent itself, researchers are now looking at “what the immune system is seeing.” Most attempts to identify mystery illnesses involve searching for a pathogen’s DNA or RNA in areas of the body such as tissue or blood. But the host-response technique involves testing expression “of all active genes” (transcriptomics) in a blood sample at any given time; this is followed by analysis using machine learning: searching for similarities between the transcriptomes of people with the illness, and those with AFM but no illness, and those with other (known) infections, including those caused by enteroviruses. Once relevant genes that directly react to AFM are identified, then perhaps treatment can follow.

DwN

Nature 13 Dec 2o18; 564: 170–171

This entry was posted in Center for Environmental Genetics. Bookmark the permalink.