Best-Yet Systemic Biomarker Predictor Of Knee Osteoarthritis Progression

By Deborah Borfitz 

February 28, 2023 | Researchers at Duke University have come up with the new “best-in-class” systemic biomarker predictive of clinically relevant knee osteoarthritis (OA) that significantly outperforms the top-rated urine test when it comes to recognizing disease progression. Development of the targeted multiple reaction monitoring proteomic panel has been a 12-year labor of love for the team, according to Virginia Byers Kraus, M.D., Ph.D., professor in the departments of medicine, pathology, and orthopedic surgery at Duke University School of Medicine. 

As newly reported in Science Advances (DOI: 10.1126/sciadv.abq50), a set of 15 blood markers achieved 73% accuracy in identifying progressors from nonprogressors versus 58% for the urinary C-terminal cross-linked telopeptide of type II collagen (uCTXII), currently the most frequently assessed marker for knee OA research. The proteomic biomarker panel was tested in nearly 600 people in the Foundation for the National Institutes of Health (FNIH) cohort, a subsample of the Osteoarthritis Initiative of the National Institute on Aging. 

By itself, uCTXII is “not very predictive,” says Kraus. Relative to blood tests, urinalyses in general have more problems being standardized because of hydration and time-of-testing effects on results. But both blood and urine tests are probably equally acceptable to patients.  

The new test will most immediately create a “new paradigm for success” for OA clinical studies, says Kraus, by identifying people with high risk of disease progression—meaning those likely to have both escalating pain and worsening joint damage identified on X-rays. “If you can enrich your trials with people who are likely to progress in the next two to four years, you cut down the number of participants to a fraction of what would otherwise be needed.” That, in turn, would dramatically lower costs and provide a “tractable pathway for developing disease modifiers that might actually do something besides just kill the pain of the disease.”   

The main possibility is to use the new blood test along with a knee X-ray to screen if a person has OA and if it is likely to worsen before deciding who to enroll in a trial and obtaining a costly MRI, she explains. Then MRI, which provides a more detailed picture than X-ray of the state of the joint, can be used to follow along to determine the ability of the drug to slow, halt, or reverse the OA process.   

The 15 markers in the panel correspond to the 13 most-informative proteins in the bloodstream, many of which are related to inflammation, Kraus says. She is particularly excited to have CRAC1, a newly emerging biomarker for OA that predicts the risk of joint replacement, in the mix. 

Detectable fragments of the inflammatory proteins and CRAC1 appear to be coming directly from cartilage breakdown and the lining of the OA joints, she adds. Vitamin D binding protein, one of the 13, is instead made by the liver and was found to have both pro and con type effects on OA. 

The proteomic panel “could definitely be useful beyond the knee,” says Kraus. In fact, the Duke team is now doing a study looking at that possibility and if the test can reveal what stage of OA, from minor to severe, a patient is in.  

Cartilages associated with all joints in the body share some common components as well as have their own unique proteins, she continues. “We used to think one size fits all, but that is not true. If we can capitalize on those unique ones, we might [also] have a specific test for the hand and for the hip, but we’re not quite there yet.” 

Getting To 15 

The knee OA blood test was studied in the context of several different types of progressors and comparators but demonstrated the best performance with 13 markers in people who had both pain worsening and X-ray progression compared to those who weren’t doing worse in either respect. However, it predicted reasonably well in the other groups, including those who had X-ray progression without pain worsening and those who had pain worsening without X-ray progression. 

When the proteomic panel was used in concert with the uCTXII test, predictive power increased a tad to 75%. But using the original top-30 “stable” set of blood biomarkers with the highest selection frequencies, together with the traditional urinary one, gender, baseline X-ray, and pain severity, the combo hit the 80% goal for gold-standard diagnostic tests, Kraus notes.  

Backward elimination was applied to the stable set of 30 markers to reduce prediction bias, leaving a group of 10 to 15 “essential” markers for distinguishing the OA patient groups. These were the smallest possible numbers without losing much predictive ability, she explains. 

The process of arriving at the marker panels began with a discovery experiment where the researchers “didn’t twist the arm of the biomarkers” but rather took the unbiased approach of looking for any differences in the joint fluid of people who were progressors compared to those who were not, continues Kraus. From the list of proteins thought to be potentially informative, they then selected the initial 177 that were also found in the bloodstream “because people don’t like their joints to be tapped if they can help it.” Based on the results in the nearly 600 people in the FNIH cohort, they narrowed the number of biomarkers down to 30 and then finally to 15. 

Further validation of the 11-biomarker panel for predicting X-ray progression with or without pain progression was done on 86 people using data and samples derived from studies conducted at Duke University. This predicted OA progression with 70% accuracy, Kraus says. She is therefore optimistic that the test will be generalizable in real-world medical scenarios.  

Work on the proteomic panel has been a longstanding collaborative effort with faculty in the Proteomics and Metabolomics Core Facility at Duke, she notes. Despite being their “biggest customer” for over a decade, this is the first publication related to the new blood test. 

Vast Potential 

Osteoarthritis is the most common joint disorder in the United States, afflicting 10% of men and 13% of women over the age of 60, and is a major cause of disability, says Kraus. It has no cure, although several promising “remittive” drugs are in development that it is hoped could one day offer effective disease treatment on the level of Humira (Abbot Laboratories) and Enbrel (Pfizer) for rheumatoid arthritis.  

The volume of OA clinical trials underway at any one time now “goes in waves,” says Kraus, reflecting a cycle of excitement over the potential for a blockbuster drug followed by disappointing results tied in part to not having enough OA progressors enrolled in studies to enable their success. But once the first drug finds its way to market, with trials powered by the right sorts of patients, she expects many others will follow in its path. 

Compounds that succeed more often employ one or more biomarkers in the drug development and clinical testing process, she notes. In a 2016 study by the Biotechnology Innovation Organization, use of biomarkers was tied to a three-fold increase, from 8% to 26%, in the chance of successfully transitioning 

a drug from phase 1 to approval by the U.S. Food and Drug Administration. At the most-expensive phase 3 trial stage, where drugs have a 50% chance of failure, use of a biomarker effectively cuts the odds of failure in half.  

Discovery of the new-and-improved systemic marker improves understanding of OA, says Kraus, referencing the involvement of so many inflammatory molecules. “This confirms what experts in the field has known for some time... that [OA] is an inflammatory disease process, at least in people who are actively progressing.” 

Ultimately, the proteomic panel will bring people having OA symptoms “peace of mind” regarding whether to worry, she adds. The test could be an indicator of when joint health needs attention versus “nuisance symptoms” potentially correctable with Tylenol.  

Kraus says she is hopeful that the same panel of markers will one day be able to identify people on a trajectory of developing joint problems before X-ray changes are detectable. Preliminary results from another soon-to-publish study suggest that it can—and up to eight years before those signals are visible via imaging.  

“Right now, we don’t have a licensing partner [for the test], but we would really welcome that,” says Kraus. Precedence suggests it could be marketed as a laboratory-developed test. 

Separately, she adds, she is collaborating with David Hunter, chair of rheumatology at the University of Sydney (Australia), to come up with metrics that could be used on MRIs, and even very fine details on X-rays that can be read by a computer, which are predictive of OA progression.