Researchers have identified cell-specific proteins in eye fluid and used AI to determine which proteins accelerated aging in particular diseases. Understanding the cellular origin of these disease-driving proteins may lead to precision treatments and more informed clinical trials.
Analyzing cells is an essential part of understanding disease mechanisms. However, in non-regenerative organs, such as the eye, taking tissue samples is impractical due to the damage it would cause. So, researchers have to get creative.
Stanford Medicine researchers did just that, developing a technique of examining cell-specific proteins found in aqueous humor, the nourishing fluid inside the front part of the eye, and using AI to determine a person’s ‘eye age’ and how it's affected by disease.
The researchers collected aqueous humor from 46 healthy patients to determine what proteins were in the fluid. Using a technique they developed called TEMPO (tracing expression of multiple protein origins), they traced proteins to a cell type where the RNA that creates that protein resides.
“The first step in developing any kind of successful therapy is understanding the molecules,” said Vinit Mahajan, corresponding author of the study. “At the molecular level, patients present different manifestations even with the same disease. With a molecular fingerprint like we’ve developed we could pick drugs that work for each patient.”
The researchers found 5,953 proteins in the fluid and fed that information to an AI algorithm to see if a subset could predict the patient’s age; they identified 26 that could do so when used as a group. Aqueous humor was also collected from people with three types of eye diseases: diabetic retinopathy, which causes blood vessels in the eye to leak, leading to vision loss; retinitis pigmentosa, which causes light-sensitive cells in the back of the eye to break down; and, uveitis, inflammation inside the eye.
Comparing diseased eye fluid with healthy fluid, the researchers found that proteins in diseased eyes indicated a higher cellular age. In patients with early-stage diabetic retinopathy, the cells were 12 years older, and in those with late-stage retinopathy, 31 years older. In patients with retinitis pigmentosa and uveitis, the cells were 29 years older.
“This is one of the best connections ever made that suggests disease triggers accelerated aging,” Mahajan said.
The AI model also found that the cells responsible for indicating increased ocular age differed across the diseases studied. In late-stage diabetic retinopathy, it was vascular cells, retinal cells in retinitis pigmentosa, and immune cells in uveitis.
The researchers found that some of these disease-affected cells are not commonly targeted by treatment, suggesting there needs to be a reevaluation of current therapies. Importantly, the researchers found that some cells showed accelerated aging before symptoms appeared, meaning that treatment could be started earlier to avoid irreparable damage. Targeting both aging and disease cells could make treatment more effective because the two act separately but simultaneously to cause eye damage, the researchers say.
They add that their findings could inform future clinical trials because those running them will have a more refined look into the cellular processes driving disease.
“It’s as if we’re holding these living cells in our hands and examining them with a magnifying glass,” said Mahajan. “We’re dialing in and getting to know our patients intimately at a molecular level, which will enable precision health and more informed clinical trials.”
The researchers plan to apply the TEMPO technique and aging clock to other organ fluids, such as live bile and joint fluid.
The study was published in the journal Cell.
Source: Stanford Medicine
