By the time a neurodegenerative condition has been diagnosed, the disease may already be causing irreversible damage. This race against time has fueled an urgent search for ways to identify biological clues earlier and track their progression.
One of these conditions with an intense need for better detection is amyotrophic lateral sclerosis (ALS), also known as Lou Gehrig’s disease, a progressive, fatal neurodegenerative disease. ALS attacks the motor neurons in the brain and spinal cord responsible for movement, leading to a progressive loss of voluntary control of movements and, eventually, paralysis.
Patients are typically diagnosed with ALS between the ages of 40 and 70, with nearly 90% of the cases arising sporadically, meaning these individuals have no family history of the disease or genetic mutation. The fact that most cases cannot be linked to any specific cause has motivated researchers to identify biomarkers that can be measured in non-invasive fluids such as blood.
Researchers from Emory University may have taken one step closer to a solution. In a recent study published in Nature, they explained how the team had identified a panel of 10 proteins that distinguished patients with ALS from healthy controls with 98% accuracy.
“The clinical impact could be in getting patients to ALS centers sooner,” Eleanor Thomas, M.D., Ph.D., assistant professor of neurology at Emory University and the study's corresponding author, told Fierce in an interview.
“It could improve clinical trial enrollment because by the time most patients see us, they aren't even eligible,” she explained. “They've had the disease too long; they've been misdiagnosed for too long.”
Researchers sampled blood from 632 ALS patients and 160 healthy individuals and ran protein panels to identify disease-associated proteins. The panels consisted of 319 proteins spanning central nervous system and inflammation markers.
The scientists then used machine learning to create a classifier of which proteins were giving the best signal for distinguishing ALS from healthy controls, Thomas explained. This narrowed the panel to 12 proteins, which they then independently validated on a cohort of 144 ALS patients and 37 healthy controls.
The final panel was further narrowed down to 10 proteins, which demonstrated 98.3% accuracy in predicting ALS patients from healthy controls. When evaluating the panel size, the researchers found that accuracy improved as more proteins were included, but the gains plateaued at 10 proteins and additional proteins no longer improved performance.
Although the researchers' panel was comprised of known ALS biomarkers, such as pTau-181—a phosphorylated version of the tau protein—neurofilament light chain (NEFL) and neurofilament heavy chain (NEFH), the rest of the proteins don't overlap with previously detailed biomarker panels for ALS.
While neurofilament is “the best ALS biomarker,” according to Thomas, “it doesn't capture every single ALS patient.”
The other proteins on the panel are tied to multiple hallmarks of ALS biology, from inflammation and neuronal damage to vascular dysfunction and cellular stress.
Thomas and her team plan to dig deeper into patient samples that more accurately reflect the ALS patients seen in the clinic. They include cases where the patient's disease is initially misidentified as myasthenia gravis.
“I think we will see that some biomarkers that are useful for disease classification will also be of prognostic importance,” she said.