ITHACA, N.Y. – When cells expire, they leave behind an activity log of sorts: RNA expelled into blood plasma that reveal changes in gene expression, cellular signaling, tissue injury and other biological processes.
Cornell University researchers developed machine-learning models that can sift through this cell-free RNA and identify key biomarkers for myalgic encephalomyelitis, also known as chronic fatigue syndrome (ME/CFS). The approach could lead to the development of diagnostic testing for a debilitating disease that has proved challenging to confirm in patients because its symptoms can be easily confused with those of other illnesses.
The findings were published Aug. 11 in Proceedings of the National Academy of Sciences. The lead author is Anne Gardella, a doctoral student in biochemistry, molecular and cell biology in the De Vlaminck lab .
The project was a collaboration between the labs of co-senior authors Iwijn De Vlaminck , associate professor of biomedical engineering in Cornell Engineering, and Maureen Hanson , Liberty Hyde Bailey Professor in the Department of Molecular Biology and Genetics in the College of Agriculture and Life Sciences.
"By reading the molecular fingerprints that cells leave behind in blood, we've taken a concrete step toward a test for ME/CFS," De Vlaminck said. "This study shows that a tube of blood can provide clues about the disease's biology."
De Vlaminck's lab previously used the cell-free RNA technique to identify the presence of Kawasaki disease and multisystem inflammatory syndrome in children (MIS-C) – puzzling inflammatory conditions that have also proved difficult to diagnose. After hearing De Vlaminck deliver a presentation about a project involving cell-free DNA, Hanson, who studies the pathophysiology of ME/CFS, reached out about a potential collaboration.
Using cell-free RNA to measure system-wide cellular turnover in patients is a relatively new concept, and it seemed particularly well-suited for unraveling the mystery of ME/CFS.
"ME/CFS affects a lot of different parts of the body," said Hanson, who directs the Cornell Center for Enervating NeuroImmune Disease (ENID). "The nervous system, immune system, cardiovascular system. Analyzing plasma gives you access to what's going on in those different parts."
There are no laboratory diagnostic tests for ME/CFS, so doctors must rely on a range of symptoms, such as exhaustion, dizziness, disturbed sleep and "brain fog."
"The problem is a lot of the symptoms that a patient might come to a primary care physician complaining about could be many different things," Hanson. "And what that primary care physician would really like to have would be a blood test."
Blood samples were collected from ME/CFS patients and a control group of healthy, albeit sedentary, people. Then De Vlaminck's team spun down the blood plasma to isolate and then sequence the RNA molecules that had been released during cellular damage and death.
They identified more than 700 significantly different transcripts between the ME/CFS cases and the control group. Those results were parsed by different machine-learning algorithms to develop a classifying tool that revealed signs of immune system dysregulation, extracellular matrix disorganization and T cell exhaustion in ME/CFS patients.
Using statistical analysis methods, they were able to map where the RNA molecules originated by deconvolving the patterns of gene expression based on known cell type-specific marker genes, as determined from a previous ME/CFS single-cell RNA sequencing study from the Grimson Lab at Cornell.
"We identified six cell types that were significantly different between ME/CFS cases and controls," Gardella said. "The topmost elevated cell type in patients is the plasmacytoid dendritic cell. These are immune cells that are involved in producing type 1 interferons, which could indicate an overactive or prolonged antiviral immune response in patients. We also observed differences in monocytes, platelets and other T cell subsets, pointing to broad immune dysregulation in ME/CFS patients"
The cell-free RNA classifier models had 77% accuracy in detecting ME/CFS – not high enough for a diagnostic test yet, but a substantial leap forward in the field. The researchers are hopeful the approach can help them understand the complex biology behind other chronic illnesses, as well as differentiating ME/CFS from long COVID.
"While long COVID has raised awareness of infection-associated chronic conditions, it's important to recognize ME/CFS, because it's actually more common and more severe than many people might realize," Gardella said.
The research was supported by the National Institutes of Health and the WE&ME Foundation.
